Citation
Walking with Lucy

Material Information

Title:
Walking with Lucy modeling mobility patterns of Australopithecus Afarensis using GIS
Added title page title:
M odeling mobility patterns of Australopithecus Afarensis using GIS
Creator:
McPherson, Rachel ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file : ;

Subjects

Subjects / Keywords:
Gait in humans ( lcsh )
Walking ( lcsh )
Geospatial data -- Modeling ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
Behavior is perhaps the most challenging component of an extinct organism to reconstruct and understand. Often in paleoanthropology, researchers primarily have fossils and paleoecological data; however, combining these into models of hominin behavior is difficult in practice. Yet for years archaeologists and wildlife biologists have been using Geographic Information Systems (GIS) to model the mobility behavior of humans and other animals. This research seeks to integrate the methodology of cost-distance modeling in GIS into paleoanthropology to understand hominin mobility, specifically investigating if the potential mobility pattern of Australopithecus afarensis can be modeled to understand how they got across Eastern Africa to their known sites. The models created for Au. afarensis, humans, and chimpanzees brought together walking time as a cost factor and modern slope as an impediment to movement. These values were input into the Cost Distance tool in ArcGIS with Laetoli as the source and tested on two study areas, Laetoli and Eastern Africa. Known Au. afarensis sites matched areas of least cost for each potential mobility pattern, which indicated that 1) none of the models could be ruled as the best potential mobility pattern for Au. afarensis, 2) Au. afarensis likely avoided steeper gradients, and 3) modern gradient data were not incompatible with the models. Despite limitations to this study, these models provide a foundation for research into hominin mobility patterns using GIS.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Rachel McPherson.

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:
on10611 ( NOTIS )
1061149836 ( OCLC )
on1061149836

Downloads

This item has the following downloads:


Full Text
WALKING WITH LUCY: MODELING MOBILITY PATTERNS OF
A USTRALOPITHECUS AFARENSIS USING GIS
by
RACHEL MCPHERSON B.S., University of Wyoming, 2014
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 Anthropology Program
2018


This thesis for the Master of Arts degree by Rachel McPherson has been approved for the Anthropology Program by
Charles Musiba, Chair Rafael Moreno Deborah Thomas Jamie Hodgkins
Date: December 24, 2017
n


McPherson, Rachel (M.A., Anthropology Program)
Walking with Lucy: Modeling Mobility Patterns of Australopithecus afarensis Using GIS Thesis directed by Associate Professor Charles Musiba
ABSTRACT
Behavior is perhaps the most challenging component of an extinct organism to reconstruct and understand. Often in paleoanthropology, researchers primarily have fossils and paleoecological data; however, combining these into models of hominin behavior is difficult in practice. Yet for years archaeologists and wildlife biologists have been using Geographic Information Systems (GIS) to model the mobility behavior of humans mid other animals. This research seeks to integrate the methodology of cost-distance modeling in GIS into paleoanthropology to understand hominin mobility, specifically investigating if the potential mobility pattern of Australopithecus afarensis can be modeled to understand how they got across Eastern Africa to their known sites. The models created for Au. afarensis, humans, and chimpanzees brought together walking time as a cost factor and modem slope as an impediment to movement. These values were input into the Cost Distance tool in ArcGIS with Laetoli as the source mid tested on two study areas, Laetoli and Eastern Africa. Known Am. afarensis sites matched areas of least cost for each potential mobility pattern, which indicated that 1) none of the models could be mled as the best potential mobility pattern for Au. afarensis, 2) Am. afarensis likely avoided steeper gradients, and 3) modem gradient data were not incompatible with the models. Despite limitations to this study, these models provide a foundation for research into hominin mobility patterns using GIS.
The form and content of this abstract are approved. I recommend its publication.
Approved: Charles Musiba
m


ACKNOWLEDGEMENTS
Many, many thanks to Rafael Moreno for being so helpful mid available, especially with the mundane or silly questions I brought to you. Our conversations were always useful and enjoyable for me. I sincerely hope you find this paper interesting and informative. Deb Thomas, thank you so much for the data, your georeferenced maps were a life saver. Charles Musiba, you kept me motivated to get my work done mid I appreciate that. To my Mams, thank you for being patient with me while I have been pursuing my Masters work and doing basically nothing else. You are the reason I have gotten this far. Connie Turner, you have been amazing. I cannot thank you enough for always keeping me apprised of deadlines, giving me encouragement, and dealing with administrative matters for me. To Liz Sweitzer and Alex Pelissero, thank you so very, very much for editing this thesis mid helping me sound smart. Finally, thank you to all of you anthropology grad students who tolerated my rants about my thesis mid gave advice mid encouragement when I struggled. You are all heroes to this kid.
IV


TABLE OF CONTENTS
CHAPTER
I. THE EVOLUTION OF A MODEL OF MOBILITY.......................................1
Introduction............................................................... 1
Literature Review...........................................................3
Australopithecus afarensis................................................3
How Optimality in Behavioral Ecology Enables Modeling of Hominin Behavior.8
The Complexities of Mobility and How to Render It Simple..................9
The Effect of the Environment on Mobility............................... 13
II. MATERIALS............................................................... 18
III. METHODS.................................................................20
The Role of GIS Cost-Distance Analysis in Models of Mobility...............20
Preparatory Work...........................................................22
Au. afarensis Localities.................................................23
Slope, The Environmental Variable........................................23
Relative Walking Speed, The Cost.........................................25
Potential Mobility Models..................................................29
IV. RESULTS..................................................................33
Laetoli...................................................................33
Eastern Africa.............................................................36
V. DISCUSSION AND CONCLUSIONS...............................................40
v


Limitations.................................................................39
Future Directions...........................................................42
Conclusions.................................................................43
REFERENCES....................................................................47
vi


LIST OF TABLES
TABLE
1: Speed and friction values on slopes for humans, chimpanzees, and Au. afarensis.28
vn


LIST OF FIGURES
FIGURE
1: Select paleoanthropological sites where Au. afarensis fossils have been found.....4
2: All localities recorded at Laetoli, Tanzania overlaying a Google Earth image....243
3: A: Digital elevation model of the Laetoli area. B: Slope created from the DEM....24
4: Flowchart of the Laetoli cost distance analysis..................................28
5: The cost raster of the Au. afarensis intermediate model created from reclassifying the
slope raster............................................................................31
6: A: The Digital Elevation Model created from merging the SRTM 1 Arc-Second Global
DEM files of the study area. B: The slope raster created by using the slope tool on the DEM. C: A slope raster reclassified into the values for the Intermediate Au. afarensis
model................................................................................32
7: The mobility models showing effective distance from each locality at Laetoli.........34
8: Mobility models for Eastern Africa showing effective distance from Laetoli...........37
9: A: Intermediate Model of mobility with the East African Rift System overlaid, showing
that the EARS is mostly along areas of lower cost but is often bounded by speckles of higher cost. B: Satellite imagery of Eastern Africa showing the topography and how it maches with the EARS and select Am. afarensis localities.......................38
vi n


CHAPTER I
THE EVOLUTION OF A MODEL OF MOBILITY Introduction
Between ~ 3.85 2.9 million years ago in Eastern Africa, the species Australopithecus afarensis (an early human ancestor) were traversing the landscape, leaving hints of their presence in the form of fossils mid footprints in volcanic ash. How did this small-bodied species get to areas in Ethiopia, Kenya, and Tanzania? Many have tried to find the answer by modeling the locomotor mechanics of their skeletons or reconstructing their environment, which has led to several exciting revelations about the way they walked and the environment in which they lived (Andrews and Bamford 2008; Selim's et al. 2005; Stem and Susman 1983). For instance, many researchers agree that like humans, Au. afarensis was a habitual, efficient biped (Sellars et al. 2005) that could live in a variety of environments as a generalized feeder (Kimbel and Delezene 2009). However, this information by itself does not suggest how Au. afarensis got from one location to another. Can the potential mobility pattern of Au. afarensis be modeled to understand how they got to their known sites?
How did behaviors common in humans, such as mobility pattern, evolve mid when? To begin to answer this question, we must understand the behavior of our hominin relatives. When investigating this of extinct hominins, researchers often have little more to go on than fossils mid ichnofossils, such as footprints (Harrison 2011; Kimbel and Delezene 2009; Sellars et al. 2005). These must be used to parse out behavior using whatever methodologies are available. Cost distance modeling is a methodology often used by archaeologists and wildlife biologists when investigating how humans or other animals move from one place to another (Davidson et al. 2013; Rahn 2005). Yet this form of investigation has made little
1


headway into paleoanthropology (but see Egeland et al. 2010), and there has been little progress towards revealing the potential mobility pattern of Au. afarensis that enabled them to get from one end of Eastern Africa to the other. The use of cost-distance modeling for Au. afarensis will bring researchers one step closer to modelling the potential mobility behavior of that species mid open up a new avenue of inquiry for scholars of other hominins.
This research sought to model the potential mobility pattern of Au. afarensis.
Potential mobility here is walking activity as it is affected by environmental conditions. The research question is built on four parts; the methodology used for the models, the walking activity modeled, the environmental condition(s) included, and the underlying assumptions. Cost-distance modeling in a GIS was the method used to model potential mobility. Relative walking time was used as a proxy for walking activity. The environmental condition adopted was slope, but due to lack of precision in paleoenvironmental reconstructions, modem elevation data was used. A key assumption is that the potential mobility pattern will be the optimal one that requires less time expenditure, mid therefore is one that avoids steeper slopes. If these assumptions hold tme and the potential mobility pattern is correct, then known sites should he in areas that would take Au. afarensis less time to get to.
This paper is divided into several parts. First, the subjects of Au. afarensis, behavioral ecology, use of models, and the interaction between environment and mobility are reviewed. Included in this section are the justifications for the decisions made during the course of this research. Then, the data acquired to build the models will be presented. Next, the methodology of least cost modeling in general and the models built for this research specifically will be discussed. Two different study areas were utilized to mn the models on, and the results are explained. Finally, the research and limitations are examined.
2


Literature Review
Australopithecus afarensis
Fossils that would later be attributed to Australopithecus afarensis were first collected by Kohl-Larsen in the 1930s in Garusi (Laetoli), Tanzania (Kimbel and Delezene 2009; Puech et al. 1986) but it was not until the International Afar Research Expedition began work at Hadar, in Ethiopia in 1973 that the fossils from both sites were recognized and given a species designation (Johanson et al. 1982). Subsequent work by paleoanthropologists have revealed more than 400Au. afarensis specimens (Kimbel and Delezene 2009) at Hadar, Dikika, and Maka in Ethiopia; East Turkana, Kenya; Laetoli, Tanzania (Kimbel 2007); Woranso-Mille, Ethiopia (Haile-Selassie et al. 2010); Kantis, Kenya (Mbua et al. 2016); and possibly Belohdelie, Ethiopia (Kimbel 2007) (
Fig. 1: Select paleoanthropological sites where Au. afarensis fossils have been
found.: Select paleoanthropological sites where Au. afarensis fossils have been found). Any understanding of Au. afarensis potential mobility patterns must be set in the context of their paleoecology at these sites mid their dietary adaptations, especially because their environment changed over time and space and yet Au. afarensis persisted and spread for nearly a million years (Bonnefille 2010; Kimbel and Delezene 2009).
Only five of the sites are included in this study; Hadar, Dikika, Woranso-Mille, and Laetoli. The oldest known site is in the Upper Laetolil Beds (ULB) at Laetoli, in which Au. afarensis dating to 3.63 -3.85 Mawas recovered (Deino 2011). Ecomorphological analysis, phytoliths, and micromammals further suggest heavy woodland-bushland environments in the older part of the ULB at Laetoli, but after Tuff 5 there is a shift towards more grassland and open woodland with increased aridity and an accompanied change from mainly C3 to
3


mainly C4 grasses (Kovarovic and Andrews 2011; Reed and Denys 2011; Rossouw and Scott 2011).
Wo r a its 0 M i l fe'0 0
Hadar
Kantis
Laetoli
250
Fig. 1: Select paleoanthropological sites where Au. afarensis fossils have been found.
Seasonal streams existed, possibly so did springs in the volcanic highlands and gallery forest in the valleys, but there was no permanent water source (Kovarovic and Andrews 2011).
Kantis has been dated to ~3.5 Ma, and is thought to have been more open and more grassland dominated than the other Au. afarensis localities based on the bovid and suid
4


assemblages (Mbua et al. 2016). However, mammalian dental carbon isotopes reveal similar values to the contemporaneous localities, possibly indicating mosaic grassland, shrubland, and woodland with perennial water bodies (Mbua et al. 2016). As of the date of this thesis, there has not been further published work on Kantis, but perhaps future research will clem' up the confusion of the environmental context at Kantis ca. 3.5 Ma.
Hadar and Dikika are close to each other spatially and temporally, with Dikika located southeast of Hadar across the Awash River. The Hadar Au. afarensis me dated to -2.90 >3.42 Ma (Reed 2008), while Dikika Am. afarensis date to 3.31 >3.4 Ma (Alemseged et al. 2005; Alemseged et al. 2006). Before 3.4 Ma, carbon mid oxygen isotopes indicate Dikika held woodland to grassland habitats, shifting to an open wooded grassland to woodland after 3.4 Ma; however, the proportions of these changed over time while conditions remained wet (Bedaso et al. 2013). Hadm also shows evidence of vegetative as well as climatic change over time (Bonnefille 2010). At 3.4 Ma, the paleoecological evidence at Hadm indicates a woodland mid shrubland mosaic, changing over time to a vmiety of habitats including mosaic woodland mid expanded wetlands along the lake margin (Reed 2008). By 2.94 Ma, Hadar had shifted to a largely shrubland or open woodland habitat. Lake Hadm expanded and regressed with changes in rainfall over time, and the climate went from lacustrine/wetland to seasonal to more mid (Reed 2008).
The last site, Woranso-Mille, is 45 kilometers north of Hadar mid contains Au. afarensis dating to 3.6 Ma and 3.2 3.3 Ma (Haile-Selassie et al. 2016). Faunal analysis indicates that this time period was characterized as wet, being a distal fluvial plain with flooding events mid seasonal dry periods (Haile-Selassie et al. 2016). At the time of the emliest Au. afarensis, habitats were mosaic, ranging from open to closed (Curran and Haile-
5


Selassie 2016). Evidence from faunal assemblages, mesowear, and ecomorphological analyses point to Woranso-Mille having a river surrounded by dense vegetation with more open habitats farther away from the river (Curran and Haile-Selassie 2016).
Fewer specimens of Au. afarensis have been found at Laetoli than at Hadar even when accounting for taphonomic processes such as carnivore activity (Su and Hams on 2008). Su and Harrison (2008) point out that there is a higher proportion of Au. afarensis in the faunal community at Hadar than at Laetoli, but Hadar was also wetter with permanent water sources and was more densely wooded. The same pattern is present in chimpanzee populations, where there is a larger population of chimpanzees in closed woodlands than open. Therefore, the authors conclude Laetoli was a marginal habitat while Hadar contained the denser woodlands in which Au. afarensis was more successful, and had more available food (Su and Harrison 2008). However, all the Au. afarensis sites became progressively more open overtime, with no apparent large consequences for Au. afarensis up to 2.9 Ma, with the exception of a period of lake extension at Hadar that led to a population decline of Au. afarensis at the site during that period (Bonnefille 2011; Reed 2008). Clearly this species was eurytopic (Bonnefille 2011; Reed 2008), although the lack of abundant water may also explain the smaller population density at Laetoli (Su and Harrison 2008).
Molar microwear analysis done on Au. afarensis teeth from Hadar and Laetoli ranging in age from 3.5 3.2 Ma demonstrate that their diet did not change significantly over time (Grine et al. 2006). Preferred foods appear to be soft mid abrasive rather than the hard and brittle items suggested by their robust masticatory apparatus, although this morphology may represent the use of fallback foods (Grine et al. 2006). However, it is impossible to conclude from this evidence if Au. afarensis found its preferred food at different habitats or if
6


it ate a variety of soft, abrasive foods (Su and Harrison 2008). Carbon isotope analysis was performed on Au. afarensis teeth from Dikika and Hadar, indicating that most individuals ate a large amount of C4/CAM foods but the range of carbon values was variable across individuals with no directional change over time or habitat (Wynn et al. 2013). Carbon isotopes from hominin teeth from Woranso-Mille confirm this diet (Curran and Haile-Selassie 2016; Haile-Selassie et al. 2016). Although Au. afarensis may have consumed food with the same mechanical properties, the carbon isotopes point to their ability to consume a range of foods (Wynn et al. 2013).
The variety of environments and climates occurring at these five sites demonstrate that Am. afarensis was likely adapted to variable conditions and was able to persist despite climatic mid vegetative change (Bonnefille 2010; Reed 2008). This alone points to Au. afarensis being a generalist rather than a specialist (Bonnefille 2010), an assertion confirmed by the isotopic and microwear evidence from their teeth (Curran and Haile-Selassie 2016; Grine et al. 2006; Wynn et al. 2013). Clearly Am. afarensis was capable of living in different environments, but how were they able to get from one site to another?
This question is directly related to morphology: Au. afarensis is characterized by a wide range of morphological variation and a mix of bipedal and arboreal characteristics. This includes the broad pelvis, bicondylar angle of the femur, and robust, adducted hallux typical of bipedal humans combined with the barrel chest, long forearms, and curved fingers which are characteristic of more arboreal primates (Kimbel and Delezene 2009). These mixed morphologies can be related to Au. afarensis potential mobility patterns through the use of the theoretical framework of Behavioral Ecology.
7


