Apes in a changing world - the effects of global warming on the behaviour and distribution of African apes

Apes in a changing world - the effects of global warming on the behaviour and distribution of African apes
of 15
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  ORIGINALARTICLE Apes in a changing world – the effectsof global warming on the behaviourand distribution of African apes Julia Lehmann 1,2 *, Amanda H. Korstjens 1,3 and Robin I. M. Dunbar 1,4 1 British Academy Centenary Research Project,School of Biological Sciences, Crown Street,University of Liverpool, Liverpool L69 7ZB,UK,  2 Department of Life Sciences, RoehamptonUniversity, London SW15 4JD, UK, 3 Conservation Sciences, BournemouthUniversity, Poole BH12 5BB, UK,  4 Institute of Cognitive and Evolutionary Anthropology,University of Oxford, Oxford OX2 6PE, UK  *Correspondence: Julia Lehmann, Life ScienceDepartment, Holybourne Avenue, RoehamptonUniversity, London SW15 4JD, UK.E-mail:  ABSTRACT Aim  In this study we use a modelling approach to identify: (1) the factorsresponsible for the differences in ape biogeography, (2) the effects that globalwarming might have on distribution patterns of African apes, (3) the underlyingmechanisms for these effects, and (4) the implications that behavioural flexibility might be expected to have for ape survival. All African apes are highly endangered, and the need for efficient conservation methods is a top priority. Theexpected changes in world climate are likely to further exacerbate the difficultiesthey face. Our study aims to further understand the mechanisms that link climaticconditions to the behaviour and biogeography of ape species. Location  Africa. Method  We use an existing validated time budgets model, derived from data on20 natural populations of gorillas ( Gorilla beringei  and  Gorilla gorilla ) andchimpanzees ( Pan troglodytes  and  Pan paniscus ), which specifies the relationshipbetween climate, group size, body weight and time available for various activities,to predict ape distribution across Africa under a uniform worst-case climatechange scenario. Results  We demonstrate that a worst-case global warming scenario is likely toalter the delicate balance between different time budget components. Our modelpoints to the importance of annual temperature variation, which was found tohave the strongest impact on ape biogeography. Our simulation indicates thatrising temperatures and changes in rainfall patterns are likely to have strongeffects on ape survival and distribution, particularly for gorillas. Even if they behaved with maximum flexibility, gorillas may not be able to survive in most of their present habitats if the climate was to undergo extreme changes. The survivalof chimpanzees was found to be strongly dependent on the minimum viablegroup size required. Main conclusions  Our model allows us to explore how climatic conditions,individual behaviour and morphological traits may interact to limit thebiogeographical distributions of these species, thereby allowing us to predictthe effects of climate change on African ape distributions under different climatechange regimes. The model suggests that climate variability (i.e. seasonality) playsa more important role than the absolute magnitude of the change, but these dataare not normally provided by climate models. Keywords Africa, apes, behavioural flexibility, biogeography, climate change, global warming,  Gorilla , habitat loss,  Pan , time budget model.  Journal of Biogeography   (  J. Biogeogr  .) (2010)  37 , 2217–2231 ª  2010 Blackwell Publishing Ltd  2217 doi:10.1111/j.1365-2699.2010.02373.x  INTRODUCTION Many analyses have demonstrated that global warming willaffect species distribution patterns as well as biodiversity ingeneral (Parmesan & Yohe, 2003; Root  et al. , 2003; Thuiller,2004). The most common models used to predict such effectsare bioclimatic envelope models (Thuiller, 2003), which aim todetermine the climate envelope that defines a species’ range by correlating its distribution patterns with selected climatevariables. Although such models are useful for predictingbiogeographical distribution patterns, they have recently beencriticized because they fail to include a number of otherimportant variables (Guisan & Thuiller, 2005; Heikkinen  et al. ,2006; Austin, 2007) and do not yield actual information aboutthe underlying mechanisms that limit a species’ distribution.Although it is widely appreciated that a species’ distribution isalso shaped by historical patterns of the