Incorporating mortality into habitat selection to identify secure and risky habitats for savannah elephants
Introduction
Using telemetry data to predict habitat utilization is a common practice in ecology, and the results are often used as a guide to protect and preserve habitats important to wildlife species (e.g. Johnson et al., 2004, Meyer et al., 1998). However, habitat utilization is only one aspect within a complex set of factors that ultimately relate to individual fitness. Selection, for example, could be maladaptive, whereby individuals select areas that ultimately increase their mortality risk or decrease their reproductive success (Battin, 2004, Delibes et al., 2001). To avoid the misclassification of highly used habitats as high-quality habitats, it is necessary to temper estimations of habitat selection with some indication of fitness or risk, particularly when habitat selection appraisals are to be used as an impetus for conservation action.
The main limitation of incorporating indices of fitness into animal occurrence models is that spatially-explicit fitness data is often difficult to obtain (Nielsen et al., 2006). Studies that incorporate offspring survival are most common for avian species, where fledging success at the nest site can readily be established (Aldridge and Boyce, 2007, Donovan and Thompson, 2001). However, with mammalian species the estimation of recruitment and survival are generally less straightforward. Changes in the demographic parameters of large mammals in particular occur over relatively long timescales because of their multi-year or multi-decade life spans and their low reproductive and mortality rates. Most large mammals are also highly mobile with large ranging patterns, making it difficult to relocate individuals to monitor survival and fecundity. Studies linking reproductive success to habitats have been most successful in ungulates on islands and other closed systems (McLoughlin et al., 2007, McLoughlin et al., 2008). Given limitations in time, resources, study species, and study area, researchers resort to data that are more readily available, such as mortality location data (Dzialak et al., 2011, Falcucci et al., 2009, Nielsen et al., 2006).
Just as live animals inform habitat selection estimations, carcass locations provide spatially explicit information on where animals die and can be used to inform the riskiness of habitats (Nielsen et al., 2006). However, mortality location data is often underutilized in the literature. African elephants, for example, have carcasses that are easily visible from the air for several years after death (Douglas-Hamilton and Hillman, 1981), and while carcasses have been used to inform local mortality rates (e.g. Dudley et al., 2001, Dunham, 2008) and CITES status (e.g. Wittemyer et al., 2013), we found no peer-reviewed study exploiting the spatial location of carcasses. Knowing where animals die can provide valuable insight into risky landscapes, which is helpful information to guide conservation and management plans. Grizzly bears mortalities in Alberta, for example, were concentrated around roads or hiking trails (Benn and Herrero, 2002), prompting calls to regulate human access in grizzly bear habitats (Alberta Sustainable Resource Development and Alberta Conservation Association, 2010). Relating mortality locations to habitat selection models has also been used to better inform habitat-based management plans (i.e. Nielsen et al., 2006).
Habitat heterogeneity contributes to the spatial pattern of use and mortalities of elephants, and describing those patterns is particularly important in Africa, where the management of savannah elephants is a continuing concern (van Aarde and Jackson, 2007). Hunting and poaching in the late 19th and early to mid 20th century reduced some populations to near extinction (Roth and Douglas-Hamilton, 1991, Whyte et al., 2003), but actions taken to decrease poaching in the mid to late-20th century were largely successful in southern Africa (Whyte et al., 2003). The decline and subsequent recovery of elephant populations also may explain changes in woodland habitat (Guldemond and van Aarde, 2008, Nasseri et al., 2011), prompting concern for elephant-related tree damage (van Aarde et al., 2008). Their role as ecosystem engineers and the susceptibility to population decline from legal and illegal hunting makes it particularly important to quantify how habitats contribute to elephant habitat selection and mortality risk.
Using the information theoretic approach, we modelled habitat selection from elephant occurrence data obtained during aerial surveys. We then created an index of habitat use by elephants. Following a similar procedure, we next modelled the relative probability of elephant mortality using the locations of elephant carcasses. Combining the relative probability of use and mortality indexes, we then defined areas of high use and low mortality as secure habitats, and areas of high use and high mortality as risky areas. By interrelating conditions where elephants live with where they die, we can begin to establish a habitat-based approach to elephant management and work towards understanding and maintaining natural regulatory processes where needed, as proposed by van Aarde and Jackson (2007) and supported by others (Chamaillé-Jammes et al., 2008). These indices, while not a direct measure of demographic sources and sinks, do provide insights for the prioritization of conservation actions and can serve as a baseline to direct future studies into elephant demography.
Section snippets
Study area
The study area in northern Botswana incorporated an area of 74,355 km2. Study area boundaries to the north and east coincide with national borders for Namibia, Zambia, and Zimbabwe, respectively (Fig. 1). Twenty percent of the study area was protected within the confines of Chobe National Park (NP), Makgadikgadi NP, Moremi Game Reserve (GR), and Nxai Pan NP, while an additional 65% occurred within Wildlife Management Areas (WMAs; Chase, 2011). WMAs were designated with a two letter code and
Results
The top-ranked model for all three habitat selection analyses (elephant vs. random, mortality vs. random, and mortality vs. elephant) was model 7 (the global model with water; see Appendix C in Supplementary material for full model results). This model had strong support in the elephant vs. random (weight = 1.00) and mortality vs. random (weight = 1.00) analyses. Conversely, in the analysis comparing carcass locations to live elephant locations, the global model with water (Model 7) had a weight of
Discussion
Our results indicate that elephant use and mortality locations were spatially separated. Elephant mortality locations were concentrated in areas close to human settlements, with 80% of all elephant mortalities occurring within 25 km of high human use areas, an area that accounts for 52% of the study area. Conversely, 50% of live elephants were observed in that same area, and live elephants selected areas of intermediate distance from people (Fig. 4). Elephant mortalities were, therefore, not
Acknowledgements
We would like to thank Elephants Without Borders, the International Fund for Animal Welfare, and the University of Pretoria for research funding. The aerial survey was sanctioned and supported by the Botswana Department of Wildlife and National Parks, through a grant administered by the Conservation Trust Fund (CTF/2010/56). Additional funding was received from the Zoological Society of San Diego, Madeleine and Jerry Delman Cohen, Mr. Brett Warren, Nathan Jamieson Memorial Fund, Abu Camp, and
References (59)
- et al.
