Growth form and reproductive output in limber pine

Material Information

Growth form and reproductive output in limber pine the cost of mutualism
Feldman, Ronald
Place of Publication:
Denver, CO
University of Colorado Denver
Publication Date:
Physical Description:
x, 46 leaves : illustrations ; 29 cm


Subjects / Keywords:
Limber pine -- Growth -- Colorado -- Rocky Mountain National Park ( lcsh )
Limber pine -- Reproduction -- Colorado -- Rocky Mountain National Park ( lcsh )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )


Includes bibliographical references.
Submitted in partial fulfillment of the requirements for the degree, Master of Arts, Department of Integrative Biology
Statement of Responsibility:
by Ronald Feldman.

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:
25352818 ( OCLC )
LD1190.L45 1991m .F44 ( lcc )


This item has the following downloads:

Full Text


GROWTH FORM AND REPRODUCTIVE OUTPUT IN LIMBER PINE: THE COST OF MUTUALISM by Ronald Feldman B.A., University of Colorado, 1981 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 Ronald Feldman has been approved for the Department of Biology by Diana F. Tomback (Emily L. Hartman Yon B. Linhart


Feldman, Ronald (M.A., Biology) Growth Form and Reproductive Output in Limber Pine: The Cost of Mutualism Thesis directed by Associate Professor Diana F. Tomback The western American limber pine (Pinus f7exi7is) may grow in clusters of two or more distinct genotypes that are often closely related --the result of multiple germination in seed caches of Clark's Nutcracker (Co7umbiana nucifraga). Kin selection may promote crowding tolerance in limber pine clusters. I tested the hypothesis that competition within clusters may lower the fitness of cluster members relative to solitary trees by measuring the reproductive output of solitary trees and of stems growing in multigenotype clusters in three limber pine stands in the Southern Rocky Mountains, Colorado. Each genet's male and female cone production and seed weights were recorded over a three-year period. Within clusters, one genet (high-yield) often produced a significantly greater number of male and female cones than co-cluster genet(s) (low-yield). The reproductive output of high-yield cluster members was often lower than that of solitary trees, while low-yield members produced significantly fewer male and female


cones than solitary trees. Also, tree cluster members had lower cumulative diameter growth rates and significantly higher height/diameter ratios than did solitary trees, a probable consequence of competition. Because the high-yield genet's reproductive output is not sufficient to meet the cost-to-benefit ratio required by kin selection theory, kin selection probably does not account for crowding tolerance. Thus, dependency on nutcrackers for seed dispersal may result in fitness costs for some limber pine trees. The form and content of this abstract are approved. I Signed recommend its publication. Diana F. Tomback iv


I For Anna We 7 7 er Rainbow Curve Crows in grey business suits, but more dapper, The nutcrackers wing down to their hidden hoards, Push bills to ground, and out spring seeds like treasure. The fledglings are gone with their begging whines, The pine cones trashed, the tourists fled. All is spent and white: a season of concentration For those who linger on this winter height. "We live on our memories."--their motto, surely. A buried seed from fabled August Recalled under February's dim light; A thousand treasure maps to a thousand caches All under a feathered cap--no bird brain this! (Nor human either.) But perfection in this world is not to be: Seeds germinate and saplings sprout Where their long memories gave out. And trees whose seeds now feed the flock Are gifts of some neglected cache Interred by some ancestral beak. "We live on our memories," these birds sing "But the future is what forgetfulness brings."


CONTENTS Figures ..... vii Tables . viii Acknowledgements i X CHAPTER 1. INTRODUCTION 1 2. METHODS 6 Study Areas .. . 6 Methods . . . 8 Statistical Analysis ............. 13 3. RESULTS AND DISCUSSION .. Cone Production by Single Trunk Trees and Tree Clusters . . . .. Single Trunk Trees and Tree Clusters: 15 15 Dbh, Height and Age . . . . . . 22 Reproductive Output per Genet: Comparisons Among Years Multi-trunk Trees: Cone and Seed Production . . Height/Diameter Ratios General Discussion .. CITATIONS . . . . vi -, 28 30 32 36 43


FIGURES Figure 2.1. Location of study areas ............. 7 3.1. Percentage of diameter (dbh), height, male cone whorls, and female cones contributed by high and low-yield tree cluster members ... 17 3.2. Mean production of numbers of male cone whorls per tree by single trunk trees and high and low-yield tree cluster members .... 19 3.3. Linear regression analysis of mean number of male cone whorls on dbh ........... 21 3.4. Mean production of number of female cones per tree by single trunk trees and high and low-yield tree cluster members ....... 23 vii


TABLES Table 2.1. Elevation, Aspect, and Slope Angle of Limber Pine Study Sites and the Sample Sizes. . . . . . . . . 3.1. Mean Male and Female Cone Production by Single Trunk Trees Compared to Production . 8 by Tree Clusters Summed Over Members in the Cluster and Averaged Over the Members ..... 16 3.2. Seed Weight and Percentage of Filled Seeds of Single and Multi-trunk Trees and Tree Cluster Members. . .......... 24 3.3. Regression of Height and Diameter (dbh) on Age for Single Trunk Trees and Tree Cluster Members. . . . . . . . . 25 3.4. Mean Dbh and Height of Single Trunk Trees and Tree Cluster Members ........... 27 3.5. Pearson Correlation Coefficients (r) for Comparisons Between Male or Female Cone Production in 1987 and 1988 and Percentage of Filled Seeds in 1988 ...... ..... 29 3.6. Yearly Production of Mean Male cCne Whorls and Female Cones by Single and Multi-trunk Trees ........ 3.7. Height/Diameter Ratios of Single and . 31 Multi-trunk Trees and Tree Cluster Members .. 33 3.8. Mean Age, Growth Rates, and Height/Diameter Ratios of Single Trunk Trees and Tree Cluster Members (All Stands Combined) ......... 35 viii


ACKNOWLEDGEMENTS There are many persons and organizations who assisted me in the process of research and writing. Above all, the members of my faculty committee: Emily Hartman, who helped me to enter the graduate program and from whom much of my botanical knowledge stems, Yon Linhart, who taught me the field and greenhouse techniques used in this study and whose advice was indispensable, and Diana Tomback, who got me safely to base camp, helped me reconnoiter the difficult terrain and, while the technical climb was in progress, gave pertinent warnings and encouragement. The University of Colorado at Denver Research and Creative Activities Committee provided my study with much needed funds through its 1987 Graduate Student Development Award Program which was supplemented by a Faculty Development Award given to Diana Tomback. David Stevens, Rocky Mountain National Park biologist, kindly permitted me free entrance to the Park for three field seasons. Jeffry Mitton at the University of Colorado at Boulder and William Schuster, now at the University of Utah, were


invaluable sources of expertise on the reproduction and genetics of limber pine. Kathy Carsey's electrophoretic analysis of the tree clusters was an absolutely essential element of the study. Marilyn Weir, Departmental Representative of English as a Second Language, Emily Griffith Opportunity School, allowed me the flexibility as an employee to combine work and study. Helen, my wife, put up with more leaky faucets, hirsute lawns, and sticky pine cones than any spouse should be expected to, and yet remained tolerant and supportive. To all of you I express my gratitude and appreciation. X


