- Email: [email protected]
Co-occurrence and potential for competition between wild and domestic large herbivores in a South American desert
Journal of Arid Environments 77 (2012) 39e44
Contents lists available at SciVerse ScienceDirect
Journal of Arid Environments journal homepage: www.elsevier.com/locate/jaridenv
Co-occurrence and potential for competition between wild and domestic large herbivores in a South American desert P. Acebes*, J. Traba, J.E. Malo Terrestrial Ecology Group-TEG, Departamento de Ecología, Facultad de Ciencias, Universidad Autónoma de Madrid, C/.Darwin, 2, E-28049 Madrid, Spain
a r t i c l e i n f o
a b s t r a c t
Article history: Received 8 July 2010 Received in revised form 28 August 2011 Accepted 8 September 2011 Available online 28 September 2011
Introduction of domestic ungulates may lead to interspecific competition with native herbivores, particularly if the species involved are of similar size and share similar foraging strategies due to a scarcity of trophic resources. This interaction has not been investigated in South American deserts in which guanaco (Lama guanicoe), the larger South American wild camelid and the most widelydistributed ungulate, co-occurs with several introduced ungulates (e.g. cattle, donkeys, sheep or red deer). We studied the occurrence and abundance patterns of guanacos, donkeys and cattle in desert areas of Argentina by dung sampling. Analyses of co-occurrence show no relationship in land use between guanaco and livestock when geographical effects are removed. Distribution and abundance models of the guanaco are strongly associated with sparse plant cover, rocky substrata and human-influence variables, guanacos appearing in areas furthest from villages. Nevertheless, while guanaco distribution is weakly related to feral livestock presence, their abundance is negatively associated with donkey presence. In contrast, livestock species appear closely associated with each other, being centred on areas of densest and most productive vegetation. These results are relevant for the management of extensive arid areas of South America, where guanacos co-occur with feral donkeys and frequently conflicts with humans activities. Ó 2011 Elsevier Ltd. All rights reserved.
Keywords: Coexistence Human impact Lama guanicoe Monte desert Protected areas Resource partitioning
1. Introduction Spatial occurrence patterns of wild species are an expression of multiscalar and multifactorial processes that act simultaneously. Such patterns are determined, at landscape or finer scales, by the evolutionary history of species, environmental factors such as distribution of trophic resources (Morris and Davidson, 2000) and by biotic interactions (Rosenzweig, 1981). Where herbivores are concerned, co-occurrence of native species may be due to evolutionary processes that have enabled coexistence through spatial or trophic resource partitioning (Voeten and Prins, 1999), although they may have competed in the past (the ’ghost of competition past’, Connell, 1980). However, the recent introduction of domestic ungulates into an assemblage of native species may culminate in interspecific competition, given that there has been insufficient time for resource partitioning to evolve (Putman, 1996; Voeten and Prins, 1999). Sympatric species of similar-sized ungulates that have similar foraging strategies may thus compete (Owen-Smith, 2002; Prins and Olff, 1998), although for this to occur there must be a high
* Corresponding author. Tel.: þ34 914978916; fax: þ34 914978001. E-mail address: [email protected] (P. Acebes). 0140-1963/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jaridenv.2011.09.003
degree of overlap in their use of spatial and trophic resources, and such resources must be scarce (Hulbert and Andersen, 2001; Prins and Olff, 1998; Putman, 1996). In addition, humans modify the use of space by wild ungulates since traditional pastoralism drives wild ungulate populations away (Bagchi et al., 2004; Mishra et al., 2004; Prins, 2000). Moreover, wild species are also displaced far from human settlements by poaching (Wilkie et al., 2000). Arid lands world-wide are resource-limited environments. Large native herbivores and livestock in such biomes often compete for space and trophic resources (Young et al., 2005). Studies of competition between native and domestic ungulates in arid and semi-arid biomes are relatively numerous for African savannas (Fritz et al., 1996; Sitters et al., 2009; Voeten and Prins, 1999; Young et al., 2005) as well as for Asian steppes (Bagchi et al., 2004; Shrestha and Wegge, 2008; Suryawanshi et al., 2010). However, very few such studies exist for similar biomes in South America, in which the guanaco (Lama guanicoe, Müller), the largest and most widely-distributed wild ungulate, stands out as a model species to study such interactions. The guanaco population prior to the Spanish conquest was between 30 and 50 million, occupying nearly all of Chile and Argentina, as well as parts of Bolivia, Peru and Paraguay (Raedeke, 1979). Current estimates point at around 600,000 guanacos, the
40
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
pronounced decline being attributed to competition with livestock, habitat loss and poaching (Baldi et al., 2008). Guanaco inhabits arid and semi-arid shrublands and grasslands and hilly, mountain and steppe environments through its geographical range (Baldi et al., 2008). It is sympatric with domestic ungulates in many of these systems but their interaction has only been examined in a few studies on the Patagonian steppes, where the guanaco is abundant and has coexisted with sheep since the latter were introduced at the end of 19th century (Baldi et al., 2001, 2004; Puig et al., 2001). In extremely arid regions, where the guanaco maintains significant populations of high conservation value (Acebes et al., 2010a), it cooccurs with both donkeys (Equus asinus) and free-ranging cattle (Bos taurus). Guanacos are highly social and monomorphic and exhibit a resource defence polygyny mating system (Franklin, 1983). During the breeding season territorial males seek out habitats that are favourable for attracting females, such as areas with abundant forage that are safe from predators, especially pumas (Puma concolor) (Bank et al., 2003). Donkeys are native to arid zones of Africa and they have colonised American and Australian deserts successfully (Grinder et al., 2006), but they remain poorly studied and few studies have evaluated their impact on South American wild fauna (Novillo and Ojeda, 2008; Ovejero et al., 2011). Their social organisation is based on solitary territorial males that have exclusive mating access within their territories, together with stable groups of females and their offspring (Moehlman, 1998). Cattle have received more attention (see for instance Fritz et al., 1996; Stewart et al., 2002) although studies of their competition with South American herbivores are lacking. The present study employed dung sampling to analyse the use of space by three large herbivores: the native guanaco, donkeys, and free-ranging cattle, in two protected areas of the hyper-arid Monte Desert (Argentina), where trophic resources are limited. We hypothesised that there would be a high degree of spatial overlap between the three species given their feeding types (guanacos and donkeys: intermediate feeders; cattle: grazers), their digestive strategies (guanacos and cattle: ruminants; donkeys: hindgut fermenters), their body weights (guanacos 100e120 kg; donkeys: 140 kg; cattle: 250 kg) and the scarcity of trophic resources. In particular, the herbaceous layer is almost nonexistent in the study area (Acebes et al., 2010b) so herbivores rely on browse-dominated diets. The potential for competition between them is thus high and it could have a negative impact on the native species. The basic objectives were a) to determine co-occurrence patterns of the three species from data on their presence and abundance, and b) to define their habitat selection requirements, evaluating the effects of abiotic and biotic factors, including those related to human presence and to the effect of potential competitors. 2. Materials and methods 2.1. Study area The study was carried out in the Ischigualasto-Talampaya World Heritage Site. This comprises two contiguous conservation units, Ischigualasto Provincial Park and Talampaya National Park, located in San Juan and La Rioja provinces respectively, in north-western Argentina (29 550 S, 68 050 W). Two villages some 10e20 km from the protected areas are major sources of cattle and donkeys. The donkeys that inhabit the protected areas and their surroundings are old animals abandoned by farmers and/or their offspring, while cattle range throughout the territory. The parks have a total area of 275,369 ha and extend along the eastern foothills of Central Andes at a mean altitude of 1300 m. The climate is typical of dry
desert with a mean annual temperature below 18 C (range 10 to 45 C) and a mean temperature in the hottest month above 22 C. Mean annual precipitation ranges from 80 to 140 mm, concentrated in summer (November to February). Monte Desert is the dominant biome (Márquez et al., 2005). The predominant vegetation is sparse shrubland, dominated by species of Zygophyllaceae (Larrea spp, Zuccagnia punctata, Bulnesia retamo and Plectrocarpa tetracantha), Fabaceae (Prosopis spp, Cercidium praecox, Geoffroea decorticans, Senna aphylla, Ramorinoa girolae) and Chenopodiacae (Atriplex spp and Suaeda divaricata). Cacti (Echinopsis spp, Tephrocactus spp and Opuntia sulphurea) and Bromeliads (Deuterocohnia longipetala and Tillandsia spp) are also frequent but to a lesser extent. Total plant cover is less than 20%, with a very sparse and seasonal herbaceous layer (Acebes et al., 2010b). 2.2. Herbivore data We used the pellet-group count technique (Neff, 1968) in order to estimate the distribution and relative abundance of the different species. In particular, we used the ‘standing-crop method’, a count of pellet groups within sample plots on the first visit. This method is suitable for arid zones, where ungulate densities are low (Putman, 1984). Bleached pellet groups were ignored, to ensure that counts reflected recent presence (Henley et al., 2007). There is ample evidence that dung counts are more reliable than aerial or ground counts when used to estimate relative habitat use (Cromsigt et al., 2009; Young et al., 2005), particularly in arid environments (Henley et al., 2007). A total of 14 line transects 5 km long were established during 2005 and 2006 taking in the maximum possible environmental diversity. GPS-georeferenced circular plots of 15 m radius were established at minimum intervals of 300 m along each transect. The circular plots totalled 219 since the difficult terrain made sampling incomplete. The abundance of guanaco, donkey and cattle dung was recorded within each plot. Scattered pellets were scored as 1, dung heaps as 2 and guanaco latrines and accumulated dung heaps from donkeys or cattle scored 3. This scale approximates to a logarithmic transformation of the pellet number and makes sampling easier, given that individual pellet counts are impossible in medium-sized or large dung piles. 2.3. Explanatory variables The following environmental variables were measured in each circular plot: gradient ( ), plant cover (%) and extent of rock or bare ground (%). Mean and maximum vegetation heights (m) were estimated and the principal plant species, the second-commonest species and other plants present were recorded. Predictor variables were calculated using ArcGIS 9.2 software on existing maps. These were the Normalised Difference Vegetation Index (NDVI) using Landsat 7 ETM þ images for a buffer zone of 50 m radius around each circular plot (named ’NDVI’ hereafter); and the minimum distance from each plot to a village (‘distance from villages’), to a small permanent spring (‘distance from springs’) and to a seasonal watercourse (‘distance from swc’), where the densest formations of Prosopis chilensis and Prosopis flexuosa of the study area occur (Acebes et al., 2010b). Distances to roads (’distance from roads’), which are used by tourists, village inhabitants and potential illegal hunters, and to tracks (’distance from tracks’), used by tourists and park wardens to visit and move around protected areas respectively, were also measured. A prior analysis of cross-correlations was employed to reduce the number of original predictor variables and to avoid colinearity between them. The resulting redundant variables (extent of bare