Habitat use by a desert ungulate: predicting effects of water availability on mountain sheep
J. Arid Environ.
(2010) - et al.
Evaluating resource selection functions
Ecol. Model.
(2002) Longitudinal studies of African elephant death and bone deposits
J. Archaeol. Sci.
(1988)- et al.
Wild meat: the bigger picture
Trends Ecol. Evol.
(2003) - et al.
Modelling the spatial distribution of human-caused grizzly bear mortalities in the Central Rockies ecosystem of Canada
Biol. Conserv.
(2004) - et al.
A habitat-based framework for grizzly bear conservation in Alberta
Biol. Conserv.
(2006) - et al.
Dynamic wildlife habitat models: seasonal foods and mortality risk predict occupancy-abundance and habitat selection in grizzly bears
Biol. Conserv.
(2010) - et al.
Megaparks for metapopulations: addressing the causes of locally high elephant numbers in southern Africa
Biol. Conserv.
(2007) - Alberta Sustainable Resource Development and Alberta Conservation Association, 2010. Status of the Grizzly Bear (Ursus...
- et al.
Linking occurrence and fitness to persistence: habitat-based approach for endangered greater sage-grouse
Ecol. Appl.
(2007)
When good animals love bad habitats: ecological traps and the conservation of animal populations
Conserv. Biol.
Grizzly bear mortality and human access in Banff and Yoho National Parks, 1971–1998
Ursus
Model Selection and Multimodel Inference: A Practical Information-theoretic Approach
Stable isotopes in elephant hair document migration patterns and diet changes
Proc. Natl. Acad. Sci. USA
Resource variability, aggregation and direct density dependence in an open context: the local regulation of an African elephant population
J. Anim. Ecol.
Landscape-scale feeding patterns of African elephant inferred from carbon isotope analysis of feces
Oecologia
Effects of an attractive sink leading into maladaptive habitat selection
Am. Nat.
Modeling the ecological trap hypothesis: a habitat and demographic analysis for migrant songbirds
Ecol. Appl.
Drought mortality of bush elephants in Hwange National Park, Zimbabwe
Afr. J. Ecol.
Detection of anthropogenic mortality in elephant Loxodonta africana populations: a long-term case study from the Sebungwe region of Zimbabwe
Oryx
The spatial pattern of demographic performance as a component of sustainable landscape management and planning
Landscape Ecol.
Assessing habitat quality for conservation using an integrated occurrence-mortality model
J. Appl. Ecol.
Linking vegetation response to seasonal precipitation in the Okavango–Kwando–Zambezi catchment of southern Africa
Int. J. Remote Sens.
A meta-analysis of the impact of African elephants on savanna vegetation
J. Wildlife Manage.
Cited by (24)
Age-sex structure of drought-driven African elephant (Loxodonta africana) mortality in Hwange National Park, Zimbabwe
2023, Scientific AfricanCitation Excerpt :Despite this, uncertainty emerges on how such mortalities impact elephant population dynamics. The monitoring of elephant mortality serves to identify demographic, the source and sink dynamics that influence elephant population [22,28] as well as assessment of the risk from extinction and response after a die-off [26]. The tools available for monitoring elephant mortality include Monitoring of Illegal Killing of Elephants (MIKE) which was instituted by the Convention on International Trade in Endangered Species (CITES) [29].
Human footprint and protected areas shape elephant range across Africa
2021, Current BiologyCitation Excerpt :This demonstrates that the areas where elephants are located (represented by the annual range area) are primarily structured by human presence and activity on the landscape, and it is only within those ranges that other environmental conditions start to influence elephant spatial patterns. The dominance of anthropogenic factors in affecting wildlife movement is increasingly common in species-level analyses.4,9,11,37–40 Previous work suggests that there might be thresholds of human density above which elephants and humans cannot coexist.14
Stakeholder attitudes toward the incentives used to mitigate human-elephant conflict in southern Africa: A news media content analysis
2021, Journal for Nature ConservationCitation Excerpt :Still, elephants will fail to use habitat where human densities are 15–20 people/km−2, or greater (Hoare & Du Toit, 1999). In Botswana, 80 % of elephant deaths occurred within 25 km of people (Roever et al., 2013). Large male elephants tend to be crop raiders, and some are repeat offenders, for example, in Amboseli, 12 % of raiders were habitual and accounted for 56 % of raids (Chiyo, Moss, Archie, Hollister-Smith, & Alberts, 2011).
Development and validation of a spatially-explicit agent-based model for space utilization by African savanna elephants (Loxodonta africana) based on determinants of movement
2021, Ecological ModellingCitation Excerpt :Other models have incorporated risky areas into habitat selection models for elephants (Roever et al., 2013a; Roever et al., 2013b). For example, by including carcass locations as representations of risky areas in a habitat selection model for elephants in northern Botswana, Roever et al. (2013a) found that protected areas were not necessarily associated with decreases in mortality risk as one would expect (Roever et al., 2013a). Spatial memory can shape animal space utilization patterns by allowing for non-random returns to locations within the landscape (Gautestad, 2011; Gautestad and Mysterud, 2010a).
Who is adjusting to whom?: Differences in elephant diel activity in wildlife corridors across different human-modified landscapes
2022, Frontiers in Conservation Science