CHAPTER 1 INTRODUCTION Individuals of the western American limber pine (Pinus f7exi7is James) are commonly found growing in clusters of 2 to 5 with stems near, contiguous, or fused at their bases. Until recently these clusters were thought to be multiple trunks of a single tree; however, protein electrophoretic studies have shown that clusters of limber pine as well as of whitebark pine (P. a7bicau7is), another western pine, are often composed of two or more distinct genotypes (Linhart and Tomback 1985; Furnier et a7. 1987; Schuster and Mitton 1991; Carsey and Tomback, in prep.). This communal growth form is a rare phenomenon among conifers which typically grow solitarily, although spruces and firs may exhibit clonal growth with some occurrences of mixed genotypes (Shea and Grant 1986). Linhart and Tomback (1985), using electrophoretic analysis of 2 to .4 allozyme loci, found 15 of 19 limber


pine clusters sampled and 5 out of 6 whitebark pine clusters to have two or more genotypes per cluster. The number of distinct genotypes in the multi-genotype clusters ranged from 2 to 5. Furnier et a1. (1987), using 11 loci, found 23 out of 35 whitebark pine clusters sampled to have two or more genotypes (range: 2 to 3). Cluster members were also more genetically similar to their own cluster members than to individuals outside the cluster, indicating family structuring within clusters. The frequency of multi-genotype clusters is variable among stands. Schuster and Mitton (in press) made an extensive genetic analysis of the entire population of an isolated limber pine stand and found only 19 of 108 clusters to contain more than one genotype. As with whitebark pine, cluster members were genetically similar and estimated to be related, on average, as almost halfsibs. Clustering in P. f1exi1is and P. a7bicau1is is the byproduct of the mutualism that both pines share with the corvid Nucifraga columbiana (Clark's Nutcracker). In late summer and fall N. columbiana preferentially harvests seeds from both P. f7exi1is and P. a1bicau7is and buries them in dispersed caches usually of 3 to 4 seeds (range: 1-15), some of which are never retrieved (Vander Wall and Balda 1977; Tomback and Kramer 1980; 2


Tomback 1980). Seedling establishment from caches is the primary means of recruitment for these two pine species. Crowding in plants leads to extreme intraspecific competition for water, soil nutrients, and sunlight, and almost invariably results in reduction in plant size and biomass or mortality (Harper 1977). In addition, crowded individuals often have lower reproductive success than isolated plants (Stern and Roche 1974; Harper 1977). That multi-genotype clusters of limber and whitebark pine survive and reproduce indicates a probable adaptive tolerance to crowding (Tomback and Linhart 1990); however, it may be an imperfect adaptation, and growing in clusters may still incur some reduction in fitness for individual trees. Thompson (1982) has argued that many mutualisms result from the amelioration over evolutionary time of an originally antagonistic relationship between two species. This implies that one or both species engaged in a mutualism may be sub-optimally adapted to the relationship either because co-evolution has been too brief or because genetic constraints prevent the elimination of all negative consequences. Loss in fitness for trees growing in clusters can be regarded as part of the cost of mutualism to the pine. The existence in stands of limber or whitebark pine of both clustered and solitary trees provides a natural 3


experiment in which the effects of intraspecific crowding can be tested. In this study solitary limber pine trees were matched with co-stand individuals growing in clusters for three limber pine stands in the Southern Rocky Mountains. The reproductive output of these trees was monitored over three years in order to test the hypothesis that individual fitness, as measured by reproductive output, is lower for individuals growing in clusters compared to those growing as solitary trees. The morphology of the two growth forms was also examined to test the hypothesis that crowding affected growth of trees in clusters. To complicate matters, starch gel protein electrophoresis of cluster members indicates that some clusters are composed of distinct genetic individuals while other "clusters'' originate from a single genotype (Linhart and Tomback 1985; Furnier et a7. '1987; Schuster and Mitton 1991). These growth forms are virtually indistinguishable in their above-ground morphology (Schuster and Mitton 1991; Carsey and Tomback, in prep.). Thus, limber pine has three growth forms: (1) single trunk, single genotype (2) multi-trunk, single genotype, and (3} tree cluster, multi-genotype. The multi-trunked growth form is usually the product of disease or mechanical injury in pines, but may be the result of a 4


genetic predisposition for this growth form in limber pine. This suggests a second hypothesis: Multi-trunk trees (single genotype) have a higher reproductive output than do single trunk (single genotype) trees. 5


CHAPTER 2 METHODS Study Areas Trees from three Pinus f7exi7is stands in the Colorado Front Range of the Southern Rocky Mountains were studied (Fig. 2.1 and Table 2.1). Two stands are in Rocky Mountain National Park; one (Rainbow stand) is near treeline on a slope face varying from a north to southeast aspect above and below the Rainbow Curve turnout on Trail Ridge Road. The second (Beaver) is 2 km east of the Rainbow stand and is located on the top and south-facing slope of an east-west trending ridge overlooking a stream and beaver ponds. The third and lowest stand (Sugarloaf) is in Roosevelt National Forest on the lower northwest-facing slope of Sugarloaf Mountain, near Boulder Canyon, 46 km southeast of the Beaver stand. P. f7exi7is predominates in all three stands and is associated with Populus tremu7oides, Pinus


a. ....., b. Da. ...., Rainbow Stand :!200m .5km :-.' ./' c.l (Sugarloaf ountain 2718m .Skm @ FIG. 2.1. ...... ...... ...... ...... N t ...... ...... 1_1 lkm ...... ...... ...... ...... -------...... ...... .... Location of study areas.