How Optimality in Behavioral Ecology Enables Modeling ofHominin Behavior
In paleoanthropology, the morphologies and behaviors of hominins are often implicitly assumed to be adaptive for the ecological context in which the hominin is found (Reed 1997; White et al. 2009), an assumption derived from evolutionary ecology (EE) (Foley 1992; Smith mid Winterhalder 1992; Winterhalder and Smith 1992). EE is a theoretical framework of adaptive modes that states that every species evolves adaptations to their environment and these adaptations help the species survive within that environment (Winterhalder and Smith 1992). Environment is defined as that which is external to the species that affects their ability to reproduce mid survive. Behavior is a part of adaptation, the study of which falls under the subset of EE called Behavioral Ecology (BE). EE and BE follow the hypothetical-deductive method of simple model creation and testing against empirical evidence (Winterhalder and Smith 1992). The purpose is to model the costs and benefits of a behavior such that the benefits outweigh the costs and would therefore be selected for (Foley 1992).
The behavior modeled generally maximizes the benefit of overall fitness (Dunbar 2001; Kelly 2007). Fitness is tracked by defining a proxy cost, such as energy or time allocation, that is optimized. The assumption of optimality is inherent within this framework; an organism will pick the optimal behavior given the tradeoff between costs and benefits associated with utilizing that behavior within a set of external constraints (Dunbar 2001; Kelly 2007). In general, behaviors favor optimization, mid this provides a way to model predictions against selective forces (Kelly 2007). However, human behavior is often found to be more complex mid not always optimal because the cultural aspects of humanity often obscure or obviate optimality (Kelly 2007). On the other hand, modern cultural behavior is
8


not thought to appear until well after Au. afarensis went extinct (Marwick 2003); therefore, cultural behavior would not necessarily affect the pursuit of optimality.
The mobility models developed for this project predict that Au. afarensis, chimpanzees, mid humans will avoid relatively more difficult terrain to expend less time on the search for food or mates. They follow BE framework and assume that behavior is optimized by limiting costs in time by avoiding the environmental condition of steep slopes. This is because steep gradients take more time mid energy to climb up or down than flat terrain (Alexander 2002; Terrier et al. 2001). Within the selective forces of the environment, different animals will avoid different terrain elements depending on their mobility pattern. These terrain elements can be modeled within a framework of specific optimality to the capabilities of that particular species, including Am. afarensis, humans, and chimpanzees.
The Complexities of Mobility and How to Render It Simple
Foley (1992) recognized early on that the best way to relate hominin characteristics to selective conditions was to use a modeling approach. To do so, behavioral adaptations must be identified along with the context of costs and benefits. For instance, the locomotor behaviors of hominins and why bipedalism evolved has been modeled within the context of the energetic advantage bipedalism gave hominins over any other form of bipedalism.
Bipedal hominins have been predicted to save more energy than chimpanzees within the same or a larger day range mid with increased body size (Foley 1992).
Despite findings such as these, a few cautions are necessary to mention. First, reality is extremely complex and not every aspect of that complexity can reasonably be modeled (Dunbar 2001). For this reason, models tend to be simplistic, a feature often criticized but necessary because it allows researchers to isolate those variables that are most important for
9


explaining a behavior. Second, models require the researcher to be explicit about his or her assumptions of how a behavior works mid its relationship to the environment (Dunbar 2001; Winterhalder and Smith 1992). Finally, models allow researchers to determine if a behavior is fitness-maximizing, assuming that the model adequately frames reality (Dunbar 2001). For this, the biology of the organism of interest and the processes being modeled must be well understood (Dunbar 2001).
In the fossil record, behavior is studied indirectly mid is therefore modeled within a cost/benefit analysis framework (Foley 1992). Because animal behaviors are generally adaptive for a particular ecological demand, Au. afarensis mobility patterns must have been adaptive as well. It remains especially important to find the ecological context to which it was adapted and model potential mobility patterns within that context. If Au. afarensis was a highly mobile species that moved across the landscape in an optimal way, then a methodology must be used that captures the interplay between environment mid mobility.
There are many approaches to modeling the potential mobility of Au. afarensis, each with its strengths and weaknesses. The comparative method is one often used to understand early hominin behavior, a tradition continued in this research. Au. afarensis are regularly compared to chimpanzees or bonobos, because the further back in time the hominin of interest is, the more similar it should be to the last common ancestor (LCA) of the Pan-Hominin lineage (Foley 1992); a species sometimes estimated to be very chimpanzee-like (Begun 2010; Leonard and Robertson 1995). Yet it is complicated by how derived chimpanzees have become since the LCA (White et al. 2009). Despite issues such as these, comparative studies between fossil hominins, nonhuman primates, mid humans are often used to reconstruct the social life mid behavior of early hominins (Koenig and Borries 2012),
10


in part because the use of analogy and proxies are a key part of paleoanthropological investigations. It is important to determine if the analogies are viable, since the process of elimination will bring researchers closer to the reality.
Analogy of similarities and differences in anatomy sparked interest in modeling the locomotor patterns of Am. afarensis (Pontzer and Wrangham 2004; Selim's et al. 2005). The efficiency of their bipedality and how arboreal they were is a highly debated issue in paleoanthropology (see Pontzer mid Wrangham 2004; Selims et al. 2003; Hunt 1994; mid Stem mid Susman 1983). Often, Am. afarensis mix of bipedal and mboreal chmacteristics are compared to chimpanzees, which have a combination of arboreal mid quadmpedal terrestrial morphologies (Pontzer and Wrangham 2004). Pontzer mid Wrangham (2004) argue that climbing in chimpanzees is maintained by selective pressure in order to increase climbing safety, specifically because their arboreal adaptations incur, rather than decrease, locomotor costs. Similarly, Pontzer and Wrangham (2004) believe that early hominins may have kept their arboreal adaptations to decrease the likelihood of falling from trees, because of environmental change, and because the need for efficient terrestrial locomotion would have created selective pressure for bipedalism, which decreases energetic walking costs (Pontzer and Wrangham 2004). The form and efficiency of the bipedalism Am. afarensis had requires further discussion to elucidate how potential mobility patterns can be modeled.
Stem mid Susman (1983) hypothesized that because of their range of mboreal traits, Am. afarensis would have had bent-knee, bent-hip (BKBH) bipedality similm to chimpanzees when walking on two legs (Sellars et al. 2005). Crompton et al. (1998) found that BHBK would not have been an effective form of locomotion, mid Selims et al. (2005) found that BHBK locomotion nearly doubled the energetic cost of locomotion. However, locomotor
11


cost does not change for chimpanzees between using quadrupedalism versus bipedalism; instead it is mechanics that increase or decrease costs (Pontzer et al. 2014). An alternative hypothesis proposed by Hunt (1994) states that the mixture of arboreal mid bipedal characteristics results from the need to collect fruit by both hanging from trees by their arms and shuffling bipedally on the ground. Sellars et al. (2005) are proponents of a third hypothesis that argues that based on their habitual bipedal characteristies, Au. afarensis would have been habitual, erect bipeds. Au. afarensis had the correct skeletal morphology for efficient and stable erect bipedalism (Selim's et al. 2005). Therefore, this species was most likely a fully competent biped, although shorter mid slower than humans (Sellars et al. 2005). This list of locomotor patterns is by no means comprehensive, these merely highlight the range of patterns hypothesized for Au. afarensis.
This debate about locomotor pattern is critical because assumptions about the type of locomotion and its efficiency greatly affects any model of potential mobility. These debates also demonstrate that testing a single model of Au. afarensis potential mobility would not suffice because the cost of walking is affected by whether they are assumed to walk like slow humans, like chimpanzees on two legs, and so on. Indeed, models should be tested for their sensitivity to walking activity adjustments and against models of other closely related species. Therefore, comparing chimpanzee, Am. afarensis, mid human potential mobility patterns is a critical part of this analysis.
Because locomotion is complex, a proxy must be used in the models to get at potential mobility pattern. For this research, walking speed was the proxy used to represented Au. afarensis humans, and chimpanzees walking activity. Relative walking speeds have been measured for humans on treadmills (Terrier et al. 2001) and in the African savanna
12


(Musiba et al. 1997). The relative speeds of wild chimpanzees have been measured by observation in a natural setting (Leonard mid Robertson 1997; Pontzer and Wrangham 2004). Au. afarensis relative walking speeds were measured indirectly by foot length measurements on footprints (Charteris et al. 1982; Tuttle et al. 1990), and by scaling down human neuromusculoskeletal models to Au. afarensis size (Nagano et al. 2005).
There are problems with all these methods of determining relative walking speed. First, treadmills do not accurately represent natural lands capes and how speeds change due to the nature of the substrate. Second, Au. afarensis was not living in an open grassland so those speeds are probably much faster than they would be through the mosaic of open to closed woodlands mid forests in which this species lived (Andrews and Bamford 2008; Kimbel mid Delezene 2009). Third, when observing chimpanzees from a distance, there is always a chance some behaviors will be missed or the chimpmizee may be lost during pursuit. Fourth, measurements from footprints may inflate or deflate the walking speed since footprint length and stride do not correlate perfectly (Charteris et al. 1982). Finally, the scaling down of a human model to Au. afarensis size assumes that both species have the same neuromusculo skeleton and they move in exactly the same manner. However, although Am. afarensis are generally accepted to be efficient bipeds (Sellars et al. 2005), they have many different morphologies from humans such as a wider pelvis (Kimbel and Delezene 2009). This research attempts to alleviate these problems by averaging walking speeds calculated using several methodologies, assuming it will even out the variance.
The Effect of the Environment on Mobility
Relative walking speed alone cannot capture how an animal moves across a dynamic and vast landscape; different conditions in the environment can have an effect on how fast an
13


animal walks. For example, one of the factors that affects bonobo distribution is large rivers, which create a geographic barrier between bonobo populations, which reduces gene flow (Eriksson et al. 2004). Barriers such as this can be impenetrable, causing isolation and speciation. However, if a continuous habitat exists on either side of a river and it is possible to cross, large mammals are generally capable of traversing it (Eriksson et al. 2004). Additionally, in humans, Hughes et al. (2007) showed with their Stepping Out model that vegetation is a critical factor to human mobility. Clearly, it is important to pinpoint landscape elements that would most affect potential mobility patterns, mid how they would be affected.
When creating a model for a 3.8 Ma hominin, the environmental features may not have been comparable to that of modem humans, so using modem environmental data can be problematic (Bailey et al. 2011). However, the environmental context of 3.8 Ma is difficult to reconstmct with any precision, leading to generalized statements that cover large areas. For example, Andrews and Bamford (2008) reconstmction at Laetoli lists the types of plants that were found but not their quantity or exact distribution. This imprecision is due to taphonomic processes such as time-averaging and poor preservation, which is unavoidable (Bonnefille 2010; Rossouw and Scott 2011).
This is a problem that plagues all environmental variables going back to such a distant age. Paleoenvironmental reconstmctions are precise enough for most purposes, but when using a model that requires more specific geographical data, they are too large scale. For example, this project would benefit from more exact knowledge of density, size, and distribution of trees, grasses, mid shmbs to enable a model of the impenetrable versus open vegetation Am. afarensis would have had to cross or avoid. Additionally, there is little data available describing how the types of vegetation that are documented affect movement of
14


humans, chimpanzees, and Au. afarensis. There are no regional scale data or models on slope or vegetation currently available, so other data were sought for this project.
The scope of this study also did not allow the incorporations of models of water banders such as the locations, sizes, depths, and salinity of rivers or lakes across Eastern Africa. However, they have a recognized role in restricting mobility in bonobos, for example (Eriksson et al. 2004), and in providing habitats and water for Au. afarensis (Kimbel and Delezene 2009; Su and Harrison 2008). While reconstructions of lake locations do exist, these do not include the other important factors that restrict or support movement, such as size and salinity. It is known that the rifting processes affecting the East African Rift System opened and closed lakes, some were saline and others freshwater, and they varied size and depth over time (Tiercelin and Lezzar 2002; Cuthbert and Ashley 2014). This is an important start, and bringing together water attribute data is a viable future direction for research.
With paleoenvironmental reconstructions not feasible at this initial stage, modem environmental data must be used. Slope is an important factor in human mobility because energy expenditure increases with slope (Minetti et al. 2002; Todd and White 2009), although humans tend to adjust their walking pattern to minimize energy costs (Alexander 2002). If steeper slopes also increase energetic expenditure for Au. afarensis, for the assumption of optimality under BE to hold, they would have had to minimize that in some way. Slope may have been important to hominin mobility in other ways; for example, King and Bailey (2006) suggest that the evolution of bipedalism may have been driven by a landscape of rough mid hilly terrain that allowed hominins to efficiently disperse to track resources or preferred habitats (Potts 1998). Slopes mid topography in general may have had
15


a drastic effect on energetic costs and the evolution of bipedalism. Therefore, the effect of slope on behavior should be tested in the mobility model.
Modem elevation data must be used cautiously. Naturally, the landscape does not look exactly the same today as it did in the past. Over time geological processes caused once-active parts of the African Rift to uplift mid become inactive, mid once they are inactive, forces of erosion obscure geological features, causing the topography to change since the time of Au. afarensis (Bailey and King 2011). Despite this, modem elevation data will be used in the potential mobility models at Laetoli mid across Eastern Africa. This approach was inspired by Egeland et al. (2010), who used modem slope data to model a route from Ubeidiya, Israel to Dmanisi, Georgia that hominins potentially would have taken during dispersal events. Egeland et al. (2010) found that their predicted route matched up with known Lower Paleolithic sites. Furthermore, after ground surveys they found 25 new Upper and Lower Paleolithic sites. Egeland et al. (2010) also stated that although paleoenvironmental data is available, it was too coarse for the predictive model, which is an issue discussed above. However, modem digital elevation data with 30-meter resolution can easily be obtained from the EartliExplorer, a spatial database maintained by the USGS. Alternatively, estimates of paleoslopes could be derived from comparisons with active parts of the African Rift (Bailey mid King 2011), but that is beyond the scope of this work.
The literature reviewed above demonstrates that Am. afarensis were eurytopic, living in highly variable, complex environment, and yet they somehow managed to get across Eastern Africa. If Au. afarensis was behaving optimally when moving across the landscape as BE would suggest, then they were likely avoiding costly terrain such as steep slopes to conserve energy. Data from biomechanical work demonstrate how quickly Au. afarensis
16


could have walked, a proxy for locomotor efficiency that changes with slope. What remains is to merge these together into models of potential mobility that predict how .4?/. afarensis would have moved across Eastern Africa if they were behaving optimally. The data incorporated into these models are described in the next section.
17


CHAPTER II
MATERIALS
Potential mobility models were created for Au. afarensis, chimpanzees, mid humans for comparability, since chimpanzees and humans are the closest living relatives of An. afarensis mid extant species mobility may lend insight into the mobility of Au. afarensis. The data required for these models include Digital Elevation Models, the spatial locations of the sites, and the relative speed at which each species walks.
Digital Elevation Models (DEMs) were downloaded from EarthExplorer (online) for areas in Ethiopia, Kenya, Tanzania, and parts of Uganda, Somalia, South Sudan, Sudan, Eritrea, Rwanda, and Burundi. These covered all of the localities and most of Eastern Africa in order to make a continuous surface for analysis.
Shapefiles containing hominin localities were not readily available, but general coordinates of localities are often provided in publications. The coordinates of localities at Laetoli containing Ah. afarensis fossils were obtained from a variety of sources, including Harrison et al. (in press), Ditchfield and Harrison (2011), Harrison and Kweka (2011), and a Garmin Oregon GPS unit that Dr. Charles Musiba used at Laetoli (
Fig. 2. All localities recorded at Laetoli, Tanzania overlaying a Google Earth
image.
Fig. 2 All localities recorded at Laetoli, Tanzania overlaying a Google Earth image). The coordinates of the Eastern African localities of Kantis, Dikika, Hadar, mid Woranso-Mille were obtained from Mbua et al. (2016), Alemseged et al. (2005) Johanson et al. (1982), and a map provided by Dr. Charles Musiba that was then georeferenced, respectively.
Laetoli was chosen as the first study area because more data on Laetoli localities was available to the author than any other site. It is also the second most productive Au. afarensis
18


fossil site and has over 20 localities (Harrison and Kweka 2011; Musiba et al. 2008), making it a good sample area. For the second study area of Eastern Africa, Laetoli was chosen as the source site because it is the oldest known Am. afarensis site (Kimbel and Delezene 2009), mid is therefore possibly at or near the place at which they first evolved. It is also possible that the area of initial evolution is in the Awash region of Ethiopia instead, since there are more localities in Ethiopia (Kimbel mid Delezene 2009) and there is no way to know exactly where and when Am. afarensis evolved. However, the oldest specimen found in that region is 3.58 Ma (Haile-Selassie et al. 2010), which Laetoli predates by 50 270 Kya, making Laetoli the best candidate. Am. afarensis was chosen as the subject because much research has gone into the study of their biomechanics, environment, and behavior, but there is much more to learn. Little has been done to model their actual movement between sites and localities even though it is of great interest. This is partially explained by the resolution; some of the localities at Laetoli are relatively close to each other (Charles Musiba, personal communication October 2017), while the known sites are all over Eastern Africa.
Data of relative walking speed used in this study was obtained for humans from Terrier et al. (2001), Musiba et al. (1997), Alexander (2002), Verb erg et al. (2004), and Moreno-Sanchez et al. (2012) for different degrees of slope. These data were generally collected for humans on treadmills with the exception of Musiba et al. (1997), who collected data on the Hadzabe and Machiguengas. Hunt (1989) mid Pontzer and Wrangham (2004) provided relative walking speeds for chimpanzees in the wild. Finally, Am. afarensis relative walking speeds were provided for Lucy by Nagano et al. (2005) mid for the G1 mid G2/3 trackways by Charteris et al. (1982).
19


CHAPTER III
METHODS
The Role of GIS Cost-Distance Analysis in Models of Mobility
The models proxy for locomotor pattern is walking speed, and for past terrain the proxy is modem elevation data. Bringing these together into potential mobility models is the role of GIS. The models created here emphasize landscape connectivity in that they account for the stmcture of the landscape as well as the mobility of the species (Adriaensen et al. 2003). Least-cost modeling in GIS focuses on ascertaining areas of least cost between points. It provides a way to incorporate proxies for environmental conditions and cost to calculate effective distance. Effective distance is the distance it takes to get somewhere while avoiding hindrances mid utilizing facilitators to movement on the landscape. This is opposed to Euclidean or straight-line distance that ignores real-world terrain phenomena. In least-cost modeling, every grid cell or landscape unit is given a value of friction based on how it helps or hinders the mobility of an animal. This is meant to represent the way the animal experiences the landscape (Adriaensen et al. 2003).
In cost distance analysis, a subset of least-cost modeling, a source layer with the starting point(s) or area(s) of interest and a friction surface composed of user-defined costs to movement, such as walking speed, are the inputs (Adriaensen et al. 2003). These costs or friction values can be based on expert experience in the relevant area, secondary published data from experts, or field data. It is important to make these as accurate as possible because they drastically affect the outcome, and are the link between the spatial information and the mobility of the animal. Every cell value in the output is calculated as the cost to reach any given cell from a source, the average cost to move through that cell, and the average cost to
20


move through the source cell. Movement through cells can happen in any direction, including diagonally for which the square root of two is multiplied by the cost. The least-cost algorithm used in this study was adopted from Adriaensen et al. (2003):
Ni+i = Ni + (ri + ri+i)/2 or
Ni+i = Ni + 2**0.5 + (n + n+i)/2
where Ni is the accumulated cost in cell i, n is the resistance value in cell i, i is the source cell, mid i + 1 is the target cell. The output is a surface of cost from the source with each cell value being the lowest cost to move to it over the friction surface. The output map contains banded patterns where all the cells in the same band have effective distance in the same class. The sensitivity of the effective distance to variation in friction values can be studied to create a robust, validated predictive model (Adriaensen et al. 2003).
While cost-distance modeling is often used to understand mobility in animals and prehistoric populations, generally these go a step beyond least-cost surfaces to making least-cost paths (Davidson et al. 2013; Egeland et al. 2010; Rahn 2005). These allow researchers to predict exactly how a population was getting from one specific place to another, so they assume an origin and destination. These paths can also be ground-truthed to find evidence that the paths are correctly predicting how a population would have moved (Davidson et al. 2013; Egeland et al. 2010; Rahn 2005). However, least-cost surfaces have been used in archaeology to understand accessibility mid the interaction between settlements mid the landscape over a large area (De Silva and Pizziolo 2001; Howey 2007; van Leusen 1998).
The differences are subtle but important. Archaeological least-cost path studies often assume humans have some knowledge about the study area and would have known exactly
21


where they wanted to go. In contrast, Au. afarensis would not have a physical destination such as Hadar in mind, but instead would have moved along areas of least resistance in their search for food and/or mates. This research explored the interaction between terrain and mobility, focusing on Au. afarensis movement across the landscape as a whole and the accessibility of known sites.
Essentially, cost distance analysis allows researchers to develop predictive models of the movement of different species across a landscape while emphasizing different landscape elements (mid their facilitating or hindering properties). Furthermore, it calculates some cost of getting from a source to anywhere in the study area, and allows researchers to gain insight into the species potential mobility patterns. In relation to the models created here, cost distance analysis allowed a visualization of a wave of movement of Au. afarensis, humans, and chimpanzees from Laetoli, Tanzania to the Awash region of Ethiopia while enabling a change in the cost parameters to test sensitivity. In this way, potential mobility patterns of Au. afarensis can be better understood through the use of this modeling technique in GIS.
Preparatory Work
The goal of this research was to explore the possibility of modeling Am. afarensis potential mobility pattern using the method Cost Distance Analysis (CDA) in ArcGIS and modem elevation and walking speed data in order to understand how they got to their known sites. Like the CDA done by Egeland et al. (2010), a match between the model and the distribution of known localities provides some support for the accuracy of the model. Several rounds of analysis were done for humans, chimpanzees, and Au. afarensis for comparison. The first used modem slope data to model potential mobility patterns in the Laetoli area only. The second analysis involved models from Laetoli across Eastern Africa, again using slope
22


data. Known Au. afarensis localities were overlaid on the second analysis to determine if they lie on least cost areas, which would indicate that 1) the potential mobility pattern could be correctly capturing Au. afarensis actual mobility pattern, 2) An. afarensis was moving optimally across the landscape by avoiding steep slopes, mid 3) the terrain has not changed so substantially that it renders modem elevation data useless for these models.
Au. afarensis Localities
Digitizing and projecting the spatial locations of sites was the first component of this analysis. The Laetoli coordinates were mostly in UTMs, but some came as Degrees, Minutes, Seconds (DMS) and had to be converted to UTMs. These were digitized into a feature class and projected in UTM Zone 36S (Fig. 2) because that is the zone in which Laetoli lies mid is most likely the projection used to obtain the spatial data. The accuracy of the Eastern African locational data is unknown because the coordinates are given in degrees and minutes without seconds, rendering them imprecise. However, in the context of the Eastern African analyses, the area is so large that any inaccuracy in the location is not likely to make a significant difference in this analysis. The Eastern African localities were projected into a coordinate system made specifically for the study area. The projection is an Albers Equal Area Conic with a central meridian of 37, which is roughly the center of the study area, and standard parallels of 7 and -1, dividing the study area roughly in one and two thirds. The purpose was to minimize the distance distortion.
Slope, The Environmental Variable
The models were tested using modem slope data alone, as discussed in the literature review. Briefly, although it is probable that the terrain has changed in the last 2-3 million
23