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
ground and mean vegetation height) were omitted from subsequent analyses. 2.4. Patterns of co-occurrence and abundance among herbivores Correlation of occurrence and abundance for the three species was evaluated by a partial Mantel test, in order to remove the spatial effect that may determine co-occurrence patterns between species (Manly, 2007). Presence and abundance of each species’ dung in each circular plot were employed as response matrices and the geographical location of each plot was taken as the control matrix. Analyses of occurrence used Jaccard distance matrices and those of abundance used Euclidean distance matrices. The said matrices represent the differences between pairs of occurrence/ abundance observations of species-pairs and the geographical distance between their locations. Significance testing employed a randomisation approach with 9999 permutations, employing the PASSaGE v.2 program (Rosenberg and Anderson, 2011). 2.5. Modelling species patterns Generalised Linear Models were used to examine the relationship between occurrence and abundance for each species with different predictor variables, assuming a binomial error distribution and a logit link function for presence/absence data and a Poisson error distribution and a log link function for abundance data (McCullagh and Nelder, 1989). Presence/absence of potentially competing species was introduced as a categorical predictor, the other variables serving as continuous predictors. Occurrence models allow the area occupied by species to be evaluated whereas abundance models inform on their preferences in particular areas. These two analytical approaches were chosen given that the environmental factors that influence species abundance may differ from those that limit their distribution (Nielsen et al., 2005). Continuous predictor variables were square-root or logtransformed to fulfil the assumption of normality. Floristic composition variables were reduced by means of a Principal Component Analysis intended to obtain independent components capable of straightforward interpretation, avoiding the colinearity problems revealed by the initial exploratory analyses. In them, the principal species was assigned 50% of total plant cover for each circular plot, the second-dominant receiving 30% and the combined remaining species 10%. Differences in the predictor variables between guanaco, donkey, cattle and all plots together were examined by means of unifactorial ANOVAs. Tukey’s honest significant difference test for unequal samples was used, a posteriori, to compare pairs of means. Model averaging (Burnham and Anderson, 2002) was employed in order to handle model uncertainty linked to multiple causality. This procedure allows a final model to be constructed in which parameters for each relevant variable are averaged across all plausible models. All possible models were thus built and ranked sequentially according to Akaike’s second order information criterion (AICc). Selection of plausible models followed the rule proposed by Burnham and Anderson (2002), in which those where Di 2 have substantial empirical support. See Gray et al. (2009), and McAlpine et al. (2008) for similar approximations. The accuracy and fit of presence models were quantified in terms of the area under the ROC curve (AUC). AUC values greater than 0.7 indicate high model accuracy (Swets, 1988). 3. Results The most widespread (most frequently encountered) ungulates were guanacos (found in 40.2% of plots), followed by cattle (36.5%)
41
Table 1 Untransformed values of the predictor variables (Means SD) in plots where guanaco (n ¼ 88), donkey (n ¼ 61) and cattle (n ¼ 80) were present and in all the sampled plots combined (n ¼ 219). The plant cover variable has been included as it is simple to interpret, despite its elimination from model averaging on account of being strongly correlated with the NDVI. Swc ¼ seasonal watercourses. The different superscripts indicate significant differences in Tukey post hoc comparisons between groups of observations. Variable
Guanacos
Donkeys
0.11 NDVI 0.10 0.02a Plant cover 15.39 14.43a 25.11 a Maximum 2.63 1.08 2.89 plant height Rock cover 62.15 32.58a 33.95 Gradient 8.81 9.60a 3.51 Distance 2.15 2.82a 0.48 from swc a Distance 5.81 3.83 6.17 from springs a Distance 3.24 2.61 3.34 from tracks Distance 9.91 6.72a 4.38 from roads Distance 27.25 10.33a 19.68 from villages
Cattle
Total plots
0.03b 0.11 0.02b 0.10 0.03a 21.49b 22.86 18.87b,c 17.32 17.05a,c 1.22a 3.04 1.21a 2.69 1.11a 36.84b 32.53 35.61b 4.61b 3.18 4.19b 0.43b 0.82 0.81b,c
46.71 37.61b 6.82 9.08a 1.55 2.41a,c
4.95a
8.75 6.07b
6.89 4.98b
2.51a
3.76 2.46a
3.38 2.51a
5.04b
5.84 5.77b,c
7.87 6.23a,c
9.24b
23.92 10.62a,b 24.53 10.14a
and donkeys (27.9%). Table 1 gives values of the predictor variables in plots in which guanacos, donkeys and cattle were recorded, as well as for the whole set of sampled plots. In general, NDVI and plant cover values were similar to availability in plots where guanacos were present, but lower in those where domestic ungulates occurred. In contrast, rock cover and gradient were much greater in plots with guanacos than in those with presence of donkeys and cattle. In addition, domestic ungulates were more often detected near to watercourses and roads.
3.1. Patterns of co-occurrence and abundance among species The partial Mantel test results showed that, after removing the geographical effect, there was no correlation between guanacos and domestic ungulates in either occurrence or abundance. In contrast, there was a significant positive correlation (p < 0.05) between donkeys and cattle for both types of data (Table 2).
3.2. Modelling species patterns Single-best models received only limited support relative to alternative models of occurrence and abundance of guanacos, donkeys and cattle since Akaike weights differed slightly between single-best models and second-ranked models. This indicates substantial uncertainty in choosing a single model and hence supports the use of model averaging (Burnham and Anderson,
Table 2 Correlation coefficients (rM) and significance (p) of partial Mantel tests for paired observations of herbivore occurrence and abundance, controlling for geographical effects. Partial Mantel test Occurrence GuanacoseDonkeys GuanacoseCattle DonkeyseCattle
rM p rM p rM p
Abundance 0.039 0.083 0.045 0.087 0.283 0.001
rM p rM p rM p
0.027 0.449 0.027 0.420 0.102 0.035
42
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
Table 3 Results of model averaging inference for the probability of occurrence and abundance of guanacos. Model-averaged parameter estimates (b) and the unconditional standard errors (SE) of the variables obtained are given. Swi represents the weight of each variable across all the plausible models. Occurrence model-averaged AUC value ¼ 0.91. Guanaco
Occurrence
Predictor
Swi
b
SE
1.000 1.000 0.840 0.236 0.131 0.075 0.072 0.063 0.060 0.056 0.056
16.661 1.326 0.644 0.218 0.044 0.020 0.012 0.014 0.019 0.006 0.109 0.008
2.059 0.175 0.151 0.130 0.049 0.030 0.019 0.023 0.037 0.013 0.374 0.029
Intercept Distance from villages Rock cover Distance from swc Distance from tracks Cattle presence Distance from springs Distance from roads Maximum plant height Donkey presence NDVI Gradient
Abundance
Table 5 Results of model averaging inference for the probability of occurrence and abundance of cattle. Model-averaged parameter estimates (b) and the unconditional standard errors (SE) of the variables obtained are given. Swi represents the weight of each variable across all the plausible models. Occurrence model-averaged AUC value ¼ 0.82. Cattle
Occurrence
Swi
b
SE
Predictor
Swi
b
SE
0.822 1.000 1.000
4.349 0.490 0.536 0.19
3.048 0.268 0.094 0.087
0.187 1.000 0.198 1.000 1.000 1.000 1.000
0.018 0.261 0.022 1.321 0.267 7.621 0.647
0.030 0.116 0.032 0.296 0.126 3.813 0.27
Intercept Distance from villages Gradient NDVI Maximum plant height Donkey presence Distance from swc Distance from tracks Rock cover Distance from roads Distance from springs Guanaco presence
1.000 0.981 0.917 0.763 0.731 0.555 0.523 0.490 0.436 0.214 0.012
15.417 1.000 1.929 16.172 0.763 0.818 0.156 0.181 0.106 0.188 0.075 0.002
6.159 0.713 0.692 7.799 0.578 0.205 0.114 0.139 0.085 0.150 0.081 0
2002). Model-averaged accuracy was satisfactory, with strong AUC values (Table 3, 4 and 5). 3.2.1. Guanacos The model selection process for guanacos stated that 13 occurrence models and four abundance models were regarded as plausible. Rock cover and distance from nearest village were given most weight (wi ¼ 1), having positive coefficients. In short, areas occupied by guanacos were characterised by rocky substrata and by being far from villages. The higher probability of guanaco presence in plots furthest from seasonal watercourses was another robust (wi > 0.8) finding (Table 3). Abundance models confirmed and complemented occurrence models, refining the occupation pattern. Three types of variables greatly influenced guanaco abundance (Table 3). Guanacos were less abundant in plots where donkeys were present (wi ¼ 1), in contrast to what the occurrence models showed. Secondly, an array of variables (wi ¼ 1 in all cases), both physical (rocky areas, of shallower gradient, close to springs but remote from seasonal watercourses) and biotic (plots with higher NDVI values, although avoiding tall vegetation), were of great significance to guanaco abundance. Finally, guanacos were most abundant far from villages (wi > 0.8). In contrast, cattle presence, and distance from roads and tracks had no relevance both to models of occurrence or abundance (wi < 0.2). The two floristic composition PCA axes did not appear in the models. Table 4 Results of model averaging inference for the probability of occurrence and abundance of donkeys. Model-averaged parameter estimates (b) and the unconditional standard errors (SE) of the variables obtained are given. Swi represents the weight of each variable across all the plausible models. Occurrence model-averaged AUC value ¼ 0.84. Donkey