Table 2.1. .stand Sugarloaf Beaver Rainbow Elevation, Aspect, and Slope Angle of Limber Pine Study Sites and the Sample Sizes. No. tagged trees single multi-Mean slope trunk trunk trees in Elevation aspect angle trees trees cl usters 2585 m 27911 14 27 7 24 ( 10 )b 2810 m 183 16 29 5 33 (16) 3300 m 161 24" 32 11 10 ( 5) 88 23 67 ( 31) a Haan numbar of tr In a cluater gre: Sugarloaf, 2.4: Beavar, 2.1; Rainbow, 2.0: all atande, 2.2 b tltmbart!l In paronth .. ue are nUIIIbare of clu11hre. contorts, Pinus ponderosa, and Picea enge7mannii at the Sugarloaf and Beaver stands; Pseudotsuga menziesii was also present at the Sugarloaf stand. In the high elevation Rainbow stand, P. f7exi1is is associated with P. contorts, P. enge1mannii and Abies 7asiocarpa. Methods The following terminology will be used based on Tomback et a1. (1990): single trunk tree refers to a single tree growing solitarily, multi-trunk tree refers to a single genet also growing solitarily but with more than one stem, and tree cluster refers to multiple genets growing in a contiguous gr.oup (and often appearing 1 ike a multi-trunk tree). When the genetic identity of the 8


latter two growth forms is unknown, the entity will be referred to as a multi-trunk tree of unknown genetic identity, or of unknown origin. Here, the term "tree" will be used synonomously with "genet." Data were gathered from June to September, 1987 to 1989. In each stand 25-30 P. f7exi7is trees growing solitarily and 30-40 P. f7exi7is stems growing in multitrunk entities of unknown origin were individually marked with numbered aluminum tree tags (Table 1). Only healthy trees bearing male or female cones and with stem diameter greater than 6 em (at 137 em above ground) were selected. The following measurements were taken for each tagged stem: dbh (at 137 em), slope aspect, and maximum height (ground to top of canopy). Trigonometric estimates of tree height were made using a suunto optical reading clinometer. In each stand I cored 20-25 of the tagged stems at breast height (137 em). After being stained, cores were examined under. a stereo microscope and annual rings counted. Age was estimated as the total number of rings per core. Since age estimates were used primarily for comparison among groups in the stands, no adjustments have been made to correct for true age. Green foliage was collected from each trunk in a entity of unknown otigin. The foliage was genetically anal.yzed using starch gel protein 9


electrophoresis at nine polymorphic loci (Carsey and Tomback, in .. The results of this analysis were used to categorize each multi-trunk entity as a multitrunk tree or as a tree cluster. When trunks had identical genotypes at the nine loci, the entity was classified as a multi-trunk tree. The probability of misclassifying as a multi-trunk tree genetically distinct trunks (coincidentally identical at the nine loci tested) depends on the degree of relatedness of the trees in the cluster. Carsey and Tomback (in prep.) calculated that the probability of two genets that are unrelated in the same cluster having the same genotype is p = 0.010 for the Rainbow population. Their calculations were based on Rainbow genotype frequencies from Schuster et a7. (1989) for the same nine polymorphic loci used in our study. Schuster and Mitton (1991) calculated the probability of two genets in a cluster having the same genotype at ten loci in their Pawnee study area. For unrelated to selfed individuals, the mean probabilities ranged from .001 to .06. Consequently, Carsey and Tomback (in prep.) estimated that fewer than two tree clusters in their study were misclassified as multi-trunk single genotype trees. Because trees in clusters are compared with solitary trees in this study to test the principal hypothesis, miscl.assification is not a major problem. 10


This problem is addressed in more detail in Carsey and Tomback (in prep.). In late June through early August, 1987 to 1989, I estimated male (pollen) cone numbers for all tagged stems as follows. First, a count of male cone whorls was made for the entire canopy of each stem using a hand counter and binoculars when necessary; second, the number of individual cones per whorl for each of 10 whorls from each stem was recorded. Since whorls located on the tips of the outer and larger branches in most cases produced more cones than whorls located proximally and on smaller branches, the 10 whorls were divided into 5 "distal" and 5 "proximal" whorls. Mean number of male strobili per distal whorl and proximal whorl were then calculated. The product of each mean and total whorls counted per stem (cones/distal whorl X total whorls/stem and cones/proximal whorl X total whorls/stem) gave an upper and lower boundary estimate of male cones per stem. The mean of this range was used as an estimate of total cones for the stem. Although this method under-or overestimates male cones if distal and proximal whorls are not in equal numbers, it was sufficiently accurate for relative comparisons among stems. In order to estimate counting error rates, counts were made of male cone whorls for 11 stems on June 27, 1990, and the counts were 1 1


repeated on the same trunks on June 30, 1990. Mean percentage differences between the two counts on a trunkto-trunk basis was However, if totals from each trunk were summed together on each count date (as they are when calculating group means), the overall difference between the two counts was only (15,793 whorls versus 16,378 whorls), since over and undercounts from tree to tree cancel out. Most comparisons were based on numbers of male cone whorls, the more accurate value, rather than estimates of numbers of male cones. In most cases, however, the two figures correlated well. In late July and August of each year I counted mature female cones per tagged stem in all stands. However, only trees at the Rainbow stand produced sufficient numbers of cones for statistical analysis. For each year of the study in late August and September, mature female cones were collected from as many tagged stems as possible in the Rainbow stand using an extendable tree pruner. The number of normal-sized seeds per cone, mean cone length, and mean seed weight of 20 seeds (weighed to the nearest 0.001 g) drawn at random were determined for each stem. The weight of the heaviest seed in the 20-seed sample was recorded as maximum seed weight for the stem. The 20 weighed seeds per stem were opened, and the percentage of empty seeds 12


was recorded. The remaining seeds were scarified with moist sterile sand and stratified in groups of five in vials of moist sand for 90 days at a temperature range of 2-8 c. Fifty seeds per genet were planted in the greenhouse in late May (1987 seeds) and mid-June (1988 seeds) to determine germination percentages. Statistical Analysis Comparisons of seed weight, mean maximum seed weight and percentage of filled seeds were made using parametric One-Way ANOVA. In all other cases, Kruskal-Wallis One Way Analysis of Variance by Ranks was used for all comparisons among three or more groups, and the Wilcoxon Test was used for all comparisons between two groups. All correlations were calculated using Pearson's product-moment correlation. For some linear regression analyses, regression coefficients for different groups were compared using an a posteriori t-test for differences between two independent coefficients (Bailey 1981 ) Because of the possibility of common influences within a tree cluster, members of the same cluster could not be considered independent. Therefore, for all tree 13


clusters, variables were either averaged or summed over all members of a cluster to produce a cluster mean or sum, or data from individual members of a cluster were divided such that only one member of each cluster contributed data to any categorical group mean, e.g. data from low-yield tree cluster members and high-yield members. For all statistical tests, dbh did not differ significantly between the groups compared. When group means differed substantially in dbh, trees were removed sequentially from the group with the higher sample size, beginning with the tree with the smallest or largest dbh, until groups did not differ significantly. When ages were known and groups showed large age differences, older or younger trees were eliminated sequentially from groups until mean ages were not significantly different. 14


CHAPTER 3 RESULTS AND DISCUSSION Cone Production by Single Trunk Trees and tree Clusters In each stand and for every year monitored, the mean numbers of male cones per single trunk tree were greater than the mean numbers of male cones per cluster member (Table 3.1). None of the comparisons between column (B) and column (C) are significant; lack of significance is caused in part by loss of information wheriaveraging over members in a cluster. In most cases cluster members were unequal in cone production, with one member often producing a disproportionate share of male or female cones despite only small differences in DBH or height (Fig. 3.1). In both the Sugarloaf and Rainbow stands high-yield trees in each cluster accounted as a group for


