Citation
Effects of disturbance on a population of California gulls

Material Information

Title:
Effects of disturbance on a population of California gulls
Creator:
Chase, Charles Alexander
Language:
English
Physical Description:
xii, 78 leaves : illustrations ; 29 cm

Subjects

Subjects / Keywords:
California gull -- Effect of habitat modification on ( lcsh )
Genre:
Academic theses. ( lcgft )
Academic theses. ( fast )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )
Academic theses ( lcgft )
Academic theses ( fast )

Notes

Thesis:
Thesis (M.A.--University of Colorado at Denver, 1991.
Bibliography:
Includes bibliographical references.
General Note:
Submitted in partial fulfillment of the requirements for the degree, Master of Arts, Department of Biology.
Statement of Responsibility:
by Charles Alexander Chase III.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
25394876 ( OCLC )
ocm25394876
Classification:
LD1190.L45 1991m .C42 ( lcc )

Downloads

This item has the following downloads:


Full Text
EFFECTS OF DISTURBANCE ON A
POPULATION OF CALIFORNIA GULLS by
Charles Alexander Chase III B.S., Colorado State University, 1979
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Arts Department of Biology 1991


This thesis for the Master of Arts degree by Charles Alexander Chase III has been approved for the Department of Biology by
Janis W. Driscoll
Ronald A. Ryder
Date
/7, m.


Chase, Charles Alexander III
Effects of Disturbance on a Population of California Gulls Thesis directed by Associate Professor Diana F. Tomback
Long-term effects of disturbances at a California Gull (Larus californicus) colony in Colorado were studied in the breeding seasons of 1981-1987. Types of disturbances included avian and mammalian predation, flooding and draining of the reservoir, and three levels of experimental human activities. Breeding adults were individually marked as chicks, and their reproductive efforts were noted each year. The major immediate effect of human disturbance was high levels of intraspecific aggression, including fighting, destruction of eggs and chicks, and occasional death of females.
Long term consequences of disturbance included reduced clutch size, fewer nests and fewer older breeding birds. While some birds had no measurable response to disturbance, most birds responded to high levels of disturbance by relocating nest sites or by not returning to the reservoir in subsequent years. Relocation often had a long term effect on an individual's fitness. Relocating birds expended more energy finding and securing new breeding sites; their subsequent breeding was delayed and often clutch size was reduced. Low level human disturbance, flooding, and draining led to increased intra-island movement; predation by owls and high level human disturbance for one or more years resulted in increased inter-island movement. Multiple disturbances resulted in the lowest proportion of adults returning to the
reservoir.


iv
Although the numbers of birds in a breeding colony did not always change after a disturbance, the demography of the colony was often dramatically affected. There were proportionately more young adults and recently established older adults in disturbed areas than would normally be expected. Investigators who gather reproductive data over one or more years without marked birds and without knowing the history of disturbance at a colony will misinterpret and misrepresent ecological data.
The form and content of this abstract are approved. I recommend its publication.
Signed


DEDICATION
I dedicate this work to three men who have had a most significant effect on my professional development.
To Dr. Ronald A. Ryder who taught me the value of thoroughness.
To Stephen J. Bissell who taught me critical questioning.
To Dr. Robert E. Carlton who provided so very many things but most especially, love.
"Mono Lake is a hundred miles in a straight line from the ocean- and between it and the ocean are one or two ranges of mountains-yet thousands of sea-gulls go there every season to lay their eggs and rear their young. One would as soon expect to find sea-gulls in Kansas..." or in Colorado.
M. Twain, 1872, Roughing It


VI
ACKNOWLEDGMENTS
I wish to express my thanks for the help, support and guidance I received during this study. First, I wish to thank the Denver Museum of Natural History for logistical support from 1979 -1985 and the World Wildlife Museum in Monument, Colorado, for financial help from 1981-1983. Second, I wish to thank my major professor Dr. Diana Tomback for her support and prodding to complete this study. Third, I wish to thank the more than 100 volunteers that helped gather data and make observations from 1979 through 1987. In particular I would like to thank the following people who helped each year on the fourth of July "Great Gull Round-up": David and Linda Blue, Walter Bruyninckx, Anna-Marie Blancquaert, Richard Bunn, Dan and Susanna Casey, Lisa Chase (Mason), Fran Enright, Laurens Halsey, Kim Norgren, Nickie Pliler, Steve and Debbie Scheffey, Stephen Vaughan, and Mary Kay Waddington.
And finally, I wish to thank Stephanie Haas, not only for all of her help in the field and for reviewing the many drafts of this work but for her continual support through many long and crazy years.


CONTENTS
CHAPTER
I. INTRODUCTION...................................... 1
II. METHODS........................................... 6
Description of Study Area...................... 6
Reproductive Biology........................... 8
Procedure......................................10
Human Disturbance Experiments...............12
Behavior of Gulls in Relation to Disturbance.15
Analyses of Experimental Disturbances.......16
Other Disturbances..........................17
III. GENERAL PATTERNS..................................19
IV HUMAN DISTURBANCE.................................28
Immediate Behavioral Response..................28
Interspecific Aggression.....................28
Intraspecific Aggression.....................30
Long-term Response.............................35
Uncontrolled Human Disturbance.................46
V . OWL DISTURBANCE...................................48
Owl Predation..................................48
Immediate Response..........................48
Long-term Response..........................52


VIII
VI. WATER LEVEL CHANGES.........................58
Draining..................................58
Immediate Response.....................58
Long-term Response.....................59
Flooding..................................61
VII. CONCLUSIONS & SUMMARY.......................64
BIBLIOGRAPHY................................71


TABLES
Table
1. Between year and island comparisons of reproductive
performance of California Gulls at Antero Res........20
2. a). Gull behavioral responses to three levels of
experimental human disturbance, b). Chi-square analysis of behavioral responses to three levels of experimental disturbance..............................32
3. a). One way ANOVA of number of young fledged per nest comparing differences between disturbance
and post-disturbance periods, b). Kruskal-Wallis test comparing disturbance treatments and the percentage of birds returning to the same site in the following year.........................................38
4. Comparison of movement responses by breeding
gulls in the year following experimental disturbance treatments on Gull Island..............................44
5. Two by two contigency analysis of a). Breeding gulls movement response to disturbance treatments.
b). Age-related response to disturbance treatment.....45
6. Summary of owl predation on California Gulls
breeding on Gull I. and Goose 1........................49
7. a). Comparison of California Gull movement patterns
in the year following different intensities of owl predation in breeding plots, b). Two by two contingency tests of movement patterns and intensities of predation................................55
8. a). Comparison between disturbance regimes and movement patterns of breeding gulls on Gull I.
b). Rank comparisons of response to disturbances........65
9. Comparisons of movement patterns of breeding gulls on Gull I. following experimental
disturbances in the previous year.......................66


X
10. Chi-square tests for between disturbance comparisons in Table 8a................................................68


FIGURES
Figure
1. California Gull breeding colonies, Antero Res., Colorado.....7
2. Location of experimental human disturbance plots on Gull I....14
3. Yearly comparisons between Gull I. and Goose I. for
a) . Total numbers of California Gull nests on each island.
b) . Average number of nests per 10 m2 plot................21
4. Comparison of California Gull nest densities on Gull I.
and Goose 1................................................22
5. Yearly variation in California Gull clutch size at Antero Res. a). Mean clutch size on Gull I. and Goose I.
b). Mean clutch size and nest density on Gull 1............24
6. Yearly variation of the percentage of 3-5 year old
California Gulls breeding on Gull I. and Goose 1...........25
7. Comparison between levels of disturbance and between years of the number of California Gull
chicks fledging per nest...................................36
8. Comparison between levels of disturbance and between years of a), percentage of adults returning to the same plot in the following year. b). percentage of
3-5 year old breeders per plot in the following year.......40
9. Comparison between level of disturbance in 1981-83 and a), percentage of birds that moved from the previous
breeding plot, b) percentage of birds that changed
islands or remained on the same island......................43
10 Major locations of chick mortality from Great Horned
Owl predation on Gull 1.....................................53


XII
11. Comparison between intensity of predation and
a), percentage of birds that moved from the previous
breeding plot. b). percentage of birds that changed
islands or remained on the same island.................54
12. Comparison of gull movement patterns influenced by
Great Horned Owl predation, a), percentage of birds that experienced direct mortality or did not experience direct mortality, b). percentage of older versus younger breeding birds in different colony sites...............56
13. Comparison of gull movement patterns influenced by draining Antero Reservoir and a), percentage of birds
that moved from the previous breeding plot. b). percentage of birds that changed islands or remained on the same island.................................................60
14. Distance moved to new breeding sites in
response to flooding...................................63


CHAPTER I
INTRODUCTION
Disturbance of breeding colonial waterbirds has been a focus of numerous studies over the past twenty years (see reviews in Burger 1981, Speich 1986). Interest in conservation management of colonial waterbirds is the driving force behind several recent works (Ellison and Cleary 1978, Anderson and Keith 1980, Hand 1980, Desgranges and Reed 1981, Boellstorf et a}. 1988, Erwin 1989). Also, researchers are concerned that disturbances cause biases in research efforts and lead to misinterpretation of results (Duffy 1979, Ellison 1979). Because of this problem, numerous papers "testing" various disturbance regimes against undisturbed "controls" have appeared (Roberts and Ralph 1975, Ellison and Cleary 1978, Shugart et aL 1981, Fetterolf 1983, Mousseau 1984).
The term disturbance has several definitions currently in use in ecology. Anderson and Keith (1980) and many others consider disturbance only in terms of human activities that result in perturbation of the "normal" condition of the colony under consideration. Sousa (1984) views disturbance as "a discrete, punctuated killing, displacement or damaging of one or more individuals (or colonies) that directly or indirectly creates an opportunity for new individuals (or colonies) to become established." The latter definition is mostly concerned with the change in fitness of individuals in the face of disturbance factors, while the former is concerned with conservation and management of colonial


2
waterbirds. Pickett and White (1985) define disturbance as "any relatively discrete event in time that disrupts ecosystem, community or population structure and changes resources, substrate availability or the physical environment." I will use this broadly encompassing statement as the working definition of disturbance in this study.
Many studies limit disturbances to human initiated effects (Anderson and Keith 1980, Hand 1980, Fetterolf 1983 and many others), when in fact disturbances are often caused by predators (Hatch 1970, Conover and Miller 1978, Southern £ia[. 1985) and natural environmental phenomena [Nero 1961 (dry lake), Scharf 1981 (erosion) and Morris and Chardine 1985 (late ice)]. It is clear that various types of disturbance are interrelated. Intrusion into a colony by humans or predators (Patterson 1965, Kury and Gochfeld 1975,) or disturbances following an environmental perturbation (Scharf 1981) can lead to increased interspecific predation as well as predation by other members of a colony on chicks and eggs. Effects of multiple disturbances are often compounding and the effect of a particular disturbance may be inseparable from other effects (Verbeek 1982, Desgranges and Reed 1981, Drapeau elai- 1984, Mousseau 1984).
Disturbances affecting breeding gulls have been described in scientific literature since the early 1900s (Dutcher 1903, Ward 1906, Herrick 1909). Interspecific predation is common at colonies where more than one species of gull occur, especially Herring Gulls (Larus argentatus) and Great Black-backed Gulls (L. marinus) (Harris 1964, Kruuk 1964, Erwin 1980). Often confused with cannibalism, intraspecific killing is the greatest post-disturbance effect in many studies (Parson 1971, Gillette! ai. 1975, Davis and Dunn 1976, Southern and Southern


3
1984). When the opportunity arises, neighbors will kill, though not eat, chicks in adjacent territories (Emlen 1956, Vermeer 1963, Patterson 1965, Hand 1980). True cannibalism is not infrequent, especially in studies where there are food shortages or where non-breeding gulls are numerous (Kirkman 1937, Paludan 1951, Tinbergen 1953, Harris 1964, Parson 1971, Southern and Southern 1984, Watanuki 1988). Herring Gulls are most commonly noted as cannibals in the above studies. Both cannibalism and intraspecific killing may occur in the same colony, especially if other disturbances have occurred. Birds that have lost their own nests or chicks often perform most attacks on neighboring adults and their chicks (Davis and Dunn 1976, Hand 1980, Pierotti and Murphy 1987, but see also Watanuki 1988).
Most studies only examine effects of disturbance on reproduction during the current season, e.g., Erwin (1989). A few multi-year studies noted whether or not an area had been abandoned (Kadlec 1971,
Buckley and Buckley 1972, Conover and Miller 1978) or if numbers or densities of breeding birds had changed (Southern and Southern 1981, Coulson £lai. 1982, DeWit and Spaans 1984). However, long-term studies of the reactions of marked animals to disturbance are few. Southern and Southern (1981) felt that changes in clutch size in a particular sub-area were good measures of pre-hatching perturbations, while drastic changes in density could indicate disturbances from previous years. Fetterolf (1983) compared between year hatching success for study plots without knowing individual birds. He recognized the confounding effects of both changing age composition (determined by plumages) and nocturnal predation on hatching success comparisons. Bergman (1986) found high site tenacity in marked Common Gulls


4
(Larus canus), except when eggs or small chicks were taken or when crows (Corvus spp.), intruded regularly.
High site fidelity is common among larids, especially those with stable breeding sites (Coulson and White 1958, McNicholl 1975, Chabrzyk and Coulson 1976, Southern and Southern 1981). Southern's (1977) colony of marked Ring-billed Gulls (Larus delawarensis) showed high site fidelity between years. Interspecific predation was low in this population. Southern £ial. (1985) presumed that high site fidelity also occurred in an unmarked disturbed colony where, in spite of yearly fox predation, a group of breeders returned to the same site over a 9 year period. Conover and Miller (1978) and Cuthbert (1985a) found low site tenacity in disturbed and unstable habitats. Austin (1951) and Cuthbert (1985b) found group fidelity in mixed species colonies of Common (.Sterna hirundo) and Caspian Terns (S. caspia). Mixed species groups moved as groups to new breeding areas after disturbances.
Since most studies of colonial waterbirds focus on study plots within a colony instead of the entire colony, it is critical to know if a particular plot is typical of the population (Coulson 1968, Southern and Southern 1981). Without knowing the history and changes that have occurred at a colony, this may be impossible to determine. Unknown biasing effects that have occurred in previous years (and may still be affecting a colony) can completely confound the interpretation of experimental studies (Jehl and Chase 1987).
It is neccessary to determine not just immediate behavioral and ecological responses to disturbance during the breeding season but also the long-term consequences. What do individual birds do in reponse to disturbances of various kinds, intensities, and frequencies? Do birds


5
have stronger reactions to one type of disturbance than another? Do they react more strongly if they have been disturbed several years in a row? What are the costs to a bird that changes breeding locations after a disturbance rather than returning to a disturbed site? Is there really such a phenomenon as an undisturbed research site?
In this study, I examined changes that occurred within and between two colonies of marked California Gulls (Larus californicus) over a seven year period. Disturbance factors in this study include predation, water-level changes, and experiments of controlled human disturbance. The main thrust of this study was to determine changes in nest site selection and reproductive success among known birds in response to various natural and human disturbance factors.


CHAPTER II
METHODS
Description of Study Area
The study population of California Gulls (Larus californicus), located at Antero Reservoir (southern end of South Park, Park County, Colorado) has been in existence approximately 30 years. The nearest town is Hartsel, 8 km to the east. Fairplay is 32 km north, and the Chaffee County dump is 56 km south. The reservoir is owned and maintained by the Denver Water Board. At an elevation of 2770 m, this is the highest gull colony in North America. Historically, there have been four breeding colonies (islands) (Fig. 1). One of the rock islands near the dam has eroded completely away and the other has had only 30-40 nests each year, most of which are disturbed by fishermen. Gull and Goose Islands are on the west end of the reservoir. Gull I. is located approximately 70 m from the shoreline; Goose I. is located 130 m northeast of Gull I. The greatest water depth between the mainland and Gull I. was 1.2 m, while the deepest point between Goose I. and the nearest mainland or Gull I. was 2 m.
Antero Reservoir has a diverse aquatic invertebrate population and supports a healthy fishery which is enhanced by annual stocking of rainbow trout. Access to the west end of the reservoir is legally restricted; fishermen must remain in boats and are not allowed to land on shore or


Fig. 1. California Gull breeding colonies, Antero Res., Colorado. A - Gull Island, B - Goose Island, C - Rock Islands.


8
on any island. This provides breeding birds some degree of protection from human disturbance
Most of the vegetation on the islands and in the surrounding basin is alkali-adapted; the shoreline is sandy gravel with alkali grasses above the waterline. The west end of the reservoir is covered with marsh plants, primarily Scirpus, J uncus and Car ex spp. Numerous grass and herbaceous species occur on both islands; however, the gulls keep most vegetation plucked close to the ground until mid-July when chicks begin fledging.
Gull Island is approximately 0.37 ha in area with an elevation rise of 2.5 m from the shore to a central plateau. Approximately 55% of the island is covered with herbaceous vegetation, 3% is a stand of sandbar willow (Salix exigua) on the northeast corner of the island, and 12% is covered with the shrub winter fat (Ceratoides lanata) on the south and west top of the rise in the middle of the island. Thirty percent of the island (the south side) is bare soil and debris.
Goose Island, approximately 2.8 ha in area, is a long narrow island with the gull breeding area located on 0.75 ha at the west end. It rises to 2 m on each end but is relatively flat throughout. The vegetation is primarily herbacecous with less than 2% covered in scattered winter fat. Similar to Gull I., the only bare ground (10% of the gull colony) has developed on the southwest edge of the island.
Reproductive Biology
California Gulls returned to Antero Reservoir in early April, with the peak of arrival in the last week of April. Many of the earliest arriving birds


9
were also the oldest and most experienced (Chase, personal observation). Territories were established upon arrival, although birds did not remain on the colony overnight until the ice on the reservoir broke up (usually by the end of April). The first eggs were typically laid during the last week of April with peak egg-laying by the end of the first week of May. While the normal clutch size was three eggs, young birds (3-5 years old) and disturbed birds typically laid only two (and rarely one). Clutches hatched in 24-26 days and chicks fledged 35-40 days later. Most of the young at Antero fledged by mid-July. Dates of clutch initiation varied by as much as two weeks, depending on when the ice broke up. Clutches initiated before ice breakup always failed because of predation by coyotes (Canis latrans).
Territories were first established in the center of a colony with later nesting birds using more peripheral and shoreline areas. Open areas of ground with a low herbaceous cover were chosen first, areas with dense herbaceous cover second, and shrubby sites and bare scoured ground were avoided by most birds. Although both sexes incubated clutches and cared for the chicks, males were involved in most fights and chick killings. Females spent more time incubating from mid-morning until midafternoon. By the time chicks were 20 days old, both parents often left to forage. Before then, at least one parent remained with or near the chicks. Adults usually abandoned chicks within two weeks after fledging, and within three weeks of fledging most adults and chicks left Antero and began migration to the Pacific coast.
While a few California Gulls remained at Antero throughout the winter, band recoveries suggested that most (99+%) wintered along the Pacific coast. Immature birds remained on the coast until their fourth year


10
(determined from band recoveries) when they returned to Antero to breed. A small percentage of third year birds and an even smaller percentage of second year birds returned to Antero. California Gulls do not reach full sexual maturity with fully developed gonads until their fourth year (Johnston 1956a and b); however, even second year birds can produce viable offspring. The above information on breeding biology at Antero was gathered as part of my studies on this population and will be presented in detail elsewhere. The breeding biology of this colony was similar to other populations of California Gulls with three-egg clutches (Behle 1958, Vermeer 1970, Raper, 1976, Pugesek 1983b).
Procedure
This study began on Gull I. in 1981 and Goose I. in 1983 and ended in 1987. Each island was divided into 10 m x 10 m north-south oriented plots marked with 0.3 m wooden stakes. I entered each island from the same 10 m2 plots each year; entry plots were excluded from all analyses. Minimizing disturbance of chicks at nest site was a primary concern. Almost every study of gulls has shown that high intraspecific and, at times, interspecific predation occurs when colonies are disturbed and chicks run from natal territories (see review in Burger 1981). Results of an earlier experiment at this colony indicated that 10 m x 20 m pens caused the least disturbance and resulted in highest fledging success (Chase, unpublished data). From 1981-1983, Gull I. was fenced into 10 m x 20 m plots with 2.5 cm x 2.5 cm x 45 cm chicken wire fence. These plots corresponded to the existing gridded plots. Fences were removed after 4 July, which corresponds to the beginning of fledging. Behle and Goates


11
(1957), Behle (1958), Vermeer (1970) and Winkler (1983) found that California Gull chicks did not use water for drinking or cooling prior to fledging so there was little concern about the fencing restricting access to water. From 1984 to1987, Gull and Goose Is. were surrounded by a single fence only during chick measurements on 4 July.
Average minimum temperatures were below 0° C from incubation through the first two weeks of brood rearing. Snowfalls, with accumulations up to 0.2 m were not uncommon through mid-June. High diurnal temperatures of 21-30° C throughout June and July were regularly interrupted by daily violent thunderstorms. Temperature variations occasionally exceed 22° C per hour and 30° C per day. Winkler (1983) conducted chick measurements at night at Mono Lake, California to reduce the problem of chicks running from nesting territories and being killed by other adults. At Antero, chicks <15 days old became hypothermic within 5-10 minutes of a researcher entering the colony at night and chicks older than 15 days ran when disturbed, regardless of day or night. From these earlier observations, I decided to limit research activities to daytime hours and to periods of moderate weather.
All movements by researchers in this study were slow and methodical; activities were as short in duration as possible. Researchers moved bent over or in a "crab walk". Once work started in a plot, workers remained in that plot until finished and did not return to that plot again until the next scheduled period. Measurements of eggs and chicks were performed by myself or an assistant. To compute error estimates I repeated every twentieth measurement.
Overall reproductive patterns were determined from data compiled by plot, island, and year and include: number of nests, average density,


12
clutch size, number of chicks fledging per nest, and percentage of eggs resulting in fledged chicks per nest. Nest numbers, density, and clutch size for all years are based on counts from 25 May. This was the average date of first hatching and 1 to 1.5 weeks prior to peak hatching. Fledging success is based on 4 July counts (average chick age - 30 days) minus dead chicks found during final mortality counts approximately 2 weeks later. Dead chicks were counted and collected each visit. Necropsies were performed on all dead chicks to ascertain cause of death.
All adults in each plot were observed from their arrival in the spring through chick fledging. Observations were made with a 15-60 power spotting scope. Band numbers and colors, plumage characteristics, sex, and nest site were noted. Initially, between 20-50% of breeding adults in each plot had been banded as chicks in earlier studies. Another 20-35% percent of adults could be aged as 3-5 year old birds by plumage variations. Thus, 40-85% of breeding birds could be aged in any particular plot. All banded adults had U.S. Fish and Wildlife Service aluminum bands. Twenty-five adults were marked with colored leg bands in 1981. All chicks fledging at Antero from 1981-1987 were marked with aluminum bands on the left leg, and chicks from 1981-1984 were also marked with colored leg bands on the right leg. Measurements of chicks included weight (300g or 1000g Pesola scales) and lengths of tarsus, culmen, 9th primary, and outer rectrix.
Human Disturbance Experiment
This experiment was run in 1981-1983; 1984-1987 were considered post-disturbance (though only in reference to this experiment). The levels of disturbance (high, medium, and low) were


13
measures of the frequency of visits to a plot. Twelve pairs of plots on Gull I. were randomly chosen and designated as high, medium or low disturbance plots. Block pairs, instead of single plots, were established so that access could occur from the island perimeter without having to cross other plots (Fig. 2). All plots on Goose I. and all other plots on Gull I. were designated as low disturbance areas. These plots are the "controls" in this experiment. It is critical to note that these are least disturbed areas, not undisturbed plots, as is commonly misstated in the literature.
All plots on both islands were visited each year on the weekends closest to 25 May and 4 July. During these visits, I used a crew of 15-20 assistants to minimize the time of disturbance. During the early visit, no more than three people were in a single plot at a time and then for less than 30 minutes per plot. Eggs were measured and nests mapped. The 4 July visits involved groups of 5-6 assistants in each penned area on both Gull and Goose Is. During 1981-1983 each group measured and banded all chicks within the experimental plots within 20 minutes. Since 1984, each island was surrounded by one fence and work on each island was completed in approximately 3 hours. All fences were removed at the end of this event. I checked chick mortality within three days of the large banding effort. A final visit was made in the third week of July, when over 95% of all birds had fledged. Final mortality counts were made at this time.
Low disturbance plots were only visited on the two weekends mentioned above. The two-visit method is similar to that of Kadlec and Drury (1968), Mineau and Weseloh (1981) and Mousseau (1984), except that the chicks in my study areas were closer to fledging when handled (majority within 10 days of fledging versus 15-20 days).