distribution of key ecological resources (Ganzhorn, 1998; Reed & Bidner, 2004;Graham  et al. , 2005; Lehman, 2006), few studies haveattempted to provide an explanation for the mechanisms thatunderpin such effects.In order to correctly predict the effects that changing(climatic) conditions might have on species survival, it isbecoming increasingly important to understand: (1) the exactmechanisms determining a species’ biogeographical distribu-tion, and (2) whether, and how, species will be able to changetheir behaviour and/or the kinds of habitats they can survivein. The latter point has largely been overlooked in previousmodels, but might become increasingly important for cogni-tively more advanced species: in birds, for example, larger-brained species cope better with seasonal changes in theirenvironments and are more resistant to extinction (Sol  et al. ,2002; Shultz  et al. , 2005), and might be able to bufferthemselves more effectively against modest levels of climatechange. Understanding the interactive effects of phenotypicaland ecological constraints and behavioural flexibility onspecies biogeographical distribution patterns and their reactionto changes in global climate should thus be accorded higherpriority. In contrast to studies investigating projected changeof habitat with climate warming, this study investigatesprojected change of behaviour with climate warming and itseffect on species survival.Africa has been identified as being the most vulnerable of allcontinents to the effects of climate change (Lovett  et al. , 2005;IPCC, 2007), and numerous cases have been reported wherespecies or populations have altered their ranges or even life-history variables in response to changes in climatic conditions(e.g. Pounds  et al. , 1999; Parmesan & Yohe, 2003). Many African primates are already highly endangered, with the mostvulnerable species being those that have highly restrictedranges, a slow life history and a large body mass (for reviewssee Cowlishaw &  Dunbar, 2000; Chapman  et al. , 2006). Africanapes, i.e. gorillas ( Gorilla  spp . ) and chimpanzees ( Pan  spp . ),show all these characteristics. Although all species of Africanapes can occur in similar habitats and have somewhatoverlapping diets (Tutin & Fernandez, 1993; Stanford & Nkurunungi, 2003; Morgan & Sanz, 2006), they differsubstantially in their biogeographical ranges.  Gorilla , whichnow has a disjunct distribution limited to relatively small areasin the western and eastern parts of central Africa, is muchmore limited in its distribution than  Pan , which occursthroughout central Africa as well as in West Africa. Thereasons for this remain unclear. On a finer scale, present-day species distribution patterns are patchy and heavily influencedby human activities, because these species often ‘compete’ forspace with humans, who are rapidly destroying ape habitats.Because the increasing human population will need moreresources (e.g. wood and meat), apes will inevitably be driveneven closer to extinction as human populations expand(Chapman  et al. , 2006). In order to implement efficientconservation measures, it is essential not only to know aboutanthropogenic effects on ape survival but also to have a muchbetter understanding of both the factors that naturally limitape distributions and the specific ecological requirements thesespecies have – or can adapt to. This might then allow us topredict more effectively how a species’ range is likely to beaffected by environmental changes.Here we use an established time budget model for Africanapes (Lehmann  et al. , 2008a) to investigate how climatewarming (in the absence of further anthropogenic factors)might affect ape survival. In addition, we assess the extent towhich behavioural flexibility might influence ape survival. Asfar as we are aware, only one previous quantitative evaluationof the potential effects of climate change on a primate speciesusing a time budget model has been published (Dunbar, 1998);this suggested dramatic effects of global warming on geladababoons ( Theropithecus gelada ), mainly due to the fact that anincrease in ambient temperature will lead to severe habitatfragmentation. Time budget models of this kind are based onindividual behaviour, i.e. on an individual’s allocation of timeto feeding, resting, travelling and socializing (Dunbar, 1992a,b,1996). A general overview of time budget models and theirbackground theory is given by  Dunbar  et al.  (2009). Theessence of this approach is that because time is limited, itcreates a constraint on the size of group (and hence populationdensity) that a species can maintain in a given habitat, thereby determining the species’ biogeographical distribution (Dunbar,1992a,b, 1998; Williamson & Dunbar, 1999; Hill & Dunbar,2002; Korstjens  et al. , 2006; Korstjens & Dunbar, 2007;Lehmann  et al. , 2007a). While time budget models predictspecies biogeographical distributions at least as well as moreconventional climate envelope models (for three differentprimate genera: Korstjens & Dunbar, 2007; Willems & Hill,2009), they have the added advantage of providing insight intothe mechanisms by which a species is prevented from usingcertain habitats as well as the level of ecological stress that aspecies faces in those areas where it can survive. In thesemodels, the amount of time an individual has to invest inessential activities depends on the ecological conditions atspecific locations, as well as on the number of competitors ( size). Because these models are based on simpleclimatological variables (which ultimately determine the J. Lehmann  et al. 2218  Journal of Biogeography   37 , 2217–2231 ª  2010 Blackwell Publishing Ltd  distribution/availability of food and other resources), they areideally suited to assessing the possible effects of climatewarming on species distribution patterns. MATERIALS AND METHODSThe model We use an existing and previously validated systems model(Lehmann  et al. , 2008a) to investigate the effects of futureclimate change on the distribution patterns of African apes.The model uses multivariate equations (see Table 1) for timebudget and behavioural ecology variables (including diet,group biomass, subgroup size and minimum viable group size)derived from data on 20 natural populations of the two speciesof gorillas ( Gorilla beringei  and  Gorilla gorilla ) and two speciesof chimpanzees ( Pan troglodytes  and  Pan paniscus ) for whichbehavioural and climatological information was available (fordetails see Lehmann  et al. , 2008a; data are provided inAppendices S1 and S2 in Supporting Information). Theseequations describe how climate influences (directly throughthermoregulation as well as indirectly through resourceavailability and distribution) individual ape behaviour. Weuse the model to find the maximum ecologically tolerablecommunity size (the largest number of individuals that can livetogether while still balancing their time, and hence energy,budgets) by using this set of equations to calculate an averageindividual’s time budget under given climate conditions ascommunity size is increased algorithmically until the sum of the time budget variables reaches 100%. We used a largeclimate database for sub-Saharan Africa to model the distri-bution of maximum ecologically tolerable community sizeacross Africa, and then used this to deduce the presence/absence of ape species on a continent-wide basis (see below fordetails). We have tested the validity of this model in fourseparate ways: (1) by showing that it correctly predictspresence/absence at a set of 639 independent sites in Africa,at about half of which apes are known to live; (2) by showingthat it correctly predicts community sizes at sites whereindividual ape species live; (3) by showing that it correctly predicts the continent-wide distribution of the two ape genera,using a matrix of 11,670 data points from the Willmott & Matsuura (2001) African climate database; and (4) by runningsensitivity analyses to determine the susceptibility of theequation parameter values to errors of estimation (seeLehmann  et al. , 2007a, 2008a,b). Time budget equations Equations to calculate feeding and moving time were derivedby applying standard multivariate model-finding procedures toobservational data from 20 study sites for which such datawere available, collated from the literature (for details, seeLehmann  et al. , 2008a). The analyses used 11 different climateindices as possible independent variables (although we haveelsewhere shown that these broadly reduce to three principaldimensions: Williamson & Dunbar, 1999); forest cover wasestimated from satellite image data (DeFries  et al. , 2000). Theequations so obtained are listed in Table 1. These indicate thatfeeding and moving time were determined by climate (mainly annual rainfall and rainfall- and temperature-seasonality), diet,body mass, foraging party size and/or community size. Forsocial time and resting time, we used a different approach andderived generic equations from analyses of a wide range of primate species. In primates, social time is an important