Table 3.1. Mean Male and Female Cone Production by Single Trunk :rrees (C) Compared to Production by Tree Clusters Summed Over Members in the Cluster (A) and Averaged over the Members (B). Stand/Year N Sugarloaf 1987 10 1988 10 1989 10 Beaver 1987 16 1988 16 1989 16 Rainbow 1987 5 1988 6 Rainbow 1987 1988 1989 5 5 5 Cluster trees male cones/ cluster (A)b 12,880 5,630 15,281 13,532 40,728 13,041 ** 1 '729 8,414 fern. cones/ cluster 32.6 10.2 7.2 male cones/ genet ( B)c 5,313 2,609 7,000 6,720 20, 150 6,416 865 4,207 fern. cones/ genet 16.3 5. 1 3.6 N 15 14 16 29 29 19 32 32 32 32 32 Single trunk trees male cones/ genet (C) 6,410 4,580 10,330 10,914 38,256 7,874 2,557 7,647 fern. cones/ genet 41.6 15.3 7.3 All comparisons between columns Band Care non-significant. Only 2 comparisons between columns A and c are significant. a All comparisons ua1ng Wilcoxon Rank-Sum Teet for Two Oroupa b Only two compar1aone batween columna A and C are c All compar1aone between columna B and c are non-a1gntflcant p (0,05 .. p (0,01 16


.... ....... SUGARILOAIF DBH HEIGHT MALE CONE 'NHORLS IB:EAVER DBH HEIGHT tviALE CONE vVHORLS DBH HEIGHT tv!ALE CONE WHORLS FEtv1ALE CONES 0% 25% 5096 75% 100% HIGH-YIELD LO\N-YlELD . FIG. 3.1. Percentage of diameter (dbh), height, male cone whorls, and female cones contributed by high and low-yield tree cluster members.


of male cone whorls produced by clusters over three years at Sugarloaf and two years at Rainbow. In the Beaver stand high-yield cluster members accounted for of cluster production of male cone whorls and in the Rainbow stand produced of cluster female cones over three years of production. Mean production of male cone whorls by high-yield cluster members was sometimes greater but not significantly different from that of co-stand single trunk trees in all stands and for all years (Fig. 3.2}. Male cone production by the high-yield tree in the cluster was apparently not affected by cluster membership. In most comparisons male reproductive output by the low-yield cluster members was significantly lower than that by single trunk trees (Fig. 3.2): Sugarloaf lowyield tree cluster members produced significantly fewer male cone whorls in all three years (Wilcoxon Rank-Sum Test for Two Groups; 1987: Z= 1.8, df= 26, p= .035; 1988: Z= 2.2, df= 26, p= .015; 1989: Z= 3.2, df= 26, p= .0007); Beaver low-yield cluster members produced significantly fewer whorls in two different years (1987: Z= 2.35, df= 44, p= 1988: Z= 2.2, df= 44, p= .013); and Rainbow low-yield cluster members produced significantly fewer total whorls in 1987 (Z= 1.94, df= 31, p= .03). 18


MEAN MALE CONE WHORLS/GENET 1987 1988 SUGARLOAF STAND 40 OO MALE CONE WHORLS/GENET 29 0 1987 1988 BEAVER STAND 17 1989 1989 MALE CONE WHORLS/GENET 5 4oor---------------27 5 2001-----1987 1988 RAINBOW STAND SINGLE TRUNK TREES HIGH-YIELD C.M. 1\<:i:f.:iid LOW-YIELD C.M. FIG. 3.2. Mean production of numbers of male cone whorls per tree by single trunk trees and high and low-yield tree cluster members. Numbers above bars are sample sizes. 19


Regression analysis of number of male cone whorls on dbh was highly significant for single trunk trees in all stands and for all years of the study (Fig. 3.3). For tree cluster members regression of male cone whorls on dbh was significant in the Beaver and Sugarloaf stands with two exceptions. However, no regressions were significant for Rainbow cluster trees. In all three production years regression coefficients were significantly higher for Sugarloaf and Beaver single trunk trees than for co-stand low-yield tree cluster members (Fig. 3.3). In the two largest male cone crop years (Sugarloaf in 1989 and Beaver in 1988) the regression coefficients for high-yield cluster members were significantly larger than that of low-yield members. Thus, as the trees grow larger, low-yield tree cluster members suffer an increasingly competitive disadvantage in male cone production with single trunk trees and also with their superior cluster members, especially in high cone production or mast years. For example, using regression equations for Sugarloaf, it is estimated that in 1989 a 20-cm diameter single trunk tree produced 2.8 times as many male cone whorls as a lowyield cluster tree of the same diameter; however, a 40-cm diameter single trunk.tree produced 4.5 times as many whorls as a 40-cm diameter low-yield tree cluster member. 20


N -IW% CONI JliOIIlS Sln,le: b = 5'.1 = .S' n"' 21 Low-,ielJ: bt' "'"'" s lW.I CONIJHOIU _Sin,le : b = 'flS = .2' 1W.1 CONI 1ROIILS "Sin,le: b = 't' =.'fS n .. 27 s t&""" """" a a v JWJ: CONIIIIOIIS D8B 0811 1987 1988 BIEAVER DBR l989 JWJ: CONIJHOIU ___________ __, Sin,le : b = SS.7 = .'tl n .. Z1 Lo"w-,ielJ: b a2tf r' ... 9, Sin,le: b='U r.z.=.'fO n::oZ7 Sin,le: b=7U r.z.=.lf-3 n:o Z7 b= ra...o n ttl Jlijh-'jic.IJ: b=IIO r:a ... 86 n.,., ID Low-,rieiJ: b=JaS r1=.7l n .. ro Low-,ieiJ: ba2t.7 r&..7Z s "a./0 a a -OBH 1989 1987 1988 SUGARLOAF Fig. 3.3. Linear regressi6n analysis of of male cone whorls on Regression lines for single trees (S) had significantly higheT slopes than regression lines for low-yield cluster members (LY). High-yield cluster members (HY) regression lines are shown when they were significantly different in slope from low-yield regression lines. For Beaver in 1989 the low-yield reqression did not have a significant Pearson's r and is not shown.


Mean production of female cones by single trunk trees was higher than that of low-yield tree cluster members in all three years in the Rainbow stand but not significantly (Fig. 3.4). Although not statistically significant, it is noteworthy that when female cone production per genet is summed over three years, single trunk trees produced twice as many cones as low-yield cluster members (Fig. 3.4). In 1987 mean seed weights and mean maximum seed weights of single trunk trees and high-yield cluster members were higher than those of low-yield tree cluster members but not significantly (Table 3.2). In 1988 mean seed weights were significantly higher (t= -2.09, df= 17, p= .026) and mean maximum seed weights were also significantly higher (t= -2.25, df= 17, p= .019) for single trunk trees than for low-yield members. In both years percentage of filled seeds did not differ significantly among groups. Single Trunk Trees and Tree Clusters: Dbh, Height and Age Regression analysis indicated that for single trunk trees regressions of dbh on age were highly significant in all three stands (Table 3.3). When data for all single trunk trees at the three stands were pooled the 22


I\) (.,) MEAN FEMALE CONES/GENET 40 -I 301 27 20 1---10 1--5 27 5 5 0 ..........__ 1987 1988 1989 3-YEAR TOTAL YEARS -SINGLE TRUNK TREES HIGH-YIELD C.M. LOW-YIELD C.M. FIG. 3.4. Mean production of number of female cones per tree by single trunk trees and high and low-yield tree cluster members. Numbers above bars are sample sizes.


Table 3.2 Seed Weight and Percentage of Filled Seeds of Single and Multi-trunk Trees and Tree Cluster Members. Variable Single Multi-High-yield Low-yield and trunk trunk cluster cluster Year tree tree member member (A) (B) (C) (D) F-valueb mean seed weight ( Q)C 1987 .062 (24) .052 (8) .057 (4) .052 (4) 1. 79 1988 .086 ( 17) .070 ( 7) .083 (3) .062 ( 2) 3. 11 mean maximum seed wt. (g) 1987 .090 .075 .088 .078 1.80 1988 .120 .098 117 .093 3.50* percentage filled seeds 1987 46" 64" 54" 50" 1. 19 1988 67" 72" 82" 80" 0.59 a All c0111par-1eona using One-Way AIIOVA b p (,05 c Number-a tn par-anthaaea ar-e sample atzea 24


Table 3.3. Regression of Height and Diameter (dbh) on Age for Single Trunk Trees and Tree Cluster Members. Regression Regression Stand coefficient df r2 p (A) Low-yield cluster trees dbh on age Sugarloaf .238 3 71 NS Beaver 127 10 .90 <.00001 Rainbow 134 3 .92 .0113 All stands .130 20 .86 <.00001 Height on age Sugarloaf 4.51 3 .56 NS Beaver 3.89 10 .77 00017 Rainbow 4.28 3 .79 .041 All stands 4.18 20 .72 (.00001 (B) High-yield cluster trees dbh on age Sugarloaf 162 4 .77 .0201 Beaver 131 11 .79 .00005 Rainbow .192 3 .66 NS All stands 142 22 .76 <.00001 Height on age Sugarloaf 5.31 4 .85 .01005 Beaver 4.04 11 .66 .0008 Rainbow 4. 14 3 .62 NS All stands 4.43 22 .64 <.00001 (C) Single trunk trees dbh on age sugarloaf .196 9 .86 .00006 Beaver 193 7 .69 .0051 Rainbow 167 9 61 .0008 All stands 186 29 71 <.00001 Height on age sugarloaf 3.27 9 .50 .015 Beaver 3.27 7 .53 .026 Rainbow 3.36 9 .72 .0008 A 11 stands 3.28 29 .52 (.0001 25


regression was very highly significant with .71 as the proportion of variation in dbh explained by the regression (r2 the coefficient of determination). Lowyield and high-yield tree cluster members showed significant regressions of dbh on age in two of three stands for both groups (Table 3.3). When stands were combined the linear regression for both low-yield and high-yield trees were very highly significant, and their coefficients of determination were higher than that for single trunk trees (r2 = .86 and .76 for low-yield and high-yield trees respectively). The regression coefficient for high-yield tree cluster members was between that for low-yield members and single trunk trees and did not differ significantly from either (Table 4). However, the regression coefficient for low-yield cluster members was significantly lower than that for single trunk trees (p< .025), indicating that tree cluster members typically have smaller diameter stems than single trunk trees of the same age, the size gap increasing with tree age. Regression of height on age for single trunk trees was also highly significant in all three stands, and coefficients were not significantly different among the stands (Table 3.3). When stands were combined the regression was very highly significant. Regressions of 26


height on age for low-and high-yield tree cluster members were significant at two of the three stands. When stand data were combined regressions for low-and high-yield members were highly significant and the regression coefficients of both groups of cluster members were higher than that for single trunk trees (Table 3.3). Although mean dbh for single trunk trees was not significantly different among the three stands for any group, mean tree height in the Rainbow stand for single trunk trees and low-and high cluster members were significantly less than the mean height of these groups in the Sugarloaf and Beaver stands (Table 3.4). Rainbow stand trees are located near treeline and their shorter stature is due most probably to exposure to greater wind speeds and lower temperatures than trees in the low elevation stands. Table 3.4. Mean Dbh and Height of Single Trunk Trees and Tree Cluster Members. Tree type Sugarloaf Beaver Rainbow p Single trunk trees dbh (em) 23.6 22.5 21.2 NS height (m) 7.43 7.00 5.38 <.01 Low-yield c. m. dbh (em) 17.0 21.7 16.4 NS height (m) 7.00 8.40 5.13 .05 High-yield c. m. dbh (em) 18.9 22.9 18.9 NS height (m) 7.95 8.63 6.15 .05 a All comparfwonw uwtng Kruakal-Wallfe One-Way ANOVA by Renka 27


Reproductive Output per Genet: Comparisons Among Years In all three stands there was significant variation in reproductive output among years for single trunk trees and tree clusters (Figs. 3.2 and 3.4). At Sugarloaf, 1989 male cone production by single trunk trees was almost double that of 1988 production; at Beaver, 1988 male cone production by single trunk trees was three times that of 1987 production; and, at Rainbow, 1987 female cone production by single trunk trees was significantly higher than in the subsequent two years. Low-and high-yield cluster members also varied from year to year in male and female cone production but with smaller amplitude. Seed weight and percentage of filled seeds also varied among years for single trunk trees (Table 3.2). Seed sample sizes for cluster trees were too small for statistical analysis. Although 1987 was a good production year for female cones, mean seed weight was significantly less than seed weight in 1988, .a relatively poor female cone crop year (t= -5.14, df= 39, p< .00001). In addition, mean maximum seed weight in 1987 was also significantly lower than 1988 maximum seed weight (t= 28