Fig. 2. Location of human disturbance plots on Gull I.
L - Low level disturbance, M - Medium, H - High. Plots without patterns experienced additional, nonexperimental disturbances and were excluded from the human disturbance experiments.
t


15
High disturbance plots were entered every 1-2 days for at least 5 minutes and up to 30 minutes per visit from the beginning of egg laying through 4 July. Medium disturbance plots were visited at 4-5 day intervals for similar periods of time. During visits to the high and medium plots, nest counts, mapping and egg measurements were performed by no more than two people at a time. Chicks were marked at hatching and measured every 1-2 days or 4-5 days depending on the plot disturbance designation. When most chicks were less than 10 days old, they were handled on or near nests. After they were 15 days old, all chicks in a pen were gathered together in one corner of the pen and contained in a smaller 2 m pen. After each chick was measured, it was released into the main pen.
Behavior of Gulls in Relation to Disturbance
Over 400 hours of gull behavioral observations were gathered on Gull I. from 1981-1983. Behavioral observations took place from a blind (1 m x 1.5 m x 1,5m) erected on the top of the hill or from open observation points on the perimeter of the colony. Behaviors were recorded 30 minutes before, during, and 30 minutes after disturbances. Five nests or family groups within the same plot were observed each period. Observations were conducted only on nests at least two plots (20+ m) away from the observers' position. New subjects in different plots were selected for the next observation period. Observations were conducted for sample periods throughout the day. Data collected included size of the brood, parental activities, especially feeding, brooding, and intraspecific aggression, as in Conover and Miller (1980). Adoptions of moving chicks by other adults were recorded and


16
continuously monitored even if the chick left the original 5 broods under study.
I recorded attack rates by adult gulls on myself or an assistant as we moved through plots performing measurements. The recording observer was seated at least 30 m away from an interaction to minimize additional disturbance. An attack consisted of a definite swoop ending within 3 m of the human intruder. The sex, age, and identity of each attacker were recorded when possible.
Analyses of Experimental Disturbances
Mortality was recorded separately for eggs, chicks, and adults. Mortality and numbers fledging per plot and disturbance type were compared between years, and islands and were compared with the subsequent breeding distribution of returning banded adults in the years following disturbances. Since California Gulls can be aged by plumage until their fifth year, any nesting young adults can be aged visually. The proportion of 3-5 year old adults within plots was compared among disturbance regimes. A one-way analysis of variance was performed on the number fledging per nest, comparing differences between human disturbance regimes during experimental disturbance years (1981-1983) and during non-disturbance years (1984-1987). A Kruskal-Wallis test was used to compare among disturbance regimes and between disturbance and post-disturbance periods for percentage of breeding birds returning in subsequent years. A Bonferoni correction was applied to adjust for repeated samples . A Dunn's multiple comparison test was used to compare among disturbance regimes for the percentage of young (3-5 year old) breeders in each plot in the subsequent year. Two by two


17
contingency analyses were used to compare disturbance with movement patterns, i.e., whether a bird moved from its previous plot in subsequent years, and if it moved, whether it remained on the same island or changed islands. Two by two contingency analyses were also used to compare movement patterns among birds of different ages and disturbance treatments.
Pearson correlation analyses compared: 1) previous breeding success with numbers returning in subsequent years, 2) adults returning to the same plot with numbers of 3-5 year breeders, and 3) nest density with fledging success. Student's t tests compared: 1) clutch size variation in 8 plots where disturbances had occurred in previous years, 2) clutch size variation between islands, and 3) frequency of attacks by adults on researchers in low and high disturbance plots. Non-parametric tests are as in Hollander and Wolfe (1973). All other tests are as in Sokal and Rohlf (1981).
Other Disturbances
Owl predation was recorded by plot, date, number, and age of birds killed from 1981-1987. Adult and chick response to owls was recorded during 5 nights that attacks were observed. Gull responses to annual water level variation were recorded in the same manner as low disturbance human experiments, i.e., nest numbers, clutch size, etc., except that after Antero was drained, I entered the colony to count nests and eggs on 5 occasions instead of 2. Movement patterns of marked birds disturbed by owls and draining were recorded in the same manner as in the human disturbance experiments. The actual distance gulls moved in subsequent years within the colony was recorded for birds


18
affected by flooding of the periphery of the islands. Two by two contingency analyses were used to compare movement patterns for owl predation, flooding and draining.


CHAPTER III
OVERALL REPRODUCTIVE PATTERNS
Overall reproductive efforts are noted by year and island in Table 1. Prior to 1983, few California Gulls nested on Goose I. There was a pattern of decreasing numbers of nests on Gull I. corresponding with increasing numbers on Goose I (Fig. 3a), with the exception of 1985. During 1985, a large influx of unmarked adults appeared and nested at Antero. This influx was attributed to gulls leaving Great Salt Lake when nesting colonies flooded. Record water levels flooded many colonies that normally survive high water periods at Great Salt Lake (Paul 1985). Colonies in Wyoming and Montana also experienced similar influxes of California Gulls (Finholdt, personal communication).
Nest numbers were converted to two types of densities (Fig. 3b and 4). Densities per hectare controlled for differences in total area between the two islands and between years (Fig. 4), while the average numbers of nests per 10 m2 plot for all plots used in a given year were compared in Figure 3b. In both cases, it was clear that there was a general decline in densities on Gull I. and an increase in densities on Goose I. Davis and Dunn (1976) and Coulson elai- (1982) found that increased density led to increased intraspecific chick killing and decreased reproductive success while Watanuki (1988) found the opposite result. Increased density was TABLE 1. Between year and island comparisons of reproductive


20
TABLE 1. Between year and island comparisons of reproductive performance of California Gulls at Antero Res. All birds are lumped within islands regardless of disturbance factors.
1981 1982 1983 1984 1985 1986 1987
Gull I.
nests 546 545 429 371 476 287 305
avg. clutch 2.42 2.49 2.4 2.0 2.31 1.67 2.38
% eggs fledged 35.9 33.2 50.5 66.9 67.5 - 66.0
avg. fledge/nest 0.86 0.83 1.21 1.33 1.56 - 1.6
fiQQ.Sfi.i-
nests 50 49 144 498 694 617 681
avg. clutch - 2.18 2.26 2.25 2.36 1.99 2.38
% eggs fledged - - 49.2 66.8 70.2 - 75.1
avg. fledge/nest - - 1.11 1.5 1.65 - 1.8


21
Year
b)
o
Q.
a>
Q.

V)
4)
6
c
a>
cn
ra
l_
a)
>
<
Year
Fig. 3. Yearly comparisons between Gull I. and Goose I. of a). Total numbers of California Gull nests on each island, b). Average number of nests per 10 m2 plot.


22
1981 1982 1983 1984 1985 1986 1987
Year
Fig. 4. Yearly comparisons of California Gull nest densities on Gull I. and Goose I.


23
strongly correlated with high fledging success (r=0.96, df=11, p<0.001) at Antero Res.
The average clutch size for least disturbed areas of the total population was 2.4 for 1981-1987. Comparing average annual clutch sizes on Gull I. and Goose I. (Fig. 5a), there were significant differences from the mean only in 1984 and 1986 (ttest: p<0.0001 for Gull I., 1984-86 and Goose I. 1986, p<0.05 for Goose I., 1984). In 1984, fishermen destroyed many eggs during egg laying, which, coupled with an increase in young breeding birds (3-5 yrs. old) (Fig. 6), resulted in a lower average clutch size of 2.0 on Gull I. In 1986, the reservoir was drained to repair the dam. By 25 May, many nests and eggs had been destroyed (see Chapter VI for more details) and clutch sizes and densities were lower than previous years on both islands in 1986 (Fig. 5b). During earlier visits (5 and 11 May) to both islands, the numbers of nests nearly equaled 1985 levels, although clutch sizes were low throughout the entire season.
I attribute the general trend of declining numbers, densities, and, to some degree, clutch size to the cumulative effects of multiple disturbances to Gull I. overtime (Fig. 3 and 5b), including human disturbance, both uncontrolled and experimental (chap. IV), owl predation (chap V), and water level changes (chap.VI). Few birds nested on Goose I. until 1983. The increase on Goose I. was caused in part by movement from Gull I. The increasing percentage of young adults breeding on Gull I. was caused by the decreasing number of older birds returning to Gull I. With fewer adults, breeding territories were available to younger birds. Numerous young gulls without nest sites were observed in 1981- 1983. Even though Goose I. was open and apparently had suitable habitat, few gulls attempted to nest prior to 1983 (25-50 each year) and then only on


24
1981 1982 1983 1984 1985 1986 1987
Year
Fig. 5. Yearly comparisons of California Gull clutch size at Antero Res. a). Average clutch size on Gull I. and Goose I. b). Average clutch size and nest density on Gull I.
Density per hectare


Percentage 3-5 yr. old breeders
25
10 H----'----â– ----1----â– ----â– ----1---->----1----1----1----'----1
1981 1982 1983 1984 1985 1986 1987
Year
Fig. 6. Yearly comparisons of the percentage of 3-5 year old California Gulls breeding on Gull I. and Goose I.


26
the western edge, nearest to Gull I. All gulls observed breeding on Goose I. before 1983 were older adults, and in fact many were known birds from Gull I. Young adults did not attempt to colonize Goose I. They either established territories on the periphery of Gull I. or did not breed at Antero. Only after greater numbers of older adults started nesting on Goose I. in 1985, did young adults start breeding there. Coulson (1968) and Coulson and Thomas (1985) found similar results with Kittiwakes (Rissa tridactyla). Because of disturbances on Goose I. (owl predation, draining of Antero, and low level researcher monitoring), by 1986 some older gulls left Antero altogether. The proportion of young adults slowly increased on Goose I., which may be attributed to continuous disturbances to the entire population at Antero.
Older gulls lay larger clutches and nest in greater densities than young breeders (Pugesek 1983, Tinbergen 1953, Coulson and Porter 1985), and this was also true at Antero. However, both clutch size and nest density decreased on Gull I. in 1984 and 1986 due to disturbances resulting in egg destruction and nest abandonment (Fig. 5b). While there was a large increase in the number of young breeding adults in 1984, there was no similar increase in 1986, though the frequency of young breeding adults remained high (Fig. 6).
Without knowing the entire, ongoing history of a colony, interpretation of reproductive variables may be highly biased. Changes in age structure and movement of individual birds result in altered colonywide reproductive performance. These changes are much more significant than simple reductions in fledging success as measured in most studies of breeding season disturbances.


27
Southern and Southern (1981) compared densities and clutch size changes between years in sub-areas (plots) of a Ring-billed Gull colony and concluded that disturbances in previous years could be detected by density changes greater than 10%. Additionally, changes in average clutch size and clutch size variance could reflect early season perturbations. At Antero, density decreased more than 10% in only 2 of 8 plots where predation had occurred the previous year; density actually increased in 1 of the 8 plots. Clutch size decreased in 6 of 8 plots, partly because 3-5 yr. old (naive) birds moved into plots after older adults failed to return. Clutch sizes were significantly different (t test: p<0.05) among 3 of the 8 plots. There were early season perturbations in only one of those plots. The Southerns' variation technique cannot predict disturbance variations for California Gulls at Antero Reservoir. This was especially true of Gull I., with its history of disturbances. Plots on Goose I. also changed significantly over time, but this change was caused by continuing growth of this colony. A decline in density or clutch size might be interpreted as an indicator of disturbance; however, it is clear that several years of monitoring are required before interpretation of any reproductive measures of success can be used in a predictive manner. If a colony has a history of disturbance, especially different types of disturbances, the predictive power of any single measure of ascertaining disturbance is greatly weakened.


CHAPTER IV
HUMAN DISTURBANCE
Immediate Behavioral Responses to Human Disturbance
Interspecific Aggression
Only responses to human intruders are considered here. Adults from this colony also responded aggressively to coyotes (Chapter VI), in panic to owls (Chapter V), and aggressively to Golden Eagles (Aquila chrysaetos) and Common Ravens (Corvus corax), whenever they approached the breeding colonies (Jehl and Chase 1987). Other species of gulls are rare at this location during the breeding season and no observations of interactions were recorded.
During egg-laying and incubation, attacks against the researchers occurred only while the researchers were moving or first sitting down to perform measurements. These attacks consisted of swoops with loud calls and occasional bill strikes to the researchers' head. Adults returned to nests within 2 m of a sitting researcher, especially during the last half of incubation. Similar results have been reported in other studies (Dutcher and Baily 1903, and Conover and Miller 1978). When chicks were old enough to run together in groups (approx. 15 days), they were captured in these groups and penned for measurement. After release, chicks would immediately run several meters and join other released chicks. Normally,


29
family groups reunited on territories within 5 minutes of chick release. Adults attacked the human intruders throughout the period that chicks were held captive. Once the last chick was released most adults settled on their territories within a couple of minutes.
From the beginning of egg-laying through the beginning of chickrearing, aggression toward human intruders escalated. Frequency of attacks during egg-laying and the first two weeks of incubation averaged less than 0.2/min. (n=50). During the third week, attacks averaged 0.5/min. (n=22), and during the last week of incubation and the first week of hatching, attacks averaged 2/min. (n=49). A high level of aggression (1-2 attacks/min.) was maintained until 30 days post-hatching (n=51). From that time until fledging (35-40 days), attacks diminished to 0.3/min. (n=38). There was no significant difference in frequency of attacks between low and high frequency disturbance plots (t test: p>0.05).
Burger (1981) and Dulude £l ai- (1987) found similar variation in aggressive behavior over the breeding season among Herring Gulls.
All birds did not attack with the same intensity and frequency. Typically, older males (>4 years) attacked more frequently (78% of all attacks where the bird could be aged and sexed, n=90 of 116) and more aggressively, often striking the human intruder repeatedly on the head. Females and younger males rarely struck the intruder (<2% of all attacks), limiting their attacks to shallow swoops and loud calling over the nest area. Certain individuals were much more aggressive than others, including one female that maintained a high attack rate from early incubation through the end of fledging. Additionally, I was identified by some birds and attacked at a greater frequency (3:1, n=32) than unknown intruders. Dutcher and Baily (1903) describe accounts of gulls


30
recognizing individual humans and behaving differently toward them than strangers. In their account, gulls attacked known individuals at a lower rate, contrary to findings in this study. My data suggest that at least some gulls will attack a known and regular intruder at a higher rate, possibly recognizing an intruder that doesn't "attack back."
Intraspecific Aggression
Many studies involving gulls, especially studies dealing with human disturbance, have found intraspecific aggression to be a major cause of egg and chick mortality (Hand 1980, Robert and Ralph 1975). Unless disturbed, at least one adult always attended a nest during incubation. In this study, clutches left unattended were frequently destroyed by neighbors and eggs were often eaten by neighbors if the nest was left untended for several minutes after a disturbance. Of 172 observed cases of egg destruction, 77% (133) were caused by neighbors and 23% (39) by immature, non-breeding marauders. Of 53 observations of eggs being eaten, 39% (21) involved an immature, non-breeding marauder, 42% (22) involved neighbors, and 19% (10) involved the male territory holder eating his own eggs. In all observed cases where the owner ate his own eggs, the eggs had been broken by an intruder, and the male ate the egg contents upon returning to his nest. Similarly, Chardine and Morris (1983) reported two males eating eggs after a presumed disturbance. Kirkman (1937) attributed cannibalism (eating eggs and chicks) to "unmated rogues," and Tinbergen (1953) attributed cannibalism to unmated 3-4 year olds. However, Parsons (1971) found that mated birds performed most of the attacks. Davis and Dunn (1976) found that most intra-specific mortality was caused by neighbors that had


31
lost their own eggs or chicks. They felt that the attacks on eggs and chicks were the result of repeated intrusions into neighbors' territories and were not simply opportunistic killings. At Antero, both young unmated marauders and neighbors destroyed large numbers of eggs and chicks. All observed intraspecific egg and chick killings were related to disturbance by humans.
Chicks that wandered into adjacent territories or were untended by parents were attacked by neighbors. Young (1963), Hunt and Hunt (1976), and Watanuki (1988) also found that unguarded chicks were subject to attack. Fetterolf (1983) reported more frequent fighting among adults, more attacks on chicks, and more chicks running during disturbances than before or after disturbances or in undisturbed conditions. From 1981-1983, I observed post-hatching behavioral responses of adults and chicks to disturbances for over 400 hours. As Fetterolf found, adults in the least disturbed plots rarely attacked chicks, chicks rarely ran, and adult fighting was less common (11 times) than in highly disturbed areas (Table 2a). In highly disturbed plots, adults attacked chicks 23 times more frequently than in least disturbed plots. The differences among the disturbance treatments for each of these behaviors were highly statistically significant (Table 2b).
Adoptions normally occurred when a chick left its parents' territory and successfuly joined another brood, sometimes permanently. Chicks rarely left their parents' territory if a parent was present. Only 4 adoptions were observed in the least disturbed areas, while 57 adoptions occurred in the highly disturbed plots. All observed adoptions occurred during and immediately after disturbances. One chick changed broods 7 times


32
TABLE 2. (a) Gull behavioral responses to three levels of experimental human (researcher) disturbance. Responses were recorded for 30 minutes before and 30 minutes after each disturbance. Four hundred hours of observations were recorded from 1981-1983. (b) Chi-square analysis of behavioral responses to three levels of experimental disturbance, df=2. Only post-disturbance periods were analyzed.
(a) Behavior, Low Medium High
no. per hour Before After Before After Before After
Adult fights with adults 0 2 0.4 15 1 22
Adult attacks on chicks 0 2 1 16 4 46
Chicks running outside terr. 0 6 0 12 2 16
No. adoptions 0 4 0 29 0 57
(b) Behavior n %2 P
Adult fights with adults 39 284.5 < 0.0001
Adult attacks on chicks 64 1066.0 < 0.0001
Chicks running outside terr. 34 22.7 < 0.0001
No. adoptions 90 858.5 <0.0001


33
before fledging, and 4 chicks changed broods 3 times each. Short-term (1 hour or less) broods of 7, 8, and 9 were observed in one plot in 1982.
It appeared that if a chick, regardless of age, could mix with a brood without a resident parent attacking it, the chick was accepted into the brood. If an adult observed a chick moving into its territory, it usually attacked immediately and vigorously drove it off, often killing young chicks. In contrast, Fetterolf (1983) found that few chicks over 10 days of age were adopted. In the highly disturbed areas on Gull I., 48% of adoptions were successful. It may be that the high frequency of disturbance caused parents to be more cautious about attacking their own broods and may have allowed a greater than normal number of successful adoptions. Alternatively, there may have been so much fighting that when the dust settled, adults simply accepted any chicks mixed with their brood that didn’t try to run away.
Parsons (1971) observed a male eating chicks but could not ascertain if that was a general occurence. At Antero, most chick attacks (80%, n=146 of 183) were performed by males. Other studies also found that males were the most aggressive sex (Pierotti 1981, Southern 1981, Southern and Southern 1982b, Dulude eiaL 1987, Watanuki 1988). Chicks less than 10 days old suffered the highest mortality per attack. An adult would pick-up a small chick and shake it and/or hammer it against the ground. Fewer than 10% of young chicks observed being killed (n=8 of 87) were subsequently eaten by the attacker. No chick older than 10 days was ever eaten after being killed by a neighbor (n=117). Older chicks were killed by repeated blows to the head. They were able to run from attacks and were more likely to survive repeated attacks than


34
younger chicks. After 30 days of age, chicks would often strike back when an adult attacked and rarely sustained serious injury.
The only gulls observed to eat chicks regularly were immature, nonbreeding marauders. They regularly preyed upon eggs and young chicks left exposed for several minutes. Cannibalism has been described in many gull studies, but only Herring, Slaty-backed (Larus schistisagus) and Western Gulls (Larus occidentalis) regularly eat chicks (Drost 1958, Brown 1967, Parsons 1971, Hunt 1972 and Watanuki 1988). However, even among these species, there are colonies where chicks were killed but not regularly eaten (Kadlec 1969). The majority of the killing of chicks at Antero appeared to be caused by the highly aggressive nature of California Gulls and not hunger (Paludan 1951), infanticide to reduce local competition, (Harris 1964),or spite (Pierotti 1980 and Pierrotti and Murphy 1987).
Most studies describe increases in both egg and chick mortalities with increasing disturbance (Hunt 1972, Hand 1980, Mousseau 1984, but see Gillette! a|- 1975). Roberts and Ralph(1975) described diminished intraspecific mortality in their most frequently disturbed area and attributed it to habituation. They observed fewer attacks in high disturbance plots as the season progressed. At Antero, there was a similar reduction in attacks, but it was a consequence of adults leaving territories after losing their clutches or broods. The frequency of attacks and chick mortality was always highest in the most disturbed areas. Approximately 50% of all losses caused by intraspecific aggression occurred in the egg stage with the other 50% happening during the chick stage, regardless of the frequency of disturbance.


35
Egg losses included only eggs that were broken or removed from the nest. The few clutches that failed due to thermal stress had been abandoned early in incubation. Diurnal temperatures never exceeded 30° C, well below temperatures described by Chappell elai- (1984) as causing heat stress to eggs in this species. Additionally, as adults were never away from nests for more than one hour (usually less than 5 minutes), it is not likely that there were significant losses due to thermally induced embryo mortality associated with human disturbance.
Three to five adult females were killed each year (1981-1983), apparently during territorial fights. There were significantly more fights in highly disturbed plots than in low disturbance plots (Table 2b). Only two of 13 females killed were in low disturbance plots. Female killings were observed at Mono Lake when male-male fights intruded on nest sites and drew females into the fight (Jehl, personal communication). The pattern of dead females at Antero suggests a similar scenario. From 1984-1987 (post-disturbance), only 7 females were found dead on Gull I.; however, the dead birds were located over the entire island unlike the locally concentrated killing in 1981-83.
Long-term Response? Tq Human DisimJaapca
The data from 12 of the 24 plots were confounded by additional factors, particularly owl predation (see Fig. 2). Only plots without additional disturbances were used in the following analysis of human disturbance.
There was a significant reduction in the number of chicks fledged per nest as the level of disturbance increased (Fig. 7), (One-way ANOVA


36
Fig. 7. Comparison between levels of disturbance and between years of the number of California Gull chicks fledging per nest


37
n=728, F=83.4, df=2, p<0.0001). After experimental disturbances stopped (1984-1985), the numbers of chicks fledged per nest were not significantly different from that of the least disturbed population mean (1.6 chicks fledge per nest) for any treatment area or year (Table 3a) (Oneway ANOVA n=363, F=1.03, df=2, p= ns). There were no significant differences in the numbers of chicks fledging per nest within each disturbance regime from year to year during either the disturbance (1981-1983) or the post-disturbance period (1984-1985). The slightly lower fledging success in 1984 for the high disturbance plots than in the low and medium plots (Fig. 7), could be explained by the higher percentage of young breeders that were then breeding in these plots.
Overall fledging successes were similar to many other studies comparing varying levels of disturbance (Shugart el a!-1981, Fetterolf 1983, Mousseau 1984). Mousseau (1984) found that different disturbance levels affected chick mortality rates but not egg mortality. In this study, both eggs and chicks were strongly affected by the changing levels of disturbance. Roberts and Ralph (1975) found the lowest mortality in least disturbed areas, but their most disturbed areas also experienced relatively high fledging success. They attributed this to habituation: chicks ran shorter distances and adults attacked less frequently. Chicks at Antero ran more frequently in highly disturbed plots than in low plots, and parental aggression was higher in high disturbance plots than in low plots (see Table 2). Increasing disturbance resulted in decreasing reproductive success. This finding is where most other disturbance studies stop.
Colonial waterbirds differ in post-disturbance responses in site fidelity. Species such as beach-nesting terns (McNicholl 1975), Ring-


38
TABLE 3. (a) One-way ANOVA of number of young fledged per nest. Sheffe's F was used to test differences between disturbance and post disturbance periods for each disturbance regime. Disturbances occurred in 1981-1983. (b) Kruskal-Wallis test comparing disturbance treatments each year (Low, Med, High) for the affects on percentage of breeders returning to same site in the following year, df=2. A Bonferoni correction was applied to correct for repeated samples.
(a) Year 81-83 Disturbance 84-85 Post-disturbance
Treatment n X SE n x SE F P
Low 264 1.68 0.079 125 1.65 0.091 0.038 NS
Med 240 1.2 0.037 133 1.66 0.053 54.56 <0.05
High 224 0.75 0.042 105 1.50 0.104 58.55 <0.05
(b) percentage adults returning n=13 plots per year
Year H P
82 5.90 0.17
83 8.61 <0.01
84 10.71 <0.001
85 3.15 0.87
%


39
billed Gulls (Conover and Miller 1978, Southern 1977, Southern and Southern 1981), and Black-legged Kittiwakes (Coulson and White 1958, Coulson and Thomas 1985) return to the same areas if undisturbed, but entire colonies often abandon disturbed areas. Bergman (1986) found that 38 of 40 successful pairs of solitary nesting Common Gull returned to the same site in the following year, while only 14 of 52 pairs that had lost chicks returned. In contrast, Kruuk (1964) records continually disturbed Black-headed Gull (Larus ridibundus) colonies that have been in the same site for over 300 years. California Gulls apparently fall between these two extremes.
There was a strong correlation between previous success (fledging) and percentage of adults returning to the same sites in the following year (r=0.86, df=23, p< 0.0001). Additionally, there was a continued decline in returning adults in the medium and high disturbance plots (Fig 8a). The percentage of birds returning in subsequent years to the same 10 m2 plot was significantly different between disturbance treatments in 2 of 3 years following disturbances (Table 3b). Each year following disturbances (1982-1984), the differences became larger and more significant. However, in 1985, two years after the experimental disturbances ceased, the percentage of adults returning to all plots approached the least disturbed population mean of 69%. The high disturbance plots still experienced somewhat lower returns, but again, this is explained by the higher proportion of young adults breeding there. A lower proportion of young adults returned (55%) than older adults (72%). This rate of return is lower than Southern elaJ. (1985) reported for Ring-billed Gulls (90.6%, n=48). I do not believe this is explained by species differences but is due to old bands (>8 years) that were lost or


40
Year
b)
o
D) (0 -•—> c o o
0.
V)
k_
d>
â– o
d)
k_
n
2
o
ra
>»
m
â– 
co
Year
Fig. 8. Comparison between levels of disturbance and between years of a). Percentage of adults returning to the same plot in the following year, b). Percentage of 3-5 year old breeders per plot in the following year.