Fig. 2. All localities recorded at Laetoli, Tanzania overlaying a Google Earth image.
years (King and Bailey 2006), there is no detailed data on past slope available. Yet, modern elevation data was successfully used to model hominin mobility from Israel to Georgia (Egeland et al. 2010). These models test if such a thing is possible for more ancient hominins despite the analysis being done on the tectonically active Eastern African Rift System (King and Bailey 2006).
Modern slope data can be derived from Digital Elevation Models (DEMs). The DEMs used were Shuttle Radar Topography Mission (SRTM) 1 Arc-Second Global in GeoTIFF format downloaded from USGS EarthExplorer. For Laetoli, the DEM was projected into UTM Zone 36S because UTMs preserve distance at larger scales, and because
24


the entire Laetoli site is within this UTM zone. Across Eastern Africa, DEMs were merged into one raster using the Mosaic to New Raster tool and the Mosaic tool and were projected into East Africa Albers Equal Area Conic. The slope tool was then used to convert the DEMs to slope percentages (Fig 3).
A B
Fig. 3. A: Digital elevation model of the Laetoli area. B: Slope created from the DEM. Relative Walking Speed, The Cost
Relative walking speed was chosen as the proxy for locomotor pattern based on published data on the walking speed of all of the species included in the analysis. Generally, energy is used as a proxy for locomotor efficiency, but the energetic output for each of the species was reported in different units that would be difficult to convert. In addition, walking speed is often used to understand accessibility in environmental impact studies (Moreno-Sanchez et al. 2012; Verburg et al. 2004), and it is a simple unit of measurement to
25


understand. However, different sources provide different relative walking speed values for each species. As Leonard and Robertson (1997) and Pontzer and Wrangham (2004) demonstrate, relative walking speed is extremely variable in chimpanzees and humans, and is a function of variables other than terrain such as height, sex, mid age. This problem was solved for this study by averaging the walking speeds provided by each source for each species (inclusive of different heights, sexes, and ages).
Secondary published literature was analyzed to determine relative speed data on flat land for humans, chimpanzees, and Au. afarensis. This was provided in either kilometers per hour or it was converted to kilometers per hour from meters per second. For humans, speed on different gradients was also discovered in the literature. Human speed for 0 24.9% slope were derived from Terrier et al. (2001) using their low speed data because it better matches speeds of modem Hadzabe and Machiguengas peoples of Tanzania mid Pem determined by Musiba et al. (1997). For slopes greater than 25%, it is more optimal to zig-zag upslope rather than go straight up; additionally, above 20% slope, cost of walking is 2.5 times that of walking on flat land for humans (Alexander 2002). Therefore, the speed for 25 49.9% were calculated at 2.5 times greater than flat land walking. The excessively difficult slope cutoff was given as 50% by Verberg et al. (2004) and Moreno-Sanchez et al. (2012), and this was adopted for this study such that anything above 50% (26.57) slope was considered an absolute barrier to movement.
While relative walking speed has been estimated on flat terrain for Au. afarensis mid chimpanzees, no speed data on other gradients was found in the literature. This is important because humans are known to walk increasingly slowly with increasing slope (Terrier et al.
26


2001). Furthermore, in order to make relative walking speeds across species on different slopes comparable, relative walking speed must be estimated for each slope category.
To determine the speed of chimpanzees andNw. afarensis on similar slopes to the data for humans, an approach derived from Taylor et al. (1972) was used. These authors state that an animal that is smaller than another uses x times more energy than the larger animal, but the additional energy to lift 1 kg of body weight per meter while running uphill will be similar for both species no matter the size difference (Taylor et al. 1972). If this can be extrapolated onto time rather than energy, then a chimpanzee walking at 2.99 km/hr walks 1.3 times more slowly on level ground than does a human walking 3.8 km/hr (3.8/2.99). Therefore, a chimpanzee may take 1.3 times more time walking on any elevation than humans so all human speeds can be divided by 1.3 to get the chimpanzee walking speed on any given slope. The base speed used for 0 4.9% slope for chimpanzees was the average of those provided for males mid females from Pontzer and Wrangham (2004), which were derived from Hunt (1989).
Au. afarensis calculations presented a unique problem because there is no certainty about how quickly they would have moved in general. Therefore, this species required a sensitivity analysis in which different parameters are used to estimate their relative walking speed. Au. afarensis base speed for 0 4.9% slope was the average of that for Lucy from Nagano et al. (2005) and for the G1 mid G2/3 trackways from Charteris et al. (1982).
Because Au. afarensis is thought by many to be a fully competent biped, the Slow Human model speed on different gradients was calculated by dividing the human speed by 1.79, the amount slowerNw. afarensis walking speed is to human speed (3.8/2.12). However, the walking speed of Au. afarensis may be even slower than chimpanzees, so another calculation
27


was done that was 1.4 times slower than chimpanzees (2.99/2.12), the Slow Chimp model. Lastly, because some paleoanthropologists believe that the locomotor capabilities of Au. afarensis were somewhere in between chimpanzees mid humans, the final speed calculations were derived using the average walking speed of humans and chimpanzees combined, the Intermediate model. All species speeds are given in Table 1.
Table 1: Speed and friction values on slopes for humans, chimpanzees, and Au. afarensis.
Slope 0-4.9% 5-9.9% 10- 14.9% 15- 24.9% 25- 49.9% >50%
Hum Speed km/h 3.81 3.61 3.31 2.81 2.3 1.82
Hum Friction Value min/m 0.01579 0.01667 0.01818 0.02143 0.02609 99
Chimp Speed km/hr 2.993 2.77 2.54 2.15 1.77 1.38
Chimp Friction Value min/m 0.02007 0.02167 0.02364 0.02786 0.03391 99
Slow human Aw. afarensis Speed km/hr 2.124 2.01 1.84 1.56 1.28 1.01
Slow human Aw. afarensis Friction Value min/m 0.02830 0.02983 0.03255 0.03836 0.04670 99
Slow chimp Aw. afarensis Speed km/hr 2.124 1.98 1.81 1.54 1.26 0.99
Slow chimp Aw. afarensis Friction Value min/m 0.02830 0.03033 0.03309 0.03900 0.04748 99
Intermediate Au. afarensis Speed km/hr 3.40 3.18 2.92 2.48 2.03 1.38
Intermediate Au. afarensis Friction Value min/m 0.01767 0.01884 0.02055 0.02422 0.02949 99
Terrier et al. 2001.
zAssuming a 0.5 km/hr drop as indicated by the previous two speed drops. 3Hunt 1989, Pontzer and Wrangham 2004.
4Charteris et al. 1982, Nagano et al. 2005.
Species speed data was converted to friction values of minutes/meter. The minutes assigned to each cell in the friction raster represent the time in minutes per meter for moving through the cell. All friction values are given in Table 1.
28


Fig. 4. Flowchart of the Laetoli cost distance analysis.
Potential Mobility Models
4 shows the flow chart for the Eastern Africa analysis, but the Laetoli analysis is the same excepting the mosaic step. Slope percentages were reclassified into the friction values that represents how slow a given species walks based on the steepness of the slope. The Reclassify tool in ArcGIS was used to convert the slope DEMs into friction surfaces
(
29


Fig. 35, Fig. 6). However, reclassify does not support floating point rasters, or rasters with decimal values. Therefore, each slope range was reclassified into the friction value multiplied by 1,000,000. Then, the times tool was applied to the resulting raster with a multiplication value of 0.000001 to get a floating point raster of the correct friction values. This process was done for each set of human, chimpanzee, and Aw. afarensis friction values to create five friction rasters for each study area. Each raster was independently input into the Cost Distance tool as the cost raster, while a vector shapefile containing the fossiliferous Laetoli localities was input as the source. This created a raster of time bands from each fossiliferous locality outward. The localities of Kantis, Dikika, Hadar, and Woranso-Mille were then overlaid onto the models of the Eastern African study area to determine if areas of least time matched up with the localities.
30


Fig. 3. The cost raster of thqAu. afarensis Intermediate model created from reclassifying the slope raster.
31


East Africa DEMs
High : 5886 Low
Fig. 4. A: The Digital Elevation Model created from merging the SRTM 1 Arc-Second Global DEM files of the study area. B: The slope raster created by using the slope tool on the DEM. C: A slope raster reclassified into the values for the Intermediate Au. afarensis model.
32


CHAPTER IV
RESULTS
These two sets of analyses resulted in effective distance surfaces in which each band of color represents the amount of time it would take to get to the cells within the band from the source. The band size can be manipulated by defining the time limits for each band. Bands are deformed by encounters with higher slopes (i.e. mountains) which cost more to walk over than to walk around. This sort of deformation manifests itself as a dip. Other deformations in the form of protrusions or bumps into other bands represent encounters with lower elevation areas, such as flat land, which cost less to walk over than around. Laetoli is relatively flat (Andrews mid Bamford 2008) so the bands should be relatively concentric. In the Eastern Africa analysis, known localities should lie on protrusions, indicating that they would take less time to walk to. Models not showing this are either inaccurate to Au. afarensis potential mobility, strongly indicate that modem slope data is not usable as a proxy for past terrain, or a marker of some other problem with the analysis. On the other hand, models showing localities on protmsions may reflect actual Au. afarensis mobility, indicate that a modem slope can be used as a proxy for past elevation, mid demonstrate that Am. afarensis did avoid steeper slopes as they moved from site to site.
Laetoli
Visual inspection of the cost distance rasters at Laetoli (Fig. 5) show that the models did turn out as expected because the bands are nearly concentric circles, confirming the fact that terrain at Laetoli is relatively flat. Inspection involved looking at the shape and width of the bands, which were standardized to take 9.23 minutes to cross. If there was a greater
33


OJ
o 1
VO vC vO 00 O'
vd vi rf *>
TT /> VO r** 00
i r~ i r" 00
vo r*-i vO £ i-
c*i
ri
O'
00
III
Fig. 5. The mobility models showing effective distance from each locality at Laetoli. The colored bands represent the amount of time in minutes it would take Au. afarensis to get from any given locality to any given location on the map. The boundary of Tanzania was obtained from Thompson (2013).
34


variety in topography, then there should be more dips or protrusions in the bands. In addition, as expected the Human Model is faster than the Intermediate Model, which is faster than the Chimpanzee Model, which in turn is faster than the Slow Human mid Slow Chimpanzee models. It is evident that the Human Model, for example, is fastest because the bands are wider than the those in the other models, thus indicating that more area can be covered in 9.23 minutes than any other model.
There are other interesting points, such as that the difference between the Slow Human model and Slow Chimpanzee model appears so minimal, it can only be seen by overlaying the layers mid turning one model on mid off in ArcGIS. For this reason, it is not necessary to use both of these models in future analyses since the information extracted is essentially the same. The differences between all the other models were drastic, which makes sense considering the larger differences in cost values.
All the Au. afarensis potential mobility models for Laetoli strongly suggest that it would not have taken very long to reach any of the given localities based on slope alone. In the slowest scenario for example, the Slow Chimp model (second down in Fig. 5), it would have taken 101 minutes to get from Locality 9 to Locality 2, which are about 3 kilometers apart as the crow flies. This is calculated by adding the number of bands between Locality 9 to Locality 2 and multiplying by the width of the band (11 bands 9.23 minutes). If day range and activity patterns are the same for Au. afarensis mid chimpanzees as some researchers have suggested (Leonard mid Robertson 1995), then Am. afarensis could have traveled between these two sites within a day, as chimpanzees spend 116 minutes per day moving (Doran 1997; Leonard and Robertson 1997; Matsumoto-Oda mid Oda 1998).
Humans could travel between these two sites in 55 minutes (6 bands 9.23 minutes), which
35


is half of the least amount of time African hunter-gatherers spent walking per day (1.9 hours) recorded by Leonard mid Robertson (1997).
Eastern Africa
The Eastern .African analyses confirmed that the Slow Human mid Slow Chimpanzee models are nearly indistinguishable and therefore it is unnecessary to use both as parameters in future analyses (Fig. 6). All of the models did show mi interesting trend, however. Each map has a series of bands which were adjusted to represent a month of walking, assuming Au. afarensis has a similar activity pattern to chimpanzees. In other words, it would take Au. afarensis one month to get to the edge of the first band from Laetoli, two to get to the edge of the second band, and so on. The number of minutes it would take to travel across each band is also included in the legend. It is enlightening to see how the different speed parameters used affect how long it would have taken Am. afarensis to get from Laetoli to the other Eastern African Localities. However, the localities in Ethiopia (the northernmost two stars) reside within a protrusion in a band no matter what parameter is used. Kantis (the star closest to Laetoli) resides in smaller protrusions in all of the analyses, which can be seen more clearly close up. Kantis currently lies on the margin of the East African Rift System (EARS), which could explain why the protrusion is not more distinct.
In addition to the dips and protrusions discussed before, these maps show different colors of speckles, or dots of color that do not match the color of the band on which they he. The colored speckles scattered across the map are small areas that have high walking costs, evidenced by the fact that the speckles are always in higher cost colors than the bands in which they reside. These are high hills or mountains that are effectively impenetrable.
36


Fig. 6. Mobility models for Eastern Africa showing effective distance from Laetoli. Four other known Au. afarensis localities were placed over the models, each corresponds to a protrusion in the bands. A reference image of Eastern Africa includes the EARS. The boundaries of Eastern African countries were obtained from Thompson (2013). East African Rift shapefile from ArcGIS Online.
37


Unfortunately, the contrast between the speckles and elevation does not show up well in image files, so no close-up image of this phenomenon was included. All known Aw. afarensis localities have neighboring speckles of high cost, seemingly restricting the possible movement of Aw. afarensis. This makes sense, considering all of the localities lay within the EARS or on the margin, which seems to have enabled Aw. afarensis to travel with relative speed within it, but it may also have effectively contained them since the edges of the EARS are associated with high walking cost. Fig. 7 shows the East African Rift System overlaid on the Intermediate Model to demonstrate this effect.
Al Ub.yyid
Jisho
ladishu)

* Laetoli
A East African Localities HEARS
Intermediate Model Months per Rand
li
Dodoma
14
15
4
5
6
7


Fig. 7. A: Intermediate Model of mobility with the East African Rift System overlaid, showing that the EARS is mostly along areas of lower cost but is often bounded by speckles of higher cost. B: Satellite imagery of Eastern Africa showing the topography and how it matches with the EARS and select Aw. afarensis localities. EARS shapefile from worldmap.harvard.edu. Background of A is an ArcGIS basemap, National Geographic World. Background of B is a color shaded relief from the Shuttle Radar Topography Mission (NASA et al. 2004), and country outlines are from Thompson (2013).
38


It is especially clem" in the northern part of the study area that the EARS seems to curve around the speckles of higher cost. While there are some higher cost speckles within the Rift System, in general it seems to be composed of lower cost landscape. This effect is the same for every model, with higher cost slopes surrounding the sites and EARS, while the sites are located on lower cost protrusions. Once again, none of the models can be decisively ruled as the correct one.
39


CHAPTER V
DISCUSSION AND CONCLUSIONS
Behavior is a complex thing to understand and construct regardless of species, but this project aimed to do just that with Am. afarensis potential mobility patterns. This is critical for understanding the relationship between hominin potential mobility behavior and past environments, and for predicting how Au. afarensis moved across the vastness of Eastern Africa. Slope is an important factor to human mobility mid may have had an equal effect in Au. afarensis. If that was so, Au. afarensis should have behaved optimally by avoiding steep slopes; this hypothesis was tested by creating potential mobility models of effective distance that demonstrated that known Au. afarensis localities do lie on areas of lesser slope. The first set of models were tested on Laetoli and showed that most of the localities would be within a day range of each other. The second set of models tested on Eastern Africa showed that Am. afarensis probably was avoiding steeper slopes as it traveled between localities. These results indicate that modem slope data can be used as an environmental factor to reconstmct Au. afarensis potential mobility, as well as the potential mobility patterns of older or younger hominins. However, none of the potential mobility patterns can be decisively mled out as the best fit for actual Au. afarensis mobility; further testing is necessary to pinpoint other important environmental conditions that affect potential mobility patterns.
An interesting finding from this study was that if Au. afarensis was behaving optimally as it seems to have been, then it may have never left the EARS. This is possible because it would take much more time to get out of the EARS, and provided that it had all of the resources Au. afarensis needed, it does not seem likely that many individuals of that species would have ventured outside of it. Furthermore, each Am. afarensis found so far is within the EARS. Each map shows that the protmsions match up to the localities, so Am.
40


afarensis was behaving optimally by avoiding steep gradients as it moved. In addition, the maps show that the EARS had terrain that was conducive to speedier walking because it had less steep slopes within, which would have helped restrict Am. afarensis movement inside the system mid funneled them from Laetoli to the other localities or vice-versa. This supports Bailey and Kings (2011) theory that there were selective forces acting on hominin evolution and mobility, and in this case, it appears one selective force may have been the steep gradients surrounding the EARS.
Limitations
This study was done to explore the possibility of using GIS cost-distance modeling with environmental and mobility data to understand potential hominin mobility patterns. As such, it is in the primary stage and therefore has many limitations that need to be addressed. These relate to the use of the comparative method, the GIS methodology, and geological and taphonomic biases.
Use of the comparative method is common, but direct comparisons between hominin other mammalian locomotion is not possible because there are no other striding bipedal mammals. Without this, the evolution of bipedalism can never be completely understood even when studying closely related animals because they do not have the same locomotor pattern (Cartmill 1990). Further, taking chimpanzees as the model for the LCA cannot explain how hominins became bipeds rather than knuckle-walking quadrupeds. Without proper analogy, paleoanthropologists are left to face the fact that they can never really know what locomotor patterns preceded bipedalism and how it evolved (Cartmill 1990).
This should not be taken as a deterrent to doing comparative research; rather, it is something to take into account and be explicit about when reporting results of these studies.
41


It is important to remember that chimpanzee mobility cannot be used as a substitute to study hominin mobility, and that conclusions will always be tentative when using this type of analogy. That is why this research used several models for Au. afarensis potential mobility using both chimpanzees mid humans for comparison as well as fossil evidence from Au. afarensis themselves. Still, none of the models can be said to represent Au. afarensis potential mobility pattern, in part because of the limitations associated with use of the comparative method.
The slope data used for the cost-distance modeling were obtained through DEMs with 30-meter resolution. This presents several problems. First, 30 m resolution is not precise enough to capture small variations in the topography. In fact, it generalizes the landscape such that if there are several changes in slope in a 30 m cell, only one slope percent will represent the whole. At a cartographic ally large scale, such as at Laetoli, this means that the concentric circles of the model outputs may not represent the actual variation of the modem landscape, and higher resolution models may look drastically different. On a carto graphic ally small scale, such as that of Eastern Africa, this problem persists with the variance being evened out and subtle changes in topography being ignored.
This limitation does have a benefit for this study, however. Given the uncertainty of how well modem elevation data approximates past elevation, a higher resolution DEM would give these models false confidence. In other words, the data is as uncertain as its representativeness. Using DEMs with higher resolution may show more dips and protmsions in the cost bands, but that does not mean that those deformations are any more likely to reflect past topography than those found with 30 m resolution DEMs.
42


Moreover, known sites may simply be on erosional landscapes that not only flattened the terrain, but also exposed fossils that could be more easily found by paleoanthropologists. Therefore, the DEMs may simply be reflecting modem erosion rather than past terrain Furthermore, places with less steep terrain are easier for researchers to get to, biasing where fossils are searched for. Finally, if Au. afarensis died mid were buried on steep slopes, as they eroded out, gravity may have pushed them down only flatter land. If any of these alternative scenarios are tme, it would lead to the false conclusion that .4?/. afarensis preferred areas of lesser slope rather than that the sites are a product of the geological process of erosion.
Faetoli provides an example of the last limitation to be discussed here, taphonomic biases. The fact that Au. afarensis fossils were found at 15 of the localities recorded for Faetoli is seemingly unsurprising given information of the relative flatness of the current landscape mid that most of the localities are within a day range of each other. This indicates that it is possible that^M. afarensis was spread over more of this area, but due to taphonomic processes, evidence of such is now gone or buried. This species only constituted a small proportion of the large mammal faunal group recovered at Faetoli (Su and Harrison 2008), which would make evidence that Au. afarensis was more widely spread difficult to find. The activity of carnivores at Faetoli only compounds this problem, especially because their size made them a target for predators and scavengers alike, creating an atmosphere in which their skeletal elements were more likely to be moved from their original position or completely removed from an assemblage. They were also too big to be completely buried by ashfall, which would have allowed their skeletal elements to escape from carnivore ravaging (Su and Harrison 2008).
43