5.48, df= 39, p< .00001}. This pattern holds true for multi-trunk trees as well. In 1988 a single trunk tree's percentage of filled seeds was significantly and negatively correlated with the total number of male cones produced by the tree in the previous year and also the number of female cones produced in the current year (Table 3.5). Filled seed percentages in 1987 were also negatively correlated with number of female cones but not significantly. The female cone crop in 1988 was also positively correlated with the tree's male cone crop for the previous year. Possibly, abundant pollen production may produce selfed seeds, which abort in higher numbers. In fact, Smith et aT. (1988) studying Pinus contorta, found a negative Table 3.5. Pearson Correlation Coefficients (r) for Comparisons Between Male or Female Cone Production 1n 1987 and 1988 and Percentage of Filled Seeds in 1988. All correlations are significant at p< .05. male cones/genet female cones/genet 1988 1987 male cones/genet 1.00 1987 male cones/genet 1988 female cones/genet 1988 per cent seeds filled 1988 .82 .69 -.71 1988 1.00 NS 1.00 NS -.78 29 per cent seeds filled 1988 1.00


correlation between an individual tree's pollen production and its percentage of filled seeds and attributed this result to higher frequencies of selfpollination in trees with large numbers of male cones. Multi-trunk Trees: Cone and Seed Production In five out of eight comparisons multi-trunk trees produced more male cone whorls per tree than single trunk trees, but the differences were not significant (Table 3.6). Yearly female cone production by single and multitrunk trees at Rainbow also did not differ significantly. When male cone production was over three years (two years at Rainbow) the production of multi-trunk trees did not differ significantly from that of single trunk trees in all stands. Three year production totals of female cones also did not differ significantly between the two growth forms (Table 3.6). In 1988 single trunk tree seed weights were significantly higher than those of seeds produced by multi-trunk trees (t= -2.46, df= 22, p= .011) and maximum seed weights of single trunk trees were also significantly higher (t= -3.19, df= 22, p= .002). Mean seed weight and mean maximum seed weight were also higher but not significantly for single trunk trees than for 30


Table 3.6 . Stand/Year Sugarloaf 1987 1988 1989 3-year total Beaver 1987 1988 1989 3-year total Rainbow 1987 1988 2-year total Rainbow 1987 1988 1989 Yearly Production of Mean Male Cone Whorls and Female Cones by Single and Multi-trunk Trees. N 27 27 27 17 17 9 20 20 20 20 20 Single trunk trees N Multi-trunk trees 735 733 1285 2753 1324 4322 1145 6791 413 948 1361 59.0 21.3 9.9 male cone whorls/genet 7 7 7 5 5 4 11 11 female cones/genet 11 11 11 1098 1050 1435 3583 1711 4240 917 6868 405 1073 1478 45.4 25.7 12.5 3-year mean total 90.2 83.6 a Sample atz and ramale conee/ganat ror etngle trunk tr are dtrrarent rrom thoee reported fn Table 2. In ordar that mean dbh ror atngle trunk tr did not dtrrar atgntrtcantly rrom mean dbh ror multt-trunk treee eome etngle trunk treae were removed ror the comparteone. 31


multi-trunk trees in 1987 (Table 3.2). Although evidence is too meager to ascribe reproductive advantages to either growth form, seed size and weight are a consideration for pine seed predators and dispersers such as squirrels and Clark's Nutcracker. (Lanner 1982; Vander Wall 1987; Tomback 1988). Given that seed predators can know and remember the differences between trees, single trunk trees with their higher weight seeds may be foraged in preferentially. Single trunk trees might also be preferred over tree clusters since low-yield cluster members produce lower weight seeds than single trunk trees (however, predators may forage preferentially in high-yield members in the cluster). Height/Diameter Ratios In the Sugarloaf and Beaver stands bo"th high-yield and low-yield cluster members had significantly higher mean height/dbh (h/d) ratios than either single trunk. or multi-trunk trees (Table 3.7). The.h/d ratios for Rainbow stand tree cluster members were also higher, though not significantly, than those of co-stand solitary trees of both growth forms. Lack of significance may be due to the shorter stature of Rainbow stand trees, which reduces mean height differences among groups. In 32


Table 3.7. Height/Diameter Ratios of Single and Multi-trunk Trees and Tree Cluster Members. Stand Sugarloaf Beaver Rainbow Single trunk tree (A) 33.0 34.8 26.9 Multitrunk tree (B) 33.4 29.9 26. 1 Low-yield cluster member (C) 46.7 40.5 31.6 High-yield cluster member (D) p 42.5 38.5 28.3 .0006 .0007 .004 .002 .004 .029 .004 .010 NS (A,C) (A,D) (B,C) (B,D) (A,C) (A, D) (B,C) (B,D) a Sfngle trunk, multf-trunk and efther hfgh-yfeld or low-yfald cluster member treaa ware compared uefng Kruakal-Wallfa One-Way ANOVA by Ranka. When dfffarancae among the three groupe were efgnfffcant the WflcoMon Rank-Sum Teat for Two Oroupe wee uaed for comparfeon between groups (rfght column). 33


all stands h/d ratios were in the order (c.m.= cluster member): low-yield c.m. > high-yield c.m. >> solitary tree Differences between low and high-yield trees were not significant. Ratios of single trunk and multi-trunk trees were similar to each other and not significantly different. Greater h/d ratios indicate that either height growth has been faster or diameter growth slower in cluster members than in solitary trees (or possibly both are occurring). Linear regression of height on age provides some evidence that height growth may be a factor. Regression coefficients of tree cluster members were higher than that for single trunk trees; however, not significantly different (Table 3.3). Other evidence indicates that unusually slow diameter growth was the primary cause of higher ratios for tree members (Table 3.8). When single trunk trees and tree cluster members of the same age were compared, neither height nor mean cumulative height growth (cm/yr) were significantly different, while mean cumulative diameter growth of cluster members was significantly less than that of single trunk trees (Table 3.8). Slower cambial growth has resulted in smaller diameter trunks in cluster members than equivalent-age single trunk trees. 34


Table 3.8. Mean Age, Growth Rates, and Height/Diameter (dbh) Ratios of Single Trunk Trees and Tree Cluster Members (All Stands Combined). See text for N Single trunk trees (A) 30 Age (yr) 101 (46.9) DBH (em) 24.6 (10.9) Height (em) 711 (217) Diameter 0.26 (.09) growth (em/yr) Height 8.2 (3.4) growth (em/yr) Height: 32.0 (11.8) Diameter ratio Tree clusters High-yield members (B) 24 103 (56.6) 21.4 (9.27) 780 (314) 0. 23 ( .07) 8.7 (2.9) 37.5 (7.42) Low-yield members (C) 22 p 101 (53.3) NS 19.3 (7.49) .05 (A,C) 737 (262) NS 0.21 (.05) .014 (A,C) 8.3 (2.9) NS 39.7 (8.92) .001 (A,B) .0002 (A,C) a All compartsone uetng Kruekal-Kallta One-Kay ANOVA by Ranke (overall compartaona), and Wilcoxon Rank-SUN Teat for Two Oroupa. b Numbera tn paranthaeae are Standard Devtatton. 35