41
that could not be read. Five and 6 year old gulls returned to least disturbed sites at an annual rate of return of 88%.
There was a significant increase in the frequency of 3-5 year old birds among the three disturbance regimes ( Dunn's multiple comparison test: p< 0.05 for low versus high 1982-1985, p<0.05 for medium versus high 1982-1983); however, this increase was almost completely attributable to increases of 3-5 year old breeders in the high disturbance plots (Fig. 4b). It was expected that older, more experienced birds would be less likely to move to an area that had been previously disturbed and that more naive birds would replace disturbed birds. When adults moved from the low and medium plots, they were replaced by other adults from other disturbed areas. The percentage of 3-5 year breeders (naive birds breeding for the first time) changed significantly only in the high disturbance plots, and this was correlated with the reduced number of returning older experienced adults (r= -0.80, df=11, p< 0.001).
Birds that moved into highly disturbed plots were new to the island or from the far side of the island (X 2 = 4.65, p<0.05), while birds that moved into the low and medium plots were as likely to come from adjacent plots as distant plots (%2 = 0.146, NS). Conover and Miller (1978) found that Ring-billed Gulls often avoided nesting in areas that had been the site of previous disturbances. Highly disturbed plots were no longer part of the central, desirable portion of the colony. At Antero, it appears that adults from immediately outside the high disturbance area could recognize and not use an area that had been highly disturbed in previous years. The age-structure shifted as highly disturbed plots became available to naive breeders. The difference between experimental plots was still evident two years later (Fig. 8b). The


42
proportion of young breeders declined in 1985 as these birds returned to the same sites in 1985 where they had bred in 1984. Many would then be classified as older breeders.
Movement patterns were compared between disturbance treatments (Fig. 9 and Table 4). The number of birds moving from the immediate area of their previous nest site was significantly different between each of the treatments (Table 5a). Additionally, birds that experienced multiple disturbances within one year or over several years moved more frequently than birds that were disturbed for only one season. Birds in the medium and high plots changed breeding islands more than birds on low plots, with many more leaving Gull I. and moving to Goose I. A higher proportion of birds was recorded as missing with increasing disturbance, suggesting that highly disturbed birds chose not to return to Antero Reservoir at all. This interpretation was confounded by birds that couldn't be relocated, birds with bands that couldn't be read or that were lost over the past year, and by birds that had died over the winter, as well as those that did not come back to Antero. In all cases, birds that experienced frequent or multiple disturbances moved from their original plot at a much higher frequency, changed islands more frequently, and were recorded missing more frequently.
Young birds moved from a previous breeding site significantly more often than older birds on the low and high disturbance areas (Table 5b). I believe there is a similar pattern for birds on the medium plots although it was not statistically significant. Birds that experienced multiple years of medium or high disturbances moved from the previous plot regardless of age. There was no significant difference between ages for birds changing islands versus those remaining on the same island regardless


43
a)
v>
â– a
k.
!5
o
0>
D>
(0
**
c
a>
o
k_
a>
CL
100
80
60
40
20
0
low + multi low medium high multi medium multi high
Level of disturbance ® Missing
£2 Moved ■ No Move
40 i
Level of disturbance
Fig. 9. Comparison between level of disturbance in 1981-83 and
a) . Percentage of birds that moved from the previous breeding plot.
b) . Percentage of birds that changed islands or remained on the same island. No Move + Move + Missing = 100%, Move = Same Island + Change Island. Multiple disturbances are birds disturbed in more than one year


44
TABLE 4. Comparison of movement responses by breeding birds in the year following experimental disturbance treatments (1982-1984) on Gull Island. Moved = birds moved to a different plot in subsequent years and either moved elsewhere on Gull I. (Same) or moved to Goose I. (Change). Move = Same Island + Change Island. Multi Low, Multi Medium and Multi High include birds that experienced more than one year of disturbance.
Disturb. Level n No Move Move Same I. Change I. Missing
Low + Multi Low 235 162 20 17 3 53
Medium 166 78 46 21 25 42
High 184 57 73 26 47 54
Multi Medium 89 28 33 10 23 28
Multi High 63 12 28 7 21 23


45
TABLE 5. (a) Two by two contingency analysis of breeding gulls movement response to disturbance treatments. Each case tested whether the second category moved more than the first, (b) Two by two contingency analysis of age-related response of breeding birds to disturbance treatment. This tested whether 3-5 year old breeders moved more frequently than older breeders. Move = bird moved from previous 10 m2 breeding site in subsequent year. Change = bird moved to Goose I. following a disturbance on Gull I. Multi Med and Multi High include birds that experienced more than one year of disturbance. Continuity correction applied to all multiple disturbances due to small cell sizes;
df=1.
(a) Move Change
X2 P X2 P
Low vs Med 29.72 <0.001 8.84 <0.01
Low vs High 73.93 <0.001 15.40 <0.001
Med vs High 9.26 <0.01 1.19 NS
Med vs Multi Med 4.83 <0.05 1.90 NS
High vs Multi High 2.43 NS 1.04 NS
(b) 3-5 yr vs. > 5 yr n=167 n=570 X2 P X 2 P
Low 10.46 <0.001 0.15 NS
Medium 1.20 NS 0.08 NS
High 5.53 <0.05 0.94 NS
Multi Med 3.34 NS 0.03 NS
Multi High 0.13 NS 0.08 NS


46
of the disturbance frequency (Table 5b).
Birds do not always abandon colonies following severe disturbances. In this study, California Gulls moved from the immediate breeding territory where disturbances occurred. Additionally, the more frequently a bird was disturbed the farther it moved from the original site.
At a certain point, birds failed to return to Antero. This point depended on the bird's age, previous experience in the colony, and individual tenacity. Young birds tended to move most frequently under any conditions, while adults displayed a greater site tenacity.
Uncontrolled Human Disturbance
On three occasions, fishermen were observed on Goose I. while I was working on Gull I. As they walked through the colony, all birds would take flight and chicks would run across the island and into the water. After the fishermen left, it took up to one hour for all chicks to return to their territories. This period was accompanied by adults fighting, eggs being pecked, and chicks being killed. On one occasion, 37 chicks (less than 15 days old) were found dead the morning after one 10 minute intrusion.
In contrast, only 5 chicks were found dead the morning after the annual 4 July banding operation when 1100 chicks were measured over a three hour period. In the latter case the chicks were all 25+ days old and human movements were controlled.
In 1984, a more serious intrusion occurred on Gull I. during the early egg-laying stage. Three people landed on the island and commenced destroying eggs throughout the entire colony. Many nests were entirely destroyed, and although re-laying started immediately, the


47
1984 average clutch size recorded on the 25 May count was 2.0 in contrast to average clutch size in undisturbed areas of 2.4. It is impossible to say how much of this reduction was due to this disturbance versus the previous years' experimental disturbance. However, it is clear that uncontrolled human activities may cause high mortality of young and rapid abandonment of entire colonies (Anderson and Keith 1980, Erwin 1980, Desgranges and Reed 1981).


CHAPTER V
OWL PREDATION
Immediate Response
At least one Great Horned Owl (Bubo virginiana) has preyed upon the Antero Reservoir California Gull population each year since 1981 (Table 6). In this study, I examined the gulls' immediate and longterm response to owl predation.
The owl nested on a hill covered with ponderosa pine (Pinus ponderosa ) approximately 2 km south of Gull I. There were no other owls nesting within 5 km of Antero. Only one adult gull was killed each night the owl attacked. All but one were killed during the incubation period and were located on either the exposed south and southeast side of Gull I. or on the south side of Goose I.
The owl began killing chicks when they became available. Owls shifting predation from adults to chicks has also been noted by Kruuk (1964), Vermeer (1970) and Jehl and Chase (at Mono Lake) (1987). No successful attacks on adults occurred on moonlit nights. Southern and Southern (1979) found that an owl preyed only in areas away from harbor lights. It is likely that adults were able to see an owl approaching and fled the colony before the owl arrived.
Attacks consisted of direct, low level flights with the owl striking a gull upon entering the colony. Of the 29 adults killed, 70% were females;


49
TABLE 6. Summary of owl predation on California Gulls breeding on Gull and Goose I. Percentage mortality is based on the number of eggs laid by 25 May.
year 1981 1982 1983 1984 1985 1986 1987
Gull I.
no. chicks killed 33 102 81 62 39 31
percent. mortality 2.4 7.5 7.9 8.4 8.1 4.3
no. adults killed Goose 1. 10 5 1 3 3
no. chicks killed - - - 37 88 7
percent mortality - - - 1.6 5.4 0.4
no. adults killed - - - 1 5 1


50
females typically incubated late at night (73-89%). I was present during 5 attacks. Adult response to intrusion by an owl was mass flight and abandonment of the island for 30 minutes to several hours. A similar response has also been reported by Emlen eiai- (1966), Southern and Southern (1978), and Southern £lai- (1982 and 1985). Hunter and Morris (1976), Southern and Southern (1978), and others report that adults often remained away from colonies until dawn. At Antero and other colonies in Wyoming and Montana, California Gulls returned to chicks well before dawn. Adults returned more rapidly following an owl intrusion when chicks were present than during incubation (X 2 = 15.85, df=2, p<0.001). Chilling of eggs and chicks was common during these periods. Nocturnal temperatures averaged 0-5° C, and chicks less than 15 days old succumbed to cold stress in about 5 minutes. Over 50% (250 of 442, 1981-1985) of all chick mortality in response to disturbance by the owl was the indirect result of thermoregulatory stress. Antero is the highest gull colony in North America (2770m) and has correspondingly cooler ambient temperatures than other colonies reporting owl predation. Adult gulls are presented with a conflict: return too early and be killed, or return too late and lose eggs or chicks to cold.
While I did not attempt to study mortality of embryos, it is likely there was at least some increased mortality and/or extension of the length of incubation. Hunter and Morris (1976) found that embryo mortality increased with age of the embryo when exposed to 10° C for 4 hours. Additionally, they found increased asynchrony of hatching and delayed hatching for chilled eggs. Hunter and Morris (1976) reported that storms during owl-caused desertions increased egg losses to approximately 90% among clutches that were near hatching. Vermeer (1970) also


51
noted increased egg losses following intrusion by owls. Nisbet (1975) found lower hatching success and longer incubation periods for Common Terns disturbed by nocturnal predators.
Chick response to an owl in the colony varied with age; young chicks crouched and remained motionless while older chicks (>15 days) ran and grouped together. Over 90% of direct chick mortality occurred among chicks over 15 days old. Nisbet (1975) found that owls mainly killed Common Tern chicks less than 9 days old. These differences may simply reflect idosyncratic behavior of individual owls (Jehl and Chase 1987).
Typically, at Antero, 2-3 broods would be killed in night attacks lasting only a few minutes. The owl would then eat the breast and viscera of one or two chicks, leaving severed wings, head, and lower torso. Adult kills often appeared uneaten (Southern and Southern 1979), but close inspection of all kills at Antero revealed that the upper breast and viscera were always missing. The owl would preferentially attack grouped chicks (51 chicks were killed in less than 30 minutes on 21 June 1982). Superstimulus killing has been reported in gull colonies for both owls and mammalian predators (see reviews in Southern and Southern 1982a and Southern elai- 1985).
Mortality was always concentrated in areas of dense nests with high visibility (Fig.10). These areas were on an open slope and hilltop directly facing the owl's approach. No adults or chicks were killed in the shrubby areas by owls. However, it is not likely that there is strong selection for shrubby habitats for breeding as selection by diurnal raptors occurrs against birds nesting in dense vegetation, at least in some breeding areas (Jehl and Chase 1987 ).


52
Long-term response
Responses of Gull I. birds to predation by owls were compared for plots that experienced: a), no predation, b). light predation, where only a few chicks were killed, typically on a single night, c). concentrated predation, where most chicks in a plot (>50%) were killed, usually over several nights, and d). multiple predation, attacks occurred in the same plot 2 or more years (Fig. 11 a, Table 7). As the intensity of predation increased, significantly more birds moved from previous breeding sites. Birds that experienced multiple disturbances moved most frequently, with only 7% returning to previous breeding sites. There were no significant differences in movements of birds from light and concentrated predation areas. If an owl preyed in an area, there was a 300-400% increase in birds moving regardless of the concentration of predation.
Predation by the owl resulted in a significant increase in the probability of changing islands (indicative of distance moved); however, there were no significant differences between different intensities of predation and whether a bird changed breeding islands or not (Fig 11b, Table 7). While there were no statistically significant differences in distances moved for birds experiencing no predation vs. light predation, I believe this is likely a statistical artifact due to small sample size.
Birds that had directly experienced loss of chicks were compared with birds that nested in the area but did not lose chicks (Fig. 12a). A significantly greater number of birds that moved from previous nesting sites had experienced loss of chicks compared to birds that did not lose chicks (X 2 =15.76, df=1, p<0.0001). Even when birds did not lose young, a significantly greater number moved from predation areas when


Fig. 10. Major locations of chick mortality from Great Horned Owl predation on Gull I.
cn
oo


54
a)
(/>
TJ
Q>
D)
re
c
a>
o
k.
0)
a.
No Predation Light Concentrated Multi-predation
Intensity of predation |§ Missing
^ Moved
| No Move
b)
V)
â– a
k.
X3
O
a)
D>
re
a>
o
k_
a>
a.
40 n
Same Island Change Island
No Predation Light Concentrated
Intensity of predation
Multi-predation
Fig. 11. Comparison between intensity of predation and a). Percentage of birds that moved from the previous breeding plot. b). Percentage of birds that changed islands or remained on the same island.
No Move + Move + Missing = 100%, Move = Same Island + Change Island. Multi-predation = predation in more than one year.


55
TABLE 7. (a) Comparison of California Gull movement patterns in the years following different intensities of owl predation in breeding plots. Light = few chicks killed, usually in one attack, Concentrated = >50% of chicks in plot killed, usually over several attacks, Multiple Predation = attacks in the same plot over more than one year, (b) Two by two contingency tests of movement patterns and intensities of predation. Move = Same Island + Change Island, NP = No predation, L = Light predation, C = Concentrated predation, MP = Multiple predation.
(a) Intensity of Predation total n No Move Move n % n % Same I n % Change I. n % Missing n %
No Pred. 150 99 66 17 11 14 9 3 2 34 23
Light 101 31 31 48 48 31 31 17 17 22 22
Concent. 84 15 18 47 56 24 29 23 27 22 26
Multiple Predation 92 6 7 57 62 29 32 28 30 29 32
(b) Intensity of Predation Move vs. No Move Change I. vs Same I.
df X2 P df X2 P
Overall differences 3 118.79 <0.0001 3 7.11 NS
NP vs. L 1 45.95 <0.0001 1 1.86 NS
NP vs C 1 65.61 <0.0001 1 5.07 <0.05
NP vs MP 1 96.80 <0.0001 1 5.33 <0.05
L vs C 1 3.56 NS 1 1.78 NS
L vs MP 1 16.06 <0.0001 1 2.0 NS
C vs MP 1 4.81 <0.05 1 0.0004 NS


56
Disturbance Direct mortality
Intensity of predation
b)
V)
T3
n
o
<1)
O)
ro
o
k.
d)
a
30 -]
20 -
10 -
0
Age of breeding birds
Fig. 12 Comparison of gull movement patterns influenced by Great Horned Owl predation, a) Disturbance = birds in predation areas but not experiencing direct mortality. Direct Mortality = birds experiencing loss of chicks due to Great Horned Owl predation, b). Age of breeding birds and breeding site location in the year following owl predation.


57
compared to movement of birds from nonpredation areas (X2 = 34.45, df=1, p<0.05). There was also a significant increase in the frequency with which birds that lost chicks changed islands (X 2 = 4.90, df=1, p<0.05).
There is a highly significant difference in movement patterns of 3-5 year old breeders versus breeders older than 5 years (Fig. 12b) (X 2 = 57.81, df=1, p<0.0001,3-5: n=38, >5: n=75). Young birds nested more frequently on the periphery of the colonies. When they moved in response to predation, they moved more frequently to other areas on the periphery of the colony. Older birds nested more frequently in the center of the colony. After eggs or young were destroyed by predators, older gulls moved to other areas in the center of the colony. There was no significant increase in older adults moving to the periphery, as would be expected if the adults had lost status and had difficulty re-establishing territories in central areas. It may also be that nest densities on Gull I., at least in some plots, were below saturation densities; however, if that were the case, a greater proportion of young breeders would be expected to move to central plots in the subsequent year. The results of the human disturbance experiments (Chapter IV), showed an increase in number of young breeders in high disturbance plots corresponding to a decrease in older adults.
Any owl predation caused both immediate and long-term responses by California Gulls, even if their chicks were not killed. Owl predation occurred in small areas of the colony, and only birds near those areas appeared to have long-term reactions to the owl. Predation over a larger area was shown to cause immediate abandonment of an island at Mono Lake, California, and much lower use in later years (Jehl and Chase 1987).


CHAPTER VI
WATER LEVEL CHANGES Draining
Immediate Response
Antero was drained in 1986 to repair the dam. By 1 May (peak initiation of egg-laying), water levels had dropped 5 meters and a land bridge had formed. Coyotes entered the colonies and ate clutches of eggs. Mass abandonment by adults for several hours occurred when a coyote entered the colony at night, whereas entering during the day resulted in mobbing attacks by the gulls. Coyotes ate relatively few eggs in comparison to those destroyed by the intraspecific aggression that followed disturbance by the coyotes. The gulls were highly agitated throughout the day and rarely incubated for more than a few hours at a time. Fights constantly occurred, and twice as many adults (mostly females) died compared to other years (11 vs 3-7). Davis and Dunn (1976) found that neighbors that destroyed other nests had already lost their own eggs. Such behavior spreads rapidly in highly disturbed colonies.
Gull I. was visited by coyotes at an earlier date and more frequently than Goose I. It was completely abandoned by 5 June. Goose I. was completely abandoned by 15 June. No chicks hatched on either island. All Canada Goose (Branta canadensis) nesting was also disrupted


59
(Chase, in prep). Twenty gull eggs were examined on Gull I. on 5 June.
All were non-viable; 8 had no signs of embryo development, 7 had small embryos, and 5 had well developed embryos. Fifty eggs were examined on Goose I. on 11 June: 22 had no embryo development, 16 had small embryos, and 12 had well developed embryos. Much of the mortality probably occurred during early incubation when temperatures were often below 0° C and the adults were off the eggs for several hours at a time.
By 30 June, there were less than 150 gulls remaining at Antero, and none of them were using the islands. They roosted in the dry lake bed approximately 12 m from the islands. This area had been used as a roost site since nest abandonment began in mid-May.
Long-Term Response
In 1987, the reservoir was refilled, and the resulting water level was
0. 6 meters higher than normal. Only 6 (15%) of older adults (total n=41) returned to previously used plots on Gull I. (Fig. 13a). At the same time,
49 (34%) of older adults (total n=143) returned to previous plots on Goose
1. (X 2 = 6.9, p<0.01). Nero (1961) reported a similar dry lake situation in 1959 in Saskatchewan at a Ring-billed and California Gull colony. Some hatching occurred but no chicks fledged. The lake filled the next year, and the colony was resettled the next year. There were no significant differences between young (total n=74) and older gulls reusing the same plots on Gull I., and there were no significant differences between young gulls from either island in any comparison. The latter finding was expected, since young birds from either island should have the same experiences at Antero and should react similarly.


60
80 -a)
3-5 YR/ GULL >5 YR/GULL >5 YR/GOOSE Age of birds/Breeding Island
50 ~i
3-5 YR/ GULL >5 YR/GULL >5 YR/GOOSE Age of birds/ Breeding Island
Fig. 13. Comparison of movement patterns influenced by draining Antero Reservoir, a). Percentage of birds that moved from the previous breeding plot. b). Percentage of birds that changed islands or remained on the same island. 3-5 YR/GULL = 3-5 year old adults returning to Gull I in subsequent years; >5 YR/GULL= adults> 5 years old returning to Gull I.; >5 YR/Goose = adults > 5 years old returning to Goose I.


61
It is interesting to note that only one older bird moved from Goose I. to Gull I. following this disturbance, while 10 (24%) of the older Gull I. birds moved to Goose I. (Fig. 13b) (X 2=29.55, p<0.0001). Experienced adults did not immigrate to the most disturbed island even when disturbed on their current breeding site. There was a significant difference between ages of Gull I. birds with respect to whether they stayed on Gull I. or moved to Goose I., with young birds remaining on Gull I. much more frequently (X2 = 10.66, p<0.001). A comparison of birds recorded missing demonstrates that a significantly higher number of older adults were missing from Gull I. after this disturbance than from earlier years (X 2 = 12.19, p< 0.001). Young adults from Gull and Goose Is. and older adults from Goose I. didn't leave Antero at significantly higher frequencies after this disturbance than in previous years. I can't attribute this difference exclusively to the affects of the disturbance caused by draining. Gull I. was subjected to numerous disturbances during the study period 1981-1987. Additionally, as explained earlier, the missing category must be interpreted very cautiously. Birds that have been disturbed almost continually since 1981 appear to have left Antero at a higher frequency than birds that experienced less cumulative disturbance, especially young adults. Axel (1956), Kruuk (1964), and Patterson (1965) found patterns of declining numbers but no abandonment, even with continual disturbance.
Flooding
Flooding affected some peripheral plots in 1985 and 1987, when water levels rose 0.6 m above normal before the beginning of egg laying.


62
Birds of all ages from both islands reacted similarly to flooded conditions. Birds that had previously nested in a plot did not spend much time searching the immediate flooded area but instead set up territories elsewhere. Only one bird returned during a flooded year and nested in the same plot as the previous year. This bird nested on the new shoreline on the unflooded edge of the plot. Most birds (71%) moved to other parts of the islands, with only 2 birds changing islands (Fig. 14) (total n=134). The number of birds recorded missing was not significantly different from least disturbed areas in earlier years. The response to this disturbance is less dramatic than other disturbances that involved loss of eggs or chicks. In this disturbance, the birds were reacting to conditions that had changed since the last breeding season instead of disturbances during the last breeding season.
No birds that had nested in plots in 1984 that were subsequently flooded in 1985 returned to those plots in 1986. Scharf (1981) reported that Common Terns nested on areas that had previously been inundated and often renested in sub-optimal habitat. In this study, only new territory holders (73% were 3-5 year olds) nested in these plots in 1986. Throughout this study, approximately 70% of all birds nesting in peripheral plots were 3-5 years old, so there was no change in age structure following this disturbance. Since this type of disturbance does little to change the overall fitness of a breeding bird, less extreme movements in reponse to this disturbance were expected.


Percentage of birds
63
NO MOVE <25M 25-50M >50M CHANGE I MISSING
Distance moved
Fig. 14. Distance moved from previous breeding season in response to flooding, 1985 and 1987; n=134. Change I. = Birds that moved from Gull I. to Goose I. after experiencing flooding.


CHAPTER VII
CONCLUSIONS AND SUMMARY
In this study I examined various types and frequencies of disturbance affecting the breeding colony of California Gulls at Antero Reservoir over a seven year period (1981 -1987). The effects of major disturbances on adult nest site choice in subsequent years are shown in Tables 8a and 9.
Annual fledging decreased from a high of 69% (1.6 per nest) in the least disturbed plots to 0% when Antero was drained and coyotes entered the colonies. The "flooded" category must be interpreted cautiously, because birds cannot nest on flooded plots and most of the birds that nest in the peripheral plots that were flooded were young birds with a lower frequency of return. A bird that moved from a plot or from another colony, such as the invasion of birds from Great Salt Lake in 1985, and successfully bred in the new area is unlikely to return to the original breeding site. On the other hand, many birds that experienced breeding failure did not immediately abandon Antero. Many returned to the same plots the next year while others relocated elsewhere on the island or moved to a nearby colony (re.Goose I.) (Tables 8 and 9).
Gulls will try to establish territories in more preferred, central colony locations (Pugesek 1983), but are limited by the presence of older gulls. The central portion of a colony is typically used by older and more


65
TABLE 8. (a) Comparison between disturbance regimes and movement patterns of breeding gulls on Gull I. (b) Rank comparisons of response to disturbances, 1 = smallest proportion of sample, 7 = greatest proportion of sample. Multi = Disturbance in more than one year.
(a) Disturbance Total n No move n/% Move n/% Missing n/%
Low 235 162/69 20/9 53/23
High 184 57/31 73/40 54/29
Multi-high Human 63 12/19 28/44 23/37
Owl 185 46/25 95/51 44/24
Multi-pred Owl 92 6/7 57/62 29/32
Drain/Coyote 115 14/12 67/58 34/30
Flood 135 2/1 100/74 33/24
(b) Disturbance No Move Move Missing
Low 7 1 1
High 6 2 4
Multi-high Human 5 3 7
Owl 4 4 2
Multi-pred Owl 2 6 6
Drain/Coyote 3 5 5
Flood 1 7 3


66
TABLE 9. Comparisons of movement patterns of breeding gulls on Gull I. from original breeding plots following disturbances in the previous year. Same I = moved to a different plot on Gull I., Change I. = moved to Goose I. Multi = Disturbance in more than one year.
Disturbance No. moved n Same I. n/% Change I n/% . Rank comparison Same I. Change I.
Low 20 17/85 3/1 6 2
High 73 26/36 47/64 2 6
Multi-high Human 28 7/25 21/75 1 7
Owl 95 55/58 40/42 4 4
Multi-pred Owl 57 29/51 28/49 3 5
Drain/Coyote 67 46/69 21/31 5 3
Flood 100 96/96 4/4 7 1


67
successful birds (e.g. Coulson 1968). In my study, as disturbances continued on Gull I., fewer adults returned to previous central nesting sites . A greater number of young adults were able to establish territories in areas that were previously unavailable to them. This would normally be advantagous; however, with continuing disturbance, no advantage was obtained, and, in fact, a loss of fitness may have resulted.
Chi square analysis (Table 10) of movement patterns (Table 8a) demonstrates highly significant differences between all disturbances and the least disturbed areas. I ranked movement responses of California Gulls at Antero to the various disturbances during 1981-1986 (Tables 8b and 9). Disturbance by the owl and the combined effects of draining and subsequent coyote predation resulted in significantly more birds moving than resulted from the most frequently repeated controlled human disturbance. However, it is interesting to note that the birds that did move changed islands more frequently as the frequency of disturbance increased, regardless of the cause of disturbance. High human disturbance plots were disturbed more frequently than were the owl disturbance plots. Similar numbers of young fledged under both disturbance regimes; however, birds changed islands more frequently when exposed to high and multiple high levels of human disturbance versus predation. This finding is counter-intuitive and may be due to sampling error or sample size problems.
There was little difference in movements due to owl predation and draining the reservoir. This result was not unexpected since both disturbances were caused by nocturnal predators (owls and coyotes, respectively) and resulted in mortality to eggs, chicks, and even adults. The frequency with which birds changed islands in response to flooding


68
TABLE 10. Chi square values and probabilities for comparisons between disturbances and percentages of birds in Table 8a. The first line of values is for Move/No Move comparisons, the second line of values are for Same l./Change I. comparisons. No comparisons were made for Move/No Move for the flood category since birds had no choice but to move. * = p<0.05.
Disturbance Low Human High Human Multi Human Owl Multi Predation Predation Drain
Low Human “
High Human 73.9 * 15.4 * -
Multi Human 67.4 * 16.8 * 2.4 NS 1.0 NS -
Owl 110.2 * 5.2 * 3.6 NS 8.2 * 0.2 NS 9.2 * -
Multi Owl 137.2 * 7.5 * 22.7 * 2.7 * 7.1 * 5.2 * 8.1 * 0.6 NS -
Drain 130.3 * 2.1 NS 15.8 * 15.3 * 2.6 NS 15.3 * 6.1 * 3.1 * 1.8 NS 4.4 * -
Flood 3.7 * 74.0 * 70.2 * 47.6 * 46.6 * 23.6 *


69
was slightly greater in the least disturbed plots but was much less than all other disturbances.
Reproductive losses can be high following human intrusion because of associated interspecific predation as well as intraspecific predation (Davis and Dunn 1976, Anderson and Keith 1980, Hand 1980). It is likely that human intrusion (in contrast to the controlled movements of this study) ellicits responses as strong as reactions to predators alone. Multiple disturbances of all types that resulted in loss of chicks also resulted in the most movement within and between islands.
California Gulls nest in relatively stable colonies (Behle 1953, Conover and Miller 1978, and Paul 1985). When colony sites are dramatically disturbed (flooding at Great Salt Lake, mammal predation in Wyoming), birds abandon the original colony and move long distances to new breeding colonies. Less intense disturbances result in less dramatic changes within a colony and less frequent movements of individual birds from one local colony to another (e.g. Gull I. and Goose I.).
Overall movement patterns and subsequent reproductive performances on Gull and Goose I. were directly related. With young breeders replacing adults, there may be no apparent change in the population demography and perhaps even a measured increase in the number of breeding birds. However, without knowing individual birds, changes in a population may be obscured.
There are many reasons to control researcher activity in breeding colonies of waterbirds as Fetterolf (1983) and others have stated. Interpretation of reproductive output without knowing the history of a colony will lead to misinterpretations of ecological questions and poor information upon which to base conservation decisions. Studying a


70
disturbance for only one year may also lead to misinterpretation. It is clear from this study that in California Gulls, there is a decrease in number of nests, clutch size, and fledging success with increasing disturbance as well as changes in individual nest site selection. There are long term effects on breeding individuals, especially as an increasing percentage of an adult's reproductive fitness is lost to disturbance.
Long-term studies are essential to understand the ecology of colonial breeding waterbirds, but it is also critical that disturbances caused by researchers, as well as other disturbances, be understood and quantified as accurately as possible. A colonial waterbird study with no researcher-caused disturbance has yet to be conducted; and, given the need to mark, weigh, and measure eggs, chicks, and adults, it is not likely to occur. Choosing critical times and methods that minimize disturbance must become a standard in colonial bird research.


LITERATURE CITED
Anderson, D., and J. O. Keith. 1980. The human influence on seabird nesting success: conservation implications. Biological Conservation 18:65-80.
Austin, O. L. 1951. Group adherence in the Common Tern (Sterna hirundo). Bird Banding 22:1-15.
Axell, R. E. 1956. Predation and protection at Dungeness Bird Reserve. British Birds 49:193-212.
Behle, W. H. 1958. The bird life of Great Salt Lake. Univ. Utah Press, Salt Lake City.
__________and W. A. Goates. 1957. Breeding biology of the California
Gull. Condor 59:235-246.
Bergman, G. 1986. Feeding habits, accommodation to men, breeding success and aspects of coloniality in the Common Gull. Ornis Fennica 63:65-78.
Boellstorf, D. E., D. W. Anderson, H. M. Ohlendorf, and E. J. O'Neil. 1988. Reproductive effects of nest-marking studies in an American White Pelican colony. Colonial Waterbirds 11:215-219.
Brown, R. G. B. 1967. Breeding success and population growth in a colony of Herring and Lesser Black-backed Gulls (Larus argentatus and L. fuscus). Behaviour 29:122-153.
Buckley, F. G. and P. A. Buckley. 1972. The breeding ecology of Royal Terns (Sterna m. maxima). Ibis 114:344-359.
Burger, J. 1980. Territory size differences in relation to reproductive stage and type of intruder in Herring Gulls (Larus argentatus). Auk 97:733-741.
__________. 1981. Effects of human disturbance on colonial species,
particularly gulls. Colonial Waterbirds 4:28-36.


72
Chabrzyk, G. and J.C. Coulson. 1976. Survival and recruitment in the Herring Gull, (Lams argentatus). Journal of Animal Ecology 45:187-203.
Chappell, M., D. Goldstein and D. Winkler. 1984. Oxygen consumption, evaporative water loss, and temperature regulation of California Gull chicks (Larus californicus) in a desert rookery. Physiological Zoology 57:204-214.
Chardine, J. W. and R. D. Morris. 1983. Herring Gull males eat their own eggs. Wilson Bulletin 95:477-478.
Conover, M. R. and D. E. Miller. 1978. Reaction of Ring-Billed Gulls to predators and human disturbance at their breeding colonies. Proceedings Colonial Waterbird Group 2:41-47.
__________and D. E. Miller. 1980. Daily activity patterns of breeding
Ring-billed and California Gulls. Journal of Field Ornithology 51:329-339.
Coulson, J. C. 1968. Differences in the quality of birds nesting in the centre and on the edges of a colony. Nature 217:478-479.
__________and E. White. 1958. The effect of age on the breeding
biology of the Kittiwake (Rissa tridactyla). Ibis 100:40-51.
__________, N. Duncan, C. S. Thomas and P. Monaghan. 1981. An
age-related difference in the bill depth of Herring Gulls. Ibis 123:499-502.
__________, N. Duncan and C. Thomas. 1982. Changes in the breeding
biology of the Herring Gull (Larus argentatus) induced by reduction in the size and density of the colony. Journal of Animal Ecology 51:739-756.
__________and J. M. Porter. 1985. Reproductive success of the
Kittiwake: the role of clutch size, chick growth rates and parental quality. Ibis 127:450-460.
__________and C. S. Thomas. 1985. Changes in the biology of the
Kittiwake: a 31-year study of a breeding colony. Journal of Animal Ecology 54:9-26.
Cuthbert, F. J. 1985a. Intraseasonal movement between colony sites by Caspian Terns in the Great Lakes. Wilson Bulletin 97:502-510.
__________. 1985b. Mate retention in Caspian Terns. Condor 87:74-78.


73
Davis, J. W. F. and E. K. Dunn. 1976. Intraspecific predation and
colonial breeding in Lesser Black-backed Gulls (Larus fuscus). Ibis 118:65-77.
Desgranges, J. L. and A. Reed. 1981. Disturbance and control of colonies of Double-Crested Cormorants in Quebec. Colonial Waterbirds 4:12-19.
Dewit, A. A. N. and A. L. Spaans. 1984. Veranderingen in de broedbiologie van de Zilvermeeuw (Larus argentatus ) door oegenomen aantallen. Limosa 57:87-90.
Drapeau, P., R. McNeil and J. Burton. 1984. Effects of human disturbance on the activity of the Double-crested Cormorant (Phalacrocorax auritus) and on the reproduction of the Great Blue Heron (Ardea herodias) in the Magdalen Islands, Canada.
Candian Field-Naturalist 98:219-222.
Drost, R. 1958. Uber die ansiedlung von jung ins Binnenland ver frachteten Silbermowen {Larus argentatus). Vogelwart 19:169-173.
Duffy, D. C. 1979. Human disturbance and breeding birds. Auk 96:815-816.
Dulude, A. M., G. Baron and R. McNeil. 1987. Role of male and female Ring-billed Gullls in the care of young and terrritorial defense. Canadian Journal of Zoology. 65:1535-1540.
Dutcher, W. and W. Baily. 1903. A contribution to the life history of the Herring Gull (Larus argentatus) in the U.S. Auk 20:417-431.
Ellison, L. N. and L. Cleary. 1978. Effects of human disturbance on the breeding of Double-crested Cormorants. Auk 95:510-517.
__________. 1979. Author's response. Auk 96:816-817.
Emlen, J. T. 1956. Juvenile mortality in a Ring-billed Gull colony. Wilson Bulletin 68:232-238.
__________, D. Miller, R. M. Evans and D. H. Thompson. 1966. Predator-
induced parental neglect in a Ring-billed Gull colony. Auk 83:677-679.
Erwin, R. M. 1980. Breeding habitat use by colonially nesting
waterbirds in two mid-Atlantic U.S. regions under different regimes of human disturbance. Biological Conservation 18:39-51.


74
__________. 1989. Responses to human intruders by birds nesting in
colonies: experimental results and management and guidelines. Colonial Waterbirds 12:104-108.
Fetterolf, P. M. 1983. Effects of investigator activity on Ring-billed Gull behavior and reproductive performance. Wilson Bulletin 95:23-41.
Gillett, W. H., J. L. Hayward, and J. F. Stout. 1975. Effects of human activity on egg and chick mortality in a Glaucous-winged Gull colony. Condor 77:492-495.
Hand, J. L. 1980. Human disturbance in Western Gull (L. occidentalis livens) colonies and possible amplification of intraspecific predation. Biological Conservation 18:59-63.
Harris, M. P. 1964. Aspects of the breeding biology of the gulls Lams argentatus, L. fuscus, and L. marinus. Ibis 106:432-456.
Hatch, J. J. 1970. Predation and piracy by gulls at a ternery in Maine. Auk 87: 244-254.
Herrick, F. 1909. Organization of the gull community. Proceedings International Zoological Congress, Boston.
Hollander, M., and D. A. Wolfe. 1973. Nonparametric statistical methods. Wiley and Sons, N.Y.
Hunt, G. L. 1972. Influence of food distribution and human disturbance on the reproductive success of Herring Gulls. Ecology 53:1051-1061.
__________and M. W. Hunt. 1975. Reproductive ecology of the Western
Gull: the importance of nesting space. Auk 92:270-279.
__________and M. W. Hunt. 1976. Gull chicks survival: the significance
of growth rates, timing of breeding and territory size. Ecology 57:62-75.
Hunter, R. A. and R. D. Morris. 1976. Nocturnal predation by a Black-crowned Night Heron at a Common Tern colony. Auk 93:629-633.
Jehl, J. and C. A. Chase. 1987. Foraging patterns and prey selection by avian predators: a comparative study in two colonies of California Gulls. Studies in Avian Biology 10:91-101.
Johnston, D. W. 1956. The annual reproductive cycle of the California Gull. I. Criteria of age and male reproductive cycle. Condor 58:134-162.


75
__________. 1956. The annual reproductive cycle of the California
Gull. II. Histology and female reroductive system. Condor 58:206-221.
Kadlec, J. A. 1971. Effects of introducing foxes and racoons in Herring Gull colonies. Journal of Wildlife Management 35:625-636.
__________, W. H. Drury and D. K. Onion. 1969. Growth and mortality of
Herring Gull chicks. Bird Banding 40:222-233.
__________and W. H. Drury. 1968. Structure of the New England Herring
Gull population. Ecology 49:644-676.
Kirkman, F. 1937. Bird behavior: a contribution based chiefly on a study of the Black-headed Gull. Nelson & Jack, London.
Kruuk, H. 1964. Predators vs. anti-predator behaviour of the Blackheaded Gull (Larus ridibundus). Behaviour Suppl. 11:1-129.
Kury, C. R. and M. Gochfeld. 1975. Human interference and gull predation in cormorant colonies. Biological Conservation 8:23-34.
McNicholl, M. K. 1975. Larid site tenacity and group adherence in relation to habitat. Auk 92:98-104.
Mineau, P. and D. U. Weseloh. 1981. Low-disturbance monitoring of Herring Gull reproductive success on the Great Lakes. Colonial Waterbirds 4:138-142.
Morris, R. D. and J. W. Chardine. 1985. The effects of ice cover over the colony site on reproductive activities of Herring Gulls. Canadian Journal of Zoology 63:607-611.
Mousseau, P. 1984. A comparison of 2 methods to assess the breeding success of Ring-billed Gulls (Larus delawarensis). Journal of Field Ornithology 55:151-159.
Nero, R. W. 1961. Dry land gull colony. Blue Jay 19:166-168.
Nisbet, I. C. T. 1975. Selective effects of predation in a tern colony. Condor 77:221-226.
Paludan, K. 1951. Contributions to the breeding biology of (L. argentatus) and (L. fuscus). Videnskabelige Meddelelser Fre Dansk Naturhistorisk Forening 114:1-128.


76
Paul, D. 1985. Current and historical breeding status of the California Gull in the Great Salt Lake Region. Report to the Los Angeles Department of Water and Power.
Parsons, J. 1971. Cannibalism in Herring Gulls. British Birds 64:528-537.
Patterson, I. J. 1965. Timing and spacing of broods in the Black-headed Gull. Ibis 107:433-459.
Pickett, S. T. A. and P. S. White, editors. 1985. The ecology of natural disturbance and patch dynamics. Academic, New York.
Pierotti, R. 1980. Spite and altruism in gulls. American Naturalist 115:290-300.
__________ 1981. Male and female parental roles in the Western Gull
under different environmental conditions. Auk 98:532-549.
__________and E. C. Murphy. 1987. Intergenerational conflicts in gulls.
Animal Behavior 35:435-444.
Pugesek, B. 1983. The relationship between parental age and reproductive effort in the California Gull. Behavioral Ecology and Sociobiology 13:161-171.
__________and K. Diem. 1983. A multivariate study of the relationship
of parental age to reproductive success on California Gulls.
Ecology 64:829-839.
Raper, E. L. 1976. Influence of the nesting habitat on the breeding success of California Gulls (Larus californicus), Bamforth Lake, Albany Co., WY. MS Thesis, Univ. Wyoming, Laramie.
Roberts, H. C. and C. J. Ralph. 1975. Effects of human disturbance on the breeding success of gulls. Condor 77:495-499.
Scharf, W. C. 1981. The significance of deteriorating man-made island habitats to Common Terns and Ring-billed Gulls in the St. Mary's River, Michigan. Colonial Waterbirds 4:155-159.
Shugart, G. W., M. A. Fitch and V. M. Shugart. 1981. Minimizing investigator disturbance in observational studies of colonial birds: access to blinds through tunnels. Wilson Bulletin 93:565-569.
Sokal, R. R., and F. J. Rohlf. 1981. Biometry. Freeman, San Fransisco.


77
Sousa, W. P. 1984. The role of disturbance in natural communities. Annual Review of Ecology and Systematics 15:353-91.
Southern, L. K. 1981. Sex-related differences in territorial aggression by Ring-billed Gulls. Auk 98:179-181.
__________, S. R. Patton and W. E. Southern. 1982. Nocturnal
predation on (Larus) gulls. Colonial Waterbirds 5:169-172.
__________and W. E. 1978. Absence of nocturnal predator defense
mechanisms in breeding gulls. Proceedings of the Colonial Waterbird Group 157-162.
Southern, W. E. 1977. Colony selection and colony site tenacity in Ringbilled Gulls at a stable colony. Auk 94:469-478.
__________, S. R. Patton, L. K. Southern and L. A. Hanners. 1985. Effects
of nine years of fox predation on two species of breeding gulls. Auk 102:827-833.
__________and L. K. Southern. 1981. Colony census results as
indicators of pre-hatching perturbations. Colonial Waterbirds 4:143-149.
__________and L. K. Southern. 1982a. Effect of habitat decimation on
Ring-billed Gull colony-site and nest site tenacity. Auk 99:328-331.
__________and L. K. Southern. 1982b. Intensification of adult Ringbilled Gull aggression during reproduction and it's possible consequences. Colonial Waterbirds 5:2-10.
__________and M. Southern. 1984. Herring Gulls specialize as Ringbilled Gull predators. Colonial Waterbirds 7:105-110.
Speich, S. M. 1986. Colonial waterbirds. Pages 387-405 in A. V.Cooperrider, et.al., editors. Inventory and Monitoring of Wildlife Habitat. U.S. Dept, of Interior, Bureau of Land Management, Denver.
Tinbergen, N. 1953. The Herring Gull's world. Colins, London.
Verbeek, N. A. M. 1982. Egg predation by Northwest Crows: its association with human and Bald Eagle activity. Auk 99:347-352.
Vermeer, K. 1963. The breeding ecology of the Glaucous-winged Gull on Mandarte Island, B.C. Occasional Papers British Columbia Provinicial Museum no.13.


78
__________. 1970. Breeding biology of California and Ring-billed Gulls: a
study of ecological adaptation to the inland habitat. Canadian Wildlife Service Report Series no. 12.
Ward, H. L. 1906. Why do Herring Gulls kill their young? Science 24:593-594.
Watanuki, Y. 1988. Intraspecific predation and chick survival:
comparison among colonies of Slaty-backed Gulls. Oikos 53:194-202.
Winkler, D. W. 1983. Ecological and behavioral determinants of clutch size: the California Gull (Larus californicus) in the Great Basin.
PhD. Diss., Univ. Calif. Berkeley, 195pp.
Young, E. C. 1963. The breeding behavior of the South Polar Skua (Catharacta maccormicki). Ibis 105:203-233.


53


Full Text

PAGE 1

EFFECTS OF DISTURBANCE ON A POPULATION OF CALIFORNIA GULLS by Charles Alexander Chase Ill B.S., Colorado State University, 1979 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Master of Arts Department of Biology 1991

PAGE 2

This thesis for the Master of Arts degree by Charles Alexander Chase Ill has been approved for the Department of Biology by Janis W. Driscoll Ronald A . Ryder Date 4., 10 I?V

PAGE 3

Chase, Charles Alexander Ill Effects of Disturbance on a Population of California Gulls Thesis directed by Associate Professor Diana F. Tomback Long-term effects of disturbances at a California Gull (Larus californicus) colony in Colorado were studied in the breeding seasons of 1981-1987. Types of disturbances included avian and mammalian predation, flooding and draining of the reservoir, and three levels of experimental human activities. Breeding adults were individually marked as chicks, and their reproductive efforts were noted each year. The major immediate effect of human disturbance was high levels of intraspecific aggression, including fighting, destruction of eggs and chicks, and occasional death of females . Long term consequences of disturbance included reduced clutch size, fewer nests and fewer older breeding birds. While some birds had no measurable response to disturbance, most birds responded to high levels of disturbance by relocating nest sites or by not returning to the reservoir in subsequent years. Relocation often had a long term effect on an individual's fitness. Relocating birds expended more energy finding and securing new breeding sites; their subsequent breeding was delayed and often clutch size was reduced. Low level human disturbance, flooding, and draining led to increased intra-island movement; predation by owls and high level human disturbance for one or more years resulted in increased inter-island movement. Multiple disturbances resulted in the lowest proportion of adults returning to the reservoir.