This is a sample of the types of taphonomic processes that make it difficult to find Au. afarensis sites and to know where they lived or died. For a study such as this, it means that there may be places that Aw. afarensis were that were not on areas of lesser slope, but because of poor preservation conditions these will never be found. If this is so, it would invalidate the conclusions reached in this research. While the limitations in this section impact the acceptability mid adequacy of these models, they also highlight parts of this research that could be improved in future phases.
Future Directions
None of the potential mobility models could be conclusively ruled as the best fit for Au. afarensis mobility. The inclusion of other environmental conditions that affect mobility, such as water bodies or substrate properties, should be included in future models to further refine them. When more precise data on past terrain is available, these should also be incorporated into the models. Additionally, there may be other important factors aside from landscape at play in Au. afarensis mobility that could be isolated, such as how much time is spent locomoting in trees and if they can avoid expending walking energy on slopes by climbing. As with any models, it is important to incorporate as many variables it takes to explain a phenomenon until including more variables ceases to add any new information or becomes too computationally complex (Dunbar 2001). Bringing in data on water bodies and climbing could be invaluable for understanding potential mobility patterns.
Adjusting the relative walking speed constructed for Au. afarensis may benefit the models as well. As discussed before, the base speeds for Au. afarensis were estimates obtained from the Laetoli footprints and from Lucy. However, performing further sensitivity analyses using relative walking speed estimates from their putative ancestor Australopithecus
44


anamensis or descendant Australopithecus africanus may be enlightening. Using energy rather than relative walking speed as the cost in the models could also lend insight into how robust the models are.
The models presented here isolate likely potential mobility patterns and at this stage they cannot be used to pinpoint where fossil hominins are likely to be found. This type of analysis could be done by creating least-cost paths or corridors in conjunction with spectral signature analysis to identify likely areas to be surveyed. The least-cost paths could reflect the parameters used in this study to create the modeled routes that An. afarensis may have taken between localities. If fossils were found along these routes, it would be one way to verify the results of the project presented here. However, they are deterministic, assuming known origins mid destinations, which implies that Am. afarensis thoughtfully and purposefully went to each known site. Yet this remains a possible future direction because it provides a way to gather evidence in support of the potential mobility models.
The research begun here will be continued with some of the future directions discussed above. Past water body and substrate data will be acquired and added as environmental conditions that affect mobility. If this makes better fit models, then the next phase would be to verify the models through groundtruthing. This would require identifying areas along least-cost paths where fossils are likely to be found, and could substantiate the potential mobility pattern modeled.
Conclusions
Despite the limitations of the models and the uncertainty surrounding which mobility model is correct, information about the Eastern African paleoenvironment and the ecosystem in which Am. afarensis lived does lend itself to possible scenarios of potential mobility
45


patterns. For example, Au. afarensis was a generalist species capable of living in anything from the dryer open woodlands of Laetoli to the riparian forests of Fladar (Kimbel mid Delezene 2009; Su and Flarrison 2008). During the Early Miocene to Late Pliocene, the rifting and volcanic activities in the EARS led to the creation of small to large ephemeral mid permanent lakes (Tiercelin mid Lezzar 2002), which could have supported the movement of Au. afarensis through the EARS by providing drinking water and suitable habitats during wet periods. During dryer periods as lake levels dropped, they may have been able to survive in surrounding refugia in a way that is similar to other species that thrive in lacustrine environments (Martens 2002). When lake levels rose again, they could have moved from such refugia to nearby lakes, thereby encountering new and previously occupied areas (Martens 2002). In this way, Au. afarensis could have gone from Laetoli to Hadar, or between any of the localities. An ability to move faster, such as in the Intermediate Model, could have enabled greater survival as Au. afarensis was traveling between refugia mid different water sources. However, if water sources were relatively stable, a slower moving Au. afarensis as in the Slow Human model could still have had great survivability.
While the models created above have limitations mid would benefit from additional data mid lines of inquiry, they do confirm that the potential mobility pattern of Au. afarensis can be modeled to help understand how they moved across Eastern Africa. Furthermore, Au. afarensis does appear to have been behaving optimally by avoiding steep slopes.
Additionally, successful models can be built off of modem slope data when past data is unavailable. Finally, this research demonstrates that GIS is an undemtilized tool in paleoanthropology with the potential to help researchers better understand hominin behavior.
46


REFERENCES
Adriaensen F, Chardon JP, De Blust G, Swinnen E, Villalba S, Gulinck H, mid Matthysen E. 2003. The application of'least-cost' modelling as afunctional landscape model. Landscape mid Urban Planning 64:233 247.
Alemseged Z, Wynn JG, Kimbel WH, Reed D, Geraads D, and Bobe R. 2005. A New
Hominin from the Basal Member of the Hadar Formation, Dikika, Ethiopia, mid its Geological Context. Journal of Human Evolution 49:499-514.
Alemseged Z, Spoor F, Kimbel WH, Bobe R, Geraads D, mid Reed D. 2006. A juvenile early hominin skeleton from Dikika, Ethiopia. Nature 443(7109):296
Alexander RM. 2002. Energetics mid optimization of human walking mid running: The 2000 Raymond Pearl Memorial lecture. American Journal of Human Biology 14:641 648.
Andrews P, mid Bamford M. 2008. Past and Present Vegetation Ecology of Laetoli, Tanzania. Journal of Human Evolution 54:78 98.
Bailey GN, Reynolds SC, mid King GCP. 2011. Landscapes of human evolution: Models and methods of tectonic geomorphology and the reconstruction of hominin landscapes. Journal of Human Evolution 60:257 280.
Bedaso ZK, Wynn JG, Alemseged Z mid Geraads D. 2013. Dietary and paleoenvironmental reconstruction using stable isotopes of herbivore tooth enamel from middle Pliocene Dikika, Ethiopia: Implicaiton for Australopithecus afarensis habitat and food resources. Journal of Human Evolution 64:21 38.
Begun DR. 2010. Miocene Hominids mid the Origins of the African Apes and Humans. Annual Review of Anthropology 39:67- 84.
Bonnefille R. 2010. Cenozoic vegetation, climate changes and hominid evolution in tropical Africa. Global mid Planetary Change 72:390 411.
Cartmill M. 1990. Human uniqueness and theoretical content in paleoanthropology. International Journal of Primatology 11(3): 173 192.
47


Charteris J, Wall JC, and Nottrodt JW. 1982. Pliocene hominid gait: New interpretations based on available footprint data from Laetoli. American Journal of Physical Anthropology 58:133 144.
Crompton RH, Wang Weijie, LY, Gunther M and Savage R. 1998. The mechanical
effectiveness of erect and "bent-hip, bent-knee" bipedal walking in Australopithecus afarensis. Journal of Human Evolution 35:55 74.
Curran S mid Haile-Selassie Y. 2016. Paleoecological reconstruction of hominin-bearing
middle Pliocene localities at Woranso-Mille, Ethiopia. Journal of Human Evolution 96:97 112.
Cuthbert MO and Ashley GM. 2014. A spring forward for hominin evolution in East Africa. PLOSone 9(9):el07358.
Davidson A, Carmel Y, and Shirli B-D. 2013. Landscape Ecology 28:1465 1478.
Deino, AL. 2011. 40Ar/39Ar Dating of Laetoli, Tanzania. In: Harrison T, editor. Paleontology and Geology of Laetoli: Human Evolution in Context. New York: Springer, p 77 97.
De Silva M, and Pizziolo G. 2001. Setting up a "human calibrated" anisotropic cost surface for archaeological landscape investigation. BAR International Series 931:279 287.
Ditchfield P, mid Harrison T. 2011. Sedimentology, Lithostratigraphy and Depositional History of the Laetoli Area. In: Harrison T, editor. Paleontology and Geology of Laetoli: Human Evolution in Context. New York: Springer, p 47 76.
Doran D. 1997. Influence of seasonality on activity patterns, feeding behavior, ranging, mid grouping patterns in Tai chimpanzees. International Journal of Primatology 18(2): 183
-206.
Dunbar RIM. 2001. Modelling Primate Behavioral Ecology. International Journal of Primatology 23(4):785 819.
Egeland CP, Nicholson CM, and Gasparian B. 2010. Using GIS and ecological variables to identify high potential areas for paleoanthropological survey: An example from Northern Armenia. Journal of Ecological Anthropology 14(1): 89 98.
48


Eriksson J, Hohmann G, Boesch C, mid Vigilant L. 2004. Rivers Influence the Population Genetic Structure of Bonobos (Pan paniscus). Molecular Ecology 13(11):3425 -
3435.
Foley RA. 1992. Evolutionary Ecology of Fossil Hominids. In: Smith EA, and Winterhalder B, editors. Evolutionary Ecology and Human Behavior. New York: Aldine de Gruyter. p 131 164.
Grine FE, Ungar PS, Teaford MF and El-Zaatari S. 2006. Molar microwear in Praeanthropus afarensis: Evidence for dietary stasis through time and under diverse paleoecological conditions. Journal of Human Evolution 51:297 319.
Haile-Selassie Y, Latimer BM, Alene M, Deino AL, Gibert L, Melillo SM, Saylor BZ, Scott GR, and Lovejoy CO. 2010. An early Australopithecus afarensis postcranium from Worano-Mille, Ethiopia. PNAS 107(27):12121-12126.
Haile-Selassie Y, Melillo SM, Ryan TM, Levin NE, Saylor BZ, Deino A, Mundil R, Scott G, Alene M mid Gibert L. 2016. Dentognathic remains of Australopithecus afarensis from Nefuraytu (Woranso-Mille, Ethiopia): Comparative description, geology, mid paleoecological context. Journal of Human Evolution 100:35 53
Harrison T. 2011. Laetoli revisited: Renewed paleontological and geological investigations at localities on the Eyasi Plateau in Northern Tanzania. In: Harrison T, editor. Paleontology and geology of Laetoli: Human evolution in context. New York: Springer, p 1 15.
Halsey LG and White CR. 2012. Comparative energetics of mammalian locomotion: Humans are not different. Journal of Human Evolution 63:718 722.
Harrison T, mid Kweka A. 2011. Paleontological localities on the Eyasi Plateau, including
Laetoli. In: Harrison T, editor. Paleontology mid geology of Laetoli: Human evolution in context. New York: Springer, p 17 45.
Howey MCL. 2007. Using multi-criteria cost surface analysis to explore past regional
landscapes: A case study of ritual activity and social interaction in Michigan, AD 1200-1600. Journal of Archaeological Science 34:1830 1846.
Hunt KD. 1989. Positional behavior of Pan troglodytes in the Mahale Mountains mid Gombe Stream National Parks, Tanzania: University of Michigan.
49


Hunt KD. 1994. The evolution of human bipedality, ecology and functional morphology. Journal of Human Evolution 26:183-202.
Johanson DC, Taieb M, and Coppens Y. 1982. Pliocene hominids from the Hadar Formation, Ethiopia (1973-1977): Stratigraphic, chronologic, and paleoenvironmental contexts, with notes on hominid morphology mid systematics. American Journal of Physical Anthropology 57:373 402.
Kelly RL. 2007. The Foraging Spectrum: Diversity in Hunter-Gatherer Lifeways. New York: Percheron Press. 446 p.
Kelly RL, and Todd LC. 1988. Coming into the Country: Early Paleoindian Hunting mid Mobility. American Antiquity 53(2):231-244.
Kimbel WH. 2007. The species and diversity of Australopiths. In: Henke W, and Tatters all I, editors. Handbook of Paleoanthropology. Berlin Heidelberg: Springer, p 1539 1573.
Kimbel WH, Lockwood CA, Ward CV, Leakey MG, Rak Y, mid Johanson DC. 2006. Was Australopithecus anamensis Ancestral to A. afarensis? A Case of Anagenesis in the Hominin Fossil Record. Journal of Human Evolution 51:134-152.
Kimbel WH, mid Delezene LK. 2009. "Lucy" redux: A review of research on
Australopithecus afarensis. Yearbook of Physical Anthropology 52:2 48.
King G, mid Bailey G. 2006. Tectonics and Human Evolution. Antiquity 80:265 286.
Koenig A, and Borries C. 2012. Hominiod Dispersal Patterns mid Human Evolution. Evolutionary Anthropology: Issues, News, and Reviews 21(3): 108 112.
Kovarovic K and Andrews P. 2011. Environmental change within the Laetoli fossiliferous sequence: Vegetation catenas mid bovid ecomorphology. In: Harrison T, editor. Paleontology and Geology of Laetoli: Human Evolution in Context. New York: Springer, p 367 380.
Leonard WR, mid Robertson ML. 1995. Energetic efficiency of human bipedality. American Journal of Physical Anthropology 97:335 338.
50


Leonard WR, mid Robertson ML. 1997. Comparative primate energetics and hominid evolution. American Journal of Physical Anthropology 102:265 281.
Martens K. 2002. Redundancy and ecosystem stability in the fluctuating environments of long-lived lakes. In: Odada EO, and Olago DO, editors. The East African Great Lakes: Limnology, palaeolimnology and biodiversity. Netherlands: Kluwer Academic Publishers, p 309 319.
Marwick B. 2003. Pleistocene exchanges networks as evidence for the evolution of language. Cambridge Archaeological Journal 13(1):67- 81.
Matsumoto-Oda A, and Oda R. 1998. Changes in the activity budget of cycling female chimpanzees. American Journal of Primatology 46:157 166.
Mbua E, Kusaka S, Kunimatsu Y, Geraads D, Sawada Y, Brown FH, Sakai T, Boisserie J-R, Saneyoshi M, Omuombo C et al. 2016. Kantis: Anew Australopithecus site on the shoulders of the Rift Valley near Nairobi, Kenya. Journal of Human Evolution 94:28 -44.
Moreno-Sanchez R, Torres-Rojo JM, Moreno-Sanchez F, Hawkins S, Little J, mid
McPartland S. 2012. National assessment of the fragmentation, accessibility and anthropogenic pressure on the forests in Mexico. Journal of Forestry Research 23(4):529 541.
Musiba CM, Tuttle RH, Hallgrimsson B, and Webb DM. 1997. Swift and sure-footed on the savanna: A study of Hadzabe gaits and feet in Northern Tanzania. American Journal of Human Biology 9:303 321.
Musiba CM, Mabula A, Selvaggio M and Magori CC. 2008. Pliocene animal trackways at Laetoli: Research and conservation potential. Ichnos 15:166 178.
Nagano A, Umberger BR, Marzke MW, and Gerritsen KGM. 2005. Neuromusculoskeletal computer modeling and simulation of upright, straight-legged, bipedal locomotion of Australopithecus afarensis (A.L. 288-1). American Journal of Physical Anthropology 126:2- 13.
NASA, JPL, mid NIMA. 2004. SRTM Data Release for Africa, Colored Height. Online. https://photojoumal.jpl.nasa.gov/catalog/PIA04965.
51


Pontzer H, and Wrangham RW. 2004. Climbing mid the Daily Energy Cost of Locomotion in Wild Chimpanzees: Implications for Hominoid Locomotor Evolution. Journal of Human Evolution 46:315 -333.
Pontzer H, Raichlen DA and Sockol MD. 2009. The metabolic cost of walking in humans, chimpanzees, mid early hominins. Journal of Human Evolution 56:43 54.
Pontzer H, Raichlen DA and Rodman PS. 2014. Bipedal and quadrupedal locomotion in chimpanzees. Journal of Human Evolution 66:64 82.
Potts R. 1998. Variability selection in hominid evolution. Evolutionary Anthropology 7:81 -96.
Puech P-F, Cianfarani F, and Roth H. 1986. Reconstruction of the maxillary dental arcade of Garusi Hominid 1. Journal of Human Evolution 15:325-332.
Rahn B. 2005. Cost surface-derived least-cost paths: A case study from Iron Age Orkney. Internet Archaeology 19.
Reed DN and Denys C. 2011. The taphonomy mid paleoenvironmentl implications of the
Laetoli micromammals. In: Harrison T, editor. Paleontology mid geology of Laetoli: Human evolution in context. New York: Springer, p 265 278.
Reed KE. 1997. Early Hominid Evolution mid Ecological Change through the African Plio-Pleistocene. Journal of Human Evolution 32:289 322).
Reed KE. 2008. Paleoecological patterns at the Hadar hominin site, Afar Regional State, Ethiopia. Journal of Human Evolution 54:743 768.
Rossouw L, and Scott L. 2011. Phytoliths and pollen, the microscopic plant remains in Pliocene volcanic sediments around Laetoli, Tanzania. In: Harrison T, editor. Paleontology and geology of Laetoli: Human evolution in context. New York: Springer, p 201 215.
Sellars WI, Cain GM, Wang W, mid Crompton RH. 2005. Stride lengths, speed mid energy costs in walking of Australopithecus afarensis using evolutionary robotics to predict locomotion of early human ancestors. Journal of the Royal Society Interface 2(431 -441).
52


Smith EA, and Winterhalder B. 1992. Evolutionary Ecology mid Human Behavior. New York: Aldine de Gruyter. p 470
Stem JT, and Susman RL. 1983. The locomotor anatomy of Australopithecus afarensis. American Journal of Physical Anthropology 60:279-317.
Su DF, mid Harrison T. 2008. Ecological implications of the relative rarity of fossil hominins at Laetoli. Journal of Human Evolution 55:672 681.
Taylor CR, Caldwell SL, and Rowntree VJ. 1972. Running up and down hills: Some consequences of size. Science 178(4065): 1096 1097.
Terrier P, Aminian K, mid Schutz Y. 2001. Can accelerometry accurately predict the energy cost of uphill/downhill walking? Ergonomics 44(1 ):48 62.
Thompson DK. 2013. Map East Africa. Online. Shapefile of Tanzanian boundary.
Tiercelin JJ, mid Lezzar KE. 2002. A 300 million years history of rift lakes in Central mid East Africa: An updated broad review. In: Odada EO, mid Olago DO, editors. The East African Great Lakes: Limnology, palaeolimnology and biodiversity. Netherlands: Kluwer Academic Publishers, p 3 60.
Todd JR, and White PS. 2009. A new cost-distance model for human accessibility mid an evaluation of accessibility bias in permanent vegetation plots in the Great Smoky Mountains National Park, USA. Journal of Vegetation Science 20:1099 1109.
Tuttle R, Webb D, Weidl E and Baksh M. 1990. Further progress on the Laetoli trails. Journal of Archaeological Science 17:347-362.
Umberger BR, Gerritsen KGM and Martin PE. 2003. A model of human muscle energy expenditure. Computer Methods in Biomechanics mid Biomedical Engineering 6(2):99 11E
Van Leusen M. 1999. Viewshed and cost surface analysis using GIS (cartographic modelling in a cell-based GIS II). BAR International Series 757:215 224.
53


Verburg PH, Overmars KP, mid Witte N. 2004. Accessibility mid land-use patterns at the forest fringe in the northeastern part of the Philippines. The Geographical Journal 170(3):238 255.
White TD, Ambrose SH, Suwa G, Su DF, DeGusta D, and al. e. 2009. Macro vertebrate
Paleontology and the Pliocene Habitat of Ardipithecus ramidus. Science 326:87-92.
Winterhalder B, and Smith EA. 1992. Evolutionary ecology mid the social sciences. In:
Smith EA, and Winterhalder B, editors. Evolutionary ecology and human behavior. New York: Aldine de Gruyter. p 3 23.
Wynn JG, Sponheimer M, Kimbel WH, Alemseged Z, Bedaso ZK and Wilson JN. 2013. Diet of Australopithecus afarensis from the Pliocene Hadar Formation, Ethiopia. PNAS 110(26): 10495 105100.
54


Full Text

PAGE 1

WALKING WITH LUCY : MODELING MOBILITY PATTERNS OF AUSTRALOPITHECUS AFARENSIS USING GIS b y RACHEL MCPHERSON B.S., University of Wyoming, 2014 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillme nt of the requirements for the degree of Master of Arts Anthropology Program 201 8

PAGE 2

ii This thesis for the Master of Arts degree by Rachel McPherson h as been approved for the Anthropology Program by Charles Musiba, Chair Rafael Moreno Deborah Thomas Jamie Hodgkins Date: December 24, 2017

PAGE 3

iii McPherson, Rachel (M.A., Anthropology Program ) Walking w ith Lucy: Modeling Mobility Patterns o f Australopithecus a farensis Using GIS Thesis directed by Associate Professor Charles Musiba ABSTRACT Behavior is perh aps the most challenging component of an extinct organism to reconstruct and understand. Often in p aleoanthropolo gy researchers primarily have fossils and paleoecological data ; however, combining the se int o models of hominin behavior is difficult in pract ice Yet for years archaeologists and wildlife biologists have been using Geographic Information Systems (GIS) to model the mobility behavior of humans and other animals. This research s eeks to integrate the methodology of c ost d istance modeling in GIS int o paleoanthropology to understand hominin mobility, specifically investigating if the potential mobility pattern of Australopithecus afarensis c a n be modeled to understand how they got across Eastern Africa to their known sites. The models created for Au. afar ensis humans, and chimpanzees brought together walking time as a cost factor and modern slope as an impediment to movement These values were input into the Cost Distance tool in ArcGIS with Laetoli as the source and tested on two study areas, Laetoli and Eastern Africa Known Au. afarensis sites matched areas of least cost for each potential mobility pattern which indicated that 1) none of the models could be ruled as the best potential mobility pattern for Au. afarensis 2) Au. afarensis like ly avoided steeper gradients, a nd 3) modern gradient data were no t incompatible with the models. Despite limitations to this st udy, these models provide a foundation for research into hominin mobility patterns using GIS The form and content of this abstr act are approved. I recommend its publication. Approved: Charles Musiba

PAGE 4

iv ACKNOWLEDGEMENTS Many, many thanks to Rafael Moreno for being so helpful and available, especially with the mundane or silly questions I brought to you. Our conversations were always useful and enjoyable for me. I sincerely hope you find this paper interesting and informative. Deb Thomas, thank you so much for the data, your georeferenced maps were a life saver. Charles Musiba, you kept me motivated to get my work done and I appreciat e that. To my Mams, thank you for be ing patient with me while I have been pursuing my Masters work and doing basically nothing else. You are the reason I have gotten this far. Connie Turner, you have been amazing. I cannot thank you enough for always keepi ng me apprised of deadlines, giving me encouragement, and dealing with administrative matters for me. To Liz Sweitzer and Alex Pelissero, thank you so very, very much for editing this thesis and helping me sound smart. Finally, thank you to all of you anth ropology grad students who tolerated my rants about my thesis and gave advice and encouragement when I struggled. You are all heroes to this kid.