General Discussion In the Front Range of the Colorado Rocky Mountains P. f7exi7is occurs predominantly in xeric and windy sites which are often steep rocky slopes or exposed ridgetops from middle elevations to treeline (Peet 1978; Hess and Alexander 1986; Peet 1988). Soils are usually poorly shallow, and coarse-textured Entisols (Hess and Alexander 1986). The extreme drought-tolerance of the pine allows it to dominate and often remain as the climax species on very xeric sites (Peet 1988). In fact, P. f7exi7is has been classified as a species exhibiting a stress-tolerant strategy (McCune 1988). At all these environmentally-extreme locations productivity is low and tree growth is very slow (Hess and Alexander 1986). Lepper (1974) conducted an ecophysiological study of 54 P. f7exi7is stands distributed over much of the species' range and concluded that the pine is limited by competition and persists only in extreme sites not colonized by other tree species. This leads to the hypothesis that tree cluster members, already growing in an inhospitable environment and probably subjected to additional competition for light, water and/or nutrients, 36


should demonstrate symptoms of stress in both morphology and reproductive-output. Slower radial growth by cluster members is the probable result of sustained water deficits during the growing season. Cambial growth in conifers is extremely sensitive to water stress. In semiarid climates up to of the variation in the width of annual rings was attributed to water stress (Fritts et a1. 1965). For many trees radial growth is usually more affected than height growth by water deficits. For Pinus species, height growth normally stops in early spring before soil moisture is depleted, while diameter growth continues through the summer (Kozlowski 1962). In addition, cambial growth has lower priority in terms of carbon allocation than branches and new foliage, new roots, and storage (Waring and Schlesinger 1985). Repressed radial growth and its release after thinning is common in dense stands of trees. Higher levels of soil moisture for thinned or widely-spaced trees have been documented in many studies, especially of pine plantations (see Zahner 1968 for numerous citations). Pinus f1exi7is at timberline near Niwot Ridge, Colorado, showed abrupt radial growth release after a forest fire occurred near the stand (Kienast and Schweingruber 1986). Tree clustering in limber pine may not be completely 37


analogous to dense forest stands, since individual trees in clusters are probably more closely-spaced and clusters more widely-spaced than would be the case for trees in a dense stand: nonetheless, rhizosphere competition for water seems the most likely cause of slower cambial growth. If single trunk trees are considered the typical limber pine growth form, then clustering has apparently caused a delayed maturation of form in tree cluster members, especially low-yield cluster members. For single trunk trees there is a significant negative correlation between height/diameter ratios and age (df =29, r= -.39, p <.05) caused by the greater diminution of height growth than diameter growth as trees age (Oliver and Larson 1990). The h/d ratios of cluster members are similar to those of younger solitary trees and are significantly greater than those of older and multi-trunk trees. Levels of reproductive output of cluster members are also similar to that of younger trees. The phenomenon of suppressed, often dwarfed trees growing under dense forest canopies is given the descriptive term "advance regeneration" in forestry (Oliver and Larson 1990). Low-yield tree cluster members are suppressed by cache cohorts many of which are probably half or full siblings (Tomback 1988: Tomback and 38


Linhart 1990; Schuster and Mitton, in 1991). For such trees suppressed by siblings rather than an older generation I propose the type name "cinderella." While oskars may be able to exploit the opening of canopy gaps to eventually grow to canopy height, it is unlikely that many cinderella trees are released since the dominant cluster member must die while leaving suppressed cluster members unharmed. It has been suggested that a clumped pattern of distribution may outweigh the negative effects of competition. Tomback and Linhart (1990) speculate on possible benefits, including increased cross-pollination and the opportunity for root-grafting, which could enhance nutrient and water absorption and provide better anchorage. Tolerance for crowding could evolve by kin se 1 ect ion among seed 1 i ngs ( Tomback and L i n.hart 1990; Schuster and Mitton 1991). Possibly, the effects of competition may be attenuated at xeric or marginal locations where seedlings and trees grow extremely slowly (Tomback 1988). The large aggregated canopy of a tree cluster may attract nutcrackers preferentially leading to greater dispersal of seeds from clusters than from solitary trees. Finally, a tree cluster, consisting of two or more genotypes growing in close proximity, is a virtual genetic mosaic "individual" that may resist 39


pathogens and parasites more effectively than a single genotype (Whitham et a1. 1984). The data may indicate some increase in crosspollination in tree clusters as evidenced by higher filled seed percentages for tree cluster members than for isolated trees (Table 3.2). The differences were not significant, however, and sample sizes were small: A definite answer requires additional observations. Rootgrafting or even entwined roots may enhance tree cluster anchorage, but this advantage is probably offset by an increased risk of stems tipping in the wind or buckling under their own weight, phenomena often occurring in trees with high h/d ratios (Halle et a1. 1978; Oliver and Larson 1990). If kin selection operates among tree cluster members then one should expect to see a restraint of competition and altruism among members (Hamilton and 1964b; Nakamura 1980). The high h/d ratios of cluster members imply that long-term competition for resources has occurred among members, and the poor reproductive output of low-yield trees may indicate a cost to their fitness. However, low-yield members may be increasing their inclusive fitness by 1) partially restraining their intake of water and nutrients to allow related highyield members more resources for reproduction, or, if 40


root-grafting and translocation occur, by 2) allocating a disproportionate amount of resources to the production of roots, foliage, or secondary metabolites, the profits from which, in terms of extra water, nutrients, photosynthate, or defensive compounds, would be shared with high-yield members a reproductive division of labor (Nakamura 1980). It can be argued that data in Table 3.1 suggest that active altruism may not be profitable enough for the donor. Column A of Table 3.1 gives mean summed production of male and female cones by tree clusters. If kin selection is operating in clusters then it is the summed production that is relevant to the inclusive fitness of the donor: its own contribution plus that of its cluster-kin group members. In only two instances were mean cluster production totals significantly greater than mean production by single trees. However, for kin selection to according to theory, the summed reproductive output of cluster members must greatly exceed that of solitary trees (Hamilton 1964ab). If cost:benefit ratios are assessed in terms of male and female reproductive output an altruistic cluster member does almost as well in fitness gains by dying, as by contributing to cluster-kin group reproduction. Thus, tolerance of crowding is probably not the result of kin selection in P. f7exi1is. 41


This study has demonstrated that most tree cluster members have an abnormal morphology indicative of suppressed growth, and that subdominant members suffer a loss of fjtness as measured by reproductive output. The total production of male or female cones by two or more trees in a cluster in most years only matched the output of a dispersed single tree. The apparent optimal dispersal for limber pine is in single-seed caches, a requirement which conflicts with Nutcracker seed caching habits (Vander Wall and Balda 1977; Tomback 1978; Tomback 1980; Hutchins and Lanner 1982). Comparisons among limber pine single trees, multi-trunk trees and tree clusters indicate that: 1. The hypothesis that tree clusters are reproductively disadvantaged in comparison to single trees is correct. 2. Multi-trunk trees have no clear reproductive advantage or disadvantage over single-trunked trees. 3. Kin selection or other hypothesized benefits of clustering do not offset the lowered reproductive output of cluster trees; and thus 4. There is a cost of mutualism for limber pine that is exacted in terms of lowered fitness for some trees. 42