factorfor group cohesion (and hence resistance to permanent fission)(Dunbar, 1988, 1991) and we thus need to be able to specify the amount of social time individuals  ought   to spend in socialactivities in order to maintain social cohesion in a group of agiven size. (What time they actually devote to socializing often Table 1  Equations used in the model to calculate individual time budgets of African apes ( Pan  spp. and  Gorilla  spp.). Variable EquationFeeding (%) 33.089 + 0.005  ·  group biomass + 0.143  ·  body weight + 0.158  ·  %fruit  )  0.006  ·  P  ann Moving (%) 18.74 + 13.92  ·  T  mo SD + 0.35  ·  prtysz  )  4.94  ·  P  2 T   + 0.32  ·  ( P  2 T  ) 2 Grooming (%) 1.01 + 0.23  ·  community sizeMin. resting (%)  ) 29.467 + 1.278  ·  T  ann  + 0.336  ·  % leaf + 5.954  ·  T  mo SDGroup biomass 4.24  ·  body weight + 29.83  ·  group size%fruit 169.429  )  50.651  ·  log(body weight)  )  0.021  ·  altitude  )  62.023  ·  moi mo mx + 0.39  ·  forest cover%leaf 100  )  %fruitParty size (chimp) 21.489 + 0.072  ·  forest cover  )  0.33  ·  P  mo  + 0.0012  ·  P  mo2 Min. party (chimp) e (2.25 ) 0.23 · ln(forest cover)+0.36 · ln( T  ann )) /3Min. group (gorilla) e (2.25 ) 0.23 · ln(forest cover)+0.36 · ln( T  ann )) /7Equations used in the model for predicting ape time budget components; all equations are based on observational data (see Appendices S1 and S2);see Lehmann  et al.  (2008a) for details on the derivation of these equations. P  ann  = mean annual rainfall (in mm);  T  ann  = mean annual mean temperature (in   C);  T  mo sd  = temperature variation between months (calculated asthe standard deviation across average values for 12 months );  P  mo  = average rainfall per month (in mm);  P  2 T   = plant productivity index [= thenumber of months in the year in which rainfall (in mm) was more than twice the average monthly temperature (Le Houe´rou, 1984)];moi mo mx = maximum monthly moisture index  (Willmott & Feddema, 1992); min. = minimum; prtysz = party size; %fruit/leaf = percentage fruit/leaves in the diet. Minimum group/party sizes are scaled to female body weight. Global warming and ape biogeography  Journal of Biogeography   37 , 2217–2231  2219 ª  2010 Blackwell Publishing Ltd  represents a compromise in managing their time budget as awhole; Dunbar  et al. , 2009.) We used the generic equationspecifying required social time as a function of social groupsize that was obtained by Lehmann  et al.  (2007b) in an analysisof a large sample of African primates ( n  = 40 species). Restingtime is less straightforward to determine than the other timebudget components because observed resting time in animalsconsists of two separate components: free (or uncommitted)time (which can be drawn on when environmental conditionsrequire more time for feeding, moving or socializing) andenforced resting time (time when animals are forced to rest toavoid heat overload or hyperthermia and/or to allow digestiveprocessing) (Korstjens  et al. , 2010). Because enforced restingtime cannot be transformed into other more urgent activities,it represents an environmentally driven reduction in theeffective length of the animal’s active day and adds animportant constraint on a species’ capacity to maintaincommunities of a minimum viable size in a given habitat(Dunbar  et al. , 2009; Korstjens  et al. , 2010). For enforcedresting time, we used a generic equation derived from ananalysis of data from 78 species of primates (Dunbar  et al. ,2009; Korstjens  et al. , 2010): enforced resting time is deter-mined by diet composition (the percentage of leaves in thediet), mean annual temperature and monthly temperaturevariation. Both of these equations are given in Table 1.  Minimum party size Although apes are generally large-bodied, predation andinfanticide (which can be viewed as within-taxon predation)still remain serious threats (Boesch & Boesch-Achermann,2000). This is especially relevant for chimpanzees, which spendmost of their time in very small subgroups. Thus, we assumethat, dependent on ecological conditions and predation risk,apes need to maintain a certain minimum party size to be safe(Dunbar, 1996; Hill & Dunbar, 1998; Shultz  et al. , 2004). Thiswill be especially important in habitats with low forest coverand a high density of predators (Lehmann & Dunbar, 2009).Minimum party size was estimated from forest cover (treecover indexes the availability of refuges) and bush cover(increased bush cover implies increased risk of being caughtunawares by a predator) using a body mass