Occurrence
Predictor
Swi
b
SE
Swi
b
SE
Intercept Distance from springs NDVI Rock cover Cattle presence Distance from roads Distance from swc Gradient Maximum plant height Guanaco presence Distance from villages Distance from tracks
1.000 1.000 1.000 1.000 1.000 1.000 0.130 0.124 0.103 0.102 0.098
19.465 1.385 21.012 0.526 0.95 0.843 0.357 0.081 0.072 0.011 0.034 0.008
4.42 0.33 8.669 0.16 0.233 0.218 0.17 0.12 0.111 0.027 0.078 0.021
1.000 1.000 1.000 1.000 1.000 0.041 0.563 0.310 0.045 0.093 0.149
7.939 0.628 8.832 0.18 0.457 0.443 0.001 0.344 0.31 0.002 0.02 0.02
1.471 0.115 3.7 0.07 0.11 0.084 0.003 0.254 0.148 0.006 0.037 0.024
Abundance Swi
b
SE
1.000 1.000 1.000 0.150 1.000
10.210 1.086 1.347 13.218 0.042 0.265
3.051 0.358 0.370 3.289 0.060 0.101
1.000 0.690 0.770 0.525 0.079
0.277 0.095 0.180 0.107 0.006
0.116 0.056 0.106 0.088 0.010
3.2.2. Donkeys Seven occurrence models and 15 abundance models were selected as plausible, and both model types showed a high degree of coincidence. One group of variables was of great importance in both cases (wi ¼ 1): presence of cattle, proximity to springs and roads, avoidance of rocky substrata and attraction to areas of greater NDVI. In contrast, proximity to watercourses was important to donkey occurrence (wi ¼ 1) but not to their abundance (Table 4). Similarly, avoidance of steep gradients was relevant to abundance models (wi > 0.55) but not in presence models (Table 4). Guanaco presence and distance from tracks or villages were unimportant to the averaged models (wi < 0.15). Once again, the two floristic composition PCA axes were not included in the plausible models. 3.2.3. Free-ranging cattle There were 55 plausible occurrence models and nine abundance models. In both model types the avoidance of steep gradients, the use of areas of high NDVI far from villages and the donkey presence (Table 5) were relevant explanatory variables. In addition, other variables of lesser importance in occurrence models pointed to the choice of areas of taller vegetation (wi > 0.7) and proximity to seasonal watercourses (wi > 0.5). Other variables that stood out in abundance models were the preference for areas further from tracks, the avoidance of rocky substrata, the dependence on water sources and proximity to roads (Table 5). Guanaco presence had little weight in the two averaged models (wi < 0.1) and the two floristic composition PCA axes did not appear in the models.
4. Discussion
Abundance
This study analyses the role of biotic and abiotic interactions in determining the spatial occupation and the abundance patterns of three ungulate species, one native and two domestic. We describe the co-occurrence of the domestic ungulates and the apparent lack of interference with occupation by the native species, although guanaco abundance may be affected by donkey presence. Furthermore, the nature of the interaction between guanacos and domestic ungulates suggests no relationship (according to partial Mantel test). In contrast, significant overlap between donkeys and cattle occurs irrespective of the analytical tools employed. In addition, human influence determines the occupation and abundance patterns of guanacos.
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
4.1. Species interactions Our results reveal an important departure from our initial hypothesis. We predicted that, given their feeding type (intermediate feeder or grazers) body weights, and the scarcity of trophic resources, the three ungulates would overlap spatially and would thus potentially compete. However, analyses of spatial co-occurrence show that both presence and abundance of guanacos are not correlated with presence or abundance of the two domestic ungulates. Competition requires overlap between spatial and trophic resources and that the latter should be scarce (Hulbert and Andersen, 2001; Putman, 1996; Prins and Olff, 1998; Voeten and Prins, 1999). Therefore, in the absence of spatial overlap, the potential for competition is low even if trophic resources are limited (Voeten and Prins, 1999). The lack of overlap that we have found may be due to the low density of guanacos (0.1e0.7 individuals/km2, Acebes et al., 2010a) and the even lower densities of donkeys and cattle (own unpublished data). Therefore an increase in domestic herbivore density could lead to greater spatial overlap, although differences in habitat selection between native and exotic species may exist (see below). In any case, the lack of a strong negative correlation between species was to be expected since this would normally appear in the case of specialist species showing coevolved resource partitioning. Finally, it is important to note that the finding of no overlap is robust in face of potential problems arising from our methodology (counting pellet groups within sample plots in the first visit). Had there been spatial overlap between species we would have been unable to discriminate whether this was due to simultaneous or asynchronous co-occurrences, but such a caveat does not apply in this case. Also, the methodology cannot detect interference competition (where one species denies resources to another by means of aggressive behaviour) but we have no evidence that such competition exists from personal observation after three years of fieldwork in the area, and it is hardly relevant to ungulates (Hulbert and Andersen, 2001; Prins, 2000; Ritchie, 2002). In contrast to the situation regarding guanacos, the domestic species spatially overlapped with each other, as occurs in cold-arid regions of Asia (Bagchi et al., 2004), and the potential for competition between them thus exists. This may be due to asynchronous occupation, as noted above, but this does not seem to be the case since donkeys and cattle were often seen together (personal observation). Overlap in areas of occupation (occurrence models) was larger than in areas of preferential use (abundance models). These differences suggest slight differences in resource use between donkeys and cattle. 4.2. Modelling occurrence and abundance patterns The use of both occurrence and abundance models in our case was informative and complementary (Nielsen et al., 2005). The most significant results of the occurrence models were i) the definition of distinct occupation patterns of guanacos and domestic ungulates; ii) the association between donkeys and cattle; and iii) the absence of any effect of domestic ungulates on the occurrence of the native species. For their part, the abundance models refined the occupation patterns of the three species, maintaining the association between donkeys and cattle, but with guanaco abundance being lower in the presence of donkeys. Guanacos showed a tendency to avoid rural settlements as a response to human pressure, occurring inside parks, which underlines the relevance of protected areas to wildlife conservation. Evolutionary factors, such as a preference for open areas offering good visibility for detecting predators (Bank et al., 2003), explain why guanacos avoid seasonal watercourses, where strips of
43
arboreal formations occur. Finally, the preference for rocky substrata over sandy sites may be associated with locomotory needs relating to the guanacos’ toenails and soft footpads, but there is no existing data with which to compare these results. The abundance models allowed the ecological and evolutionary preferences of guanacos to be defined more precisely. Guanacos showed a preference for areas offering a trade-off between trophic resource availability and high visibility. As in other ungulates, such as elk Cervus elaphus (Clair and Farrest, 2009), guanaco occurrence and abundance were unaffected by tourist pressure, as suggested by the lack of any change in landscape use related to distance from the tourist circuit. However, guanacos showed some tendency to avoid roads (Table 1), which are used not only by tourists but also by local people, who may exert other pressures, such as poaching (Wilkie et al., 2000). A lower abundance of guanacos in the presence of donkeys suggests the following interpretations: (i) differences in habitat selection; (ii) competition resulting in the native species being displaced by donkeys, and; (iii) a combination of both effects. Stewart et al. (2002) showed resource partitioning between cattle and native herbivores as a result of competitive displacement of the natives by domestic species. In contrast, Sitters et al. (2009) recently found spatial redistribution between cattle and native herbivores without their being any evidence of displacement, their different ecological requirements being suggested as responsible for this effect. Guanacos occur in nearly all habitats (Acebes et al., 2010a) and prefer slightly different ones to those used by donkeys (Table 1), although the latter may force guanacos to make less use of certain habitats. Therefore, an increase in donkey density may increase the likelihood of spatial overlap between the species, leading to exploitation competition (Owen-Smith, 2002; Putman, 1996). Differences between occurrence and abundance patterns in guanacos are explained in terms of their social behaviour and reproductive strategies, which have implications on the use of spatial and trophic resources by polygynous ungulates (CluttonBrock et al., 1982). Guanacos show resource defence polygyny, a mating system in which territorial males seek habitats that are favourable for attracting females (Franklin, 1983). This behaviour involves a trade-off between choosing abundant trophic resources and selecting habitats that are both suitable for mating and that offer good visibility for avoiding the chief predator, the puma (Bank et al., 2003; Franklin, 1983). In addition to territorial family harems, the populations include non-breeding male groups and solitary males in search of new territories (Franklin, 1983; González et al., 2006). Solitary males and male herds are more mobile than family groups and they are relegated to less favourable habitats by intraspecific competition (Sarno et al., 2003). Family groups carry greater weight in abundance models since they concentrate higher densities of animals and have a large number of latrines within their territories. On the other hand, solitary males and male groups are highlighted in occurrence models since they show more scattered defecation patterns. Factors defining the occurrence and abundance of domestic species did not coincide with those for guanacos, revealing differences in their ecological requirements. This helps to explain why the predicted spatial overlap between them did not occur. Unlike guanacos, both domestic ungulates were most abundant where plant productivity was greatest. Donkeys generally include springs in their territories (Saltz et al., 2000), as revealed by our models. The tendency for cattle to appear far from springs has been recently described (Sitters et al., 2009), and may be related to livestock management as well as to the lower trophic availability near springs due to overgrazing (Owen-Smith, 2002; Sitters et al., 2009). Proximity to seasonal watercourses indicates the greater dependence of cattle on trophic and water resources. Stewart et al. (2002) also found that