PAGE 4

iv Although the numbers of birds in a breeding colony did not always change after a disturbance, the demography of the colony was often dramatically affected. There were proportionately more young adults and recently established older adults in disturbed areas than would normally be expected. Investigators who gather reproductive data over one or more years without marked birds and without knowing the history of disturbance at a colony will misinterpret and misrepresent ecological data. The form and content of this abstract are approved. I recommend its publication. Signed ________________________________ __ __

PAGE 5

DEDICATION I dedicate this work to three men who have had a most significant effect on my professional development. To Dr. Ronald A. Ryder who taught me the value of thoroughness. To Stephen J. Bissell who taught me critical questioning. To Dr. Robert E. Carlton who provided so very many things but most especially, love. "Mono Lake is a hundred miles in a straight line from the oceanand between it and the ocean are one or two ranges of mountains-yet thousands of sea-gulls go there every season to lay their eggs and rear their young. One would as soon expect to find sea-gulls in Kansas ... " or in Colorado. M . Twain, 1872, Roughing It

PAGE 6

vi ACKNOWLEDGMENTS I wish to express my thanks for the help, support and guidance I received during this study. First, I wish to thank the Denver Museum of Natural History for logistical support from 1979 -1985 and the World Wildlife Museum in Monument, Colorado, for financial help from 1981-1983. Second, I wish to thank my major professor Dr. Diana Tomback for her support and prodding to complete this study. Third, I wish to thank the more than 100 volunteers that helped gather data and make observations from 1979 through 1987 . In particular I would like to thank the following people who helped each year on the fourth of July "Great Gull Round-up": David and Linda Blue, Walter Bruyninckx, Anna-Marie Blancquaert, Richard Bunn , Dan and Susanna Casey, Lisa Chase (Mason) , Fran Enright , Laurens Halsey , Kim Norgren , Nickie Pliler, Steve and Debbie Scheffey, Stephen Vaughan, and Mary Kay Waddington. And finally, I wish to thank Stephanie Haas, not only for all of her help in the field and for reviewing the many drafts of this work but for her continual support through many long and crazy years.

PAGE 7

CONTENTS CHAPTER I. INTRODUCTION .................................................................. 1 II. METHODS...... ............... ................. . ...... . ................... . ....... .... 6 Description of Study Area..... ..... ............ . . ..... ............. .... 6 Reproductive Biology ................. . . . .................................. 8 Procedure ......................................................................... 1 0 Human Disturbance Experiments ............ .... ..... ....... 12 Behavior of Gulls in Relation to Disturbance ......... 15 Analyses of Experimental Disturbances ...... ... ........ 16 Other Disturbances ......... ........................................... . 17 Ill. GENERAL PATIERNS ................... ............ . . . ......... . ........... . 19 IV HUMAN DISTURBANCE. .......................... ....... ...... ............ 28 Immediate Behavioral Response ................................ . 28 Interspecific Aggression ............................ ........... . .... 28 Intraspecific Aggression ........ ........................... . . ....... 30 Long-term Response ..... .... ....... . ....... . ................. ........... . 35 Uncontrolled Human Disturbance ................... . ..... ..... .46 V. OWL DISTURBANCE. ........... . ............................................ .48 Owl Predation ................................................................. .48 Immediate Response ......... . ............... . ...................... .48 Long-term Response . ..... .................................... . ...... 52

PAGE 8

viii VI. WATER LEVEL CHANGES ............................................... 58 Draining ........................................................................... 58 Immediate Response ............................................... 58 Long-term Response ...................................... . ........ 59 Flooding ........ . . ........................................ . ............... . ...... . 61 VII. CONCLUSIONS & SUMMARY ........... . ........................... 64 BIBLIOGRAPHY .......... . .............. ...... ................... . ............. . 71

PAGE 9

TABLES Table 1. Between year and island comparisons of reproductive performance of California Gulls at Antero Res ........ .......... . 20 2 . a). Gull behavioral responses to three levels of experimental human disturbance. b). Chi-square analysis of behavioral responses to three levels of experimental disturbance ........... .. .................. ........ . .............. 32 3 . a ) . One way ANOVA of number of young fledged per nest comparing differences between disturbance and post-disturbance periods. b). Kruskai-Wallis test comparing disturbance treatments and the percentage of birds returning to the same s i te in the following year ............................. .......... .................................... . 38 4 . Comparison of movement responses by breed ing gulls in the year following experimental disturbance treatments on Gull Island .................... ...... .............................. .44 5. Two by two contigency analysis of a). Breeding gulls movement response to disturbance treatments . b). Age-related response to disturbance treatment.. .......... 45 6. Summary of owl predation on California Gulls breeding on Gulli. and Goose 1 .......................................... .. .49 7. a). Comparison of California Gull movement patterns in the year following different intensities of owl predation in breeding plots . b). Two by two contingency tests of movement patterns and intensities of predation ...... ...... ............ .................... ...... ... ... ..... 55 8. a). Comparison between disturbance regimes and movement patterns of breeding gulls on Gull I. b). Rank comparisons of response to disturbances ...... ...... 65 9 . Comparisons of movement patterns of breeding gulls on Gull I. following experimental disturbances in the previous year .......................................... 66

PAGE 10

X 1 0. Chi-square tests for between disturbance comparisons in Table 8a .................................................................................. 68

PAGE 11

FIGURES Figure 1. California Gull breeding colonies, Antero Res., Colorado .......... 7 2. Location of experimental human disturbance plots on Gull 1 .... 14 3. Yearly comparisons between Gull I. and Goose I. for a). Total numbers of California Gull nests on each island. b). Average number of nests per 10 m2 plot ............................... 21 4. Comparison of California Gull nest densities on Gull I. and Goose 1 ............. ................... .............................. ... ....................... 22 5. Yearly variation in California Gull clutch size at Antero Res. a). Mean clutch size on Gull I. and Goose I. b). Mean clutch size and nest density on Gulll .......... .................. 24 6. Yearly variation of the percentage of 3-5 year old California Gulls breeding on Gull I. and Goose 1.. ....................... 25 7. Comparison between levels of disturbance and between years of the number of California Gull chicks fledging per nest. .............................................. ..................... 36 8. Comparison between levels of disturbance and between years of a). percentage of adults returning to the same plot in the following year. b). percentage of 3 5 year old breeders per plot in the following year ........ ............ 40 9. Comparison between level of disturbance in 1981-83 and a). percentage of birds that moved from the previous breeding plot. b) percentage of birds that changed islands or remained on the same island ................................ ...... .43 10 Major locations of chick mortality from Great Horned Owl predation on Gull 1.. ..................... ..................................... . . ....... 53

PAGE 12

11. Comparison between intensity of predation and a). percentage of birds that moved from the prev i ous breeding plot. b). percentage of birds that changed islands or remained on the same island ............ .......... ...... .......... 54 12 . Comparison of gull movement patterns influenced by Great Horned Owl predat ion. a). percentage of birds that experienced direct mortality or did not experience direct mortal ity. b). percentage of older versus younger breeding birds in different colony sites ........................................ . 56 13. Comparison of gull movement patterns influenced by dra i ning Antero Reservo i r and a). percentage of birds that moved from the previous breeding plot. b). percentage of birds that changed islands or remained on the same island ....... . ........... ........ . . . . ...... . ... ...... ......... . ..... ............... .......... . . . ......... 60 14. D i stance moved to new breeding s i tes in response to flooding ......................... ........... ............... . ..................... 63 xii

PAGE 13

CHAPTER I INTRODUCTION Disturbance of breeding colonial waterbirds has been a focus of numerous studies over the past twenty years (see reviews in Burger 1981, Speich 1986) . Interest in conservation management of colonial waterbirds is the driving force behind several recent works (Ellison and Cleary 1978, Anderson and Keith 1980, Hand 1980 , Desgranges and Reed 1981, Boellstorf m ru. 1988, Erwin 1989). Also , researchers are concerned that disturbances cause biases in research efforts and lead to misinterpretation of results (Duffy 1979, Ellison 1979) . Because of this problem, numerous papers "testing" various disturbance regimes against undisturbed "controls" have appeared (Roberts and Ralph 1975 , Ellison and Cleary 1978, Shugart m gl. 1981 I Fetterolf 1983 , Mousseau 1984) . The term disturbance has several defin i tions currently in use in ecology. Anderson and Keith (1980) and many others consider disturbance only in terms of human activities that result in perturbation of the "normal" condition of the colony under considerat i on. Sousa (1984) views disturbance as "a discrete, punctuated killing , displacement or damaging of one or more individuals (or colonies) that directly or indirectly creates an opportunity for new individuals (or colonies) to become established ." The latter definition is mostly concerned with the change in fitness of individuals in the face of disturbance factors , while the former is concerned with conservation and management of colonial

PAGE 14

2 waterbirds. Pickett and White (1985) define disturbance as "any relatively discrete event in time that disrupts ecosystem, community or population structure and changes resources, substrate availability or the physical environment." I will use this broadly encompassing statement as the working definition of disturbance in this study. Many studies limit disturbances to human initiated effects (Anderson and Keith 1980, Hand 1980, Fetterolf 1983 and many others), when in fact disturbances are often caused by predators (Hatch 1970, Conover and Miller 1978, Southern m .at. 1985) and natural environmental phenomena [Nero 1961 (dry lake), Scharf 1981 (erosion) and Morris and Chardine 1985 (late ice)]. It is clear that various types of disturbance are interrelated. Intrusion into a colony by humans or predators (Patterson 1965, Kury and Gochfeld 1975,) or disturbances following an environmental perturbation (Scharf 1981) can lead to increased interspecific predation as well as predation by other members of a colony on chicks and eggs. Effects of multiple disturbances are often compounding and the effect of a particular disturbance may be inseparable from other effects (Verbeek 1982, Desgranges and Reed 1981, Drapeau .ID .a[. 1984, Mousseau 1984). Disturbances affecting breeding gulls have been described in scientific literature since the early 1900s (Dutcher 1903, Ward 1906, Herrick 1909). Interspecific predation is common at colonies where more than one species of gull occur, especially Herring Gulls (Larus argentatus) and Great Black-backed Gulls (L. marinus) (Harris 1964 , Kruuk 1964, Erwin 1980). Often confused with cannibalism, intraspecific killing is the greatest post-disturbance effect in many studies (Parson 1971, Gillett .ID .at. 1975 , Davis and Dunn 1976, Southern and Southern

PAGE 15

3 1984). When the opportunity arises, neighbors will kill, though not eat, chicks in adjacent territories (Emlen 1956, Vermeer 1963 , Patterson 1965, Hand 1980). True cannibalism is not infrequent, especially in studies where there are food shortages or where non-breeding gulls are numerous (Kirkman 1937, Paludan 1951, Tinbergen 1953, Harris 1964, Parson 1971, Southern and Southern 1984, Watanuki 1988). Herring Gulls are most commonly noted as cannibals in the above studies . Both cannibalism and intraspecific killing may occur in the same colony, especially if other disturbances have occurred. Birds that have lost their own nests or chicks often perform most attacks on neighboring adults and their chicks (Davis and Dunn 1976, Hand 1980, Pierotti and Murphy 1987, but see also Watanuki 1988). Most studies only examine effects of disturbance on reproduction during the current season, e.g., Erwin (1989). A few multi-year studies noted whether or not an area had been abamdoned (Kadlec 1971, Buckley and Buckley 1972, Conover and Miller 1978) or if numbers or densities of breeding birds had changed (Southern and Southern 1981, Coulson .e.t al. 1982, DeWit and Spaans 1984) . However, long-term studies of the reactions of marked animals to disturbance are few. Southern and Southern (1981) felt that changes in clutch size in a particular sub-area were good measures of pre-hatching perturbations, while drastic changes in density could indicate disturbances from previous years. Fetterolf (1983) compared between year hatching success for study plots without knowing individual birds. He recognized the confounding effects of both changing age composition (determined by plumages) and nocturnal predation on hatching success comparisons. Bergman (1986) found high site tenacity in marked Common Gulls

PAGE 16

(Larus canus), except when eggs or small chicks were taken or when crows (Corvus spp.), intruded regularly. 4 High site fidelity is common among larids, especially those with stable breeding sites (Coulson and White 1958, McNichol! 1975 , Chabrzyk and Coulson 1976 , Southern and Southern 1981 ). Southern's (1977) colony of marked Ring-billed Gulls (Larus delawaren s is) showed high site fidelity between years. Interspecific predation was low in this population. Southern .ftl.al. (1985) presumed that high site fidelity also occurred in an unmarked disturbed colony where , i n sp i te of yearly fox predation , a group of breeders returned to the same site ove r a 9 year period. Conover and Miller (1978) and Cuthbert (1985a) found low site tenacity in disturbed and unstable habitats. Austin (1951) and Cuthbert (1985b) found group fidelity in mixed species colon ies of Common (St e r n a hirundo) and Casp i an Terns ( S . c a s p ia). Mixed species groups moved as groups to new breeding areas after disturbances. Since most studies of co l onial waterbirds focus on study plots within a colony instead of the ent ire colony, it is critical to know if a part i cular plot is typ i cal of the population (Coulson 1968, Southern and Southern 1981 ) . Without knowing the history and changes that have occurred at a colony , this may be impossible to determine. Unknown bias ing effects that have occurred in previous years (and may still be affecting a colony) can completely confound the interpretation of experimen tal studies ( Jehl and Chase 1987) . It i s neccessary to determ ine not just immed i ate behav i ora l and ecolog i cal responses to disturbance during the breeding season but also the long term consequences. What do individual b irds do i n reponse to disturbances of various kinds, intensities , and frequencies? Do birds

PAGE 17

5 have stronger reactions to one type of disturbance than another? Do they react more strongly if they have been disturbed several years in a row? What are the costs to a bird that changes breeding locations after a disturbance rather than returning to a disturbed site? Is there really such a phenomenon as an undisturbed research site? In this study, I examined changes that occurred within and between two colonies of marked California Gulls (Larus californicus) over a seven year period. Disturbance factors in this study include predation, water-level changes, and experiments of controlled human disturbance. The main thrust of this study was to determine changes in nest site selection and reproductive success among known birds in response to various natural and human disturbance factors .

PAGE 18

CHAPTER II METHODS Description of Study Area The study population of California Gulls (Larus californicus), located at Antero Reservoir (southern end of South Park, Park County , Colorado) has been in existence approximately 30 years. The nearest town is Hartsel, 8 km to the east. Fairplay is 32 km north, and the Chaffee County dump is 56 km south. The reservoir is owned and maintained by the Denver Water Board. At an elevation of 2770 m , this is the highest gull colony in North America. Historically, there have been four breeding colonies (islands) (Fig. 1 ) . One of the rock islands near the dam has eroded completely away and the other has had only 30-40 nests each year, most of which are disturbed by fishermen . Gull and Goose Islands are on the west end of the reservoir. Gull I. is located approximately 70 m from the shoreline; Goose I. is located 130 m northeast of Gull I. The greatest water depth between the mainland and Gull I. was 1.2 m, while the deepest point between Goose I. and the nearest mainland or Gull I. was 2m. Antero Reservoir has a diverse aquatic invertebrate population and supports a healthy fishery which is enhanced by annual stocking of rainbow trout. Access to the west end of the reservoir is legally restricted ; fishermen must remain in boats and are not allowed to land on shore or

PAGE 19

Antero Reservoir Fig. 1. California Gull breeding colonies, Antero Res., Color ado. A-Gull Island, B-Goose Island , C-Rock Islands. p?o c

PAGE 20

on any island. This provides breeding birds some degree of protection from human disturbance 8 Most of the vegetation on the islands and in the surrounding basin is alkali-adapted; the shoreline is sandy gravel with alkali grasses above the waterline. The west end of the reservoir is covered with marsh plants, primarily Scirpus, Juncus and Carex spp. Numerous grass and herbaceous species occur on both islands; however, the gulls keep most vegetation plucked close to the ground until mid-July when chicks begin fledging. Gull Island is approximately 0.37 ha in area with an elevation rise of 2.5 m from the shore to a central plateau. Approximately 55% of the island is covered with herbaceous vegetation, 3% is a stand of sandbar willow (Salix exigua) on the northeast corner of the island, and 12% is covered with the shrub winter fat (Ceratoides lanata) on the south and west top of the rise in the middle of the island. Thirty percent of the island (the south side) is bare soil and debris. Goose Island , approximately 2.8 ha in area, is a long narrow island with the gull breeding area located on 0.75 ha at the west end. It rises to 2 m on each end but is relatively flat throughout. The vegetation is primarily herbacecous with less than 2% covered in scattered winter fat. Similar to Gull 1., the only bare ground (1 0% of the gull colony) has developed on the southwest edge of the island. Reproductive Biology California Gulls returned to Antero Reservoir in early April, with the peak of arrival in the last week of April. Many of the earliest arriving birds

PAGE 21

9 were also the oldest and most experienced (Chase, personal observation). Territories were established upon arrival, although birds did not remain on the colony overnight until the ice on the reservoir broke up (usually by the end of April). The first eggs were typically laid during the last week of April with peak egg-laying by the end of the first week of May. While the normal clutch size was three eggs , young birds (3-5 years old) and disturbed birds typically laid only two (and rarely one). Clutches hatched in 24-26 days and chicks fledged 35-40 days later. Most of the young at Antero fledged by mid-July. Dates of clutch initiation varied by as much as two weeks, depending on when the ice broke up. Clutches initiated before ice breakup always failed because of predation by coyotes (Canis latrans). Territories were first established in the center of a colony with later nesting birds using more peripheral and shoreline areas . Open areas of ground with a low herbaceous cover were chosen first , areas with dense herbaceous cover second , and shrubby sites and bare scoured ground were avoided by most birds. Although both sexes incubated clutches and cared for the chicks, males were involved in most fights and chick killings. Females spent more time incubating from mid-morning until mid afternoon. By the time chicks were 20 days old, both parents often left to forage. Before then, at least one parent remained with or near the chicks. Adults usually abandoned chicks within two weeks after fledging, and within three weeks of fledging most adults and chicks left Antero and began migration to the Pacific coast. While a few California Gulls remained at Antero throughout the winter, band recoveries suggested that most (99+%) wintered along the Pacific coast. Immature birds remained on the coast until their fourth year

PAGE 22

10 (determined from band recoveries) when they returned to Antero to breed. A small percentage of third year birds and an even smaller percentage of second year birds returned to Antero. California Gulls do not reach full sexual maturity with fully developed gonads until the i r fourth year (Johnston 1956a and b); however, even second year birds can produce viable offspring. The above information on breeding biology at Antero was gathered as part of my studies on this population and will be presented in detail elsewhere . The breeding biology of this colony was similar to other populations of California Gulls with three-egg clutches (Behle 1958 , Vermeer 1970, Raper, 1976 , Pugesek 1983b). Procedure This study began on Gull I. in 1981 and Goose I. in 1983 and ended in 1987 . Each island was divided into 10 m x 10 m north-south oriented plots marked w i th 0 . 3 m wooden stakes. I entered each island f r om the same 1 0 m2 plots each year ; entry plots were excluded from all analyses. Minimizing disturbance of chicks at nest site was a primary concern . Almost every study of gulls has shown that high intraspecific and , at times , interspec ific predat ion occurs when colonies are disturbed and chicks run from natal territor i es (see review in Burger 1981 ). Results of an earlier experiment at th i s colony ind i cated that 10 m x 20 m pens caused the least d i sturbance and resulted in highest fledging success (Chase, unpublished data) . From 1981-1983, Gull I. was fenced into 10m x 20 m plots with 2.5 em x 2 . 5 em x 45 em chicken wire fence . These plots corresponded to the existing gridded plots . Fences were removed after 4 July, which corresponds to the beginning of fledging . Behle and Goates

PAGE 23

11 (1957), Behle (1958), Vermeer (1970) and Winkler (1983) found that California Gull chicks did not use water for drinking or cooling prior to fledging so there was little concern about the fencing restricting access to water . From 1984 to1987, Gull and Goose Is. were surrounded by a single fence only during chick measurements on 4 July. Average minimum temperatures were below oo C from incubation through the first two weeks of brood rearing. Snowfalls, with accumulations up to 0.2 m were not uncommon through mid-June . High diurnal temperatures of 21-300 C throughout June and July were regularly interrupted by daily violent thunderstorms . Temperature variations occasionally exceed 220 C per hour and 300 C per day. Winkler (1983) conducted chick measurements at night at Mono Lake , California to reduce the problem of chicks running from nesting territories and being killed by other adults. At Antero, chicks <15 days old became hypothermic within 5-10 minutes of a researcher entering the co l ony at night and chicks older than 15 days ran when disturbed, regardless of day or night. From these earlier observations, I decided to limit research activities to daytime hours and to periods of moderate weather. All movements by researchers in this study were slow and methodical; activities were as short in duration as possible . Researchers moved bent over or in a "crab walk". Once work started in a plot , workers remained in that plot until finished and did not return to that plot again until the next scheduled period. Measurements of eggs and chicks were performed by myself or an assistant. To compute error estimates I . repeated every twentieth measurement. Overall reproductive patterns were determined from data compiled by plot, island, and year and include : number of nests, average density,

PAGE 24

12 clutch size, number of chicks fledging per nest, and percentage of eggs resulting in fledged chicks per nest. Nest numbers, density, and clutch size for all years are based on counts from 25 May. This was the average date of first hatching and 1 to 1.5 weeks prior to peak hatching. Fledging success is based on 4 July counts (average chick age 30 days) minus dead chicks found during final mortality counts approximately 2 weeks later. Dead chicks were counted and collected each visit. Necropsies were performed on all dead chicks to ascertain cause of death. All adults in each plot were observed from their arrival in the spring through chick fledging. Observations were made with a 15-60 power spotting scope. Band numbers and colors, plumage characteristics, sex, and nest site were noted. Initially, between 20-50% of breeding adults in each plot had been banded as chicks in earlier studies. Another 20-35% percent of adults could be aged as 3-5 year old birds by plumage variations. Thus, 40-85% of breeding birds could be aged in any particular plot. All banded adults had U.S. Fish and Wildlife Service aluminum bands. Twenty-five adults were marked with colored leg bands in 1981. All chicks fledging at Antero from 1981-1987 were marked with aluminum bands on the left leg, and chicks from 1981-1984 were also marked with colored leg bands on the right leg. Measurements of chicks included weight (300g or 1 OOOg Pesola scales) and lengths of tarsus, culmen, 9th primary, and outer rectrix. Human Disturbance Experiment This experiment was run in 1981-1983; 1984-1987 were considered post-disturbance (though only in reference to this experiment). The levels of disturbance (high, medium, and low) were

PAGE 25

13 measures of the frequency of visits to a plot. Twelve pairs of plots on Gull I. were randomly chosen and designated as high, medium or low disturbance plots. Block pairs, instead of single plots, were established so that access could occur from the island perimeter without having to cross other plots (Fig. 2). All plots on Goose I. and all other plots on Gull I. were designated as low disturbance areas. These plots are the "controls" in this experiment. It is critical to note that these are least disturbed areas, not undisturbed plots, as is commonly misstated in the literature. All plots on both islands were visited each year on the weekends closest to 25 May and 4 July. During these visits, I used a crew of 15-20 assistants to minimize the time of disturbance. During the early visit, no more than three people were in a single plot at a time and then for less than 30 minutes per plot. Eggs were measured and nests mapped. The 4 July visits involved groups of 5-6 assistants in each penned area on both Gull and Goose Is. During 1981-1983 each group measured and banded all chicks within the experimental plots within 20 minutes . Since 1984, each island was surrounded by one fence and work on each island was completed in approximately 3 hours. All fences were removed at the end of this event. I checked chick mortality within three days of the large banding effort. A final visit was made in the third week of July, when over 95% of all birds had fledged. Final mortality counts were made at this time. Low disturbance plots were only visited on the two weekends mentioned above . The two-visit method is similar to that of Kadlec and Drury (1968), Mineau and Weseloh (1981) and Mousseau (1984), except that the chicks in my study areas were closer to fledging when handled (majority within 10 days of fledging versus 15-20 days) .

PAGE 26

1 G F E D c B A 2 3 4 5 6 7 8 9 Fig. 2. Location of human disturbance plots on Gulli. L-Low level disturbance, M-Medium, H-High. Plots without patterns experienced additional, nonexperimental disturbances and were excluded from the human disturbance experiments. 10 11

PAGE 27

15 High disturbance plots were entered every 1-2 days for at least 5 minutes and up to 30 minutes per visit from the beginning of egg laying through 4 July. Medium disturbance plots were visited at 4-5 day intervals for similar periods of time. During visits to the high and medium plots, nest counts, mapping and egg measurements were performed by no more than two people at a time. Chicks were marked at hatching and measured every 1-2 days or 4-5 days depending on the plot disturbance designation. When most chicks were less than 1 0 days old, they were handled on or near nests. After they were 15 days old, all chicks in a pen were gathered together in one corner of the pen and contained in a smaller 2 m pen. After each chick was measured, it was released into the main pen . Behavior of Gulls in Relation to Disturbance Over 400 hours of gull behavioral observations were gathered on Gull I. from 1981-1983. Behavioral observations took place from a blind (1 m x 1.5 m x 1.5m) erected on the top of the hill or from open observation points on the perimeter of the colony. Behaviors were recorded 30 minutes before, during, and 30 minutes after disturbances. Five nests or family groups within the same plot were observed each period. Observations were conducted only on nests at least two plots (20+ m) away from the observers' position. New subjects in different plots were selected for the next observatioon period. Observations were conducted for sample periods throughout the day . Data collected included size of the brood, parental activities, especially feeding , brooding, and intraspecific aggression, as in Conover and Miller (1980). Adoptions of moving chicks by other adults were recorded and

PAGE 28

16 continuously monitored even if the chick left the original 5 broods under study. I recorded attack rates by adult gulls on myself or an assistant as we moved through plots performing measurements. The recording observer was seated at least 30 m away from an interaction to minimize additional disturbance. An attack consisted of a definite swoop ending within 3 m of the human intruder. The sex, age, and identity of each attacker were recorded when possible. Analyses of Experimental Disturbances Mortality was recorded separately for eggs, chicks, and adults. Mortality and numbers fledging per plot and disturbance type were compared between years , and islands and were compared with the subsequent breeding distribution of returning banded adults in the years following disturbances. Since California Gulls can be aged by plumage until their fifth year , any nesting young adults can be aged visually. The proportion of 3-5 year old adults within plots was compared among disturbance regimes. A one-way analysis of variance was performed on the number fledging per nest, comparing differences between human disturbance regimes during experimental disturbance years (1981-1983) and during non-disturbance years (1984-1987). A Kruskai-Wallis test was used to compare among disturbance regimes and between disturbance and post-disturbance periods for percentage of breeding birds returning in subsequent years. A Bonferoni correction was applied to adjust for repeated samples . A Dunn's multiple comparison test was used to compare among disturbance regimes for the percentage of young (3-5 year old) breeders in each plot in the subsequent year. Two by two

PAGE 29

17 contingency analyses were used to compare disturbance with movement patterns, i.e., whether a bird moved from its previous plot in subsequent years, and if it moved, whether it remained on the same island or changed islands. Two by two contingency analyses were also used to compare movement patterns among birds of different ages and disturbance treatments. Pearson correlation analyses compared: 1) previous breeding success with numbers returning in subsequent years, 2) adults returning to the same plot with numbers of 3-5 year breeders, and 3) nest density with fledging success . Student's t tests compared: 1) clutch size variation in 8 plots where disturbances had occurred in previous years , 2) clutch size variation between islands, and 3) frequency of attacks by adults on researchers in low and high disturbance plots. Non-parametric tests are as in Hollander and Wolfe (1973). All other tests are as in Sokal and Rohlf (1981 ). Other Disturbances Owl predation was recorded by plot , date, number, and age of birds killed from 1981-1987. Adult and chick response to owls was recorded during 5 nights that attacks were observed. Gull responses to annual water level variation were recorded in the same manner as low disturbance human experiments, i.e., nest numbers , clutch size, etc., except that after Antero was drained, I entered the colony to count nests and eggs on 5 occasions instead of 2. Movement patterns of marked birds disturbed by owls and draining were recorded in the same manner as in the human disturbance experiments. The actual distance gulls moved in subsequent years within the colony was recorded for birds

PAGE 30

18 affected by flooding of the periphery of the islands. Two by two contingency analyses were used to compare movement patterns for owl predation, flooding and draining.

PAGE 31

CHAPTER Ill OVERALL REPRODUCTIVE PATIERNS Overall reproductive efforts are noted by year and island in Table 1. Prior to 1983, few California Gulls nested on Goose I. There was a pattern of decreasing numbers of nests on Gull I. corresponding with increasing numbers on Goose I (Fig. 3a), with the exception of 1985. During 1985, a large influx of unmarked adults appeared and nested at Antero. This influx was attributed to gulls leaving Great Salt Lake when nesting colonies flooded. Record water levels flooded many colonies that normally survive high water periods at Great Salt Lake (Paul 1985). Colonies in Wyoming and Montana also experienced similar influxes of California Gulls (Finholdt, personal communication). Nest numbers were converted to two types of densities (Fig. 3b and 4). Densities per hectare controlled for differences in total area between the two islands and between years (Fig. 4), while the average numbers of nests per 10 m2 plot for all plots used in a given year were compared in Figure 3b. In both cases, it was clear that there was a general decline in densities on Gull I. and an increase in densities on Goose I. Davis and Dunn (1976) and Coulson .al. (1982) found that increased density led to increased intraspecific chick killing and decreased reproductive success while Watanuki (1988) found the opposite result. Increased density was TABLE 1. Between year and island comparisons of reproductive

PAGE 32

20 TABLE 1 . Between year and island comparisons of reproductive performance of California Gulls at Antero Res. All birds are lumped within islands regardless of disturbance factors. 1981 1982 1983 1984 1985 1986 1987 Gulli. nests 546 545 429 371 476 287 305 avg. clutch 2.42 2.49 2.4 2.0 2.31 1.67 2.38 %eggs 35.9 33.2 50.5 66.9 67.5 66.0 fledged avg. 0.86 0.83 1.21 1.33 1.56 1.6 fledge/nest Goose I. nests 50 49 144 498 694 617 681 avg. clutch 2.18 2.26 2.25 2.36 1.99 2.38 %eggs 49. 2 66.8 70.2 75 . 1 fledged avg. 1.11 1.5 1.65 1.8 fledge/nest

PAGE 33

a) 800 600 f/) f/) G) c 400 0 0 z 200 b) 15 0 14 c. ... G) 13 c. f/) -12 f/) G) c 0 11 c G) 10 C) ctl ... G) 9 > < 8 --oGull • Goose 1981 1982 1983 1984 1985 1986 1987 Year --o• 1981 1982 1983 1984 1985 1986 1987 Year 21 Gull Goose Fig. 3 . Yearly comparisons between Gulli. and Goose I. of a). Total numbers of California Gull nests on each island. b). Average number of nests per 1 0 m2 plot.

PAGE 34

2000 G) ... < 1500 (.) G) .c::: ... G) a. tn 1000 tn G) c -0 >-500 = tn c G) c 0 1981 1982 • Gull Goose 1983 1984 Year 22 1985 1986 1987 Fig. 4 . Yearly comparisons of California Gull nest densities on Gulli. and Goose I.

PAGE 35

23 strongly correlated with high fledging success (r=0.96 , df=11, p<0.001) at Antero Res. The average clutch size for least disturbed areas of the total population was 2.4 for 1981-1987. Comparing average annual clutch sizes on Gulli. and Goose I. (Fig. Sa), there were significant differences from the mean only in 1984 and 1986 (t test: p<0.0001 for Gull 1., 1984-86 and Goose I. 1986 , p<0.05 for Goose 1., 1984). In 1984, fishermen destroyed many eggs during egg laying, which, coupled with an increase in young breeding birds (3-5 yrs. old) (Fig. 6), resulted in a lower average clutch size of 2.0 on Gull I. In 1986, the reservoir was drained to repair the dam. By 25 May, many nests and eggs had been destroyed (see Chapter VI for more details) and clutch sizes and densities were lower than previous years on both islands in 1986 (Fig . 5b). During earlier visits (5 and 11 May) to both islands, the numbers of nests nearly equaled 1985 levels, although clutch sizes were low throughout the entire season. I attribute the general trend of declining numbers, densities, and, to some degree, clutch size to the cumulative effects of multiple disturbances to Gull I. over time (Fig. 3 and 5b), including human disturbance, both uncontrolled and experimental (chap . IV), owl predation (chap V), and water level changes (chap . VI). Few birds nested on Goose I. until 1983. The increase on Goose I. was caused in part by movement from Gull I. The increasing percentage of young adults breeding on Gull I. was caused by the decreasing number of older birds returning to Gull I. With fewer adults, breeding territories were available to younger birds. Numerous young gulls without nest sites were observed in 19811983. Even though Goose I. was open and apparently had suitable habitat, few gulls attempted to nest prior to 1983 (25-50 each year) and then only on

PAGE 36

24 2 . 6 a) 2.4 C1) (/) 2 . 2 ..c 0 ::J 0 2.0 C1) C) C'O --o-Gull ... C1) 1.8 > • Goose < 1.6 1981 1982 1983 1984 1985 1986 1987 Year b) 2 . 4 1400 C1) 2.2 1200 N (/) ..c 0 2 . 0 1000 ::J 0 C1) C) C'O 1 . 8 800 ... C1) > --a-Clutch size < • Density 600 1981 1982 1983 1984 1985 1986 1987 Year Fig . 5 . Yearly comparisons of Californ i a Gull clutch size at Antero Res. a). Average clutch s ize on Gull I. and Goose I. b). Average clutch size and nest density on Gull I. C1) ... C'O 0 C1) ..c ... C1) a. >. ;::: (/) c:: C1) c

PAGE 37

U) ... Cl) 40 '0 30 Cl) Cl) Cl) 0) ... ns.o C'O Bo ... Cl)..: D. >-20 10 I M 25 --o--Gull • Goose 1981 1982 1983 1984 1985 1986 1987 Year Fig. 6. Yearly comparisons of the percentage of 3-5 year old California Gulls breeding on Gull I. and Goose I.

PAGE 38

26 the western edge, nearest to Gull I. All gulls observed breeding on Goose I. before 1983 were older adults, and in fact many were known birds from Gulli. Young adults did not attempt to colonize Goose I. They either established territories on the periphery of Gull I. or did not breed at Antero . Only after greater numbers of older adults started nesting on Goose I. in 1985, did young adults start breeding there. Coulson (1968) and Coulson and Thomas (1985) found similar results with Kittiwakes (Rissa tridactyla) . Because of disturbances on Goose I. (owl predation , draining of Antero, and low level researcher monitor i ng), by 1986 some older gulls left Antero altogether. The proportion of young adults slowly increased on Goose 1., which may be attributed to continuous disturbances to the entire population at Antero . Older gulls lay larger clutches and nest in greater densities than young breeders (Pugesek 1983, Tinbergen 1953, Coulson and Porter 1985) , and this was also true at Antero. However, both clutch size and nest density decreased on Gull I. in 1984 and 1986 due to disturbances resulting in egg destruction and nest abandonment (Fig . 5b). While there was a large increase in the number of young breeding adults in 1984, there was no similar increase in 1986, though the frequency of young breeding adults remained high (Fig. 6). Without knowing the entire, ongoing history of a colony, interpretation of reproductive variables may be highly biased. Changes in age structure and movement of individual birds result in altered colony wide reproductive performance. These changes are much more significant than simple reductions in fledging success as measured in most studies of breeding season disturbances.

PAGE 39

27 Southern and Southern (1981) compared densities and clutch size changes between years in sub-areas (plots) of a Ring-billed Gull colony and concluded that disturbances in previous years could be detected by density changes greater than 10% . Additionally, changes in average clutch size and clutch size variance could reflect early season perturbations. At Antero, density decreased more than 1 0% in only 2 of 8 plots where predation had occurred the previous year; density actually increased in 1 of the 8 plots. Clutch size decreased in 6 of 8 plots, partly because 3-5 yr. old (naive) birds moved into plots after older adults failed to return. Clutch sizes were significantly different (t test: p<0.05) among 3 of the 8 plots . There were early season perturbations in only one of those plots. The Southerns' variation technique cannot predict disturbance variations for California Gulls at Antero Reservoir. This was especially true of Gulli., with its history of disturbances. Plots on Goose I. also changed significantly over time, but this change was caused by continuing growth of this colony. A decline in density or clutch size might be interpreted as an indicator of disturbance; however, it is clear that several years of monitoring are required before interpretation of any reproductive measures of success can be used in a predictive manner. If a colony has a history of disturbance, especially different types of disturbances , the predictive power of any single measure of ascertaining disturbance is greatly weakened .

PAGE 40

CHAPTER IV HUMAN DISTURBANCE Immediate Behavioral Responses to Human Disturbance Interspecific Aggression Only responses to human intruders are considered here. Adults from this colony also responded aggressively to coyotes (Chapter VI), in panic to owls (Chapter V), and aggressively to Golden Eagles (Aquila chrysaetos) and Common Ravens (Corvus corax), whenever they approached the breeding colonies (Jehl and Chase 1987). Other species of gulls are rare at this location during the breeding season and no observations of interactions were recorded. During egg-laying and incubation, attacks against the researchers occurred only while the researchers were moving or first sitting down to perform measurements. These attacks consisted of swoops with loud calls and occasional bill strikes to the researchers' head. Adults returned to nests within 2 m of a sitting researcher, especially during the last half of incubation. Similar results have been reported in other studies (Dutcher and Baily 1903, and Conover and Miller 1978). When chicks were old enough to run together in groups (approx. 15 days), they were captured in these groups and penned for measurement. After release, chicks would immediately run several meters and join other released chicks. Normally,

PAGE 41

29 family groups reunited on territories within 5 minutes of chick release. Adults attacked the human intruders throughout the period that chicks were held captive. Once the last chick was released most adults settled on their territories within a couple of minutes. From the beginning of egg-laying through the beginning of chick rearing, aggression toward human intruders escalated. Frequency of attacks during egg-laying and the first two weeks of incubation averaged less than 0.2/min. (n=50). During the third week, attacks averaged 0.5/min. (n=22), and during the last week of incubation and the first week of hatching, attacks averaged 2/min. (n=49). A high level of aggression (1-2 attacks/min.) was maintained until 30 days post-hatching (n=51 ). From that time until fledging (35-40 days), attacks diminished to 0.3/min. (n=38). There was no significant difference in frequency of attacks between low and high frequency disturbance plots (t test: p>0.05). Burger (1981) and Dulude .e.t .a!. (1987) found similar variation in aggressive behavior over the breeding season among Herring Gulls. All birds did not attack with the same intensity and frequency. Typically, older males (>4 years) attacked more frequently (78% of all attacks where the bird could be aged and sexed, n=90 of 116) and more aggressively, often striking the human intruder repeatedly on the head. Females and younger males rarely struck the intruder (<2% of all attacks), limiting their attacks to shallow swoops and loud calling over the nest area. Certain individuals were much more aggressive than others, including one female that maintained a high attack rate from early incubation through the end of fledging. Additionally, I was identified by some birds and attacked at a greater frequency (3:1, n=32) than unknown intruders. Dutcher and Baily (1903) describe accounts of gulls

PAGE 42

30 recognizing individual humans and behaving differently toward them than strangers. In their account, gulls attacked known individuals at a lower rate, contrary to findings in this study. My data suggest that at least some gulls will attack a known and regular intruder at a higher rate, possibly recognizing an intruder that doesn't "attack back." Intraspecific Aggression Many studies involving gulls, especially studies dealing with human disturbance, have found intraspecific aggression to be a major cause of egg and chick mortality (Hand 1980, Robert and Ralph 1975). Unless disturbed, at least one adult always attended a nest during incubation. In this study, clutches left unattended were frequently destroyed by neighbors and eggs were often eaten by neighbors if the nest was left untended for several minutes after a disturbance . Of 172 observed cases of egg destruction, 77% (133) were caused by neighbors and 23% (39) by immature, non-breeding marauders. Of 53 observations of eggs being eaten, 39% (21) involved an immature, non-breeding marauder, 42% (22) involved neighbors, and 19% (1 0) involved the male territory holder eating his own eggs . In all observed cases where the owner ate his own eggs, the eggs had been broken by an intruder, and the male ate the egg contents upon returning to his nest. Similarly, Chardine and Morris (1983) reported two males eating eggs after a presumed disturbance. Kirkman (1937) attributed cannibalism (eating eggs and chicks) to "unmated rogues," and Tinbergen (1953) attributed cannibalism to unmated 3-4 year olds. However, Parsons (1971) found that mated birds performed most of the attacks. Davis and Dunn (1976) found that most intra-specific mortality was caused by neighbors that had

PAGE 43

31 lost their own eggs or chicks . They felt that the attacks on eggs and chicks were the result of repeated intrusions into neighbors' territories and were not simply opportunistic killings. At Antero , both young unmated marauders and neighbors destroyed large numbers of eggs and chi cks . All observed intraspecific egg and chick killings were related to disturbance by humans. Chicks that wandered into adjacent territories or were untended by parents were attacked by neighbors. Young (1963) , Hunt and Hunt (1976) , and Watanuki (1988) also found that unguarded chicks were subject to attack . Fetterolf (1983) reported more frequent fighting among adults , more attacks on chicks , and more chicks running during disturbances than before or after disturbances or in undisturbed conditions. From 1981-1983 , I observed post-hatching behavioral responses of adults and chicks to disturbances for over 400 hours . As Fetterolf found, adults in the least disturbed plots rarely attacked chicks, chicks rarely ran, and adult fighting was less common (11 times) than in highly disturbed areas (Table 2a) . In high l y disturbed plots, adu l ts attacked chicks 23 times more frequently than in least d i sturbed plots. The differences among the disturbance treatments for each of these behaviors were highly statistically significant (Table 2b) . Adoptions normally occurred when a chick left its parents' territory and successfuly joined another brood , sometimes permanently . Chicks rarely left their parents' territory if a parent was present. Only 4 adoptions were observed in the least disturbed areas , while 57 adoptions occurred in the highly disturbed plots . All observed adoptions occurred during and immediately after disturbances . One chick changed broods 7 times

PAGE 44

TABLE 2. (a) Gull behavioral responses to three levels of experimental human (researcher) disturbance. Responses were recorded for 30 minutes before and 30 minutes after each disturbance. Four hundred hours of observations were recorded from 1981-1983. (b) Chi-square analysis of behavioral responses to three levels of experimental disturbance, df=2. Only post-disturbance periods were analyzed . (a) Behavior, Low no. per hour Before After Adult fights 0 2 with adults Adu It attacks 0 2 on chicks Chicks running 0 6 outside terr. No. adoptions 0 4 (b) Behavior Adult fights with adults Adult attacks on chicks Chicks running outside terr. No . adoptions n 39 64 34 90 Medium Before After 0.4 15 1 16 0 12 0 29 284.5 1066.0 22.7 858. 5 High Before After 1 22 4 46 2 16 0 57 p < 0.0001 < 0.0001 < 0.0001 <0.0001 32

PAGE 45

33 before fledging, and 4 chicks changed broods 3 times each. Short-term (1 hour or less) broods of 7, 8 , and 9 were observed in one plot i n 1982. It appeared that if a chick, regardless of age, could mix with a brood without a resident parent attacking it, the chick was accepted into the brood. If an adult observed a chick moving into its territory, it usually attacked immediately and vigorously drove it off, often killing young chicks. In contrast , Fetterolf (1983) found that few chicks over 10 days of age were adopted. In the highly disturbed areas on Gull 1., 48% of adoptions were successful. It may be that the high frequency of disturbance caused parents to be more cautious about attacking their own broods and may have allowed a greater than normal number of successful adoptions . Alternatively, there may have been so much fighting that when the dust settled, adults simply accepted any chicks mixed with their brood that didn't try to run away . Parsons (1971) observed a male eating chicks but could not ascertain if that was a general occurence. At Antero, most chick attacks (80%, n=146 of 183) were performed by males. Other studies also found that males were the most aggressive sex (Pierotti 1981, Southern 1981, Southern and Southern 1982b, Dulude .e..t .aL. 1987 , Watanuki 1988). Chicks less than 10 days old suffered the highest mortality per attack . An adult would pick-up a small chick and shake it and/or hammer it against the ground. Fewer than 10% of young chicks observed being killed (n=8 of 87) were subsequently eaten by the attacker. No chick older than 1 0 days was ever eaten after being killed by a neighbor (n=117). Older chicks were killed by repeated blows to the head. They were able to run from attacks and were more likely to survive repeated attacks than

PAGE 46

younger chicks. After 30 days of age , chicks would often strike back when an adult attacked and rarely sustained serious injury. 34 The only gulls observed to eat chicks regularly were immature, non breeding marauders. They regularly preyed upon eggs and young chicks left exposed for several minutes. Cannibalism has been described in many gull studies, but only Herring, Slaty-backed (Larus schistisagus) and Western Gulls (Larus occidentalis) regularly eat chicks (Drost 1958, Brown 1967, Parsons 1971, Hunt 1972 and Watanuki 1988). However, even among these species, there are colonies where chicks were killed but not regularly eaten (Kadlec 1969). The majority of the killing of chicks at Antero appeared to be caused by the highly aggressive nature of California Gulls and not hunger (Paludan 1951 ) , infanticide to reduce local competition, (Harris 1964),or spite (Pierotti 1980 and Pierrotti and Murphy 1987). Most studies describe increases in both egg and chick mortalities with increasing disturbance (Hunt 1972 , Hand 1980, Mousseau 1984, but see Gillett mal. 1975). Roberts and Ralph(1975) described diminished intraspecific mortality in their most frequently disturbed area and attributed it to habituation. They observed fewer attacks in high disturbance plots as the season progressed. At Antero, there was a similar reduction in attacks, but it was a consequence of adults leaving territories after losing their clutches or broods. The frequency of attacks and chick mortality was always highest in the most disturbed areas . Approximately 50% of all losses caused by intraspecific aggression occurred in the egg stage with the other 50% happening during the chick stage, regardless of the frequency of disturbance.

PAGE 47

35 Egg losses included only eggs that were broken or removed from the nest. The few clutches that failed due to thermal stress had been abandoned early in incubation. Diurnal temperatures never exceeded 300 C, well below temperatures described by Chappell m .a!.. (1984) as causing heat stress to eggs in this species. Additionally, as adults were never away from nests for more than one hour (usually less than 5 minutes), it is not likely that there were significant losses due to thermally induced embryo mortality associated with human disturbance. Three to five adult females were killed each year (1981-1983), apparently during territorial fights. There were significantly more fights in highly disturbed plots than in low disturbance plots (Table 2b). Only two of 13 females killed were in low disturbance plots. Female killings were observed at Mono Lake when male-male fights intruded on nest sites and drew females into the fight (Jehl, personal communication). The pattern of dead females at Antero suggests a similar scenario. From 1984-1987 (post-disturbance), only 7 females were found dead on Gull 1.; however, the dead birds were located over the entire island unlike the locally concentrated killing in 1981-83. Long-term Responses To Human Disturbance The data from 12 of the 24 plots were confounded by additional factors, particularly owl predation (see Fig . 2). Only plots without additional disturbances were used in the following analysis of human disturbance. There was a significant reduction in the number of chicks fledged per nest as the level of disturbance increased (Fig. 7), (One-way ANOVA

PAGE 48

36 2.5 End of Disturbance 2 . 0 tn C1) 1 . 5 tnl: (.)a.. -(1) "fie. "(1) 1 . 0 00) z"O C1) ---o--Low 6 Medium 0 . 5 • High n=65 plots 1981 1982 1983 1984 1985 Year Fig. 7. Comparison between levels of disturbance and between years of the number of California Gull chicks fledging per nest

PAGE 49

37 n=728, F=83.4, df=2, p<0.0001 ). After experimental disturbances stopped (1984-1985), the numbers of chicks fledged per nest were not significantly different from that of the least disturbed population mean (1.6 chicks fledge per nest) for any treatment area or year (Table 3a) (One way ANOVA n=363, F=1.03, df=2, p= ns). There were no significant differences in the numbers of chicks fledging per nest within each disturbance regime from year to year during either the disturbance (19811983) or the post-disturbance period (1984-1985). The slightly lower fledging success in 1984 for the high disturbance plots than in the low and medium plots (Fig. 7), could be explained by the higher percentage of young breeders that were then breeding in these plots. Overall fledging successes were similar to many other studies comparing varying levels of disturbance (Shugart .e..t .al.1981, Fetterolf 1983, Mousseau 1984). Mousseau (1984) found that different disturbance levels affected chick mortality rates but not egg mortality. In this study, both eggs and chicks were strongly affected by the changing levels of disturbance. Roberts and Ralph (1975) found the lowest mortality in least disturbed areas, but their most disturbed areas also experienced relatively high fledging success. They attributed this to habituation: chicks ran shorter distances and adults attacked less frequently. Chicks at Antero ran more frequently in highly disturbed plots than in low plots, and parental aggression was higher in high disturbance plots than in low plots (see Table 2). Increasing disturbance resulted in decreasing reproductive success. This finding is where most other disturbance studies stop. Colonial waterbirds differ in post-disturbance responses in site fidelity. Species such as beach-nesting terns (McNichol! 1975), Ring-

PAGE 50

38 TABLE 3. (a) One-way ANOVA of number of young fledged per nest. Sheffe's F was used to test differences between disturbance and post disturbance periods for each disturbance regime. Disturbances occurred in 1981-1983. (b) Kruskai-Wallis test comparing disturbance treatments each year (Low, Med, High) for the affects on percentage of breeders returning to same site in the following year, df=2. A Bonferoni correction was applied to correct for repeated samples. (a) Year 81-83 84-85 Disturbance Post-disturbance Treatment n X SE n X SE F p Low 264 1.68 0.079 125 1.65 0.091 0.038 NS Med 240 1 . 2 0.037 133 1 .66 0.053 54.56 <0 . 05 High 224 0.75 0.042 105 1.50 0.1 04 58.55 <0.05 (b) percentage adults returning n=13 plots per year Year H p 82 5.90 0.17 83 8.61 <0.01 84 10.71 <0 .001 85 3.15 0.87

PAGE 51

39 billed Gulls (Conover and Miller 1978, Southern 1977, Southern and Southern 1981 ), and Black-legged Kittiwakes (Coulson and White 1958, Coulson and Thomas 1985) return to the same areas if undisturbed, but entire colonies often abandon disturbed areas. Bergman (1986) found that 38 of 40 successful pairs of solitary nesting Common Gull returned to the same site in the following year, while only 14 of 52 pairs that had lost chicks returned. In contrast, Kruuk (1964) records continually disturbed Black-headed Gull (Larus ridibundus) colonies that have been in the same site for over 300 years. California Gulls apparently fall between these two extremes. There was a strong correlation between previous success (fledging) and percentage of adults returning to the same sites in the following year (r=0 .86, df=23, p< 0.0001 ). Additionally, there was a continued decline in returning adults in the medium and high disturbance plots (Fig 8a). The percentage of birds returning in subsequent years to the same 10 m2 plot was significantly different between disturbance treatments in 2 of 3 years following disturbances (Table 3b). Each year following disturbances (1982-1984), the differences became larger and more significant. However, in 1985, two years after the experimental disturbances ceased , the percentage of adults returning to all plots approached the least disturbed population mean of 69%. The high disturbance plots still experienced somewhat lower returns, but again, this is explained by the higher proportion of young adults breeding there. A lower proportion of young adults returned (55%) than older adults (72%). This rate of return is lower than Southern .e..t ru. (1985) reported for Ring-billed Gulls (90.6%, n=48). I do not believe this is explained by species differences but is due to old bands (>8 years) that were lost or

PAGE 52

40 a) 80 End of Disturbance 70 a c 60 c ... G) ::l a-coeD 50 _ ... tn u= ... ::l 40 Q)"' D..ca --o--Low * Medium • High 30 0 n=52 plots 20 J81 1982 1983 1984 1985 Year 50 End of Disturbance b) tn 40 ... G) "0 G) G) 30 G) ... a.C c-G) 0 (.)._ 20 ... ca G) G) D.>--o--Low 10 10 * Medium I • High M n=65 plots 0 1981 1982 1983 1984 1985 Year Fig. 8. Comparison between levels of disturbance and between years of a). Percentage of adults returning to the same plot in the following year. b). Percentage of 3-5 year old breeders per plot in the following year.

PAGE 53

that could not be read. Five and 6 year old gulls returned to least disturbed sites at an annual rate of return of 88%. 41 There was a significant increase in the frequency of 3-5 year old birds among the three disturbance regimes ( Dunn's multiple comparison test: p< 0.05 for low versus high 1982-1985, p<0.05 for medium versus high 1982-1983); however, this increase was almost completely attributable to increases of 3-5 year old breeders in the high disturbance plots (Fig. 4b) . It was expected that older, more experienced birds would be less likely to move to an area that had been previously disturbed and that more naive birds would replace disturbed birds. When adults moved from the low and medium plots, they were replaced by other adults from other disturbed areas. The percentage of 3-5 year breeders (naive birds breeding for the first time) changed significantly only in the high disturbance plots, and this was correlated with the reduced number of returning older experienced adults (r= -0.80, df=11, p< 0.001 ). Birds that moved into highly disturbed plots were new to the island or from the far side of the island (X 2 = 4.65, p<0.05), while birds that moved into the low and medium plots were as likely to come from adjacent plots as distant plots (X 2 = 0.146, NS). Conover and Miller (1978) found that Ring-billed Gulls often avoided nesting in areas that had been the site of previous disturbances. Highly disturbed plots were no longer part of the central, desirable portion of the colony. At Antero, it appears that adults from immediately outside the high disturbance area could recognize and not use an area that had been highly disturbed in previous years. The age-structure shifted as highly disturbed plots became available to naive breeders. The difference between experimental plots was still evident two years later (Fig. 8b). The

PAGE 54

42 proportion of young breeders declined in 1985 as these birds returned to the same sites in 1985 where they had bred in 1984. Many would then be classified as older breeders. Movement patterns were compared between disturbance treatments (Fig. 9 and Table 4). The number of birds moving from the immediate area of their previous nest site was significantly different between each of the treatments (Table Sa). Additionally, birds that experienced multiple disturbances within one year or over several years moved more frequently than birds that were disturbed for only one season. Birds in the medium and high plots changed breeding islands more than birds on low plots, with many more leaving Gull I. and moving to Goose I. A higher proportion of birds was recorded as missing with increasing disturbance, suggesting that highly disturbed birds chose not to return to Antero Reservoir at all. This interpretation was confounded by birds that couldn't be relocated, birds with bands that couldn't be read or that were lost over the past year , and by birds that had died over the winter, as well as those that did not come back to Antero. In all cases, birds that experienced frequent or multiple disturbances moved from their original plot at a much higher frequency, changed islands more frequently, and were recorded missing more frequently . Young birds moved from a previous breeding site significantly more often than older birds on the low and high disturbance areas (Table 5b). I believe there is a similar pattern for birds on the medium plots although it was not statistically significant. Birds that experienced multiple years of medium or high disturbances moved from the previous plot regardless of age . There was no significant difference between ages for birds changing islands versus those remaining on the same island regardless

PAGE 55

100 a) 80 VI "C .... .c 60 0 (I) a CIS 40 r::: (I) (,) .... (I) c. 20 0 40 b) VI 30 "C .... .c 0 (I) 20 a CIS r::: (I) (,) .... (I) 10 c. 0 low+ multi low medium high Level of disturbance • Same Island Change Island low+ multi low medium high multi medium multi high tm Missing Moved • No Move multi medium multi high Level of disturbance Fig. 9. Comparison between level of disturbance in 1981-83 and a). Percentage of birds that moved from the previous breeding plot. b). Percentage of birds that changed islands or remained on the same island. No Move + Move + Missing = 1 00%, Move = Same Island + Change Island. Multiple disturbances are birds disturbed in more than one year 43

PAGE 56

44 TABLE 4. Comparison of movement responses by breeding birds in the year following experimental disturbance treatments (1982-1984) on Gull Island. Moved = birds moved to a different plot in subsequent years and either moved elsewhere on Gull I. (Same) or moved to Goose I. (Change). Move =Same Island + Change Island. Multi Low, Multi Medium and Multi High include birds that experienced more than one year of disturbance. Disturb. Level n No Move Move Same I. Change I. Missing LOW+ 235 162 20 17 3 53 Multi Low Medium 166 78 46 21 25 42 High 184 57 73 26 47 54 Multi 89 28 33 10 23 28 Medium Multi 63 12 28 7 21 23 High

PAGE 57

45 TABLE 5. (a) Two by two contingency analysis of breeding gulls movement response to disturbance treatments. Each case tested whether the second category moved more than the first. (b) Two by two contingency analysis of age-related response of breeding birds to disturbance treatment. This tested whether 3-5 year old breeders moved more frequently than older breeders. Move= bird moved from previous 10 m2 breeding site in subsequent year. Change = bird moved to Goose I. following a disturbance on Gull I. Multi Med and Multi High include birds that experienced more than one year of disturbance. Continuity correction applied to all multiple disturbances due to small cell sizes; df=1. (a) Low vs Med Low vs High Med vs High Med vs Multi Med High vs Multi High (b) 3-5 yr vs. > 5 yr n=167 n=570 Low Medium High Multi Med Multi High Move p 29.72 <0.001 73.93 <0.001 9.26 <0.01 4.83 <0.05 2.43 NS X2 p 10.46 <0.001 1.20 NS 5.53 <0.05 3.34 0.13 NS NS Change X2 p 8.84 <0.01 15.40 <0.001 1.19 NS 1.90 NS 1.04 X2 0.15 0.08 0.94 0.03 0.08 NS p NS NS NS NS NS

PAGE 58

46 of the disturbance frequency (Table 5b). Birds do not always abandon colonies following severe disturbances. In this study, California Gulls moved from the immediate breeding territory where disturbances occurred. Additionally, the more frequently a bird was disturbed the farther it moved from the original site. At a certain point, birds failed to return to Antero. This point depended on the bird's age, previous experience in the colony, and individual tenacity. Young birds tended to move most frequently under any conditions, while adults displayed a greater site tenacity. Uncontrolled Human Disturbance On three occasions, fishermen were observed on Goose I. while I was working on Gull I. As they walked through the colony, all birds would take flight and chicks would run across the island and into the water. After the fishermen left, it took up to one hour for all chicks to return to their territories. This period was accompanied by adults fighting, eggs being pecked, and chicks being killed. On one occasion, 37 chicks (less than 15 days old) were found dead the morning after one 10 minute intrusion. In contrast, only 5 chicks were found dead the morning after the annual 4 July banding operation when 1100 chicks were measured over a three hour period. In the latter case the chicks were all 25+ days old and human movements were controlled . In 1984, a more serious intrusion occurred on Gull I. during the early egg-laying stage. Three people landed on the island and commenced destroying eggs throughout the entire colony. Many nests were entirely destroyed, and although re-laying started immediately, the

PAGE 59

47 1984 average clutch size recorded on the 25 May count was 2 . 0 i n contrast to average clutch size in undisturbed areas of 2.4. It is impossible to say how much of this reduction was due to this disturbance versus the previous years' experimental disturbance. However, it is clear that uncontrolled human activities may cause high mortality of young and rapid abandonment of entire colonies (Anderson and Keith 1980 , Erwin 1980, Desgranges and Reed 1981 ) .

PAGE 60

CHAPTERV OWL PREDATION Immediate Response At least one Great Horned Owl (Bubo virginiana) has preyed upon the Antero Reservoir California Gull population each year since 1981 (Table 6). In this study, I examined the gulls' immediate and long term response to owl predation. The owl nested on a hill covered with ponderosa pine (Pinus ponderosa) approximately 2 km south of Gull I. There were no other owls nesting within 5 km of Antero. Only one adult gull was killed each night the owl attacked. All but one were killed during the incubation period and were located on either the exposed south and southeast side of Gull I. or on the south side of Goose I. The owl began killing chicks when they became available. Owls shifting predation from adults to chicks has also been noted by Kruuk (1964), Vermeer (1970) and Jehl and Chase (at Mono Lake) (1987}. No successful attacks on adults occurred on moonlit nights . Southern and Southern (1979) found that an owl preyed only in areas away from harbor lights. It is likely that adults were able to see an owl approaching and fled the colony before the owl arrived . Attacks consisted of direct , low level flights with the owl striking a gull upon entering the colony. Of the 29 adults killed, 70% were females;

PAGE 61

49 TABLE 6. Summary of owl predation on California Gulls breeding on Gull and Goose I. Percentage mortality is based on the number of eggs laid by 25 May . year 1981 1982 1983 1984 1985 1986 1987 Gulli. no. chicks 33 102 81 62 39 31 killed percent. 2.4 7.5 7.9 8.4 8.1 4.3 mortality no. adults 10 5 1 3 3 killed Goose I. no. chicks 37 88 7 killed percent 1.6 5.4 0.4 mortality no. adults 1 5 1 killed

PAGE 62

50 females typically incubated late at night (73-89%). I was present during 5 attacks. Adult response to intrusion by an owl was mass flight and abandonment of the island for 30 minutes to several hours. A similar response has also been reported by Emlen .et.al. (1966), Southern and Southern (1978), and Southern .et.al. (1982 and 1985). Hunter and Morris (1976), Southern and Southern (1978), and others report that adults often remained away from colonies until dawn. At Antero and other colonies in Wyoming and Montana, California Gulls returned to chicks well before dawn. Adults returned more rapidly following an owl intrusion when chicks were present than during incubation (X 2 = 15.85, df=2 , p<0.001 ). Chilling of eggs and chicks was common during these periods . Nocturnal temperatures averaged o-so C , and chicks less than 15 days old succumbed to cold stress in about 5 minutes. Over 50% (250 of 442, 1981-1985) of all chick mortality in response to disturbance by the owl was the indirect result of thermoregulatory stress. Antero is the highest gull colony in North America (2770m} and has correspondingly cooler ambient temperatures than other colonies reporting owl predation . Adult gulls are presented with a conflict: return too early and be killed, or return too late and lose eggs or chicks to cold. While I did not attempt to study mortality of embryos, it is likely there was at least some increased mortality and / or extension of the length of incubation . Hunter and Morris (1976} found that embryo mortality increased with age of the embryo when exposed to 1 oo C for 4 hours. Additionally, they found increased asynchrony of hatching and delayed hatching for chilled eggs . Hunter and Morris (1976) reported that storms during owl-caused desertions increased egg losses to approximately 90% among clutches that were near hatching. Vermeer (1970) also

PAGE 63

51 noted increased egg losses following intrusion by owls. Nisbet (1975) found lower hatching success and longer incubation periods for Common Terns disturbed by nocturnal predators . Chick response to an owl in the colony varied with age; young chicks crouched and remained motionless while older chicks (> 15 days) ran and grouped together. Over 90% of direct chick mortality occurred among chicks over 15 days old. Nisbet (1975) found that owls mainly killed Common Tern chicks less than 9 days old. These differences may simply reflect idosyncratic behavior of individual owls (Jehl and Chase 1987). Typically, at Antero, 2-3 broods would be killed in night attacks lasting only a few minutes. The owl would then eat the breast and viscera of one or two chicks, leaving severed wings, head, and lower torso. Adult kills often appeared uneaten (Southern and Southern 1979) , but close inspection of all kills at Antero revealed that the upper breast and viscera were always missing. The owl would preferentially attack grouped chicks (51 chicks were killed in less than 30 minutes on 21 June 1982} . Super stimulus killing has been reported in gull colonies for both owls and mammalian predators (see reviews in Southern and Southern 1982a and Southern .e1.al. 1985) . Mortality was always concentrated in areas of dense nests with high visibility (Fig . 1 0). These areas were on an open slope and hilltop directly facing the owl's approach. No adults or chicks were killed in the shrubby areas by owls . However, it is not likely that there is strong selection for shrubby habitats for breeding as selection by diurnal raptors occurrs against birds nesting in dense vegetation , at least in some breeding areas (Jehl and Chase 1987 ) .

PAGE 64

52 Long-term response Responses of Gull I. birds to predation by owls were compared for plots that experienced: a). no predation, b). light predation, where only a few chicks were killed, typically on a single night, c). concentrated predation, where most chicks in a plot (>50%) were killed, usually over several nights, and d). multiple predation, attacks occurred in the same plot 2 or more years (Fig. 11 a, Table 7). As the intensity of predation increased, significantly more birds moved from previous breeding sites. Birds that experienced multiple disturbances moved most frequently, with only 7% returning to previous breeding sites. There were no s i gnificant differences in movements of b i rds from light and concentrated predation areas. If an owl preyed in an area, there was a 300-400% increase in birds moving regardless of the concentration of predation. Predation by the owl resulted in a significant increase in the probability of changing islands (indicative of distance moved); however, there were no significant differences between different intensities of predation and whether a bird changed breeding islands or not ( Fig 11 b, Table 7). While there were no statistically significant differences in distances moved for birds experiencing no predation vs. light predation , I believe this is likely a statistical artifact due to small sample size . Birds that had directly experienced loss of chicks were compared with birds that nested in the area but did not lose chicks (Fig. 12a). A significantly greater number of birds that moved from previous nesting sites had experienced loss of chicks compared to birds that did not lose chicks (X 2 =15.76 , df=1, p<0 .0001 ). Even when birds did not lose young, a significantly greater number moved from predation areas when

PAGE 65

1 G F E D c B A ' 2 3 4 5 6 7 8 9 10 11 Fig . 1 0. Major locations of chick mortality from Great Horned Owl predation on Gull I. (}1 w

PAGE 66

a) Ul 't:J ... .c 0 (I) C) cu t: (I) (.) ... (I) ll. 100 80 60 40 20 0 b) 40 30 ... 0 (I) 20 C) cu t: (I) (.) ... 8:. 10 0 54 No Predation Light Concentrated Multi predation Intensity of predation • Same Island Change Island m1 Missing Moved • No Move No Predation Light Concentrated Multi-predation Intensity of predation Fig.11. Comparison between intensity of predation and a). Percentage of birds that moved from the previous breeding plot. b). Percentage of birds that changed islands or remained on the same island. No Move + Move + Missing = 1 00%, Move = Same Island + Change Island. Multi-predation = predation in more than one year.

PAGE 67

TABLE 7. (a) Comparison of California Gull movement patterns i n the years following different intensit i es of owl predation i n breeding plots. Light= few chicks killed, usually in one attack , Concentrated= >50% of chicks in plot killed , usually over several attacks, Multiple Predat i on = attacks in the same plot over more than one year . (b) Two by two contingency tests of movement patterns and intensit i es of predat i on . Move= Same Island +Change Island , NP = No predation, L = Light predation , C = Concentrated predation, MP = Mult i p l e predation . (a) Intensity of total No Move Move Same I Change I. Missing Predation n n % n % n % n % n % No Pred . Light Concent. Multiple Predation (b) 150 99 66 1 7 11 1 4 9 3 2 34 23 1 01 31 31 48 48 31 31 17 17 22 22 84 15 18 4 7 56 24 29 23 27 22 26 92 6 7 57 62 29 32 28 30 29 32 Intensity of Move vs. No Move Predation Change I. vs Same I. df Overall differences 3 NP vs . L 1 NP vs C 1 NP vs MP 1 L vs C 1 L vs MP 1 C vs MP 1 X2 p 118 .79 <0.0001 45 . 95 <0.0001 65 .61 <0.0001 96 .80 <0.0001 3.56 NS 16.06 <0.0001 4.81 <0. 05 df p 3 7.11 NS 1 1.86 NS 1 5 . 07 <0 . 05 1 5 . 33 <0.05 1 1.78 NS 1 2.0 NS 1 0.0004 NS 55

PAGE 68

56 a) 40 0 • No Move "0 t;2J Same I. ... .0 30 II Change I. 0 C1) 20 C) ns s::: C1) (.) ... 10 C1) c.. 0 Disturbance Direct mortality Intensity of predation b) 30 • Center 0 "0 EJ Periphery ... 20 .0 1m Change I. 0 C1) C) ns 10 s::: C1) (.) ... C1) c.. 0 > 5 yr old 3-5 yr old Age of breeding birds Fig. 12 Comparison of gull movement patterns influenced by Great Horned Owl predation . a) Disturbance = birds in predation areas but not experiencing direct mortality. Direct Mortality = birds experiencing loss of chicks due to Great Horned Owl predation . b). Age of breeding birds and breeding site location in the year following owl predation .

PAGE 69

57 compared to movement of birds from nonpredation areas (X 2 = 34.45, df=1, p<0.05). There was also a significant increase in the frequency with which birds that lost chicks changed islands (X 2 = 4.90, df=1, p5: n=75). Young birds nested more frequently on the periphery of the colonies. When they moved in response to predation, they moved more frequently to other areas on the periphery of the colony. Older birds nested more frequently in the center of the colony. After eggs or young were destroyed by predators, older gulls moved to other areas in the center of the colony . There was no significant increase in older adults moving to the periphery, as would be expected if the adults had lost status and had difficulty re-establishing territories in central areas. It may also be that nest densities on Gull 1., at least in some plots, were below saturation densities; however, if that were the case, a greater proportion of young breeders would be expected to move to central plots in the subsequent year. The results of the human disturbance experiments (Chapter IV), showed an increase in number of young breeders in high disturbance plots corresponding to a decrease in older adults. Any owl predation caused both immediate and long-term responses by California Gulls , even if their chicks were not killed. Owl predation occurred in small areas of the colony, and only birds near those areas appeared to have long-term reactions to the owl. Predation over a larger area was shown to cause immediate abandonment of an island at Mono Lake, California, and much lower use in later years (Jehl and Chase 1987).

PAGE 70

CHAPTER VI WATER LEVEL CHANGES Draining Immediate Response Antero was drained in 1986 to repair the dam. By 1 May (peak initiation of egg-laying) , water levels had dropped 5 meters and a land bridge had formed . Coyotes entered the colonies and ate clutches of eggs. Mass abandonment by adults for several hours occurred when a coyote entered the colony at night, whereas entering during the day resulted in mobbing attacks by the gulls. Coyotes ate relatively few eggs in comparison to those destroyed by the intraspecific aggression that followed disturbance by the coyotes. The gulls were highly agitated throughout the day and rarely incubated for more than a few hours at a time. Fights constantly occurred, and twice as many adults (mostly females) died compared to other years (11 vs 3-7). Davis and Dunn (1976) found that neighbors that destroyed other nests had already lost their own eggs . Such behavior spreads rapidly in highly disturbed colonies . Gull I. was visited by coyotes at an earlier date and more frequently than Goose I. It was completely abandoned by 5 June . Goose I. was completely abandoned by 15 June . No chicks hatched on either island. All Canada Goose (Branta canadensis) nesting was also disrupted

PAGE 71

59 (Chase, in prep). Twenty gull eggs were examined on Gull I. on 5 June. All were non-viable; 8 had no signs of embryo development, 7 had small embryos, and 5 had well developed embryos. Fifty eggs were examined on Goose I. on 11 June: 22 had no embryo development, 16 had small embryos, and 12 had well developed embryos. Much of the mortality probably occurred during early incubation when temperatures were often below oo C and the adults were off the eggs for several hours at a time. By 30 June, there were less than 150 gulls remaining at Antero, and none of them were using the islands. They roosted in the dry lake bed approximately 12 m from the islands. This area had been used as a roost site since nest abandonment began in mid-May. Long-Term Response In 1987, the reservoir was refilled, and the resulting water level was 0.6 meters higher than normal. Only 6 (15%) of older adults (total n=41) returned to previously used plots on Gull I. (Fig. 13a). At the same time, 49 (34%) of older adults (total n=143) returned to previous plots on Goose I. (X 2 = 6.9, p<0.01 ). Nero (1961) reported a similar dry lake situation in 1959 in Saskatchewan at a Bing-billed and California Gull colony. Some hatching occurred but no chicks fledged. The lake filled the next year, and the colony was resettled the next year. There were no significant differences between young (total n=74) and older gulls reusing the same plots on Gull 1., and there were no significant differences between young gulls from either island in any comparison. The latter finding was expected, since young birds from either island should have the same experiences at Antero and should react similarly.

PAGE 72

80 a) 60 U) '0 ... .0 0 40 G) C) ns c: G) 0 20 ... G) n. 0 50 b) 40 30 ... 0 G) C) ns c: G) 0 ... G) D. 20 10 0 60 • No Move Moved Em Missing 3-5 YR! GULL >5 YR!GULL >5 YR/GOOSE Age of birds/Breeding Island 3-5 YR/ GULL >5 YR / GULL II Same Island Change Island >5 YR/GOOSE Age of birds/ Breeding Island Fig. 13. Comparison of movement patterns influenced by draining Antero Reservoir. a). Percentage of birds that moved from the previous breeding plot. b). Percentage of birds that changed islands or remained on the same island. 3-5 YR/GULL = 3-5 year old adults returning to Gull I. in subsequent years; >5 YR/GULL= adults> 5 years old returning to Gull 1.; >5 YR/Goose =adults> 5 years old returning to Goose I.