PAGE 5

v TABLE OF CONTENTS CHAPTER I THE EVOLUTION OF A MODEL OF MOBIL ITY ................................ ........................... 1 Introduction ................................ ................................ ................................ ........................... 1 Literature Review ................................ ................................ ................................ ................. 3 Au stralopithecus afarensis ................................ ................................ ................................ 3 How Optimality in Behavioral Ecology Enables Modeling of Hominin Behavior ........... 8 The Complexities of Mobility and How to Render It Simple ................................ ............ 9 The Effect of the Environment on Mobility ................................ ................................ ..... 13 II. MATERIALS ................................ ................................ ................................ ..................... 18 III. METHODS ................................ ................................ ................................ ....................... 20 The Role of GIS Cost Distance Analysis in Models of Mobility ................................ ....... 20 Preparatory Work ................................ ................................ ................................ ................ 22 Au. afarensis Localities ................................ ................................ ................................ ... 23 Slope The Environmental Variable ................................ ................................ ................ 23 Relative Walking Speed The Cost ................................ ................................ .................. 25 Potential Mobility Models ................................ ................................ ................................ .. 29 IV. RESULTS ................................ ................................ ................................ ......................... 33 Laetoli ................................ ................................ ................................ ................................ 33 Eastern Africa ................................ ................................ ................................ ..................... 36 V. DISCUSSION AND CONCL USIONS ................................ ................................ ............. 40

PAGE 6

vi Limitations ................................ ................................ ................................ .......................... 3 9 Future Directions ................................ ................................ ................................ ................ 4 2 Conclusions ................................ ................................ ................................ ......................... 4 3 REFERENCES ................................ ................................ ................................ ....................... 47

PAGE 7

vii LIST OF TABLES TABLE 1 : Speed and friction values on slopes for humans, chimpanzees, and Au. afarensis ....... 28

PAGE 8

viii LIST OF FIGURES FIGURE 1 : Select p aleoanthropological s ites w here A u. afarensis f ossils h ave b een f ound .............. 4 2 : All l ocalities r ecorded at Laetoli, Tanzania overlaying a Google Earth image. ........... 24 3 3 : A: Digital e levation m odel of the Laetoli area. B: Slope created from the DEM .......... 2 4 4 : Flowchart of the Laetoli c ost d istance a nalysis ................................ .............................. 2 8 5 : The cost raster of the Au. afare nsis i ntermediate model created from reclassifying the slope raster ................................ ................................ ................................ ...................... 31 6 : A: The Digital Elevation Model created from merging the SRTM 1 Arc Second Global DEM files of the study area. B: The slope raster created by using the slope tool on the DEM. C: A slope raster reclassified into the values for the Intermediate Au. afarensis model ................................ ................................ ................................ ............................... 32 7 : The mobility models showing effe ctive distance from each locality at Laetoli ............. 34 8 : Mobility models for Eastern Africa showing effective distance from Laetoli ............... 37 9 : A: Intermediate Model of mobility with the East African Rift System overlaid, showing that the EARS is mostly along areas of lower cost but is often bounded by speckles of higher cost. B: Satellite imagery of Eastern Africa showing the topogr aphy and how it maches with the EARS and select Au. afarensis localities ................................ ............. 38

PAGE 9

1 CHAPTER I THE EVOLUTION OF A MODEL OF MOBILITY Introduction Between ~ 3.85 2.9 million years ag o in Eastern Africa, the s pecies Australopith ecus afarensis (an early human ancestor) were traversing the landscape, lea ving hints of their presence in the form of fossils and footprints in volcanic ash. How did this small bodied species get to areas in Ethiopia, Kenya, and Tanzania? Many have tried t o find the answer by modeling the locomotor mechanics of their skeletons or reconstructing their environment, which has led to several exciting revelations about the way they walked a nd the environment in which they lived (Andrews and Bamford 2008; Sellar s et al. 2005; Stern and Susman 1983) For instance, many researcher s agree that like humans, Au. afarensis was a habitual, efficient biped (Sellars et al. 2005) that could live in a variety of environments as a generalized feeder ( Kimbel and Deleze ne 2009) However, this informa tion by itself does not suggest how Au. afarensis go t fro m one location to another. Can the potential mobility pattern of Au. afarensis be modeled to understand how they got to their known sites? How did behaviors common in humans such as mobility pattern, evolve and when? To begin to answer this question, we must understand the behavior of our hominin relatives. When investigating th is of extinct hominins, researchers often have little more to go on than fossils and ichnof ossils such as footprints (Harrison 2011; Kimbel and Delezene 2009; Sellars et al. 2005) T hese must be used to parse out behavior using whatever methodologies are available. Cost distance modeling is a methodology often used by archaeologists and wildlif e biologists when investigat ing how humans or other animals move from one place to another (Davidson et al. 2013; Rahn 2005) Yet this form of investigation has made little

PAGE 10

2 headway into pal eoanthropology (but see Egelan d et al. 2010) and there has been little progress towards revealing the potential mobility pattern of Au. afarensis that enabled them to get from one end of Eastern Africa to the other. The use of cost distance modeling for Au. afarensis will bring resea rchers one step closer to modelling the potential mobility behavior of that species and open up a new avenue of i nquiry for scholars of other hominins. This research sought to model the potential mobility pattern of Au. afarensis P otential mobility here is walking activity as it is affected by environmental conditions. The research question is built on four parts; the methodology used for the models, the walking activity modeled, the environmental condition(s) included, and the underlying assumptions. Co st distance modeling in a GIS was the method used to model potential mobility. Relative walking time was used as a proxy for walking activity. The environmental condition adopted was slope, but due to lack of precision in paleoenvironmental re constructions, modern elevation data was used. A key assumption is that the potential mobility pattern will be the optimal one that requires less time expenditure and therefore is one that avoids steeper slopes If these assumptions hold true and the pote ntial mobility pattern is correct, then known sites should lie i n areas that would take Au. afarensis less time to get to. This paper is divided into several parts. First, the subjects of Au. afarensis behavioral ecology, use of models, and the interacti on between environment and mobility are reviewed. Included in this section are the justifications for the decisions made during the course of this research. Then, the data acquired to build the models will be presented. Next, the methodology of least cost modeling in general and the models built for this research specifically will be discussed. Two different study areas were utilized to run the models on, and the results are explained. Finally, the research and limitations are examined.

PAGE 11

3 Literature Review A ustralopithecus afarensis Fossils that would later be attributed to Au stralopithecus afarensis were first collected b y K o h l L a r s e n in the 1930s in Garusi (Laetoli) Tanzania ( Kimbel and Delezene 2009; Puech et al. 1986) but it was not until the International Afar Research Expedition began work at Hadar, i n Ethiopia in 1973 that the fossils from both sites were recognized and given a species designation (Johanson et al. 1982) Subsequent work by paleoanthropologists have revealed more than 400 Au. afarensis specimens ( Kimbel and Delezene 2009) at Hadar, Dikika and Maka in Ethiopia; East Turkana, Kenya; Laetoli, Tanzania (Kimbel 2007) ; Worans o Mille, Ethiopia (Haile Selassie et al. 2010) ; Kantis, Kenya (Mbua et al. 2016) ; and possibly Belohdelie, Ethiopia (Kimbel 2007) ( Fig. 1 : Select p aleoanthropological s ites w here Au. afarensis f ossils h ave b een f ound. : S e l e c t p a l e o a n t h r o p o l o g i c a l s i t e s w h e r e A u a f a r e n s i s f o s s i l s h a v e b e e n f o u n d ) Any understanding of Au. afarensis potential mobility patterns must be set in the cont ext of their paleoecology at these sites and their dietary adaptations, espe cially because their environment changed over time and space and yet Au. afarensis persisted and spread for nearly a million years (Bonnefille 2010 ; Kimbel and Delezene 2009). On ly five of the sites are included in this study; Hadar, Dikika Woranso Mille and Laetoli. The oldest known site is in the Upper Laetoli l Beds (ULB) at Laetoli in which Au. afarensis dat i n g to 3.63 3. 85 Ma w a s r e c o v e r e d (Deino 2011). Ecomorphological analysis, ph ytoliths, and micromammals further suggest heavy woodland bushland environments in the older part of the ULB a t L a e t o l i but after Tuff 5 there is a shift towards more grassland and open woodland with increased aridity a n d an accompanied change from mainly C 3 to

PAGE 12

4 mai nly C 4 grasses (Kovarovic and Andrews 2011; Reed and Denys 2011; Rossouw and Scott 2011). Fig. 1 : Select p aleoanthropological s ites w here Au. afarensis f ossils h ave b een f ound. S easonal streams existed, possibly so did spri ngs in the volcanic highlands and gallery forest in the valleys, but there was no permanent water source (Kovarovic and Andrews 2011). Kantis has been dated to ~3.5 Ma, and is thought to have been more open and more grassland dominated than the other Au. a farensis localities based on the bovid and suid

PAGE 13

5 assemblages (Mbua et al. 2016) However, mammalian dental carbon isotopes reveal similar values to the contemporaneous localiti es, possibly indicating mosaic grassland, shrubland, and woodland with perennial water bodies (Mbua et al. 2016) As of the date of this thesis, there has not been further published work on Kantis, but perhaps future research will clear up the confusion of the environmental context at Kantis ca. ~ 3.5 Ma. Hadar and Dikika are close t o each other spatially and temporally with Dikika located southeast of Hadar across the Awash River. The Hadar Au. afarensis are dated to ~2.90 > 3.42 Ma (Reed 2008), while Dikika Au. afarensis date to 3.31 > 3.4 Ma (Alemseged et al. 2005; Alemseged et al. 2006) Before 3.4 Ma, carbon and oxygen isotopes indicate Dikika held woodland to grassland habitats, shifting to an open wooded grassland to woodland after 3.4 Ma; however, the proportions of these changed over time while conditions remained wet (Beda so et al. 2013 ). Hadar also shows evidence of vegetati ve as well as climat ic change over time (Bonnefille 2010) At ~3.4 Ma the paleoecological evidence at Hadar indicates a woodland and shrubland mosaic changing over time to a variety of habitats includ ing mosaic woodland and expanded wetlands along the lake margin (Reed 2008) B y ~2.94 Ma, Hadar had shifted to a largely shrubland or open woodland habitat La ke Hadar expanded and regressed with changes in rainfall over time, and the climate went from lac ustrine/wetland to seasonal to more arid (Reed 2008) The last site, Woranso Mille, is 45 kilometers north of Hadar and contains Au. afarensis dating to 3.6 Ma and 3.2 3.3 Ma (Haile Selassie et al. 2016). Faunal analysis indicates that this time per iod was characterized as wet being a distal fluvial plain with flooding events and seasonal dry periods (Haile Selassie et al. 2016). At the time of the earliest Au. afarensis habitats were mosaic, ranging from open to closed (Curran and Haile

PAGE 14

6 Selassie 2 016). Evidence from faunal assemblages, mesowear, and ecomorphological analyses point to Woranso Mille having a river surrounded by dense vegetation with more open habitats farther away from the river (Curran and Haile Selassie 2016). Fewer specimens of Au. afarensis have been found at Laetoli than at Hadar even when accounting for taphonomic processes such as carnivore activity (Su and Harrison 2008) Su and Harrison (2008) point out that there is a higher proportion of Au. afarensis in the faunal commun ity at Had ar than at Laetoli, but Hadar was also wetter with permanent water sources and was more densely wooded. The same pattern is present in chimpanzee populations, where there is a larger population of chimpanzees in closed woodlands than open. Theref ore, the authors conclude Laetoli was a marginal habitat while Hadar contained the denser woodlands in which Au. afarensis was more successful, and had more available food (Su and Harrison 2008) However, all the Au. afarensis sites bec a me progressively mo re open over time, with no apparent large consequences for Au. afarensis up to ~2.9 Ma with the exception of a period of lake extension at Hadar that led to a population decline of Au. afarensis at the site during that period (Bonnefille 2011; Reed 2008). Clearly this species was eurytopic (Bonnefille 2011; Reed 2008), although the lack of abundant water may a l s o explain the smaller population density at Laetoli (Su and Harrison 2008). Molar microwear analysis done on Au. afarensis teeth from Hadar and Laetol i ranging in age from 3.5 3.2 Ma demonstrate that their diet did not change significantly over time (Grine et al. 2006) Pref erred foods appear to be soft and abrasive rather than the hard and brittle items suggested by their robust masti catory apparatus although this morphology may represent the use of fallback foods (Grine et al. 2006) However, it is impossible to conclude from this evidence if Au. afarensis found its preferred food at different habitats or if

PAGE 15

7 it ate a variety of soft, abrasive foods (Su and Harrison 2008). C arbon isotope analysis was performed on Au. afarensis teeth from Dikika and Hadar, indicating that most individuals ate a large amount of C 4 /CAM foods but the range of carbon values was variable across individuals with no direction al change over time or habitat (Wynn et al. 2013). Carbon isotopes from hominin teeth from Woranso Mille confirm this diet (Curran and Haile Selassie 2016; Haile Selassie et al. 2016). Although Au. afarensis may have consumed food with the same mechanical properties, the carbon isotopes point to their ability to consume a range of foods (Wynn et al. 2013) The variety of environments and climates occurring at these five sites demonstrate that Au. afarensis was likely adapted to variable conditions and was able to persist despite climatic and vegetative change (Bonnefille 2010 ; Reed 2008 ). This alone points to Au. afarensis being a generalist rather than a specialist (Bonnefille 2010), an assertion confirmed by the isotopic and microwear evidence from their teeth ( Curran and Haile Selassie 2016; Grine et al. 2006; Wynn et al. 2013 ). Clearly Au. afarensis was capable of living in d i f f e r e n t environments, but how were they able to get from one site to another? This question is directly related to morphology : Au. afarensis is characterized by a wide range of morphological variation and a mix of bipedal and arboreal characteristics. This includes the broad pelvis, bicondylar angle of the femur, and robust, adducted hallux typical of bipedal humans combined wit h the barrel chest, long forearms, and curved fingers which are characteristic of more arboreal primates (Kimbel and De lezene 2009) These mixed morphologies can be related to Au. afarensis potential mobility patterns through the use of the theoretical framework of Behavioral Ecology.

PAGE 16

8 How Optimality in Behavioral Ecology Enables Modeling of Hominin Behavior I n paleoanthr opology, the morphologies and behaviors of hominins are often implicitly assumed to be adaptive for the ecological contex t in which the hominin is found (Reed 1997; White et al. 2009) an assumption deriv ed from evolutionary ecology (EE) (Foley 1992; Smith and Winterhalder 1992; Winterhalder and Smith 1992) EE is a theoretic al framework of adaptive m o d e s that states that every species evolves adaptations to their environment and these adaptations help the species survive within that environment (Winterhalder and Smith 1992) Environment is defined as that which is external to the species that affects their ability to reproduce and survive. Behavior is a part of ad aptation, the study of which falls under the subset of EE called Behavioral Ecology (BE). EE and BE follow the hypothetic a l deductive method of simple model creation and testing against empirical evidence (Winterhalder and Smith 1992) The purpose is to mod el the costs and benefits of a behavior such that the benefits outweigh the costs and would therefore be selected for (Foley 1992). The behavior modeled generally maximizes the benefit of overall fi tness (Dunbar 2001; Kelly 2007) Fitness is trac ked by defining a proxy cost such a s energy or time allocation that is optimized The assumption of optimality is inherent within this framework; an organism will pick the optima l behavior given t he tradeoff between costs and benefits associated with utilizing that behavior within a set of external constraints (Dunbar 2001; Kelly 2007). I n general, behaviors favor optimization, and this provides a way to model predictions against selective forces ( Kelly 2007). However, human behavior is often found to be more complex and not always optimal because the cultural aspects of humanity often obscure or obviate optimality (Kelly 2007). On the other hand modern cultural behavior is

PAGE 17

9 not thought to appear un til well after Au. afarensis went extinct (Marwick 2003) ; therefore, cultural behavior would not necessarily affect the pursuit of optimality T he mobility models developed for this project predict that Au. afarensis chimpanzees, and humans will avoid relatively more difficu lt terrain to expend less time on the search for food or mates They follow BE framework and assume that behavior is optimized by limiting costs in time by avoiding the environmental condition of steep slopes This is because ste ep gradients t ake more time and energy to climb up or down than flat terrain (Alexander 2002; Terrier et al. 2001). Within the selective forces of the environment, different animals will avoid different terrain elements depending on their mobility pattern. These terrain elements can be modeled within a framework of specific optimality to the capabilities of that particular species, including Au. afarensis humans, and chimpanzees. The Complexities of Mobility and How to Render It Simple Foley (1992) recognized early on that the best way to relate hominin characteristics to selective conditions was to use a modeling approach. To do so, behavioral adaptations must be identified along with the context of costs and benefits. For instance, the locomotor behavior s of hominins and why bipedalism evolved has been modeled within the context of the energetic advantage bipedalism gave hominins over any other form of bipedalism. Bi pedal hominins have been predicted to save more energy than chimpanzees within the same or a larger day range and with increased body size (Foley 1992). Despite findings such as these a few cautions are necessary to mention. First, r eality is extremely complex and not every aspect of that complexity can reasonably be modeled (Dunbar 2001). For this reason, models tend to be simplistic, a feature often criticized but necessary because it allows researchers to isolate those variables that are most import ant for