CITATIONS Bailey, N. T. J. 1981. Statistical methods in biology. John Wiley and Sons, New York, New York, U.S.A. Carsey, K. s., and D. F. Tomback. In prep. The relationship between growth form and elevation in limber pine in the Colorado Front Range. Thesis. University of Colorado at Denver, U.S.A. Fritts, H. c., D. G. Smith, J. W. Cardis, and c. A. Budelsky. 1965. Tree-ring characteristics along a vegetation gradient in northern Arizona. Ecology 46:393-401. Furnier, G. R., P. Knowles, M.A. Clyde, and P. Dancik. 1987. Effects of avian seed dispersal on the genetic structure of whitebark pine populations. Evolution 41:607-612. Halle, F., R. A. A. Oldeman, and P. B. Tomlinson. 1978. Tropical trees and forests. Springer-Verlag, New York, New York, U.S.A. Harper, J. 1977. Population biology of plants. Academic Press, New York, New York, U.S.A. Hamilton, W. D. 1964a. The genetical evolution of social behavior. I. Journal of Theoretical Biology 7:1-16. Hamilton, W. D. 1964b. The genetical evolution of social behavior. II. Journal of Theoretical Biology 7:17-52. 43


Hess, K., and R. R. Alexander. 1986. Forest vegetation of the Arapahoe and Roosevelt National Forests in Central ColOrado: a habitat type classification. Research paper RM-266. Rocky Mountain Forest and Range Experiment Station, Ft. Collins, U.S. Department of Agriculture. Hutchins, H. E., and R. M. Lanner. 1982. The central role of Clark's Nutcracker in the dispersal and establishment of whitebark pine. Oecologia (Berlin) 55:192-201. Kamil, A. C., and R. P. Balda. 1985. Cache recovery and spatial memory in Clark's Nutcrackers (Nucifraga co7umbiana). Journal of Experimental Psychology: Animal Behavioral Processes 11:95-111. Kienast, F., and F. H. Schweingruber. 1986. Dendroecological studies in the Front Range, Colorado, U.S.A. Arctic and Alpine Research, 18:277 -288. Kozlowski, T. T. 1962. Photosynthesis, climate, and tree growth. Pages 149-170 in T. T. Kozlowski, editor. Tree growth. The Ronald Press, New York, New York, U.S.A. Lanner, R. M. 1982. Adaptations of whitebark pine for seed dispersal by Clark's Nutcracker. Journal of Forestry 78:637-639. Lepper, M. G. 1974. Pinus f7exi7is James, and its environmental relationships. University of California, Davis, California, U.S.A. Linhart, Y. B., and D. F. Tomback. 1985. Seed dispersal by nutcrackers causes multi-trunk growth form in pines. Oecologia 67:107-110. McCune, B. pines. 1988. Ecological diversity in North American American Journal of Botany 75(3):353-368. Nakamura, R. R. 1980. Plant kin selection. Evolutionary Theory 5:113-117. Oliver, C. D., and C. C. Larson. 1990. Forest stand dynamics. McGraw-Hill, Inc.New York, New York, U.S.A. 44


Peet, R. K. 1978. Latitudinal variation in southern Rocky Mountain forests. Journal of Biogeography 5:275-289. Peet, R. K. 1988. Forests of the Rocky Mountains. Pages 65-101 in Barbour, M.G., and W. D. Billings, editors. North American terrestrial vegetation. Cambridge University Press, New York, New York, U.S.A. Schuster, W. S. F., D. L. Alles, and J. B. Mitton. 1989. Gene flow in limber pine: evidence from pollination phenology and genetic differentiation along an elevational transect. American Journal of Botany 76(9):1395-1403. Schuster, W. S. F., and J. B. Mitton. 1991. Discerning individuality: genetic analysis of multi-trunk trees and fused clusters of trees in limber pine. Journal of Heredity, in press. Shea, K. L., and M. c. Grant. 1986. Clonal growth in spire-shaped Engelmann spruce and subalpine fir trees. Canadian Journal of Botany 64:255-261 . Smith, c. c., J. L. Hamrick, and C. L. Kramer. 1988. The effects of stand density on frequency of filled seeds and fecundity in lodgepole pine (Pinus contorta Doug.). Canadian Journal of Forestry Research 18:453-460. Stern, K., and L. Roche. 1974. Genetics of forest ecosystems. Springer-Verlag, New York, New York, U.S.A. Thompson, J. N. 1982. Interaction and coevolution. John Wiley and Sons, New York, New York, U.S.A. Tomback, D. F. 1978. Foraging strategies of Clark's Nutcracker. Living Bird 16:123-161. Tomback, D. F. 1980. How nutcrackers find their seed stores. Condor 82:10-19. Tomback, D. F. 1986. Post-fire regeneration of krummholz whitebark pine: a consequence of nutcracker seed caching. Madrono 33:100-110. 45


Tomback, D. F. 1988. Nutcracker-pine mutualisms: trees and seed size. Pages 518-527 in H. Ouellet, editor. Acta XIX Congressus Internationalis Ornithologici, Vol. I. University of Ottawa Press, Ottawa, Canada. Tomback D. F., and K. A. Kramer. 1980. Limber pine seed harvest by Clark's Nutcracker: timing and foraging behavior. Condor 82:467-468. Tomback, D. F., L. A. Hoffman, and S. K. Sund. 1990. Coevolution of whitebark pine and nutcrackers: implications for forest regeneration. Pages 118-129 in Proceedings--Symposium on whitebark pine ecosystems: ecology and management of a high mountain resource. USDA Forest Service Intermountain Research Station, Ogden, Utah, General Technical Report, INT-270. Tomback, D. F., andY. B. Linhart. 1990. The evolution of bird-dispersed pines. Evolutionary Ecology 4:185-219. Vander Wall, S. B. 1987. Foraging of Clark's Nutcrackers on rapidly changing pine seed resources. The Condor 90:621-631. Vander Wall, S. B., and R. P. Balda. 1977. Coadaptions of the Clark's Nutcracker and the pinon pine for efficient seed harvest and dispersal. Ecological Monographs 47:89-111. Waring, R. H., and W. H. Schlesinger. 1985. Forest ecosystems. Press, Orlando, Florida, U.S.A. Whitham, T. G., A. G. Williams, and A.M. Robinson. 1984. Individual planti as temporal and spatial mosaics of resistance. Pages 15-51 in P. W. Price, C. N. Slobodchikoff, and W. S. Gaud, editors. A new ecology: novel approaches to interactive systems. J. Wiley and Sons, New York, New York, USA. Zahner, R. 1968. Water deficits and growth of trees. Pages 191-254 in T.T. Kozlowski, editor. Water deficits and plant growth: Vol. II. Academic Press, New York, New York, 46