44
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
cattle choose areas near to water sources and riparian habitats, avoiding steep locations. Finally, domestic ungulates are more closely linked to humans, as revealed by their closer proximity to rural settlements (donkeys) or to roads (donkeys and cattle), even using the latter for moving about their territories (pers. obs.). To conclude, our analyses of co-occurrence show no evidence of overlap between guanacos and domestic ungulates. Nevertheless, abundance models suggest that increases in the density of donkeys may alter this situation, having a negative impact on guanacos. Management policies for controlling domestic ungulates are thus recommended in locations with higher domestic stock densities. In addition, human presence clearly influences guanaco distribution, displacing them towards areas furthest from rural settlements. Finally, these results may be relevant for the management of arid lands across large expanses of Argentina, Chile and Peru, where guanacos co-occur with donkeys and other domestic species close to areas of human activity. Acknowledgements We thank the staff at the Ischigualasto Provincial Park for their collaboration and help and R. Ovejero and F. Suarez for field work. This research was funded by a Biological Conservation project from the BBVA Foundation (INTERMARG Project). Partial support for UAM researchers is provided by the REMEDINAL-2 research network (S-2009/AMB/1783). This article is dedicated to the memory of Dr. Francisco Suárez. References Acebes, P., Traba, J., Malo, J.E., Ovejero, R., Borghi, C.E., 2010a. Density and habitat use at different spatial scales of a guanaco population (Lama guanicoe) in the Monte desert of Argentina. Mammalia 74, 57e62. Acebes, P., Traba, J., Peco, B., Reus, L., Giannoni, S.M., Malo, J.E., 2010b. Abiotic gradients drive floristic composition and structure of plant communities in the Monte Desert. Revista Chilena de Historia Natural 83, 395e407. Bagchi, S., Mishra, C., Bhatnagar, Y.V., 2004. Conflicts between traditional pastoralism and conservation of Himalayan ibex (Capra sibirica) in the TransHimalayan mountains. Animal Conservation 7, 121e128. Baldi, R., Albon, S., Elston, D., 2001. Guanacos and sheep: evidence for continuing competition in arid Patagonia. Oecologia 129, 561e570. Baldi, R., Pelliza-Sbriller, A., Elston, D., Albon, S., 2004. High potential for competition between guanacos and sheep in Patagonia. Journal of Wildlife Management 68, 924e938. Baldi, B., Lichtenstein, G., González, B., Funes, M., Cuellar, E., Villalba, L., Hoces, D., Puig, S., 2008. Lama Guanicoe. URL:. IUCN 2010 IUCN Red List of Threatened Species. Bank, M.S., Sarno, R.J., Franklin, W.L., 2003. Spatial distribution of guanaco mating sites in southern Chile: conservation implications. Biological Conservation 112, 427e434. Burnham, K.P., Anderson, D.R., 2002. Model Selection and Multimodel Inference: A Practice Information-Theoretic Approach. Spring Verlag, New York. Clair, C.C., Farrest, A., 2009. Impacts of vehicle traffic on the distribution and behaviour of rutting elk, Cervus elaphus. Behaviour 146, 393e413. Clutton-Brock, T.H., Guinness, F.E., Albon, S.D., 1982. Red Deer. Behavior and Ecology of Two Sexes. Edinburgh University Press, Edinburgh. Connell, J.H., 1980. Diversity and the coevolution of competitors, or the ghost of competition past. Oikos 35, 131e138. Cromsigt, J.P.G.M., Prins, H.H.T., Olff, H., 2009. Habitat heterogeneity as a driver of ungulate diversity and distribution patterns: interaction of body mass and digestive strategy. Diversity and Distributions 15, 513e522. Franklin, W.L., 1983. Contrasting socioecologies of South American’s wild camelids: the vicuña and guanaco. Special Publication No. 7. In: Eisenberg, J.F., Kleiman, D.G. (Eds.), Advances in the Study of Mammalian Behavior. American Society of Mammalogists, Allen Press, Lawrence, Kansas, pp. 573e629. Fritz, H., de Garine-Wichatitsky, M., Letessier, G., 1996. Habitat use by sympatric wild and domestic herbivores in an African savanna woodland: the influence of cattle spatial behaviour. Journal of Applied Ecology 33, 589e598. González, B.A., Palma, E., Zapata, B., Marín, J.C., 2006. Taxonomic and biogeographical status of Lama guanicoe (Artyodactila, Camelidae). Mammal Review, 157e178. Gray, T.N.E., Borey, R., Hout, S.K., Chamman, H., Collar, N.J., Dolman, P.M., 2009. Generality of models that predict the distribution of species: conservation activity and reduction of model transferability for a threatened bustard. Conservation Biology 23, 433e439. Grinder, M.I., Krausman, P.R., Hoffmann, R.S., 2006. Equus asinus. Mammalian Species 794, 1e9.
Henley, S.R., Ward, D., Schmidt, I., 2007. Habitat selection by two desert-adapted ungulates. Journal of Arid Environments 70, 39e48. Hulbert, I.A.R., Andersen, R., 2001. Food competition between a large ruminant and a small hindgut fermentor: the case of the roe deer and mountain hare. Oecologia 128, 499e508. Manly, B.F.J., 2007. Randomization, Bootstrap and Monte Carlo Methods in Biology, third ed. Chapman and Hall, London. Márquez, J., Martínez Carretero, E., Dalmasso, A., Pastrán, G., Ortiz, S., 2005. Las áreas protegidas de la provincia de San Juan (Argentina) II. La vegetación del Parque Provincial de Ischigualasto. Multequina. Latin American Journal of Natural Resources 14, 1e27. McAlpine, C.A., Rhodes, J.R., Bowen, M.E., Lunney, D., Callaghan, J.G., Mitchell, D.L., Possingham, H.P., 2008. Can multiscale models of species’ distribution be generalized from region to region? A case study of the koala. Journal of Applied Ecology 45, 558e567. McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models. Chapman and Hall, New York. Mishra, C., Van Wieren, S.E., Ketner, P., Heitkönig, I.M.A., Prins, H.H.T., 2004. Competition between domestic livestock and wild bharal Pseudois nayaur in the Indian Trans-Himalaya. Journal of Applied Ecology 41, 344e354. Moehlman, P.D., 1998. Feral asses (Equus africanus): intraspecific variation in social organization in arid and mesic habitats. Applied Animal Behaviour Science 60, 171e195. Morris, D.W., Davidson, D.L., 2000. Optimally foraging mice match patch use with habitat differences in fitness. Ecology 81, 2061e2066. Neff, D.J., 1968. The pellet-group count technique for big game trend, census, and distribution: a review. Journal of Wildlife Management 32, 597e614. Nielsen, S.E., Johnson, C.J., Heard, D.C., Boyce, M.S., 2005. Can models of presenceabsence be used to scaled abundance? Two case studies considering extremes in life history. Ecography 28, 1e12. Novillo, A., Ojeda, R.A., 2008. The exotic mammals of Argentina. Biological Invasions 10, 1333e1344. Ovejero, R., Acebes, P., Malo, J.E., Traba, J., Mosca Torres, M.E., Borghi, C.E., 2011. Lack of feral livestock interference on native guanaco during the dry season in a South American desert. European Journal of Wildlife Research. doi:10.1007/ s10344-011-0511-4. Owen-Smith, N., 2002. Adaptative Herbivore Ecology. From Resources to Populations in Variable Environments. Wits University Press, Johannesburg. Prins, H.H.T., 2000. Competition between wildlife and livestock in Africa. In: Prins, H.H.T., Grootenhuis, J.G., Dolan, T.T. (Eds.), Wildlife Conservation by Sustainable Use. Kluwer Academic Publishers, Boston, pp. 51e80. Prins, H.H.T., Olff, H., 1998. Species richness of African grazer assemblages: towards a functional explanation. In: Newbery, D., Prins, H.H.T., Brown, N.D. (Eds.), Dynamics of Tropical Communities. Blackwell, Oxford, pp. 448e490. Puig, S., Videla, F., Cona, M.I., Monge, S.A., 2001. Use of food availability by guanacos (Lama guanicoe) and livestock in Northern Patagonia (Mendoza, Argentina). Journal of Arid Environments 47, 291e308. Putman, R.J., 1984. Facts from faeces. Mammal Review 14, 79e97. Putman, R.J., 1996. Competition and Resource Partitioning in Temperate Ungulate Assemblies. Chapman and Hall, London. Raedeke, K., 1979. Population dynamics and socioecology of the guanaco (Lama guanicoe) of Magallanes, Chile. PhD thesis, College of Forest Resources, University of Washington, Washington, USA. Ritchie, M.E., 2002. Competition and coexistence in mobile animals. In: Sommer, U., Worm, B. (Eds.), Competition and Coexistence. Springer, Berlin, pp. 127e141. Rosenberg, M.S., Anderson, C.D., 2011. PASSaGE: Pattern Analysis, Spatial Statistics and Geographic Exegesis. Version 2. Methods in Ecology and Evolution 2, 229e232. Rosenzweig, M.L., 1981. A theory of habitat selection. Ecology 62, 327e335. Saltz, D., Rowen, M., Rubenstein, D.I., 2000. The effect of space-use patterns of reintroduced asiatic wild ass on effective population size. Conservation Biology 14, 1852e1861. Sarno, R.J., Bank, M.S., Stern, H.A., Franklin, W.L., 2003. Forced dispersal of juvenile guanacos (Lama guanicoe): causes, variation, and fates of individuals dispersing at different times. Behavioral Ecology and Sociobiology 54, 22e29. Shrestha, R., Wegge, P., 2008. Habitat relationships between wild and domestic ungulates in Nepalese trans-Himalaya. Journal of Arid Environments 72, 914e925. Sitters, J., Heitkönig, I.M.A., Holmgren, M., Ojwang, G.S.O., 2009. Herded cattle and wild grazers partition water but share forage resources during dry years in East African savannas. Biological Conservation 142, 738e750. Stewart, K.M., Bowyer, R.T., Kie, J.G., Cimon, N.J., Johnson, B.K., 2002. Temporospatial distributions of elk, mule deer, and cattle: resource partitioning and competitive displacement. Journal of Mammalogy 83, 229e244. Suryawanshi, K.R., Bhatnagar, Y.V., Mishra, C., 2010. Why should a grazer browse? Livestock impact on winter resource use by bharal Pseudois nayaur. Oecologia 162, 453e462. Swets, J.A., 1988. Measuring the accuracy of diagnostic systems. Science 240, 1285e1293. Voeten, M.M., Prins, H.H.T., 1999. Resource partitioning between sympatric wild and domestic herbivores in the Tarangire region of Tanzania. Oecologia 120, 287e294. Wilkie, D., Shaw, E., Rotberg, F., Morelli, G., Auzel, P., 2000. Roads, development, and conservation in the Congo Basin. Conservation Biology 14, 1614e1622. Young, T.P., Palmer, T.M., Gadd, M.E., 2005. Competition and compensation among cattle, zebras and elephants in a semi-arid savanna in Laikipia, Kenya. Biological Conservation 122, 351e359.
Contents lists available at SciVerse ScienceDirect
Journal of Arid Environments journal homepage: www.elsevier.com/locate/jaridenv
Co-occurrence and potential for competition between wild and domestic large herbivores in a South American desert P. Acebes*, J. Traba, J.E. Malo Terrestrial Ecology Group-TEG, Departamento de Ecología, Facultad de Ciencias, Universidad Autónoma de Madrid, C/.Darwin, 2, E-28049 Madrid, Spain
a r t i c l e i n f o
a b s t r a c t
Article history: Received 8 July 2010 Received in revised form 28 August 2011 Accepted 8 September 2011 Available online 28 September 2011
Introduction of domestic ungulates may lead to interspecific competition with native herbivores, particularly if the species involved are of similar size and share similar foraging strategies due to a scarcity of trophic resources. This interaction has not been investigated in South American deserts in which guanaco (Lama guanicoe), the larger South American wild camelid and the most widelydistributed ungulate, co-occurs with several introduced ungulates (e.g. cattle, donkeys, sheep or red deer). We studied the occurrence and abundance patterns of guanacos, donkeys and cattle in desert areas of Argentina by dung sampling. Analyses of co-occurrence show no relationship in land use between guanaco and livestock when geographical effects are removed. Distribution and abundance models of the guanaco are strongly associated with sparse plant cover, rocky substrata and human-influence variables, guanacos appearing in areas furthest from villages. Nevertheless, while guanaco distribution is weakly related to feral livestock presence, their abundance is negatively associated with donkey presence. In contrast, livestock species appear closely associated with each other, being centred on areas of densest and most productive vegetation. These results are relevant for the management of extensive arid areas of South America, where guanacos co-occur with feral donkeys and frequently conflicts with humans activities. Ó 2011 Elsevier Ltd. All rights reserved.
Keywords: Coexistence Human impact Lama guanicoe Monte desert Protected areas Resource partitioning
1. Introduction Spatial occurrence patterns of wild species are an expression of multiscalar and multifactorial processes that act simultaneously. Such patterns are determined, at landscape or finer scales, by the evolutionary history of species, environmental factors such as distribution of trophic resources (Morris and Davidson, 2000) and by biotic interactions (Rosenzweig, 1981). Where herbivores are concerned, co-occurrence of native species may be due to evolutionary processes that have enabled coexistence through spatial or trophic resource partitioning (Voeten and Prins, 1999), although they may have competed in the past (the ’ghost of competition past’, Connell, 1980). However, the recent introduction of domestic ungulates into an assemblage of native species may culminate in interspecific competition, given that there has been insufficient time for resource partitioning to evolve (Putman, 1996; Voeten and Prins, 1999). Sympatric species of similar-sized ungulates that have similar foraging strategies may thus compete (Owen-Smith, 2002; Prins and Olff, 1998), although for this to occur there must be a high
* Corresponding author. Tel.: þ34 914978916; fax: þ34 914978001. E-mail address: [email protected] (P. Acebes). 0140-1963/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jaridenv.2011.09.003
degree of overlap in their use of spatial and trophic resources, and such resources must be scarce (Hulbert and Andersen, 2001; Prins and Olff, 1998; Putman, 1996). In addition, humans modify the use of space by wild ungulates since traditional pastoralism drives wild ungulate populations away (Bagchi et al., 2004; Mishra et al., 2004; Prins, 2000). Moreover, wild species are also displaced far from human settlements by poaching (Wilkie et al., 2000). Arid lands world-wide are resource-limited environments. Large native herbivores and livestock in such biomes often compete for space and trophic resources (Young et al., 2005). Studies of competition between native and domestic ungulates in arid and semi-arid biomes are relatively numerous for African savannas (Fritz et al., 1996; Sitters et al., 2009; Voeten and Prins, 1999; Young et al., 2005) as well as for Asian steppes (Bagchi et al., 2004; Shrestha and Wegge, 2008; Suryawanshi et al., 2010). However, very few such studies exist for similar biomes in South America, in which the guanaco (Lama guanicoe, Müller), the largest and most widely-distributed wild ungulate, stands out as a model species to study such interactions. The guanaco population prior to the Spanish conquest was between 30 and 50 million, occupying nearly all of Chile and Argentina, as well as parts of Bolivia, Peru and Paraguay (Raedeke, 1979). Current estimates point at around 600,000 guanacos, the
40
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
pronounced decline being attributed to competition with livestock, habitat loss and poaching (Baldi et al., 2008). Guanaco inhabits arid and semi-arid shrublands and grasslands and hilly, mountain and steppe environments through its geographical range (Baldi et al., 2008). It is sympatric with domestic ungulates in many of these systems but their interaction has only been examined in a few studies on the Patagonian steppes, where the guanaco is abundant and has coexisted with sheep since the latter were introduced at the end of 19th century (Baldi et al., 2001, 2004; Puig et al., 2001). In extremely arid regions, where the guanaco maintains significant populations of high conservation value (Acebes et al., 2010a), it cooccurs with both donkeys (Equus asinus) and free-ranging cattle (Bos taurus). Guanacos are highly social and monomorphic and exhibit a resource defence polygyny mating system (Franklin, 1983). During the breeding season territorial males seek out habitats that are favourable for attracting females, such as areas with abundant forage that are safe from predators, especially pumas (Puma concolor) (Bank et al., 2003). Donkeys are native to arid zones of Africa and they have colonised American and Australian deserts successfully (Grinder et al., 2006), but they remain poorly studied and few studies have evaluated their impact on South American wild fauna (Novillo and Ojeda, 2008; Ovejero et al., 2011). Their social organisation is based on solitary territorial males that have exclusive mating access within their territories, together with stable groups of females and their offspring (Moehlman, 1998). Cattle have received more attention (see for instance Fritz et al., 1996; Stewart et al., 2002) although studies of their competition with South American herbivores are lacking. The present study employed dung sampling to analyse the use of space by three large herbivores: the native guanaco, donkeys, and free-ranging cattle, in two protected areas of the hyper-arid Monte Desert (Argentina), where trophic resources are limited. We hypothesised that there would be a high degree of spatial overlap between the three species given their feeding types (guanacos and donkeys: intermediate feeders; cattle: grazers), their digestive strategies (guanacos and cattle: ruminants; donkeys: hindgut fermenters), their body weights (guanacos 100e120 kg; donkeys: 140 kg; cattle: 250 kg) and the scarcity of trophic resources. In particular, the herbaceous layer is almost nonexistent in the study area (Acebes et al., 2010b) so herbivores rely on browse-dominated diets. The potential for competition between them is thus high and it could have a negative impact on the native species. The basic objectives were a) to determine co-occurrence patterns of the three species from data on their presence and abundance, and b) to define their habitat selection requirements, evaluating the effects of abiotic and biotic factors, including those related to human presence and to the effect of potential competitors. 2. Materials and methods 2.1. Study area The study was carried out in the Ischigualasto-Talampaya World Heritage Site. This comprises two contiguous conservation units, Ischigualasto Provincial Park and Talampaya National Park, located in San Juan and La Rioja provinces respectively, in north-western Argentina (29 550 S, 68 050 W). Two villages some 10e20 km from the protected areas are major sources of cattle and donkeys. The donkeys that inhabit the protected areas and their surroundings are old animals abandoned by farmers and/or their offspring, while cattle range throughout the territory. The parks have a total area of 275,369 ha and extend along the eastern foothills of Central Andes at a mean altitude of 1300 m. The climate is typical of dry
desert with a mean annual temperature below 18 C (range 10 to 45 C) and a mean temperature in the hottest month above 22 C. Mean annual precipitation ranges from 80 to 140 mm, concentrated in summer (November to February). Monte Desert is the dominant biome (Márquez et al., 2005). The predominant vegetation is sparse shrubland, dominated by species of Zygophyllaceae (Larrea spp, Zuccagnia punctata, Bulnesia retamo and Plectrocarpa tetracantha), Fabaceae (Prosopis spp, Cercidium praecox, Geoffroea decorticans, Senna aphylla, Ramorinoa girolae) and Chenopodiacae (Atriplex spp and Suaeda divaricata). Cacti (Echinopsis spp, Tephrocactus spp and Opuntia sulphurea) and Bromeliads (Deuterocohnia longipetala and Tillandsia spp) are also frequent but to a lesser extent. Total plant cover is less than 20%, with a very sparse and seasonal herbaceous layer (Acebes et al., 2010b). 2.2. Herbivore data We used the pellet-group count technique (Neff, 1968) in order to estimate the distribution and relative abundance of the different species. In particular, we used the ‘standing-crop method’, a count of pellet groups within sample plots on the first visit. This method is suitable for arid zones, where ungulate densities are low (Putman, 1984). Bleached pellet groups were ignored, to ensure that counts reflected recent presence (Henley et al., 2007). There is ample evidence that dung counts are more reliable than aerial or ground counts when used to estimate relative habitat use (Cromsigt et al., 2009; Young et al., 2005), particularly in arid environments (Henley et al., 2007). A total of 14 line transects 5 km long were established during 2005 and 2006 taking in the maximum possible environmental diversity. GPS-georeferenced circular plots of 15 m radius were established at minimum intervals of 300 m along each transect. The circular plots totalled 219 since the difficult terrain made sampling incomplete. The abundance of guanaco, donkey and cattle dung was recorded within each plot. Scattered pellets were scored as 1, dung heaps as 2 and guanaco latrines and accumulated dung heaps from donkeys or cattle scored 3. This scale approximates to a logarithmic transformation of the pellet number and makes sampling easier, given that individual pellet counts are impossible in medium-sized or large dung piles. 2.3. Explanatory variables The following environmental variables were measured in each circular plot: gradient ( ), plant cover (%) and extent of rock or bare ground (%). Mean and maximum vegetation heights (m) were estimated and the principal plant species, the second-commonest species and other plants present were recorded. Predictor variables were calculated using ArcGIS 9.2 software on existing maps. These were the Normalised Difference Vegetation Index (NDVI) using Landsat 7 ETM þ images for a buffer zone of 50 m radius around each circular plot (named ’NDVI’ hereafter); and the minimum distance from each plot to a village (‘distance from villages’), to a small permanent spring (‘distance from springs’) and to a seasonal watercourse (‘distance from swc’), where the densest formations of Prosopis chilensis and Prosopis flexuosa of the study area occur (Acebes et al., 2010b). Distances to roads (’distance from roads’), which are used by tourists, village inhabitants and potential illegal hunters, and to tracks (’distance from tracks’), used by tourists and park wardens to visit and move around protected areas respectively, were also measured. A prior analysis of cross-correlations was employed to reduce the number of original predictor variables and to avoid colinearity between them. The resulting redundant variables (extent of bare
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
ground and mean vegetation height) were omitted from subsequent analyses. 2.4. Patterns of co-occurrence and abundance among herbivores Correlation of occurrence and abundance for the three species was evaluated by a partial Mantel test, in order to remove the spatial effect that may determine co-occurrence patterns between species (Manly, 2007). Presence and abundance of each species’ dung in each circular plot were employed as response matrices and the geographical location of each plot was taken as the control matrix. Analyses of occurrence used Jaccard distance matrices and those of abundance used Euclidean distance matrices. The said matrices represent the differences between pairs of occurrence/ abundance observations of species-pairs and the geographical distance between their locations. Significance testing employed a randomisation approach with 9999 permutations, employing the PASSaGE v.2 program (Rosenberg and Anderson, 2011). 2.5. Modelling species patterns Generalised Linear Models were used to examine the relationship between occurrence and abundance for each species with different predictor variables, assuming a binomial error distribution and a logit link function for presence/absence data and a Poisson error distribution and a log link function for abundance data (McCullagh and Nelder, 1989). Presence/absence of potentially competing species was introduced as a categorical predictor, the other variables serving as continuous predictors. Occurrence models allow the area occupied by species to be evaluated whereas abundance models inform on their preferences in particular areas. These two analytical approaches were chosen given that the environmental factors that influence species abundance may differ from those that limit their distribution (Nielsen et al., 2005). Continuous predictor variables were square-root or logtransformed to fulfil the assumption of normality. Floristic composition variables were reduced by means of a Principal Component Analysis intended to obtain independent components capable of straightforward interpretation, avoiding the colinearity problems revealed by the initial exploratory analyses. In them, the principal species was assigned 50% of total plant cover for each circular plot, the second-dominant receiving 30% and the combined remaining species 10%. Differences in the predictor variables between guanaco, donkey, cattle and all plots together were examined by means of unifactorial ANOVAs. Tukey’s honest significant difference test for unequal samples was used, a posteriori, to compare pairs of means. Model averaging (Burnham and Anderson, 2002) was employed in order to handle model uncertainty linked to multiple causality. This procedure allows a final model to be constructed in which parameters for each relevant variable are averaged across all plausible models. All possible models were thus built and ranked sequentially according to Akaike’s second order information criterion (AICc). Selection of plausible models followed the rule proposed by Burnham and Anderson (2002), in which those where Di 2 have substantial empirical support. See Gray et al. (2009), and McAlpine et al. (2008) for similar approximations. The accuracy and fit of presence models were quantified in terms of the area under the ROC curve (AUC). AUC values greater than 0.7 indicate high model accuracy (Swets, 1988). 3. Results The most widespread (most frequently encountered) ungulates were guanacos (found in 40.2% of plots), followed by cattle (36.5%)
41
Table 1 Untransformed values of the predictor variables (Means SD) in plots where guanaco (n ¼ 88), donkey (n ¼ 61) and cattle (n ¼ 80) were present and in all the sampled plots combined (n ¼ 219). The plant cover variable has been included as it is simple to interpret, despite its elimination from model averaging on account of being strongly correlated with the NDVI. Swc ¼ seasonal watercourses. The different superscripts indicate significant differences in Tukey post hoc comparisons between groups of observations. Variable
Guanacos
Donkeys
0.11 NDVI 0.10 0.02a Plant cover 15.39 14.43a 25.11 a Maximum 2.63 1.08 2.89 plant height Rock cover 62.15 32.58a 33.95 Gradient 8.81 9.60a 3.51 Distance 2.15 2.82a 0.48 from swc a Distance 5.81 3.83 6.17 from springs a Distance 3.24 2.61 3.34 from tracks Distance 9.91 6.72a 4.38 from roads Distance 27.25 10.33a 19.68 from villages
Cattle
Total plots
0.03b 0.11 0.02b 0.10 0.03a 21.49b 22.86 18.87b,c 17.32 17.05a,c 1.22a 3.04 1.21a 2.69 1.11a 36.84b 32.53 35.61b 4.61b 3.18 4.19b 0.43b 0.82 0.81b,c
46.71 37.61b 6.82 9.08a 1.55 2.41a,c
4.95a
8.75 6.07b
6.89 4.98b
2.51a
3.76 2.46a
3.38 2.51a
5.04b
5.84 5.77b,c
7.87 6.23a,c
9.24b
23.92 10.62a,b 24.53 10.14a
and donkeys (27.9%). Table 1 gives values of the predictor variables in plots in which guanacos, donkeys and cattle were recorded, as well as for the whole set of sampled plots. In general, NDVI and plant cover values were similar to availability in plots where guanacos were present, but lower in those where domestic ungulates occurred. In contrast, rock cover and gradient were much greater in plots with guanacos than in those with presence of donkeys and cattle. In addition, domestic ungulates were more often detected near to watercourses and roads.
3.1. Patterns of co-occurrence and abundance among species The partial Mantel test results showed that, after removing the geographical effect, there was no correlation between guanacos and domestic ungulates in either occurrence or abundance. In contrast, there was a significant positive correlation (p < 0.05) between donkeys and cattle for both types of data (Table 2).
3.2. Modelling species patterns Single-best models received only limited support relative to alternative models of occurrence and abundance of guanacos, donkeys and cattle since Akaike weights differed slightly between single-best models and second-ranked models. This indicates substantial uncertainty in choosing a single model and hence supports the use of model averaging (Burnham and Anderson,
Table 2 Correlation coefficients (rM) and significance (p) of partial Mantel tests for paired observations of herbivore occurrence and abundance, controlling for geographical effects. Partial Mantel test Occurrence GuanacoseDonkeys GuanacoseCattle DonkeyseCattle
rM p rM p rM p
Abundance 0.039 0.083 0.045 0.087 0.283 0.001
rM p rM p rM p
0.027 0.449 0.027 0.420 0.102 0.035
42
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
Table 3 Results of model averaging inference for the probability of occurrence and abundance of guanacos. Model-averaged parameter estimates (b) and the unconditional standard errors (SE) of the variables obtained are given. Swi represents the weight of each variable across all the plausible models. Occurrence model-averaged AUC value ¼ 0.91. Guanaco
Occurrence
Predictor
Swi
b
SE
1.000 1.000 0.840 0.236 0.131 0.075 0.072 0.063 0.060 0.056 0.056
16.661 1.326 0.644 0.218 0.044 0.020 0.012 0.014 0.019 0.006 0.109 0.008
2.059 0.175 0.151 0.130 0.049 0.030 0.019 0.023 0.037 0.013 0.374 0.029
Intercept Distance from villages Rock cover Distance from swc Distance from tracks Cattle presence Distance from springs Distance from roads Maximum plant height Donkey presence NDVI Gradient
Abundance
Table 5 Results of model averaging inference for the probability of occurrence and abundance of cattle. Model-averaged parameter estimates (b) and the unconditional standard errors (SE) of the variables obtained are given. Swi represents the weight of each variable across all the plausible models. Occurrence model-averaged AUC value ¼ 0.82. Cattle
Occurrence
Swi
b
SE
Predictor
Swi
b
SE
0.822 1.000 1.000
4.349 0.490 0.536 0.19
3.048 0.268 0.094 0.087
0.187 1.000 0.198 1.000 1.000 1.000 1.000
0.018 0.261 0.022 1.321 0.267 7.621 0.647
0.030 0.116 0.032 0.296 0.126 3.813 0.27
Intercept Distance from villages Gradient NDVI Maximum plant height Donkey presence Distance from swc Distance from tracks Rock cover Distance from roads Distance from springs Guanaco presence
1.000 0.981 0.917 0.763 0.731 0.555 0.523 0.490 0.436 0.214 0.012
15.417 1.000 1.929 16.172 0.763 0.818 0.156 0.181 0.106 0.188 0.075 0.002
6.159 0.713 0.692 7.799 0.578 0.205 0.114 0.139 0.085 0.150 0.081 0
2002). Model-averaged accuracy was satisfactory, with strong AUC values (Table 3, 4 and 5). 3.2.1. Guanacos The model selection process for guanacos stated that 13 occurrence models and four abundance models were regarded as plausible. Rock cover and distance from nearest village were given most weight (wi ¼ 1), having positive coefficients. In short, areas occupied by guanacos were characterised by rocky substrata and by being far from villages. The higher probability of guanaco presence in plots furthest from seasonal watercourses was another robust (wi > 0.8) finding (Table 3). Abundance models confirmed and complemented occurrence models, refining the occupation pattern. Three types of variables greatly influenced guanaco abundance (Table 3). Guanacos were less abundant in plots where donkeys were present (wi ¼ 1), in contrast to what the occurrence models showed. Secondly, an array of variables (wi ¼ 1 in all cases), both physical (rocky areas, of shallower gradient, close to springs but remote from seasonal watercourses) and biotic (plots with higher NDVI values, although avoiding tall vegetation), were of great significance to guanaco abundance. Finally, guanacos were most abundant far from villages (wi > 0.8). In contrast, cattle presence, and distance from roads and tracks had no relevance both to models of occurrence or abundance (wi < 0.2). The two floristic composition PCA axes did not appear in the models. Table 4 Results of model averaging inference for the probability of occurrence and abundance of donkeys. Model-averaged parameter estimates (b) and the unconditional standard errors (SE) of the variables obtained are given. Swi represents the weight of each variable across all the plausible models. Occurrence model-averaged AUC value ¼ 0.84. Donkey
Occurrence
Predictor
Swi
b
SE
Swi
b
SE
Intercept Distance from springs NDVI Rock cover Cattle presence Distance from roads Distance from swc Gradient Maximum plant height Guanaco presence Distance from villages Distance from tracks
1.000 1.000 1.000 1.000 1.000 1.000 0.130 0.124 0.103 0.102 0.098
19.465 1.385 21.012 0.526 0.95 0.843 0.357 0.081 0.072 0.011 0.034 0.008
4.42 0.33 8.669 0.16 0.233 0.218 0.17 0.12 0.111 0.027 0.078 0.021
1.000 1.000 1.000 1.000 1.000 0.041 0.563 0.310 0.045 0.093 0.149
7.939 0.628 8.832 0.18 0.457 0.443 0.001 0.344 0.31 0.002 0.02 0.02
1.471 0.115 3.7 0.07 0.11 0.084 0.003 0.254 0.148 0.006 0.037 0.024
Abundance Swi
b
SE
1.000 1.000 1.000 0.150 1.000
10.210 1.086 1.347 13.218 0.042 0.265
3.051 0.358 0.370 3.289 0.060 0.101
1.000 0.690 0.770 0.525 0.079
0.277 0.095 0.180 0.107 0.006
0.116 0.056 0.106 0.088 0.010
3.2.2. Donkeys Seven occurrence models and 15 abundance models were selected as plausible, and both model types showed a high degree of coincidence. One group of variables was of great importance in both cases (wi ¼ 1): presence of cattle, proximity to springs and roads, avoidance of rocky substrata and attraction to areas of greater NDVI. In contrast, proximity to watercourses was important to donkey occurrence (wi ¼ 1) but not to their abundance (Table 4). Similarly, avoidance of steep gradients was relevant to abundance models (wi > 0.55) but not in presence models (Table 4). Guanaco presence and distance from tracks or villages were unimportant to the averaged models (wi < 0.15). Once again, the two floristic composition PCA axes were not included in the plausible models. 3.2.3. Free-ranging cattle There were 55 plausible occurrence models and nine abundance models. In both model types the avoidance of steep gradients, the use of areas of high NDVI far from villages and the donkey presence (Table 5) were relevant explanatory variables. In addition, other variables of lesser importance in occurrence models pointed to the choice of areas of taller vegetation (wi > 0.7) and proximity to seasonal watercourses (wi > 0.5). Other variables that stood out in abundance models were the preference for areas further from tracks, the avoidance of rocky substrata, the dependence on water sources and proximity to roads (Table 5). Guanaco presence had little weight in the two averaged models (wi < 0.1) and the two floristic composition PCA axes did not appear in the models.
4. Discussion
Abundance
This study analyses the role of biotic and abiotic interactions in determining the spatial occupation and the abundance patterns of three ungulate species, one native and two domestic. We describe the co-occurrence of the domestic ungulates and the apparent lack of interference with occupation by the native species, although guanaco abundance may be affected by donkey presence. Furthermore, the nature of the interaction between guanacos and domestic ungulates suggests no relationship (according to partial Mantel test). In contrast, significant overlap between donkeys and cattle occurs irrespective of the analytical tools employed. In addition, human influence determines the occupation and abundance patterns of guanacos.
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
4.1. Species interactions Our results reveal an important departure from our initial hypothesis. We predicted that, given their feeding type (intermediate feeder or grazers) body weights, and the scarcity of trophic resources, the three ungulates would overlap spatially and would thus potentially compete. However, analyses of spatial co-occurrence show that both presence and abundance of guanacos are not correlated with presence or abundance of the two domestic ungulates. Competition requires overlap between spatial and trophic resources and that the latter should be scarce (Hulbert and Andersen, 2001; Putman, 1996; Prins and Olff, 1998; Voeten and Prins, 1999). Therefore, in the absence of spatial overlap, the potential for competition is low even if trophic resources are limited (Voeten and Prins, 1999). The lack of overlap that we have found may be due to the low density of guanacos (0.1e0.7 individuals/km2, Acebes et al., 2010a) and the even lower densities of donkeys and cattle (own unpublished data). Therefore an increase in domestic herbivore density could lead to greater spatial overlap, although differences in habitat selection between native and exotic species may exist (see below). In any case, the lack of a strong negative correlation between species was to be expected since this would normally appear in the case of specialist species showing coevolved resource partitioning. Finally, it is important to note that the finding of no overlap is robust in face of potential problems arising from our methodology (counting pellet groups within sample plots in the first visit). Had there been spatial overlap between species we would have been unable to discriminate whether this was due to simultaneous or asynchronous co-occurrences, but such a caveat does not apply in this case. Also, the methodology cannot detect interference competition (where one species denies resources to another by means of aggressive behaviour) but we have no evidence that such competition exists from personal observation after three years of fieldwork in the area, and it is hardly relevant to ungulates (Hulbert and Andersen, 2001; Prins, 2000; Ritchie, 2002). In contrast to the situation regarding guanacos, the domestic species spatially overlapped with each other, as occurs in cold-arid regions of Asia (Bagchi et al., 2004), and the potential for competition between them thus exists. This may be due to asynchronous occupation, as noted above, but this does not seem to be the case since donkeys and cattle were often seen together (personal observation). Overlap in areas of occupation (occurrence models) was larger than in areas of preferential use (abundance models). These differences suggest slight differences in resource use between donkeys and cattle. 4.2. Modelling occurrence and abundance patterns The use of both occurrence and abundance models in our case was informative and complementary (Nielsen et al., 2005). The most significant results of the occurrence models were i) the definition of distinct occupation patterns of guanacos and domestic ungulates; ii) the association between donkeys and cattle; and iii) the absence of any effect of domestic ungulates on the occurrence of the native species. For their part, the abundance models refined the occupation patterns of the three species, maintaining the association between donkeys and cattle, but with guanaco abundance being lower in the presence of donkeys. Guanacos showed a tendency to avoid rural settlements as a response to human pressure, occurring inside parks, which underlines the relevance of protected areas to wildlife conservation. Evolutionary factors, such as a preference for open areas offering good visibility for detecting predators (Bank et al., 2003), explain why guanacos avoid seasonal watercourses, where strips of
43
arboreal formations occur. Finally, the preference for rocky substrata over sandy sites may be associated with locomotory needs relating to the guanacos’ toenails and soft footpads, but there is no existing data with which to compare these results. The abundance models allowed the ecological and evolutionary preferences of guanacos to be defined more precisely. Guanacos showed a preference for areas offering a trade-off between trophic resource availability and high visibility. As in other ungulates, such as elk Cervus elaphus (Clair and Farrest, 2009), guanaco occurrence and abundance were unaffected by tourist pressure, as suggested by the lack of any change in landscape use related to distance from the tourist circuit. However, guanacos showed some tendency to avoid roads (Table 1), which are used not only by tourists but also by local people, who may exert other pressures, such as poaching (Wilkie et al., 2000). A lower abundance of guanacos in the presence of donkeys suggests the following interpretations: (i) differences in habitat selection; (ii) competition resulting in the native species being displaced by donkeys, and; (iii) a combination of both effects. Stewart et al. (2002) showed resource partitioning between cattle and native herbivores as a result of competitive displacement of the natives by domestic species. In contrast, Sitters et al. (2009) recently found spatial redistribution between cattle and native herbivores without their being any evidence of displacement, their different ecological requirements being suggested as responsible for this effect. Guanacos occur in nearly all habitats (Acebes et al., 2010a) and prefer slightly different ones to those used by donkeys (Table 1), although the latter may force guanacos to make less use of certain habitats. Therefore, an increase in donkey density may increase the likelihood of spatial overlap between the species, leading to exploitation competition (Owen-Smith, 2002; Putman, 1996). Differences between occurrence and abundance patterns in guanacos are explained in terms of their social behaviour and reproductive strategies, which have implications on the use of spatial and trophic resources by polygynous ungulates (CluttonBrock et al., 1982). Guanacos show resource defence polygyny, a mating system in which territorial males seek habitats that are favourable for attracting females (Franklin, 1983). This behaviour involves a trade-off between choosing abundant trophic resources and selecting habitats that are both suitable for mating and that offer good visibility for avoiding the chief predator, the puma (Bank et al., 2003; Franklin, 1983). In addition to territorial family harems, the populations include non-breeding male groups and solitary males in search of new territories (Franklin, 1983; González et al., 2006). Solitary males and male herds are more mobile than family groups and they are relegated to less favourable habitats by intraspecific competition (Sarno et al., 2003). Family groups carry greater weight in abundance models since they concentrate higher densities of animals and have a large number of latrines within their territories. On the other hand, solitary males and male groups are highlighted in occurrence models since they show more scattered defecation patterns. Factors defining the occurrence and abundance of domestic species did not coincide with those for guanacos, revealing differences in their ecological requirements. This helps to explain why the predicted spatial overlap between them did not occur. Unlike guanacos, both domestic ungulates were most abundant where plant productivity was greatest. Donkeys generally include springs in their territories (Saltz et al., 2000), as revealed by our models. The tendency for cattle to appear far from springs has been recently described (Sitters et al., 2009), and may be related to livestock management as well as to the lower trophic availability near springs due to overgrazing (Owen-Smith, 2002; Sitters et al., 2009). Proximity to seasonal watercourses indicates the greater dependence of cattle on trophic and water resources. Stewart et al. (2002) also found that
44
P. Acebes et al. / Journal of Arid Environments 77 (2012) 39e44
cattle choose areas near to water sources and riparian habitats, avoiding steep locations. Finally, domestic ungulates are more closely linked to humans, as revealed by their closer proximity to rural settlements (donkeys) or to roads (donkeys and cattle), even using the latter for moving about their territories (pers. obs.). To conclude, our analyses of co-occurrence show no evidence of overlap between guanacos and domestic ungulates. Nevertheless, abundance models suggest that increases in the density of donkeys may alter this situation, having a negative impact on guanacos. Management policies for controlling domestic ungulates are thus recommended in locations with higher domestic stock densities. In addition, human presence clearly influences guanaco distribution, displacing them towards areas furthest from rural settlements. Finally, these results may be relevant for the management of arid lands across large expanses of Argentina, Chile and Peru, where guanacos co-occur with donkeys and other domestic species close to areas of human activity. Acknowledgements We thank the staff at the Ischigualasto Provincial Park for their collaboration and help and R. Ovejero and F. Suarez for field work. This research was funded by a Biological Conservation project from the BBVA Foundation (INTERMARG Project). Partial support for UAM researchers is provided by the REMEDINAL-2 research network (S-2009/AMB/1783). This article is dedicated to the memory of Dr. Francisco Suárez. References Acebes, P., Traba, J., Malo, J.E., Ovejero, R., Borghi, C.E., 2010a. Density and habitat use at different spatial scales of a guanaco population (Lama guanicoe) in the Monte desert of Argentina. Mammalia 74, 57e62. Acebes, P., Traba, J., Peco, B., Reus, L., Giannoni, S.M., Malo, J.E., 2010b. Abiotic gradients drive floristic composition and structure of plant communities in the Monte Desert. Revista Chilena de Historia Natural 83, 395e407. Bagchi, S., Mishra, C., Bhatnagar, Y.V., 2004. Conflicts between traditional pastoralism and conservation of Himalayan ibex (Capra sibirica) in the TransHimalayan mountains. Animal Conservation 7, 121e128. Baldi, R., Albon, S., Elston, D., 2001. Guanacos and sheep: evidence for continuing competition in arid Patagonia. Oecologia 129, 561e570. Baldi, R., Pelliza-Sbriller, A., Elston, D., Albon, S., 2004. High potential for competition between guanacos and sheep in Patagonia. Journal of Wildlife Management 68, 924e938. Baldi, B., Lichtenstein, G., González, B., Funes, M., Cuellar, E., Villalba, L., Hoces, D., Puig, S., 2008. Lama Guanicoe. URL:. IUCN 2010
Henley, S.R., Ward, D., Schmidt, I., 2007. Habitat selection by two desert-adapted ungulates. Journal of Arid Environments 70, 39e48. Hulbert, I.A.R., Andersen, R., 2001. Food competition between a large ruminant and a small hindgut fermentor: the case of the roe deer and mountain hare. Oecologia 128, 499e508. Manly, B.F.J., 2007. Randomization, Bootstrap and Monte Carlo Methods in Biology, third ed. Chapman and Hall, London. Márquez, J., Martínez Carretero, E., Dalmasso, A., Pastrán, G., Ortiz, S., 2005. Las áreas protegidas de la provincia de San Juan (Argentina) II. La vegetación del Parque Provincial de Ischigualasto. Multequina. Latin American Journal of Natural Resources 14, 1e27. McAlpine, C.A., Rhodes, J.R., Bowen, M.E., Lunney, D., Callaghan, J.G., Mitchell, D.L., Possingham, H.P., 2008. Can multiscale models of species’ distribution be generalized from region to region? A case study of the koala. Journal of Applied Ecology 45, 558e567. McCullagh, P., Nelder, J.A., 1989. Generalized Linear Models. Chapman and Hall, New York. Mishra, C., Van Wieren, S.E., Ketner, P., Heitkönig, I.M.A., Prins, H.H.T., 2004. Competition between domestic livestock and wild bharal Pseudois nayaur in the Indian Trans-Himalaya. Journal of Applied Ecology 41, 344e354. Moehlman, P.D., 1998. Feral asses (Equus africanus): intraspecific variation in social organization in arid and mesic habitats. Applied Animal Behaviour Science 60, 171e195. Morris, D.W., Davidson, D.L., 2000. Optimally foraging mice match patch use with habitat differences in fitness. Ecology 81, 2061e2066. Neff, D.J., 1968. The pellet-group count technique for big game trend, census, and distribution: a review. Journal of Wildlife Management 32, 597e614. Nielsen, S.E., Johnson, C.J., Heard, D.C., Boyce, M.S., 2005. Can models of presenceabsence be used to scaled abundance? Two case studies considering extremes in life history. Ecography 28, 1e12. Novillo, A., Ojeda, R.A., 2008. The exotic mammals of Argentina. Biological Invasions 10, 1333e1344. Ovejero, R., Acebes, P., Malo, J.E., Traba, J., Mosca Torres, M.E., Borghi, C.E., 2011. Lack of feral livestock interference on native guanaco during the dry season in a South American desert. European Journal of Wildlife Research. doi:10.1007/ s10344-011-0511-4. Owen-Smith, N., 2002. Adaptative Herbivore Ecology. From Resources to Populations in Variable Environments. Wits University Press, Johannesburg. Prins, H.H.T., 2000. Competition between wildlife and livestock in Africa. In: Prins, H.H.T., Grootenhuis, J.G., Dolan, T.T. (Eds.), Wildlife Conservation by Sustainable Use. Kluwer Academic Publishers, Boston, pp. 51e80. Prins, H.H.T., Olff, H., 1998. Species richness of African grazer assemblages: towards a functional explanation. In: Newbery, D., Prins, H.H.T., Brown, N.D. (Eds.), Dynamics of Tropical Communities. Blackwell, Oxford, pp. 448e490. Puig, S., Videla, F., Cona, M.I., Monge, S.A., 2001. Use of food availability by guanacos (Lama guanicoe) and livestock in Northern Patagonia (Mendoza, Argentina). Journal of Arid Environments 47, 291e308. Putman, R.J., 1984. Facts from faeces. Mammal Review 14, 79e97. Putman, R.J., 1996. Competition and Resource Partitioning in Temperate Ungulate Assemblies. Chapman and Hall, London. Raedeke, K., 1979. Population dynamics and socioecology of the guanaco (Lama guanicoe) of Magallanes, Chile. PhD thesis, College of Forest Resources, University of Washington, Washington, USA. Ritchie, M.E., 2002. Competition and coexistence in mobile animals. In: Sommer, U., Worm, B. (Eds.), Competition and Coexistence. Springer, Berlin, pp. 127e141. Rosenberg, M.S., Anderson, C.D., 2011. PASSaGE: Pattern Analysis, Spatial Statistics and Geographic Exegesis. Version 2. Methods in Ecology and Evolution 2, 229e232. Rosenzweig, M.L., 1981. A theory of habitat selection. Ecology 62, 327e335. Saltz, D., Rowen, M., Rubenstein, D.I., 2000. The effect of space-use patterns of reintroduced asiatic wild ass on effective population size. Conservation Biology 14, 1852e1861. Sarno, R.J., Bank, M.S., Stern, H.A., Franklin, W.L., 2003. Forced dispersal of juvenile guanacos (Lama guanicoe): causes, variation, and fates of individuals dispersing at different times. Behavioral Ecology and Sociobiology 54, 22e29. Shrestha, R., Wegge, P., 2008. Habitat relationships between wild and domestic ungulates in Nepalese trans-Himalaya. Journal of Arid Environments 72, 914e925. Sitters, J., Heitkönig, I.M.A., Holmgren, M., Ojwang, G.S.O., 2009. Herded cattle and wild grazers partition water but share forage resources during dry years in East African savannas. Biological Conservation 142, 738e750. Stewart, K.M., Bowyer, R.T., Kie, J.G., Cimon, N.J., Johnson, B.K., 2002. Temporospatial distributions of elk, mule deer, and cattle: resource partitioning and competitive displacement. Journal of Mammalogy 83, 229e244. Suryawanshi, K.R., Bhatnagar, Y.V., Mishra, C., 2010. Why should a grazer browse? Livestock impact on winter resource use by bharal Pseudois nayaur. Oecologia 162, 453e462. Swets, J.A., 1988. Measuring the accuracy of diagnostic systems. Science 240, 1285e1293. Voeten, M.M., Prins, H.H.T., 1999. Resource partitioning between sympatric wild and domestic herbivores in the Tarangire region of Tanzania. Oecologia 120, 287e294. Wilkie, D., Shaw, E., Rotberg, F., Morelli, G., Auzel, P., 2000. Roads, development, and conservation in the Congo Basin. Conservation Biology 14, 1614e1622. Young, T.P., Palmer, T.M., Gadd, M.E., 2005. Competition and compensation among cattle, zebras and elephants in a semi-arid savanna in Laikipia, Kenya. Biological Conservation 122, 351e359.