PAGE 18

10 explaining a behavior. Second, m odels require the researcher to be explicit about his or her assump tions of how a behavior works and its relationship to the environment (Dunbar 2001; Winterhal der and Smith 1992). Finally models allow researchers to determine if a behavior is fitness maximizing, assuming that the model adequately frames reality (Dunbar 2001 ). For this, the biology of the organism of interest and the processes being modeled must be well understood (Dunbar 2001) In the fossil record, behavior is studied indirectly and is therefore modeled within a cost/benefit analysis framework (Foley 1 992). Because animal behaviors are generally adaptive for a particular ecological demand Au. afarensis mobility patterns must have be en adaptive as well. It remains especially important to find the ecological context to which it was adapted and model pote ntial mobility pattern s within that context. If Au. afarensis was a highly m obile species that moved across the la ndscape in an optimal way, then a methodology must be used that capture s the interplay between environment and mobility There are many appro aches to modeling the potential mobility of Au. afarensis each with its strengths and weaknesses. The comparative method is one often used to understand early hominin behavior a tradition continued in this research Au. afarensis are regularly compared t o chimpa nzees or bonobos, because the fu rther back in time the hominin of interest is, the more similar it should be to the l ast common ancestor (LCA) of the Pan H ominin lineage (Foley 1992); a species sometimes estimated to be v ery chimpanzee like (Begun 2010; Leonard and Robertson 1995) Yet it is complicated by how derived chimpanzees have become since the LCA (White et al. 2009). D e s p i t e i s s u e s s u c h a s t h e s e c omparative studies between fossil hominins, nonhuman primates, and humans are often used to reconstruct the social life and behavior of early ho minins (Koenig and Borries 2012)

PAGE 19

11 in part because the use of analogy and proxies are a key part of pal eoanthropological investigation s It is important to determine if the analogies are viable since the process of elimination will bring researchers closer to the reality. Analogy of similarities and differences in anatomy sparked interest in modeling th e locomotor patterns of Au. afarensis (Pontzer and Wrangham 2004 ; Sellars et al. 2005 ) The efficiency of their bipedality and how arboreal they were is a highly debated issue in paleoanthropology (see Pontzer and Wrangham 2 004 ; Sellars et al. 2003 ; Hunt 1 994 ; and Stern and Susman 1983 ) Often, Au. afarensis mix of bipedal and arboreal characteristics are compared to chimpanzees, which have a combination of arboreal and quadrupedal terrestrial morphologies (Pontzer and Wrangham 2004) Pontzer and Wrangham (2004) argue that climbing in chimpanzees is maintained by selective pressure in order to increase climbing safety, speci fically because their arbo real adaptations incur rather than decrease locomotor costs. Similarly, Pontzer and Wrangham (2004) believe that early hominins may have kept their arboreal adaptations to decrease the likelihood of falling from trees, because of environmental change an d because the need for efficient terrestrial locomotion would have created selective pressure for bipedalism, which decreases energetic walking costs (Pontzer and Wrangham 2004) T he form and efficiency of the bipedalism Au. afarensis had requires further discussion to elucidate how potential mobility patterns can be modeled Stern and Susman (1983) hypothesized that because of their range of arboreal traits, Au. afarensis would have had bent knee, bent hip (BKBH) bipedality similar to chimpanzees when walking on two legs (Sellars et al. 2005) Crompton et al. (1998) found t hat BHBK would not have been an effective form of locomotion, and Sellars et al. (2005) found that BHBK locomotion nearly doubled the energetic cost of locomotion. However, locomotor

PAGE 20

12 cost does not change for chimpanzees between using quadrupedalism versus bi pedalism; instead it is mechanics that increase or decrease costs (Pontzer et al. 2014). An alternative hypothesis proposed by Hunt (1994) states that the mixture of arboreal and bipedal characteristics results from the need to collect fruit by both hanging from trees by the ir arms and shuffling bipedally on the ground. Sellars et al. (2005) are proponents of a third hypothesis that argues that based on their habitu al bipedal characteristics, Au. afarensis would have been habitual, erect bipeds Au. afarensis had the correct skeletal morphology for efficient and stable erect bipedalism (Sellars et al. 2005) Therefore, this species was most likely a fully co mpetent biped, although shorter and slower than humans (Sellars et al. 2005) This list of locomotor patterns is by no means compr ehensive, these merely highlight the range of patterns hypothesized for Au. afarensis Th i s debate about locomotor pattern is critical because assumptions about the type of locomotion and i t s efficiency greatly affects any model of potential mobility. The se d ebates also demonstrate that testing a single model of Au. afarensis potential mobility wou ld not suffice because t h e c o s t o f walking is a ffected by whether they are assumed to walk like slow humans, like chimpanzees on two legs, and so on. Indeed, models should be tested for their se nsitivity to walking a c t i v i t y adjustments and against models of other closely related species Therefore, comparing chimpanzee Au. afarensis and human potential mobility patterns is a cr itical part of this analysis Because locomotion is complex, a proxy must be used in the models to get at potential mobility pattern For this research walking speed was the proxy used to represented Au. afarensis humans and chimpanzees walking activity Relative walkin g speeds have been measured f or humans on treadmills (Terrier et al. 2001) and in the African savanna

PAGE 21

13 (Musiba et al. 1997) The relative speeds of wild chimpanzees have been measured by observation in a natural setting (Leonard and Robertson 1997; Pontzer and Wrangham 2004) Au. af arensis relative walking speeds were measured indirectly by foot length measure ments on footprints (Charteris et al. 1982 ; Tuttle et al. 1990 ) and by scaling down human neuromusculoskeletal models to Au. afarensis size (Nagano et al. 2005) There are problems with all these methods of determ ining relative walking speed. First, treadmills do not accurately represent natural landscapes and how speeds change due to the nature of the substrate Second, Au. afarensis was not living in an open grassland so those speeds are probably much fas ter than they would be through the mosaic of open to closed woodlands and forests in which this species lived (Andrews and Bamford 2008; Kimbel and Delezene 2009) Third, when observing chimpanzees from a di stance, there is always a chance some behaviors will be missed or the chimpanzee may be lost during pur suit. Fourth, measurements from footprints may inflate or deflate the walking speed since footprint length and stride do not correlate perfectly (Charteris et al. 1982) Finally, the scaling down of a human model to Au. afarensis size assumes that both species have the same neuromusculoskeleton and t hey move in exac tly the same manner However, although Au. afarensis are generally accepted to be efficient bipeds (Sellars et al. 2005) they hav e many different mor phologies from humans such as a wide r pelvis ( Kimbel and Delezene 2009) This research attempts to alleviate these problems by averaging walking speeds calculated using several methodologies, assuming it will even out the variance The Effect of the Environment on Mobility Relative w alking speed alone cannot capture how an animal moves across a dynam ic and vast landscape; d ifferent conditions in the environment can have a n effect on how fast an

PAGE 22

14 animal walks For example, one of the factors that affects bonobo distribution is l arge rivers which create a geographic barrier between bonobo populations w hich reduces ge ne flow (Eriksson et al. 2004) Barriers such as this can be impenetrable, causing isolation and speciation. However, if a continuous habitat exists on either side of a river and it is possible to cross, large mammals are generally capable of traver sing it (Eriksson et al. 2004) Additionally, i n humans, Hughes et al. (2007) showed with their Stepping Out model that vegetation is a critical factor to hum an mobility Clear ly, it is important to pinpoint landscape elements that would most affect potential mobility patterns, and how they would be affected. When creating a model for a ~3.8 Ma hominin, the environmental features may not have been comparable to that o f modern humans so using modern environmental data can be problematic (Bailey et al. 2011) However, the environmental context of 3.8 Ma is difficult to reconstruct with any precision, leading t o generalized statements that cover large areas For example, Andrews and Bamford (2008) reconstruction at Laetoli lists the types of plants that were found but not their quantity or exac t distribution. This imprecision is due to taphonomic processes such as time averaging and poor preservation which is unavoidable (Bonnefille 2010 ; Rossouw and Scott 2011 ) This is a problem that plagues all environmental variables going bac k to such a d istant age. Paleoenvironmental r econstructions are precise enough for most purposes, but when using a model that requires more specific geographical data, they are too large scale For example, this project would benefit from more exact knowledge of densit y, size, and distribution of trees, grasses, and shrubs to enable a model of the impenet rable versus open vegetation Au. afarensis would have had to cross or avoid Additionally, there is little data available describing how the types of vegetation that ar e documented affect movement of

PAGE 23

15 humans, chimpanzees, and Au. afarensis There are no regional scale data or models on slope or vegetation currently available, so other data were sought for this project. The scope of this s t u d y a l s o did not allow the incorpo rations of models of water b arr iers such as the locations, sizes, depths, and salinity of rivers or lakes across Eastern Africa. However, they have a recognized role in restricting mobility in bonobos, for example (Eriksson et al. 2004) and in providing habitats and water for Au. afarensis ( Kimbel and Delezene 2009; Su and Harrison 2008) While reconstructions of lake locations do exist, these do not includ e the other important factors that restrict or sup port movement, such as size and salinity. It is known that the rifting processes affecting the East African Rift Syste m opened and closed lakes some were saline and others freshwater, and they varied size and depth over time (Tiercelin and Lezzar 2002 ; Cuthbert and Ashley 2014 ) T his is a n important start, and bringing together water attribute data is a viable future direction for research. With paleoenvironmental reconstructions not feasible a t t h i s i n i t i a l s t a g e modern environmental data must be used. Slope is an important factor in human mobility because energy expenditure increases with slope ( Minetti et al. 2002; Todd and White 2009) although humans tend to adjust their walking pattern to minimize energy costs (Alexander 2002). If steeper slopes al so increase energetic expenditure for Au. afarensis for the assumption of optimality under BE to hold, they would have had to minimize that in some way. Slope may have been important to hominin mobility in other ways ; for example, King and Bailey (2006) s uggest that the evolution of bipedalism may have been driven by a landscape of rough and hilly terrain that a l l o w e d hominins t o efficiently disperse to track resources or preferred habitats (Potts 1998) S lopes and topography in general may have had

PAGE 24

16 a drastic effect on energetic costs and the evolution of bip edalism. Therefore, the effect of slope on behavior should be tested in the mobility model. Modern elevation data must be used cautiously Naturally, the landscape does not look exactly the same today as it did in the past. O ver time geological processes cause d once active parts of the African Rift to uplift and become in active, and once they are inactive, forces of erosion obscure geological features, causing the topography to change since the time of Au. afarensis (Bailey and King 2011). Despite this, mo dern elev ation data will be used in the potential mobility models at Laetoli and across Eastern Africa. This approach was inspired by Egeland et al. (2010) who used modern slope data to model a route from Ubeidiya, Israel to Dmanisi, Georgia that hominins potentially would have take n dur ing dispersal events Egeland et al. (2010) found that their predicted route matched up wit h known Lower Paleolithic sites. Furthermore after ground surveys they found 25 new Upper and Lower Paleolithic sites. Egeland et al. (2010) also stated that a lthough paleoenvironmental data is available, it was too coarse for the predictive model which is an issue discussed above However, modern d igital elevation data with 30 meter resolution can easily be obtained from the EarthExplorer, a spatial database m aintained by the USGS Alternatively, estimates of paleoslopes could be derived from c omparisons with active parts of t he African Rift (Bailey and King 2011), but that is beyond the scope of this work T he literature reviewed above demonstrates that Au. a farensis were eurytopic living in highly variable, complex environment and yet they someho w managed to get across Eastern Africa. If Au. afarensis was behaving optimally when moving across the landscape as BE would sug gest, then they were likely avoiding costly terrain such as steep slopes to conserve energy Data from biomechanical work demonstrate how quickly Au. afarensis

PAGE 25

17 c ould have walked, a proxy for locomotor efficiency that changes with slope. What remains is to merge these together into models of potential mobility that predict how Au. afarensis would have moved across Eastern Africa if they were behaving optimally. The data incorporated into these models are described in the next section.

PAGE 26

18 CHAPTER II MATERIALS Potential mobility m odels were crea ted for Au. afarensis chimpanzees, and humans for comparability, since chimpanzees and humans are the closest living relatives of Au. afarensis and extant species mobility may lend insight into the mobility of Au. afarensis The data required for these mo dels include Digital Elevation Models, the spatial locations of the sites, and the relative speed at which each species walks Digital Elevation Models (DEMs) were downloaded from EarthExplorer (online) for areas in Ethiopia, Kenya, Tanzania, and parts of Uganda, Somalia, South Sudan, Sudan, Eritrea, Rwanda, and Burundi. These covered all of the localities and most of Eastern Africa in order to make a continuous surface for analysis. Shapefiles containing hominin localities were not readily available, but general coordinates of localities are often provided in publications. The coordinates of localities at Laetoli containing Au. afarensis fossils were obtained from a variety of sources, including Harrison et al (in press), Ditchfield and Harrison (2011) Harrison and Kweka (2011) and a Garmin Oregon GPS unit that Dr. Charles Musiba used at Laetoli ( Fig. 2 All l ocalities r ecorded at Laetoli Tanzan ia o verlaying a Google Earth i mage Fig 2 A l l l o c a l i t i e s r e c o r d e d a t L a e t o l i T a n z a n i a o v e r l a y i n g a G o o g l e E a r t h i m a g e ). The coordinates of the Eastern African localities of Kantis, Dikika, Hadar, and Woranso Mille were obtained from Mbua et al. (2016) Alemseged et al. (2005) Johanson et al. ( 1982) and a map provided by Dr. Charles Musiba that was then georeferenced, respectively. Laetoli was chosen as the first study area because more data on Laetoli localities was available to the author than any other site. It is also the second most produ ctive Au. afarensis

PAGE 27

19 fossil site and has over 20 localities (Harrison and Kweka 2011; Musiba et al. 2008) making it a good sample area. For the second study area of Eastern Africa Laetoli was chosen as the source site because it is the oldest known Au. afarensis site (Kimbel and Delezene 2009) and is therefore possibly at or near the place at which they first evolved It is also po ssible that the area of initial evolution is in the Awash region of Ethiopia instead, since there are more localities in Ethiopia (Kimbel and Delezene 2009) and there is no way to know exactly where and when Au. afarensis evolved However, the oldest specimen found in that region is 3.58 Ma (Haile Selassie et al. 2010) which Laetoli predates by 50 270 K y a making Laetoli the best candidate Au. afarensis was chosen as the subject because much research has gone into the study of their biomechanics, environment, and behavior, but there is much more to learn. L ittle has been done to model their actual movement between sites and localities even t hough it is of great interest. This is parti ally explained by the resolution ; some of the localities at Laetoli are relatively close to each other (Charles Musiba, personal communication October 2017) while the known sites are all over Eastern Africa Data of relative walking speed used in this st udy was obtained for humans from T errier et al. (2001) Musiba et al. (19 97) Alexander ( 2002) Verberg et al. (2004) and Moreno Sanchez et al. (2012) for differ ent degrees of slope. These data were generally collected for humans on treadmills with the exception of Musiba et al. (1997), who collected data on the Hadzabe and Machiguengas. Hunt ( 1989 ) and Pontzer and Wrangham ( 2004 ) provided relative walking speeds for chimpanzees in the wild. Finally, Au. afarensis relative walking speeds were provided for Lucy by Nagano et al. (2005) and f or the G1 and G2 /3 trackways by Charteris et al. (1982)

PAGE 28

20 CHAPTER III METHODS The Role of GIS Cost Distance Analysis in Models of Mobility The models proxy for locom otor pattern is walking speed, and for past terrain the proxy is modern elevation data. Bringing these together into p otential mobility models is the role of GIS. The models created here emphasize landscape connectivity in that they account for the structu re of the landscape as well as the mobility of the species (Adriaensen e t al. 2003) Least cost modeling in GIS focuses on ascertaining areas of least cost between points. It pro vides a way to incorporate proxies for environmental conditions and cost to calculate effective distance. Effective distance is the distance it takes to get somewhere while avoiding hindrances and utilizing facilitators to movement on the landscape. This is opposed to Euclidean or straight line distance that ignores real world terrain phenomena. In least cost modeling, every grid cell or landscape unit is given a value of friction based on how it helps or hinders the mobility of an animal. This is meant to represent the way the animal experiences the landscape (Adriaensen et al. 2003) In cost distance analysis, a subset of least cost modeling, a source layer with the starting point(s) or area(s) of interest and a fricti on surface composed of user defined costs to movement such as walking speed, are the inputs (Adriaensen et al. 2003) These costs or friction values can be based on expert experience in the relevant area, secondary published data from experts, or field data. It is important to make these as accurate as possible because they drastically affect the outcome, and are the link between the spatial information and the mobility of the animal. E very cell value in the output is calculated as the cost to reach any given cell from a source, the average cost to move through that cell, and the average cost to

PAGE 29

21 move through the source cell. Movement through cells can happen in any direction, including diagonally for which the square root of two is multiplied by the cost The least cost algorithm used in this study was adopted from Adriaensen et al. (2003): N i+1 = N i + (r 1 + r i+1 )/2 or N i+1 = N i + 2**0.5 + (r 1 + r i+1 )/2 where N i is the accumulated cost in cell i, r i is the resistance value in cell i i is the source cell, and i + 1 is the target cell. The output is a surface of cost from th e source with each cell value being the lowest cost to move to it over the friction surface. The output map contains banded patterns where all the cells in the same band have effective distance in the same class. The sensitivity of the effective distance t o variation in friction values can be studied to create a robust, validated predictive model (Adriaensen et al. 2003) While cost distance modeling is often used to understand mobility in animals and prehistoric populations, generally these go a step beyond least cost surfaces to making least cost paths (Davidson et al. 2013; Egeland et al. 2010; Rahn 2005) These allow researchers to predict exactly how a population was getting from one specific place to another so they assume an origin and des tination. These paths can also be ground truthed to find evidence that the paths are correctly predicting how a population would have moved (Davidson et al. 2013; Egeland et al. 2010; Rahn 20 05) However, least cost surfaces have been used in archaeology to understand accessibility and the interaction between settlements and the landscape over a large area (De Silva and Pizziolo 2001; Howey 2007; van Leusen 1998). The differences are subtle but important. Archaeological least cost path studies often assume humans have some knowledge about the study area and would have known exactly

PAGE 30

22 where they wanted to go. In contrast, Au. afarensis would not have a physical destination such as Hadar in mind, but instead would have moved along areas of least resistance in their search for food and/or mates. This research explored the interaction between terrain and mobility, focusing on Au. afarensis movement across the landscape as a whole and the accessibili ty of known sites. Essentially, cost distance analysis allows researchers to develop predictive model s of the movement of different species across a landscape while emphasizing different landscape elements ( and their facilitating or hindering properties ) Furthermore, i t calculates some cost of getting from a source to anywhere in the study area, and allows researchers to gain insight into the species potential mobility patterns In relation to the models created here, cost distance analysis allowed a vi sualization of a wave of movement of Au. afarensis humans, and chimpanzees from Laetoli, Tanzania to the Awash region of Ethiopia while enabling a change in the cost parameters to test sensitivity. In this way, potential mobility patterns of Au. afarensis can be better understood through the use of th is modeling technique in GIS. Preparatory Work The goal of this research was to explore the possibility of modeling Au. afarensis potential mobility pattern u sing the method Cost Distance A nalysis (CDA) in A rcGIS and modern elevation and walking speed data in order to understand how they got to their known sites Like the CDA done by Egeland et al. (2010), a match between the model and the distribution of known localities p rovides some support for the accurac y of the model. Several rounds of analysis were done for humans, chimpanzees, and Au. afarensis for comparison The first used modern slope data to model potential mobility patterns in the Laetoli are a only The second analysis involved model s from Laetoli across Eastern Africa again using slope

PAGE 31

23 data Known Au. afarensis localities were overlaid on the second analysis to determine if they lie on least cost areas, which would indicate that 1) the potential mobility pattern could be correctly capturing Au. a farensis 2 ) Au. afarensis was moving optimally across the landscape by avoiding steep slopes, and 3 ) the terrain has not changed so subst antially that it renders modern elevation data useless for these model s Au. afarensis Lo calities Digitizing and projecting the spatial locations of sites was the first component of this analysis The Laetoli coordinates were mostly in UTMs, but some came as Degrees, Minutes, Seconds (DMS) and had to be converted to UTMs. These were digitized into a feature class and projected in UTM Zone 36S (Fig. 2) because that is the zone in which Laetoli lies and is most like ly the projection used to obtain the spatial data The accuracy of the Eastern African location al data is unknown because the coordin ates are given in degrees and minutes without seconds rendering them imprecise However, in the context of the Eastern African analyses, the area is so large that any inaccuracy in the location is not likely to make a significant difference in this analysis. The Eastern African localities were projected into a coordinate system made specifically for the study area. The projection is an Alb ers Equal Area Conic with a central meridian of 37 which is roughly the center of the study area and standard parallels of 7 and 1, dividing the study area roughly in one and two thirds. The purpose was to minimize the distance distortion. Slope The En vironmental Variable The models were tested using modern slope data alone as discussed in the literature review Briefly, although it is probable that the terrain has changed in the last 2 3 million

PAGE 32

24 Fig. 2 All l ocalities r ecorded at Laetoli Tanzan ia o verlaying a Google Earth i mage years (King and Bailey 2006), there is no detailed dat a on past slope available. Yet, modern elevation data was successfully used to model hominin mobility from Israel to Georgia (Egeland et al. 2010) T hese models test if such a thing is possible for more ancient hominins despite the analysis being done on the tectonically active Easter n African Rift System (King and Bailey 2006) M odern slope data can be derived from Digital Elevation Models (DEMs). The DEMs used were Shuttle Radar Topography Mission (SRTM) 1 Arc Second Global in GeoTIFF format downloaded from USGS EarthExplorer. For L aetoli, the DEM was projected into UTM Zone 36S because UTMs preserve distance at larger scales, and because

PAGE 33

25 the entire Laetoli site is within this UTM zone. Across Eastern Africa, DEMs were merged into one raster using the Mosaic to New Raster tool and th e Mosaic tool and were projected into East Africa Albers Equal Area Conic. The slope tool was then used to convert the DEM s to slope percentages (Fig 3) Fig. 3. A: Digital e levation m odel of the Laetoli area. B: Slope created from the DEM. Relativ e Walking Speed The Cost Relative w alking speed was chosen as the proxy for locomotor pattern based on published data on the walking speed of all of the species included in the analysis. Generally, energy is used as a proxy for locomotor efficiency, but t he energetic output for each of the species was reported in different units that would be difficult to convert. In addition, walking speed is often used to understand accessibility in environmental impact studies (Moreno Sanchez et al. 2012; Verburg et al. 2004) and it is a simple uni t of measurement to

PAGE 34

26 understand. However, different sources provide different relative walking speed values for each species. As Leonard and Robertson (1997) and Pontzer and Wrangham (2004) demonstrate, relative walking speed is extremely variable in chimpanzees and humans, and is a function of variables other than terrain such as height, sex, and age. This problem was solved for this study by averaging the walkin g speeds provided by each source for each species ( inclusive of different heights, sexes, and ages ) Secondary published literature was analyzed to determine relative speed data on flat land for humans, chimpanzees, and Au. afarensis This was provided in either kilometers per hour or it was converted to kilometers per hour from meters per second. For humans, speed on different gradients was also discovered in the literature. Human speed for 0 24.9% slope were derived from Terrier et al. (2001) using their low speed data because it better matches speeds of modern Hadzabe and Machiguengas peoples of Tanzania and Peru determined by Musiba et al. (1997) For slopes greater than 25%, it is more optimal to zig zag upslope rather than go straight up; additionally, above 20% slope, cost of walking is 2.5 times that of walking on fla t land for humans (Alexander 2002) Therefore, the speed for 25 49.9% were calculated at 2.5 times greater than flat land walking. The excessively difficult slope cutoff was given as 50% by Verberg et al. (2004) and Moreno Sanchez et al. (2012) and this was adopted for this study such that anything above 50% (26.57) slope was considered an absolute barrier to movement. While relative walking speed has been estimated on flat terrain for Au. afarensis and chimpanzees, no speed data on other gradients w as found in the literature This is important because humans are known to walk increasingly slowly with increasing slope (Terrier et al.

PAGE 35

27 2001). Furthermore, in order to make relative walking speeds across species on different slopes comparable, relative walking speed must be estimated for each slope category. To determine the speed of chimpanzee s and Au. afarensis on similar slopes to the data for humans, an approach derived from Taylor et al. (1972) was used. These authors state that an animal that is smaller than another uses x times more energy than the larger animal, but the additional energy to lift 1 kg of body weight per meter while running uphill will be similar for both species no matter the size difference (Taylor et al. 1972) If this can be extrapolated onto time rather than energy, then a chimpanzee wal king at 2.99 km/hr walks 1.3 times more slowly on level ground than does a human walking 3.8 km/hr (3.8/2.99). Therefore, a chimpanzee may take 1.3 times more time walking on any elevation than humans so all human speeds can be divided by 1.3 to get the ch impanzee walking speed on any given slope. The base speed used for 0 4.9% slope for chimpanzees was the average of those provided for males and females from Pontzer and Wrangham (2004) which were derived from Hunt (1989) Au. afarensis calculations presented a unique pr oblem because there is no certainty about how quickly they would have moved in general Therefore, this species required a sensitivity analysis in which different parameters are used to estimate their relative walking speed. Au. afarensis base speed for 0 4.9% slope was the average of that for Lucy from Nagano et al. (2005) and for the G1 and G2 /3 trackways from Charteris et al. (1982) Because Au. afarensis is thought by many to be a fully competent biped, the model speed on different gradients was calculated by dividing the hu man speed by 1.79, the amount slower Au. afarensis walking speed is to human speed (3.8/2.12). However, the walking speed of Au. afarensis may be even slower than chimpanzees, so another calculation

PAGE 36

28 was done that was 1.4 times slower than chimpanzees (2.99 Lastly, because some paleoanthropologists believe that the locomotor capabilities of Au. afarensis were somewhere in between chimpanzees and humans, the final speed calculations were derived using the average walking speed o f humans and chimpanzees combined, the Table 1 : Speed and friction values on slopes for humans, chimpanzees, and Au. afarensis Slope 0 4.9% 5 9.9% 10 14.9% 15 24.9% 25 49.9% >50% Hum Speed km/h 3.8 1 3.6 1 3.3 1 2.8 1 2.3 1.8 2 Hum Friction Value min/m 0.01579 0.01667 0.01818 0.02143 0.02609 99 Chimp Speed km/hr 2.99 3 2.77 2.54 2.15 1.77 1.38 Chimp Friction Value min/m 0.02007 0.02167 0.02364 0.02786 0.03391 99 Slow hum an Au. afarensis Speed km/hr 2.12 4 2.01 1.84 1.56 1.28 1.01 Slow human Au. afarensis Friction Value min/m 0.02830 0.02983 0.03255 0.03836 0.04670 99 Slow chimp Au. afarensis Speed km/hr 2.12 4 1.98 1.81 1.54 1.26 0.99 Slow chimp Au. afarensis Friction Va lue min/m 0.02830 0.03033 0.03309 0.03900 0.04748 99 Intermediate Au. afarensis Speed km/hr 3.40 3.18 2.92 2.48 2.03 1.38 Intermediate Au. afarensis Friction Value min/m 0.01767 0.01884 0.02055 0.02422 0.02949 99 1 Terrier et al. 2001. 2 Assuming a 0.5 km /hr drop as indicated by the previous two speed drops 3 Hunt 1989, Pontzer and Wrangham 2004. 4 Charteris et al. 1982, Nagano et al. 2005. Species speed data was converted to friction values of minutes/meter. The minutes assigned to each cell in the fricti on raster represent the time in minutes per meter for moving through the cell. All friction values are given in Table 1

PAGE 37

29 F i g 4 F l o w c h a r t o f t h e L a e t o l i c o s t d i s t a n c e a n a l y s i s Potential Mobility Models 4 shows the flow chart for th e E a s t e r n A f r i c a analysis b u t t h e L a e t o l i a n a l y s i s i s t h e s a m e e x c e p t i n g t h e m o s a i c s t e p S lope percentages were reclassifie d into the friction values that represents how slow a given species walks based on the steepness of the slope. The Reclassify tool in ArcGIS was used to convert the slope DEMs into friction surfaces (

PAGE 38

30 Fig. 3 5 Fig. 6 ). However, reclassify doe s not support floating point rasters, or rasters with decimal values. Therefore, each slope range was reclassified into the friction value multiplied by 1,000,000. Then, the times tool was applied to the resulting raster with a multiplication value of 0.00 0001 to get a floating point raster of the correct friction values. This process was done for each set of human, chimpanzee, and Au. afarensis friction values to create five friction rasters for each study area Each raster was independently input into the Cost Distance tool as the cost raster, while a vector shapefile containing the fossiliferous Laetoli localities was input as the source. This created a raster of time band s from each fossiliferous locality outward. The localities of Kantis, Dikika, Hada r, and Woranso Mille were then overlaid onto the models of the Eastern African study area to determine if a r e a s of least time matched up with the localities.

PAGE 39

31 Fig. 3 The cost raster of the Au. afarensis Intermediate model created from reclassifying the slope raster.

PAGE 40

32 Fig. 4 A: The Digital Elevation Model created from merging the SRTM 1 Arc Second Global DEM files of the study area B: The slope raster created by using the slope tool on the DEM. C: A slope raster reclassified into the values for the Intermediate Au. afarensis model.

PAGE 41

33 CHAPTER I V RESULTS These two sets of analyses resulted in effective distance s urfaces in which each band of color represent s the amount of time it would take to get to the cells within the band from the source. The band size can be manipulated by defining the time li mits for each band. Bands are deformed by encounters with higher sl opes (i.e. mountains) which cost more to walk over than to walk around. This sort of deformation manifests itself as a dip. Other deformations in the form of protrusions or bumps into other bands represent encounters with lower elevation areas, such as fla t land, which cost less to walk over than around. Laetoli is relatively flat (Andrews and Bamford 2008) so the bands should be relatively concentric. In the Eastern Africa analysis, known localities should lie on protrusions, indicating that they would take less time to wa lk to. Models not showing this are either inaccurate to Au. afarensis potential mobility, strongly indicat e that modern slope data is not usable as a proxy for past terrain, or a marker of some other problem with the analysis. On the other hand, models sho wing localities on protrusions may reflect actual Au. afarensis mobility, indicat e that a modern slope can be used as a proxy for past elevation, and demonstrate that Au. afarensis did avoid steeper slopes as they moved from site to site Laetoli Visual inspection of the cost distance rasters at Laetoli ( Fig. 5 ) show that t he models did turn out as expected because the bands are nearly concentric circles confirming the fact that terrain at Laetoli is relative ly flat Inspection invo lved looking at the shape and width of the bands, which were standardized to take 9.23 minutes to cross. If there was a greater

PAGE 42

34 Fig. 5 The mobility models showing effective distance from each locality at Laetoli. The co lored bands represent the amount of time in minutes it would take Au. afarensis to get from any given locality to any given location on the map. The boundary of Tanzania was obtained from Thompson (2013)

PAGE 43

35 variety in topography, then there should be more dips or protrusions in the bands. In addition as expected the Human Model is faster than the Intermediate Model, which is faster than the Chimpanzee Model, which in turn is faster than the Slow Human and Slow Chimp anzee models. It is evident that the Human Model, for example, is fastest because the bands are wider than the those in the other models, thus indicating that more area can be covered in 9.23 minutes than any other model. There are o ther interesting points such as that t he difference between the Slow H uman model and Slow C himpanzee model appears so minimal, it can only be seen by overlaying the layers and turning one model on and off in ArcGIS For this reason, it is not necessary to use both of these models in f uture analyses since the information extracted is essentially the same. The di fferences between all the other models were drastic, which makes sense considering the larger difference s in cost values. All the Au. afarensis potential mobility models for Laetoli strongly suggest that it would not have taken very long to reach any of the given localities based on slope alone. In the slowest scenario for example, t h e S l o w C h i m p m o d e l (second down in Fig. 5 ), it would have taken ~101 minutes to get from Locality 9 to Locality 2 which are about 3 kilome ters apart as the crow flies This is calculated by adding the number of bands between Locality 9 to Locality 2 and multiplying by the width of the band (11 bands 9.23 minutes). If day range and activity patterns are the same for Au. afarensis and chimpa nzees as some researchers have suggested (Leonard and Robertson 1995) then Au. afarensis could have traveled between these two sites within a day, as chimpanzees spend ~116 minutes per day moving (Doran 1997; Leonard and Robertson 199 7; Matsumoto Oda and Oda 1998) Humans could trav el between these two sites in 55 minutes (6 bands 9.23 minutes) which

PAGE 44

36 is half of the least amount of time African hunter gatherers spent walking per day (1.9 hours) recorded by Leonard and Robertson (1997) Eastern Africa The Eastern African analyses confirmed that the Slow Human and Slow Chimpan zee models are nearly indistinguishable and therefore it is unnecessary to use both as parameters in future analyses ( Fig. 6 ) All of the m odel s did show an interesting trend, however. Each map has a series of bands which were adjusted to represent a month of walking assuming Au. afarensis has a similar activity pattern to chimpanzees. In other words, it would take Au. afarensis one month to get to the edge of the first band from Laetoli two to get to the edge of the second band and so on. The number of minutes it would take to travel across each band is also included in the legend. It is enlightening to see how the different speed paramete rs used affect how long it would have taken Au. afarensis to get from Laetoli to the other Eastern African Localities. However, the localities in Ethiopia (the northernmost two stars) reside within a protrusion in a band no matter what parameter is used. K antis (the star closest to Laetoli) resides in smaller protrusions in all of the analyses which can be seen m o r e clearly c l o s e u p Kantis currently lies on the margin of the East African Rift System ( EARS ) which could explain why the protru sion is not more distinct. In addition to the dips and protrusions discussed before, these maps show different colors of speckles, or dots of color that do not match the color of the band on which they lie. The colored speckles scattered across the map a re small areas that have high walking costs, evidenced by the fact that the speckles are always in higher cost colors than the bands in which they reside. These are high hills or mountains that are effectively impenetrable.

PAGE 45

37 Fig. 6 Mobility models for Eastern Africa showing effective distance from Laetoli. Four other known Au. afarensis localities were placed over the models, each correspon ds to a protrusion in the bands. A reference image of Eastern Africa includes the EA RS. The boundaries of Eastern African countries were obtained from Thompson (2013). East African Rift shapefile from ArcGIS Online.

PAGE 46

38 Unfortunately, the contrast between the speckles and elevation does not show up well in image files, so no close up image o f this phenomenon was included. All known Au. afarensis localities have neighboring speckles of high cost, seemingly restricting the possible movement of Au. afarensis This makes sense, considering all of the localities lay within the EARS or on the margi n, which seems to have enabled Au. afarensis to travel with relative speed within it, but it may also have effectively contained them since the edges of the EARS are associated with high walking cost. Fig. 7 shows the East African Rif t System overlaid on the Intermediate Model to demonstrate this effect. Fig. 7 A: Intermediate Model of mobility with the East African Rift System overlaid, showing that the EARS is mostly along areas of lower cost but is often bounded by speckles of higher cost. B: Satellite imagery of Eastern Africa showing the topography and how it ma t ches with the EARS and select Au. afarensis localities. E ARS shapefile from worldmap.harvard.edu B ackground of A is an ArcGIS basemap, Na tional Geographic World Background of B is a color shaded relief from the Shuttle Radar Topography Mission (NASA et al. 2004), and country outlines are from Thompson (2013).

PAGE 47

39 It is especially clear in the northern part of the study area that the EARS see ms to curve around the speckles of higher cost. While there are some higher cost speckles within the Rift System, in general it seems to be composed of lower cost landscape. This effect is the same for every model, with higher cost slopes surrounding the s ites and EARS, while the sites are located on lower cost protrusions. Once again, none of the models can be decisively ruled as the correct one.

PAGE 48

40 CHAPTER V DISCUSSION AND CONCLUSIONS Behavior is a complex thing to understand and construct regardless of species, but this project aimed to do just that with Au. afarensis potential mobility patterns. This is critical for understanding the relationship between hominin potential mobility behavior and past environments, and for predicting how Au. afarensis move d across the vastness of Eastern Africa. Slope is an important factor to human mobility and may have had an equal effect in Au. afarensis If that was so, Au. afarensis should have behave d opt imally by avoiding steep slopes; this hypothesis was tested by c reating potential mobility models of effective distance that demonstrate d that known Au. afarensis localities do lie on areas of lesser slope. The first set of models were tested on Laetoli and showed that most of the localities would be within a day range of each other. The second set of models tested on Eastern Africa showed that Au. afarensis probably was avoiding steeper slopes as it traveled between localities. These results indicate that modern slope data can be used as an environmental factor to reco nstruct Au. afarensis potential mobility, as well as the potential mobility patterns of older or younger hominins However, none of the potential mobility patterns can be decisively ruled out as the best fit for actual Au. afarensis mobility; further testi ng is necessary to pinpoint other important environmental conditions that affect potential mobility patterns. An interesting finding from this study was that if Au. afarensis was behaving optimally as it seems to have been, then it may have never left th e EARS. This is possible because it would take much more time to get out of the EARS, and provided that it had all of the resources Au. afarensis needed, it does not seem likely that many individuals of that species would have ventured outside of it. Furth ermore, each Au. afarensis found so far is within the EARS. Each map shows that the protrusions match up to the localities, so Au.

PAGE 49

41 afarensis was behaving optimally by avoiding steep gradients as it moved. In addition, the maps show that the EARS had terr ain that was conducive to speedier walking because it had less steep slopes w i t h i n which would have helped restrict Au. afarensis movement inside the system and funneled them from Laetoli to the other localities or vice versa. This supports 11) theory that there were selective forces acting on hominin evolution and mobility, and in this case, it appears one selective force may have been the steep gradients surrounding the EARS. Limitations This study was done to explore the possibility of u sing GIS cost distance modeling with environmental and mobility data to understand potential hominin mobility patterns. As such, it is in the primary stage and therefore has many limitations that need to be addressed. These relate to the use of the compara tive method, the GIS methodology, and geological and taphonomic biases. Use of the comparative method is common, but d irect comparisons between hominin other mammalian locomotion is not possible because there are no other striding bipedal mammals Withou t this the evolution of bipedalism can never be completely understood even when studying closely related animals because they do not have the same locomotor pattern (Cartmill 1990). Furthe r taking chimpanzees as the model for the LCA cannot explain how h ominins became bipeds rather than knuckle walking quadrupeds. Without proper analogy paleoanthropologists are left to face the fact that they can never really know what locomotor patterns preceded bipedalism and how it evolved (Cartmill 1990). This shoul d not be taken as a deterrent to doing comparative research; rather, it is something to take into account and be explicit about when reporting results of these studies.

PAGE 50

42 I t is important to remember that chimpanzee mobility cannot be used as a substitute t o study hominin mobility, and that conclusions will always be tentative when using this type of analogy. That is why this research used several models for Au. afarensis potential mobility using both chimpanzees and humans for comparison as well as fossil e vidence from Au. afarensis themselves. Still, none of the models can be said to represent Au. afarensis potential mobility pattern, in part because of the limitations associated with use of the comparative method. The slope data used for the cost distance modeling were obtained through DEMs with 30 meter resolution. This presents several problems. First, 30 m resolution is not precise enough to capture small variations in the topography. In fact, it generalizes the landscape such that if there are several changes in slope in a 30 m cell, only one slope percent will represent the whole. At a cartographically large scale, such as at Laetoli, this means that the concentric circles of the model outputs may not represent the actual variation of the modern landsc ape, and higher resolution models may look drastically different. On a cartographically small scale, such as that of Eastern Africa, this problem persists with the variance being evened out and subtle changes in topography being ignored. This limitation do es have a b e n e f i t f o r t h i s s t u d y however. Given the uncertainty of how well modern elevation data approximates past elevation, a higher resolution DEM would give these models false confidence. In other words, the data is as uncertain as its representativeness. Using DEMs with higher resolution may show more dips and protrusions in the cost bands, but that does not mean that those deformations are any more likely to reflect past topography than those found with 30 m resolution DEMs.

PAGE 51

43 Moreover, known sites may simply be on erosional landscapes that not only flattened the terrain, but also exposed fossils that could be more easily found by paleoanthropologists. Therefore, the DEMs may simply be reflecting modern erosion rather than past terrain Furthermore, places with le ss steep terrain are easier for researchers to get to, biasing where fossils are searched for. Finally, if Au. afarensis died and were buried on steep slopes, as they eroded out, gravity may have pushed them down only flatter land. If any of these alternat ive scenarios are true, it would lead to the false conclusion that Au. afarensis preferred areas of lesser slope rather than that the sites are a product of the geological process of erosion. Laetoli provides an example of the last limitation t o b e discussed he re, taphonomic biases. The fact that Au. afarensis fossils were found at 15 of the localities recorded for Laetoli is seemingly unsurprising given information of the relative flatness of the current land scape and that most of the localities are within a da y range of each other. Th is indicate s that it is possible that Au. afarensis was spread over more of this area, but due to taphonomic processes, evidence of such is now gone or buried. This species only constituted a small proportion of the large mammal fa unal group recovered at Laetoli (Su and Harrison 2008) which would make evidence that Au. afarensis was more widely spread difficult to find. The activity of carnivores at Laetoli only compounds this problem, especially because their size m ade them a target for predators and scavengers alike, creating an atmosphere in which their skeletal elements were more likely to be moved from their original position or completely removed from an assemblage. They were also too big to be completely buried by ashfall, which would have allowed their skeletal elements to escape from carnivore ravaging (Su and Harrison 2008)

PAGE 52

44 This is a sample of the types of taphonomic processes that make it difficult to find Au. afarensis sites and to know whe re they lived or died. For a study such as this, it means that there may b e places that Au. afarensis were that were not on areas of lesser slope, but because of poor preservation conditions these will never be found. If this is so, it would invalidate the co nclusions reached in this research. While the limitations in this section impact the acceptability and adequacy of these models, they also highlight parts of this research that could be improved in future phases. Future Directions None of the potential mobility models could be conclusively ruled as the best fit for Au. afarensis mobility The inclusion of other environmental conditions that affect mobility, such as water bodies or substrate properties should be included in future models to further refin e them. When more precise data on past terrain is available, these should also be incorporated into the models. Additionally, t here may be other important factors aside from landscape at play in Au. afarensis mobility that could be isolated such as how mu ch time is spent locomoting in trees and if they can avoid expending walking energy on slopes by climbing. As with any models, it is important to incorporate as many variables it takes to explain a phenomenon until including more variables ceases to add an y new information or becomes to o computationally complex (Dunbar 2001) Bringing in data on water bodies and climbing could be invaluable for understanding potential mobility patterns. Adjusting the relative walking speed constructed for Au. afarensis may benefit the models as well. As discussed before, the base speeds for Au. afarensis were estimates obtained from the Laetoli footprints and from Lucy However, performing further sensitivity analyses using relative walking speed estimates from their putativ e ancestor Australopithecus

PAGE 53

45 anamensis or descendant Australopithecus africanus may be enlightening. Using energy rather than relative walking speed as the cost in the models could also lend insight into how robust the models are. T he models presented here isolate likely potential mobility patterns and at this stage they cannot be used to pinpoint where fossil hominins are likely to be found. This type of analysis could be done by c reating least cost paths or corridors in conjunction with spectral signature analysis t o identify likely areas to be surve yed The least cost paths could re flect the parameters used in this study to create the modeled routes that Au. afarensis may have taken between localities. If fossils were found along these routes, it would be one way to verify the results of the project presented here. However, they are deterministic, assum ing known origins and destinations, which implies that Au. afarensis thoughtfully and purposefully went to each known site. Yet this remains a possible futu re direction because it provides a way to gather evidence in support of the potential mobility models. The research begun here will be continued with some of the future directions discussed above. Past water body and substrate data w i l l b e a c q u i r e d a n d add e d as environmental conditions that affect mobility If this makes better fit models, then the next phase would be to verify the models through groundtruthing. This would require identifying areas along least cost paths where fossils are likely to be found, and could substantiate the potential mobility pattern modeled. Conclusions Despite the limitations of the models and the uncertainty surrounding which mobility model is correct, information about the Eastern African paleoenvironme nt and the ecosystem in which Au. afarensis lived does lend itself to possible scenarios of potential mobility

PAGE 54

46 patterns For example, Au. afarensis was a generalist species capable of living in anything from the dryer open woodlands of Laetoli to the ripar ian forests of Hadar ( Kimbel and Delezene 2009; Su and Harrison 2008) During the Early Miocene to Late Pliocene, the rifting and volcanic activities in the EARS led t o the creation of small to large ephemeral and permanent lakes (Tiercelin and Lezzar 2002) which could have supported the movement of Au. afarensis through the EARS by providing drinking w ater and suitable habitats during wet periods. During dryer periods as lake levels dropped, they may have been able to survive in surrounding refugia in a way that is similar to other species that thrive in lacustrine environments (Martens 2002) W hen lake levels rose again, they could have moved from such refugia to nearby lakes, thereby encountering new and previously occupied areas (Martens 2002) In this way, Au. afarensis could have gone from Laetoli to Hadar, or between any of the loca lities. An ability to move faster, such as in the Intermediate Model, could have enabled greater survival as Au. afarensis was traveling between refugia and different water sources. However, if water sources were relatively stable, a slower moving Au. afar ensis as in the Slow Human model could still have had great survivability. While the models created above have limitations and would benefit from addition a l data and lines of inquiry, they do confirm that the potential mobility pattern of Au. afarensis can b e modeled to help understand how they moved across Eastern Africa. Furthermore, Au. afarensis does appear to have been behaving optimally by avoiding steep slopes. Additionally, successful models can be built off of modern slope data when past data is unav ailable. Finally, this research demonstrates that GIS is an underutilized tool in paleoanthropology with the potential to help researchers better understand hominin b e h a v i o r

PAGE 55

47 REFERENCES Adriaensen F, Chardon JP, De Blust G, Swinnen E, Villalba S, Gulinck H, and Matthysen E. 2003. The application of 'least cost' modelling as a functional landscape model. Landscape and Urban Planning 64:233 247. Alemseged Z, Wynn JG, Kimbel WH, Reed D, Geraads D, and Bobe R. 2005. A New Hominin from the Basal Member of the Hadar Formation, Dikika, Ethiopia, and its Geological Context. Journal of Human Evolution 49:499 514. Alemseged Z, Spoor F, Kimbel WH, Bobe R, Geraads D, and Reed D. 2006. A juvenile early hominin skeleton from Dik ika, Ethiopia. Nature 443(7109):296 Alexander RM. 2002. Energetics and optimization of human walking and running: The 2000 Raymond Pearl Memorial lecture. American Journal of Human Biology 14:641 648. Andrews P, and Bamford M. 2008. Past and Present Veg etation Ecology of Laetoli, Tanzania. Journal of Human Evolution 54:78 98. Bailey GN, Reynolds SC, and King GCP. 2011. Landscapes of human evolution: Models and methods of tectonic geomorphology and the reconstruction of hominin landscapes. Journal of Hu man Evolution 60:257 280. Bedaso ZK, Wynn JG, Alemseged Z and Geraads D. 2013. Dietary and paleoenvironmental reconstruction using stable isotopes of herbivore tooth enamel from middle Pliocene Dikika, Ethiopia: Implicaiton for Australopithecus afarensis habitat and food resources. Journal of Human Evolution 64:21 38. Begun DR. 2010. Miocene Hominids and the Origins of the African Apes and Humans. Annual Review of Anthropology 39:67 84. Bonnefille R. 2010. Cenozoic vegetation, climate changes and hom inid evolution in tropical Africa. Global and Planetary Change 72:390 411. Cartmill M. 1990. Human uniqueness and theoretical content in paleoanthropology. International Journal of Primatology 11(3):173 192.

PAGE 56

48 Charteris J, Wall JC, and Nottrodt JW. 1982 Pliocene hominid gait: New interpretations based on available footprint data from Laetoli. American Journal of Physical Anthropology 58:133 144. Crompton RH, Wang Weijie, LY, Gnther M and Savage R. 1998. The mechanical effectiveness of erect and "bent hip, bent knee" bipedal walking in Australopithecus afarensis Journal of Human Evolution 35:55 74. Curran S and Haile Selassie Y. 2016. Paleoecological reconstruction of hominin bearing middle Pliocene localities at Woranso Mille, Ethiopia. Journal o f Human Evolution 96:97 112. Cuthbert MO and Ashley GM. 2014. A spring forward for hominin evolution in East Africa. PLOSone 9(9):e107358. Davidson A, Carmel Y, and Shirli B D. 2013. Landscape Ecology 28:1465 1478. Deino, AL. 2011. 40 Ar/ 39 Ar Dating of Laetoli, Tanzania. In: Harrison T, editor. Paleontology and Geology of Laetoli: Human Evolution in Context. New York: Springer. p 77 97. De Silva M, and Pizziolo G. 2001. Setting up a "human calibrated" anisotropic cost surface for archaeological land scape investigation. BAR International Series 931:279 287. Ditchfield P, and Harrison T. 2011. Sedimentology, Lithostratigraphy and Depositional History of the Laetoli Area. In: Harrison T, editor. Paleontology and Geology of Laetoli: Human Evolution in Context. New York: Springer. p 47 76. Doran D. 1997. Influence of seasonality on activity patterns, feeding behavior, ranging, and grouping patterns in Tai chimpanzees. International Journal of Primatology 18(2):183 206. Dunbar RIM. 2001. Modelling Pr imate Behavioral Ecology. International Journal of Primatology 23(4):785 819. Egeland CP, Nicholson CM, and Gasparian B. 2010. Using GIS and ecological variables to identify high potential areas for paleoanthropological survey: An example from Northern A rmenia. Journal of Ecological Anthropology 14(1):89 98.

PAGE 57

49 Eriksson J, Hohmann G, Boesch C, and Vigilant L. 2004. Rivers Influence the Population Genetic Structure of Bonobos ( Pan paniscus ). Molecular Ecology 13(11):3425 3435. Foley RA. 1992. Evolutionary Ecology of Fossil Hominids. In: Smith EA, and Winterhalder B, editors. Evolutionary Ecology and Human Behavior. New York: Aldine de Gruyter. p 131 164. Grine FE, Ungar PS, Teaford MF and El Zaatari S. 2006. Molar microwear in Praeanthropus afarensis : Ev idence for dietary stasis through time and under diverse paleoecological conditions. Journal of Human Evolution 51:297 319. Haile Selassie Y, Latimer BM, Alene M, Deino AL, Gibert L, Melillo SM, Saylor BZ, Scott GR, and Lovejoy CO. 2010. An early Austra lopithecus afarensis postcranium from Worano Mille, Ethiopia. PNAS 107(27):12121 12126. Haile Selassie Y, Melillo SM, Ryan TM, Levin NE, Saylor BZ, Deino A, Mundil R, Scott G, Alene M and Gibert L. 2016. Dentognathic remains of Australopithecus afarensis f rom Nefuraytu (Woranso Mille, Ethiopia): Comparative description, geology, and paleoecological context. Journal of Human Evolution 100:35 53 Harrison T. 2011. Laetoli revisited: Renewed paleontological and geological investigations at localities on the E yasi Plateau in Northern Tanzania. In: Harrison T, editor. Paleontology and geology of Laetoli: Human evolution in context. New York: Springer. p 1 15. Halsey LG and White CR. 2012. Comparative energetics of mammalian locomotion: Humans are not differen t. Journal of Human Evolution 63:718 722. Harrison T, and Kweka A. 2011. Paleontological localities on the Eyasi Plateau, including Laetoli. In: Harrison T, editor. Paleontology and geology of Laetoli: Human evolution in context. New York: Springer. p 1 7 45. Howey MCL. 2007. Using multi criteria cost surface analysis to explore past regional landscapes: A case study of ritual activity and social interaction in Michigan, AD 1200 1600. Journal of Archaeological Science 34:1830 1846. Hunt KD. 1989. Pos itional behavior of Pan troglodytes in the Mahale Mountains and Gombe Stream National Parks, Tanzania: University of Michigan.

PAGE 58

50 Hunt KD. 1994. The evolution of human bipedality, ecology and functional morphology. Journal of Human Evolution 26:183 202. Johan son DC, Taieb M, and Coppens Y. 1982. Pliocene hominids from the Hadar Formation, Ethiopia (1973 1977): Stratigraphic, chronologic, and paleoenvironmental contexts, with notes on hominid morphology and systematics. American Journal of Physical Anthropology 57:373 402. Kelly RL. 2007. The Foraging Spectrum: Diversity in Hunter Gatherer Lifeways. New York: Percheron Press. 446 p. Kelly RL, and Todd LC. 1988. Coming into the Country: Early Paleoindian Hunting and Mobility. American Antiquity 53(2):231 244. K imbel WH. 2007. The species and diversity of Australopiths. In: Henke W, and Tattersall I, editors. Handbook of Paleoanthropology. Berlin Heidelberg: Springer. p 1539 1573. Kimbel WH, Lockwood CA, Ward CV, Leakey MG, Rak Y, and Johanson DC. 2006. Was Aus tralopithecus anamensis Ancestral to A. afarensis ? A Case of Anagenesis in the Hominin Fossil Record. Journal of Human Evolution 51:134 152. Kimbel WH, and Delezene LK. 2009. "Lucy" redux: A review of research on Australopithecus afarensis Yearbook of Phy sical Anthropology 52:2 48. King G, and Bailey G. 2006. Tectonics and Human Evolution. Antiquity 80:265 286. Koenig A, and Borries C. 2012. Hominiod Dispersal Patterns and Human Evolution. Evolutionary Anthropology: Issues, News, and Reviews 21(3):108 112. Kovarovic K and Andrews P. 2011. Environmental change within the Laetoli fossiliferous sequence: Vegetation catenas and bovid ecomorphology. In: Harrison T, editor. Paleontology and Geology of Laetoli: Human Evolution in Contex t. New York: Springer. p 367 380. Leonard WR, and Robertson ML. 1995. Energetic efficiency of human bipedality. American Journal of Physical Anthropology 97:335 338.

PAGE 59

51 Leonard WR, and Robertson ML. 1997. Comparative primate energetics and hominid evolution. American Journal of Physical Anthropology 102:265 281. Martens K. 2002. Redundancy and ecosystem stability in the fluctuating environments of long lived lakes. In: Odada EO, and Olago DO, editors. The East African Great Lakes: Limnology, palaeolimnology and biodiversity. Netherlands: Kluwer Academic Publishers. p 309 319. Marwick B. 2003. Pleistocene exchanges networks as evidence for the evolution of language. Cambridge Archaeological Journal 13(1):67 81. Matsumoto Oda A, and Oda R. 1998. Changes in the activity budg et of cycling female chimpanzees. American Journal of Primatology 46:157 166. Mbua E, Kusaka S, Kunimatsu Y, Geraads D, Sawada Y, Brown FH, Sakai T, Boisserie J R, Saneyoshi M, Omuombo C et al. 2016. Kantis: A new Australopithecus site on the shoulders of the Rift Valley near Nairobi, Kenya. Journal of Human Evolution 94:28 44. Moreno Sanchez R, Torres Rojo JM, Moreno Sanchez F, Hawkins S, Little J, and McPartland S. 2012. National assessment of the fragmentation, accessibility and anthropogenic press ure on the forests in Mexico. Journal of Forestry Research 23(4):529 541. Musiba CM, Tuttle RH, Hallgrimsson B, and Webb DM. 1997. Swift and sure footed on the savanna: A study of Hadzabe gaits and feet in Northern Tanzania. American Journal of Human Bio logy 9:303 321. Musiba CM, Mabula A, Selvaggio M and Magori CC. 2008. Pliocene animal trackways at Laetoli: Research and conservation potential. Ichnos 15:166 178. Nagano A, Umberger BR, Marzke MW, and Gerritsen KGM. 2005. Neuromusculoskeletal computer modeling and simulation of upright, straight legged, bipedal locomotion of Australopithecus afarensis (A.L. 288 1). American Journal of Physical Anthropology 126:2 13. NASA, JPL, and NIMA. 2004. SRTM Data Release for Africa, Colored Height. Online. http s://photojournal.jpl.nasa.gov/catalog/PIA04965

PAGE 60

52 Pontzer H, and Wrangham RW. 2004. Climbing and the Daily Energy Cost of Locomotion in Wild Chimpanzees: Implications for Hominoid Locomotor Evolution. Journal of Human Evolution 46:315 333. Pontzer H, Raich len DA and Sockol MD. 2009. The metabolic cost of walking in humans, chimpanzees, and early hominins. Journal of Human Evolution 56:43 54. Pontzer H, Raichlen DA and Rodman PS. 2014. Bipedal and quadrupedal locomotion in chimpanzees. Journal of Human Ev olution 66:64 82. Potts R. 1998. Variability selection in hominid evolution. Evolutionary Anthropology 7:81 96. Puech P F, Cianfarani F, and Roth H. 1986. Reconstruction of the maxillary dental arcade of Garusi Hominid 1. Journal of Human Evolution 15 :325 332. Rahn B. 2005. Cost surface derived least cost paths: A case study from Iron Age Orkney. Internet Archaeology 19. Reed DN and Denys C. 2011. The taphonomy and paleoenvironmentl implications of the Laetoli micromammals. In: Harrison T, editor. Pale ontology and geology of Laetoli: Human evolution in context New York: Springer. p 265 278. Reed KE. 1997. Early Hominid Evolution and Ecological Change through the African Plio Pleistocene. Journal of Human Evolution 32: 289 322). Reed KE. 2008. Paleoe cological patterns at the Hadar hominin site, Afar Regional State, Ethiopia. Journal of Human Evolution 54:743 768. Rossouw L, and Scott L. 2011. Phytoliths and pollen, the microscopic plant remains in Pliocene volcanic sediments around Laetoli, Tanzani a. In: Harrison T, editor. Paleontology and geology of Laetoli: Human evolution in context New York: Springer. p 201 215. Sellars WI, Cain GM, Wang W, and Crompton RH. 2005. Stride lengths, speed and energy costs in walking of Australopithecus afarensi s using evolutionary robotics to predict locomotion of early human ancestors. Journal of the Royal Society Interface 2(431 441).

PAGE 61

53 Smith EA, and Winterhalder B. 1992. Evolutionary Ecology and Human Behavior. New York: Aldine de Gruyter. p 470 Stern JT, an d Susman RL. 1983. The locomotor anatomy of Australopithecus afarensis American Journal of Physical Anthropology 60:279 317. Su DF, and Harrison T. 2008. Ecological implications of the relative rarity of fossil hominins at Laetoli. Journal of Human Evolut ion 55:672 681. Taylor CR, Caldwell SL, and Rowntree VJ. 1972. Running up and down hills: Some consequences of size. Science 178(4065):1096 1097. Terrier P, Aminian K, and Schutz Y. 2001. Can accelerometry accurately predict the energy cost of uphill/d ownhill walking? Ergonomics 44(1):48 62. Thompson DK. 2 013. Map East Africa. Online. Shapefile of Tanzanian boundary. Tiercelin JJ, and Lezzar KE. 2002. A 300 million years history of rift lakes in Central and East Africa: An updated broad review. In: Od ada EO, and Olago DO, editors. The East African Great Lakes: Limnology, palaeolimnology and biodiversity. Netherlands: Kluwer Academic Publishers. p 3 60. Todd JR, and White PS. 2009. A new cost distance model for human accessibility and an evaluation of accessibility bias in permanent vegetation plots in the Great Smoky Mountains National Park, USA. Journal of Vegetation Science 20:1099 1109. Tuttle R, Webb D, Weidl E and Baksh M. 1990. Further progress on the Laetoli trails. Journal of Archaeological Science 17:347 362. Umberger BR, Gerritsen KGM and Martin PE. 2003. A model of human muscle energy expenditure. Computer Methods in Biomechanics and Biomedical Engineering 6(2):99 111. Van Leusen M. 1999. Viewshed and cost surface analysis using GIS ( cartographic modelling in a cell based GIS II). BAR International Series 757:215 224.

PAGE 62

54 Verburg PH, Overmars KP, and Witte N. 2004. Accessibility and land use patterns at the forest fringe in the northeastern part of the Philippines. The Geographical Jou rnal 170(3):238 255. White TD, Ambrose SH, Suwa G, Su DF, DeGusta D, and al. e. 2009. Macrovertebrate Paleontology and the Pliocene Habitat of Ardipithecus ramidus. Science 326:87 92. Winterhalder B, and Smith EA. 1992. Evolutionary ecology and the socia l sciences. In: Smith EA, and Winterhalder B, editors. Evolutionary ecology and human behavior. New York: Aldine de Gruyter. p 3 23. Wynn JG, Sponheimer M, Kimbel WH, Alemseged Z, Bedaso ZK and Wilson JN. 2013. Diet of Australopithecus afarensis from th e Pliocene Hadar Formation, Ethiopia. PNAS 110(26):10495 105100.