PAGE 73

61 It is interesting to note that only one older bird moved from Goose I. to Gull I. following this disturbance, while 10 {24%) of the older Gull I. birds moved to Goose I. (Fig. 13b) (X 2=29.55, p<0.0001 ). Experienced adults did not immigrate to the most disturbed island even when disturbed on their current breeding site. There was a significant difference between ages of Gull I. birds with respect to whether they stayed on Gull I. or moved to Goose 1., with young birds remaining on Gull I. much more frequently (X 2 = 1 0.66, p<0.001 ). A comparison of birds recorded missing demonstrates that a significantly higher number of older adults were missing from Gull I. after this disturbance than from earlier years (X 2 = 12.19, p< 0.001 ). Young adults from Gull and Goose Is. and older adults from Goose I. didn't leave Antero at significantly higher frequencies after this disturbance than in previous years. I can't attribute this difference exclusively to the affects of the disturbance caused by draining . Gull I. was subjected to numerous disturbances during the study period 1981-1987. Additionally, as explained earlier, the missing category must be interpreted very cautiously. Birds that have been disturbed almost continually since 1981 appear to have left Antero at a higher frequency than birds that experienced less cumulative disturbance, especially young adults. Axel (1956), Kruuk {1964), and Patterson (1965) found patterns of declining numbers but no abandonment, even with continual disturbance. Flooding Flooding affected some peripheral plots in 1985 and 1987, when water levels rose 0.6 m above normal before the beginning of egg laying.

PAGE 74

62 Birds of all ages from both islands reacted similarly to flooded conditions. Birds that had previously nested in a plot did not spend much time searching the immediate flooded area but instead set up territories elsewhere. Only one bird returned during a flooded year and nested in the same plot as the previous year. This bird nested on the new shoreline on the unflooded edge of the plot. Most birds (71 %) moved to other parts of the islands, with only 2 birds changing islands (Fig . 14) (total n=134). The number of birds recorded missing was not signif i cantly different from least disturbed areas in earlier years . The response to this disturbance is less dramatic than other disturbances that involved loss of eggs or chicks. In this disturbance , the birds were reacting to conditions that had changed since the last breed ing season instead of disturbances during the last breeding season. No birds that had nested in plots in 1984 that were subsequently flooded in 1985 returned to those plots in 1986. Scharf (1981) reported that Common Terns nested on areas that had previously been inundated and often renested in sub-optimal habitat. In this study, only new territory holders (73% were 3-5 year olds) nested in these plots in 1986. Throughout this study , approximately 70% of all birds nesting in peripheral plots were 3-5 years old, so there was no change in age structure following this disturbance. Since this type of disturbance does little to change the overall fitness of a breeding bird , less extreme movements in reponse to this disturbance were expected.

PAGE 75

25 VI 20 '0 ... ..0 0 15 Q) C) < 10 (.) ... Q) c. 5 63 NO MOVE <25M 25-SOM >50M CHANGE I MISSING Distance moved Fig. 14. Distance moved from previous breeding season in response to flooding, 1985 and 1987; n=134. Change I. = Birds that moved from Gull I. to Goose I. after experiencing flooding.

PAGE 76

CHAPTER VII CONCLUSIONS AND SUMMARY In this study I examined various types and frequencies of disturbance affecting the breeding colony of California Gulls at Antero Reservoir over a seven year period (1981-1987). The effects of major disturbances on adult nest site choice in subsequent years are shown in Tables 8a and 9. Annual fledg ing decreased from a high of 69% (1.6 per nest) in the least disturbed plots to 0% when Antero was drained and coyotes entered the colonies. The "fl ooded" category must be interpreted cautiously, because birds cannot nest on flooded plots and most of the birds that nest in the peripheral plots that were flooded were young birds with a lower frequency of return. A bird that moved from a plot or from another colony , such as the invasion of birds from Great Salt Lake in 1985, and successfully bred in the new area is unlikely to return to the original breeding site. On the other hand , many birds that experienced breeding failure did not immediately abandon Antero. Many returned to the same plots the next year while others relocated elsewhere on the island or moved to a nearby colony (re.Goose 1.) (Tables 8 and 9). Gulls will try to establish territories in more preferred, central colony locations (Pugesek 1983), but are limited by the presence of older gulls. The central portion of a colony is typically used by older and more

PAGE 77

65 TABLE 8. (a) Comparison between disturbance regimes and movement patterns of breeding gulls on Gull I. (b) Rank comparisons of response to disturbances, 1 = smallest proportion of sample, 7 = greatest proportion of sample. Multi = Disturbance in more than one year. (a) Total No move Move Missing Disturbance n n/% n/% n/% Low 235 162/69 20/9 53/23 High 184 57/31 73/40 54/29 Multi-high Human 63 12/19 28/44 23/37 Owl 185 46/25 95/51 44/24 Multi-pred Owl 92 6/7 57/62 29 / 32 Drain/Coyote 115 14/12 67/58 34/30 Flood 135 2/1 100/74 33/24 (b) Disturbance No Move Move Missing Low 7 1 1 High 6 2 4 Multi-high Human 5 3 7 Owl 4 4 2 Multi-pred Owl 2 6 6 Drain/Coyote 3 5 5 Flood 1 7 3

PAGE 78

66 TABLE 9. Comparisons of movement patterns of breeding gulls on Gull I. from original breed ing plots following disturbances in the previous year . Same I= moved to a different plot on Gulli., Change I.= moved to Goose I. Multi = Disturbance in more than one year . No. moved Same I. Change I. Rank comparison Disturbance n n / % n/% Same I. Change I. Low 20 17/ 85 3/1 6 2 High 73 26/36 47/64 2 6 Multi-high Human 28 7/25 21/75 1 7 Owl 95 55/58 40142 4 4 Multi-pred 57 29/51 28 / 49 3 5 Owl Drain/Coyote 67 46/69 21/31 5 3 Flood 100 96/96 4/4 7 1

PAGE 79

67 successful birds (e.g. Coulson 1968). In my study, as disturbances continued on Gull 1., fewer adults returned to previous central nesting sites . A greater number of young adults were able to establish territories in areas that were previously unavailable to them. This would normally be advantagous; however, with continuing disturbance, no advantage was obtained, and, in fact, a loss of fitness may have resulted. Chi square analysis (Table 1 0) of movement patterns (Table Sa) demonstrates highly significant differences between all disturbances and the least disturbed areas. I ranked movement responses of California Gulls at Antero to the various disturbances during 1981-1986 (Tables 8b and 9). Disturbance by the owl and the combined effects of draining and subsequent coyote predation resulted in significantly more birds moving than resulted from the most frequently repeated controlled human disturbance. However, it is interesting to note that the birds that did move changed islands more frequently as the frequency of disturbance increased, regardless of the cause of disturbance. High human disturbance plots were disturbed more frequently than were the owl disturbance plots. Similar numbers of young fledged under both disturbance regimes; however, birds changed islands more frequently when exposed to high and multiple high levels of human disturbance versus predation. This finding is counter-intuitive and may be due to sampling error or sample size problems. There was little difference in movements due to owl predation and draining the reservoir. This result was not unexpected since both disturbances were caused by nocturnal predators (owls and coyotes, respectively) and resulted in mortality to eggs, chicks, and even adults. The frequency with which birds changed islands in response to flooding

PAGE 80

68 TABLE 10. Chi square values and probabilities for comparisons between disturbances and percentages of birds in Table 8a. The first line of values is for Move/No Move comparisons, the second line of values are for Same I./Change I. comparisons. No comparisons were made for Move/No Move for the flood category since birds had no choice but to move. * = p<0.05. Disturbance Low High Multi Owl Multi Drain Low Human High Human Multi Human Owl Multi Owl Drain Flood Human Human Human Predation Predation 73.9 * 15.4 * 67.4 * 16.8 * 110.2 * 5.2 * 2.4 NS 1.0 NS 3.6 NS 0.2 NS 8 . 2 * 9 . 2 * 137.2 * 22.7 * 7.1 * 5.2 * 7.5 * 2.7 * 8.1 * 0.6 NS 130.3 * 15.8 * 2.6 NS 6.1 * 2.1 NS 15.3 * 15.3 * 3.1 * 3 . 7 * 74.0 * 70.2 * 47.6 * 1.8 NS 4.4 * 46.6 * 23.6 *

PAGE 81

69 was slightly greater in the least disturbed plots but was much less than all other disturbances. Reproductive losses can be high following human intrusion because of associated interspecific predation as well as intraspecific predation (Davis and Dunn 1976, Anderson and Keith 1980, Hand 1980). It is likely that human intrusion (in contrast to the controlled movements of this study) ellicits responses as strong as reactions to predators alone. Multiple disturbances of all types that resulted in loss of chicks also resulted in the most movement within and between islands. California Gulls nest in relatively stable colonies (Behle 1953 , Conover and Miller 1978, and Paul 1985). When colony sites are dramatically disturbed (flooding at Great Salt Lake, mammal predation in Wyoming), birds abandon the original colony and move long distances to new breeding colonies . Less intense disturbances result in less dramatic changes within a colony and less frequent movements of individual birds from one local colony to another (e.g. Gull I. and Goose 1.). Overall movement patterns and subsequent reproductive performances on Gull and Goose I. were directly related. With young breeders replacing adults, there may be no apparent change in the population demography and perhaps even a measured increase in the number of breeding birds . However, without knowing i ndividual birds , changes in a population may be obscured . There are many reasons to control researcher activity in breeding colonies of waterbirds as Fetterolf (1983) and others have stated . Interpretation of reproductive output without knowing the history of a colony will lead to misinterpretations of ecological questions and poor information upon which to base conservation decisions. Studying a

PAGE 82

70 disturbance for only one year may also lead to misinterpretation. It is clear from this study that in California Gulls, there is a decrease in number of nests, clutch size, and fledging success with increasing disturbance as well as changes in individual nest site selection. There are long term effects on breeding individuals, especially as an increasing percentage of an adult's reproductive fitness is lost to disturbance. Long-term studies are essential to understand the ecology of colonial breeding waterbirds, but it is also critical that disturbances caused by researchers, as well as other disturbances, be understood and quantified as accurately as possible. A colonial waterbird study with no researcher-caused disturbance has yet to be conducted; and, given the need to mark, weigh, and measure eggs, chicks, and adults, it is not likely to occur. Choosing critical times and methods that minimize disturbance must become a standard in colonial bird research.

PAGE 83

LITERATURE CITED Anderson, D., and J. 0. Keith. 1980. The human influence on seabird nesting success: conservation implications . Biological Conservation 18:65-80. Austin, 0. L. 1951. Group adherence in the Common Tern (Sterna hirundo). Bird Banding 22:1-15. Axell, R. E. 1956 . Predation and protection at Dungeness Bird Reserve. British Birds 49:193-212 . Behle , W. H . 1958 . The bird life of Great Salt Lake . Univ. Utah Press, Salt Lake City. ---=--:-:----:and W. A. Goates. 1957. Breeding biology of the California Gull. Condor 59:235-246 . Bergman, G. 1986. Feeding habits, accommodation to men, breeding success and aspects of coloniality in the Common Gull. Ornis Fennica 63:65-78. Boellstorf, D. E., D . W. Anderson, H. M. Ohlendorf, and E . J. O'Neil. 1988. Reproductive effects of nest-marking studies in an American White Pelican colony. Colonial Waterbirds 11 :215-219. Brown, R. G. B. 1967. Breeding success and population growth in a colony of Herring and Lesser Black-backed Gulls (Larus argentatus and L. fuscus). Behaviour 29:122-153. Buckley , F . G. and P . A. Buckley . 1972. The breeding ecology of Royal Terns (Sterna m. maxima). Ibis 114 : 344-359 . Burger, J. 1980 . Territory size differences in relation to reproductive stage and type of intruder in Herring Gulls (Larus argentatus). Auk 97 : 733-7 41. ____ . 1981. Effects of human disturbance on colonial spec ies, particularly gulls . Colonial Waterbirds 4 : 28-36.

PAGE 84

Chabrzyk, G. and J.C. Coulson. 1976. Survival and recruitment in the Herring Gull, (Larus argentatus). Journal of Animal Ecology 45:187-203. 72 Chappell, M., D. Goldstein and D. Winkler. 1984. Oxygen consumption, evaporative water loss, and temperature regulation of California Gull chicks (Larus californicus) in a desert rookery . Physiological Zoology 57:204-214. Chardine, J. W. and R. D. Morris . 1983. Herring Gull males eat their own eggs. Wilson Bulletin 95:477-478. Conover, M. R. and D. E. Miller. 1978. Reaction of Ring-Billed Gulls to predators and human disturbance at their breeding colonies. Proceedings Colonial Waterbird Group 2:41-47. ________ and D. E. Miller. 1980. Daily activity patterns of breeding Ring-billed and California Gulls. Journal of Field Ornithology 51 :329339. Coulson, J . C. 1968. Differences in the quality of birds nesting in the centre and on the edges of a colony. Nature 217:478-479. ____ and E. White. 1958 . The effect of age on the breeding biology of the Kittiwake (Rissa tridactyla). Ibis 100:40-51. ____ , N. Duncan, C. S. Thomas and P. Monaghan. 1981. An age-related difference in the bill depth of Herring Gulls. Ibis 123:499502. ____ , N. Duncan and C. Thomas. 1982. Changes in the breeding biology of the Herring Gull (Larus argentatus) induced by reduction in the size and density of the colony. Journal of Animal Ecology 51 :739-756. ____ and J. M. Porter. 1985. Reproductive success of the Kittiwake: the role of clutch size, chick growth rates and parental quality. Ibis 127:450-460. _____ and C. S. Thomas. 1985. Changes in the biology of the Kittiwake: a 31-year study of a breeding colony. Journal of Animal Ecology 54:9-26. Cuthbert, F. J. 1985a. lntraseasonal movement between colony sites by Caspian Terns in the Great Lakes. Wilson Bulletin 97:502 510. ____ . 1985b . Mate retention in Caspian Terns. Condor 87:74-78.

PAGE 85

73 Davis, J. W. F. and E. K. Dunn. 1976. Intraspecific predation and colonial breeding in Lesser Black-backed Gulls (Larus fuscus). Ibis 118:65-77. Desgranges, J. L. and A. Reed. 1981. Disturbance and control of colonies of Double-Crested Cormorants in Quebec. Colonial Waterbirds 4:12-19. Dewit, A. A. N. and A. L. Spaans. 1984. Veranderingen in de broedbiologie van de Zilvermeeuw (Larus argentatus) door oegenomen aantallen. Limosa 57:87-90 . Drapeau, P., R. McNeil and J. Burton. 1984. Effects of human disturbance on the activity of the Double-crested Cormorant (Phalacrocorax auritus) and on the reproduction of the Great Blue Heron (Ardea herodias) in the Magdalen Islands , Canada. Candian Field-Naturalist 98:219-222. Drost, R. 1958. Uber die ansiedlung von jung ins Binnenland ver frachteten Silbermowen (Larus argentatus). Vogelwart 19:169173. Duffy, D. C. 1979. Human disturbance and breeding birds. Auk 96:815 816 . Dulude, A. M., G. Baron and R. McNeil. 1987. Role of male and female Ring-billed Gullls in the care of young and terrritorial defense. Canadian Journal of Zoology. 65:1535-1540. Dutcher, W. and W . Baily. 1903. A contribution to the life history of the Herring Gull (Larus argentatus) in the U.S . Auk 20:417-431. Ellison, L. N. and L. Cleary. 1978. Effects of human disturbance on the breeding of Double-crested Cormorants. Auk 95:510-517. ____ . 1979. Author's response. Auk 96:816-817 . Emlen, J . T. 1956. Juvenile mortality in a Ring-billed Gull colony. Wilson Bulletin 68:232-238. ____ , D . Miller, R. M . Evans and D. H. Thompson. 1966. Predator induced parental neglect in a Ring-billed Gull colony. Auk 83:677679. Erwin, R. M. 1980 . Breeding habitat use by colonially nesting waterbirds in two mid-Atlantic U.S. regions under different regimes of human disturbance. Biological Conservation 18 : 39-51.

PAGE 86

____ . 1989 . Responses to human intruders by birds nesting in colonies : experimental results and management and guidelines. Colonial Waterbirds 12:104-108 . Fetterolf , P. M. 1983 . Effects of investigator activity on Ring-b i lled Gull behavior and reproductive performance. Wilson Bulletin 95:23-41 . 74 Gillett, W. H., J . L. Hayward, and J. F. Stout. 1975. Effects of human activity on egg and chick mortality in a Glaucous-winged Gull colony. Condor 77:492-495. Hand , J . L. 1980 . Human disturbance in Western Gull (L. occi dent a l i s li v en s) colonies and possible amplification of i ntraspecific predation . Biological Conservation 18 : 59-63 . Harris , M. P . 1964. Aspects of the breeding biology of the gulls L arus a r ge ntat us, L. f u se u s , and L. rnarinus. Ibis 106:432-456 . Hatch, J . J. 1970 . Predation and piracy by gulls at a ternery in Maine . Auk 87 : 244-254 . Herrick , F. 1909. Organization of the gull community. Proceedings International Zoological Congress, Boston. Hollander , M., and D. A. Wolfe. 1973. Nonparametr i c stat i stica l methods . Wiley and Son s , N .Y. Hunt , G . L. 1972 . Influence of food distr i bution and human disturbance on the reproduct i ve success of Herring Gulls. Ecology 53 :1051-1061. --=--:-:----and M . W . Hunt. 1975. Reproductive ecology of the Western Gull: the importance of nesting space. Auk 92:270 279. ---=---and M. W. Hunt. 1976 . Gull chicks surv i val: the sign i ficance of growth rates , timing of breeding and territory size . Ecology 57 : 62 75 . Hunter , R . A. and R. D . Morris . 1976 . Nocturnal predation by a Black crowned Nigh t Heron at a Common Tern colony . Auk 93:629-633. Jehl , J . and C. A. Chase. 1987 . Foraging patterns and prey selection by avian predators: a comparative study in two colonies of California Gulls. Studies in Avian Biology 10 : 91-101. Johnston, D. W. 1956. The annual reproductive cycle of the Ca l ifornia Gull. I. Criter i a o f age and male reproductive cycle . Condor 58:134162 .

PAGE 87

---::-....,..,....-....,. 1956. The annual reproductive cycle of the California Gull. II. Histology and female reroductive system. Condor 58:206221. Kadlec, J. A. 1971. Effects of introducing foxes and racoons in Herring Gull colonies. Journal of Wildlife Management 35:625-636. -----=--' W. H. Drury and D . K. Onion. 1969. Growth and mortality of Herring Gull chicks. Bird Banding 40 : 222-233. 75 --=---:-:---and W. H. Drury. 1968. Structure of the New England Herring Gull population. Ecology 49:644-676 . Kirkman, F. 1937 . Bird behavior: a contribution based chiefly on a study of the Black-headed Gull. Nelson & Jack, London. Kruuk, H. 1964 . Predators vs. anti-predator behaviour of the Black headed Gull ( L a rus ridibundus ) . Behaviour Suppl. 11:1-129. Kury, C. R. and M . Gochfeld. 1975. Human interference and gull predation in cormorant colonies . Biological Conservation 8:23-34. McNichol!, M . K. 1975. Larid site tenacity and group adherence in relation to habitat. Auk 92:98-104 . Mineau , P . and D . U. Weseloh. 1981. Low disturbance monitoring of Herring Gull reproductive success on the Great Lakes. Colonial Waterbirds 4:138-142. Morris, R. D. and J. W. Chardine. 1985. The effects of ice cover over the colony site on reproductive activities of Herring Gulls. Canadian Journal of Zoology 63:607-611 . Mousseau, P. 1984. A comparison of 2 methods to assess the breeding success of Ring-billed Gulls (Larus delawaren s is). Journal of Field Ornithology 55:151-159. Nero, R. W . 1961. Dry land gull colony. Blue Jay 1 9 : 166-168. Nisbet, I. C. T. 1975. Selective effects of predation in a tern colony . Condor 77:221-226. Paludan, K. 1951. Contributions to the breeding biology of ( L. argentatus) and (L. fusc u s). Videnskabelige Meddelelser Fre Dansk Naturhistorisk Forening 114:1-128.

PAGE 88

Paul, D. 1985. Current and historical breeding status of the California Gull in the Great Salt Lake Region. Report to the Los Angeles Department of Water and Power. Parsons, J . 1971. Cannibalism in Herring Gulls. British Birds 64:528537. 76 Patterson, I. J. 1965. Timing and spacing of broods in the Black-headed Gull. Ibis 107:433-459. Pickett, S . T. A . and P. S. White, editors. 1985. The ecology of natural disturbance and patch dynamics. Academic, New York . Pierotti , R. 1980 . Spite and altruism in gulls. American Naturalist 115:290-300 . 1981 . Male and female parental roles in the Western Gull under different environmental conditions. Auk 98:532-549 . ---=---=---.and E. C. Murphy. 1987. lntergenerational conflicts in gl.l'lls. Animal Behav ior 35:435-444. Pugesek, B . 1983. The relationship between parental age and reproductive effort in the California Gull. Behavioral Ecology and Sociobiology 13:161 171. ____ and K . D i em. 1983 . A multivariate study of the relationship of parental age to reproductive success on California Gulls . Ecology 64:829-839. Raper, E. L. 1976. Influence of the nesting habitat on the breeding success of California Gulls (Larus californi c us) , Bamforth Lake , Albany Co. , WY. MS Thesis, Univ. Wyoming, Laramie. Roberts , H. C. and C. J. Ralph . 1975. Effects of human disturbance on the breeding success of gulls. Condor 77:495-499 . Scharf, W. C. 1981 . The significance of deteriorating man-made island habitats to Common Terns and Ring-billed Gulls in the St. Mary's River , Michigan . Colonial Waterbirds 4:155-159 . Shugart , G. W., M .A. Fitch and V . M . Shugart. 1981 . Minimizing investigator d i sturbance i n observational studies of colonial birds: access to blinds through tunnels . Wilson Bulletin 93:565 569. Sakal , R. R., and F. J. Rohlf. 1981. Biometry. Freeman , San Fransisco.

PAGE 89

77 Sousa , W. P. 1984. The role of disturbance in natural commun iti es. Annual Review of Ecology and Systematics 15:353-91 . Southern , L. K . 1981. Sex-related differences in territorial aggression by Ring-billed Gulls . Auk 98:179-181. ____ , S . R . Patton and W . E. Southern. 1982. Nocturnal predation on ( Larus) gulls. Colonial Waterbirds 5:169 172 . ____ and W. E. 1978. Absence of nocturnal predator defense mechanisms in breeding gulls. Proceedings of the Colonia l Waterbird Group 157 162 . Southern , W . E . 1977. Colony selection and colony s i te tenacity i n Ring billed Gulls at a stable colony. Auk 94 : 469-478 . ____ , S . R. Patton , L. K. Southern and L.A. Hanners. 1985 . Effects of nine years of fox predat ion on two species of breeding gulls . Auk 102 : 827-833. ____ and L. K . Southern . 1981. Colony census results as indicators of pre hatching perturbations. Colon i al Waterbirds 4:143 149 . --=--and L. K. Southern. 1982a. Effect of habitat decimation on Ring-billed Gull colony-site and nest site tenacity . Auk 99:328 331. -....,......,..,,............,.-.and L. K. Southern. 1982b. Intensification of adult R i ng billed Gull aggression during reproduction and it ' s possible consequences . Colonial Waterbirds 5 : 2-10 . -....,......,..,-_,and M . Southern. 1984. Herring Gulls specialize as Ring billed Gull predators. Colonial Waterbirds 7:105-110. Speich , S. M. 1986. Colonial waterbirds. Pages 387-405 in A . V.Cooperrider, et.al., editors . Inventory and Monitoring of Wildl ife Habitat. U.S. Dept. of Interior , Bureau of Land Management, Denver. Tinbergen, N. 1953 . The Herring Gull's world. Colins , London. Verbeek, N. A. M. 1982. Egg predation by Northwest Crows: its association with human and Bald Eagle activity. Auk 99 : 347 352 . Vermeer , K. 1963 . The breed ing ecology of the Glaucous-winged Gull on Mandarte Island , B . C . Occasional Papers British Columb i a Provinicial Museum no.13 .

PAGE 90

78 ____ . 1970. Breeding biology of California and Ring-billed Gulls: a study of ecological adaptation to the inland habitat. Canadian Wildlife Service Report Series no. 12. Ward, H. L. 1906. Why do Herring Gulls kill their young? Science 24:593-594. Watanuki, Y. 1988. Intraspecific predation and chick survival: comparison among colonies of Slaty-backed Gulls. Oikos 53:194202. Winkler, D. W. 1983 . Ecological and behavioral determinants of clutch size: the California Gull (Larus californicus) in the Great Basin. PhD. Diss., Univ. Calif. Berkeley, 195pp. Young, E. C. 1963. The breeding behavior of the South Polar Skua (Catharacta maccormicki). Ibis 105:203-233.

PAGE 91

53