corrected versionof the equation given by Dunbar (1996) (Table 1). For apes tosurvive at a given location, the model requires that averageparty size be equal to or larger than minimum party size (fordetails see Dunbar  et al. , 2009; Lehmann & Dunbar, 2009). Diet composition Although gorillas and chimpanzees show some dietary overlap(Tutin & Fernandez, 1993; Stanford & Nkurunungi, 2003;Morgan & Sanz, 2006), gorillas consume less fruit thanchimpanzees. We therefore used the data on ape diets from thesample of study sites to derive a general equation for thepercentage of feeding time devoted to fruit versus leaves as afunction of ecological conditions and body mass (Lehmann et al. , 2008a) (see Table 1). Although these values allow apes tobe more flexible than has been observed in the field, thisapproach is justified by the fact that we do not know where thereal limits of dietary flexibility are for apes. However, for thepurposes of the model, we set limits on dietary flexibility at 10–100% folivory. Body mass The two taxa,  Pan  and  Gorilla , are represented in the model by two distinct weight categories (40 and 120 kg), which roughly correspond to the mean weight of male and female chimpan-zees and gorillas, respectively (Smith & Jungers, 1997; Calde-cott & Miles, 2005). Although ape subspecies can differ in theirbody masses (Caldecott & Miles, 2005) and there is significantsexual dimorphism in body mass in both ape taxa, we chosethis approach for the sake of simplicity, because presentingseparate models for males, females and subspecies would resultin unnecessary complexity. Social demography  We draw a distinction between two types of groups thatcharacterize ape societies. These are the community (the set of individuals who share a common range area, and whosemembership is relatively stable over time) and the foragingsubgroup or party (the set of individuals who happen to betogether at any given moment, and whose composition isusually unstable over time, but whose membership is invari-ably drawn from one particular community). Gorillas live insmaller and more stable groups than chimpanzees, and thecommunity and the party are usually one and the same in theircase. In some equations, we use group biomass as analternative to including both body mass and group sizeseparately. The equation used to calculate group biomassreflects species-specific changes in group composition whenoverall group size increases, and was derived by Lehmann  et al. (2008a); it provides a conservative estimate of overall groupbiomass rather than using a simple multiplication of group sizeand mean individual body mass. Climate and ape biogeographical distribution We used a large (10,075 locations) climate database for Africa(obtained from Willmott & Matsuura, 2001) to providecontinuous climate data on a grid of 0.5   latitude and 0.5  longitude across Africa. A linear program in dBase (1994,Borland International Inc., Scotts Valley, CA, USA) used thesevalues and the equations given in Table 1 to calculate themaximum ecologically tolerable community size for eachlocation in the dataset under given climate conditions. Themodel assumes that apes are able to live at a particular site if:(1) average party size is larger than the minimum party size,and (2) predicted maximum ecologically tolerable community size is larger than a set minimum, which varies as a function of body mass. In the initial analyses, minimum community size J. Lehmann  et al. 2220  Journal of Biogeography   37 , 2217–2231 ª  2010 Blackwell Publishing Ltd  for  Pan  was set to 10 individuals (because almost all knownchimpanzee populations live in communities larger than this:Lehmann  et al. , 2007a), while for gorillas this limit was set tofive individuals (the minimum stable group size for knowngorilla groups: Anderson  et al. , 2002; Lehmann  et al. , 2008b).In addition, we also ran the model, using a less conservativeand more ecologically driven approach to the problem of minimum size (i.e. using information about forest cover weestimated a habitat-dependent minimum party size that isrequired to ensure sufficient protection against predation) (seeTable 1 and below for further details). Apes were then assumedto be able to survive within a given habitat if their ecologically tolerable community size exceeded the minimum party size.We then use our model to assess the likely effects that anincrease in temperature and rainfall will have on ape groupsizes and biogeographical distributions. Because (1) the exactextent to which climate will change is not known and heavily debated, and (2) our interests are in determining the worst-case scenario to inform conservation planning, we decided touse a ‘worst-case scenario’ estimate for the predicted level of climate change. Current ‘best guess’ estimates of climatechange for Africa range between 3.5   C and 6.5   C (see IPCC,2007) and a 15% increase in rainfall by 2100 (corresponding toa conservative assumption of a 3% increase in precipitation perdegree kelvin; see, e.g. Wentz  et al. , 2007). We have thereforeused a value for overall increase in temperature of 5.2   C – avalue in the upper half of this range, but conservatively wellbelow the upper limit – and an increase in rainfall of 15%. Forreasons of simplicity we decided to use a uniform climatechange scenario; it is generally assumed that temperatures willincrease across Africa. With regard to rainfall, some modelsactually predict a regional decrease in rainfall; however, as weare interested in the overall pattern of how climate variableswill affect ape distributions, we opted to model an increase inrainfall as a worst-case scenario, especially as this is thepredicted pattern across the current distribution range of apes Table 2  (a) Predicted distribution of maximum ecologically tol-erable community sizes for chimpanzees (40 kg), depending onmean annual temperature variation ( T  mo SD) and annual rainfall( P  ann ). Dashes indicate that these particular climatic conditionsdid not occur on the African continent. Shading indicates climaticconditions at which chimpanzees have been observed. (b) Pre-dicted distribution of maximum ecologically tolerable group sizesfor gorillas (120 kg), depending on mean annual temperaturevariation ( T  mo SD) and annual rainfall ( P  ann ). Dashes indicate thatthese particular climatic conditions did not occur on the Africancontinent. Shading indicates climatic conditions at which gorillashave been observed. (a) T  mo SD(  C) P  ann  (mm)0–500 500–1000 1000–1500 1500–2000 2000–2500 2500–30000–0.5 – 79.0 86.0 90.0 88.0 89.00.50–1 19.5 49.0 61.0 72.0 61.0 59.01–1.5 3.0 25.0 35.0 43.0 48.5 49.01.5–2 0.0 3.0 10.0 11.5 19.0 7.02–2.5 0.0 0.0 0.0 0.0 0.0 3.52.5–3 0.0 0.0 0.0 0.0 0.0 –3–3.5 0.0 0.0 0.0 0.0 – –(b) T  mo SD(  C) P  ann  (mm)0–500 500–1000 1000–1500 1500–2000 2000–2500 2500–30000–0.5 – 20.0 23.0 24.0 25.0 29.00.50–1 0.0 4.0 10.0 15.0 11.0 12.01–1.5 0.0 0.0 0.0 2.0 4.0 6.01.5–2 0.0 0.0 0.0 0.0 0.0 0.02–2.5 0.0 0.0 0.0 0.0 0.0 0.02.5–3 0.0 0.0 0.0 0.0 0.0 –3–3.5 0.0 0.0 0.0 0.0 – – Table 3  Equations used to calculate the effect of an increase in temperature and rainfall on the remaining climate variables in the model of African ape biogeography. Climate variables  r F P P  2 T   ) 2.49 + 3.752  ·  log10( P  ann )  )  0.004  ·  ( T  ann ) 2 0.89 19,951 *** T  mo SD  ) 7.891  )  0.002  ·  P ann  + 11.05  ·  log10( T  ann )  )  0.004  ·  ( T  ann ) 2 + 0.034  ·  lat 0.78 3791 ***moi mo mx   ) 0.027  )  0.001  ·  ( T  ann ) 2 + 0.001  ·  lat + 0.001  ·  P  ann  0.86 9336 *** P  mo  4.137 + 0.081  ·  P  ann  0.97 18,879 ***moi mo av   ) 0.11 + 0.001  ·  P  ann  )  0.028  ·  T  ann  )  0.0000117  ·  altitude  )  0.000000139  ·  ( P  ann ) 2 0.97 40,547 ***ml 100  12.644  )  0.007  ·  P  ann  + 0.00000091  ·  ( P  ann ) 2 0.95 48,501 ***Frcover 88.046  )  0.121  ·  P  mo  + 22.944  ·  moi mo av   )  5.511  ·  ml 100  0.8 5943 ***All equations are based on linear regression and curvilinear estimation procedures estimating the effects of temperature and rainfall on each of theother climate variables of interest. P  2 T   = plant productivity index [the number of months in the year in which rainfall (in mm) was more twice the average monthly temperature (LeHoue´rou, 1984)];  P  ann  = mean annual rainfall (in mm);  T  ann  = mean annual mean temperature (in   C); lat = latitude;  T  mo SD = temperaturevariation between months (calculated as the standard deviation across average values for 12 months );  P  mo  = average rainfall per month (in mm);ml 100  = number of months per year with < 100 mm of rainfall; Frcover = forest cover; moi mo av = average monthly moisture index (Willmott & Feddema, 1992); moi mo mx = maximum monthly moisture index (Willmott & Feddema, 1992);  r   = correlation coefficient of the regressionmodel;  F   =  F  -value of the equation;  P   = level of significance.*** P   < 0.001. Global warming and ape biogeography  Journal of Biogeography   37 , 2217–2231  2221 ª  2010 Blackwell Publishing Ltd
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks