Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Deep ecology wikipedia , lookup
Human impact on the nitrogen cycle wikipedia , lookup
Landscape ecology wikipedia , lookup
Cultural ecology wikipedia , lookup
Plant defense against herbivory wikipedia , lookup
Reforestation wikipedia , lookup
Perovskia atriplicifolia wikipedia , lookup
Theoretical ecology wikipedia , lookup
Tree shaping wikipedia , lookup
Journal of Animal Ecology 2011, 80, 875–883 doi: 10.1111/j.1365-2656.2011.01819.x Soil nutrient status determines how elephant utilize trees and shape environments Yolanda Pretorius*1, Willem F. de Boer1, Cornelis van der Waal1, Henjo J. de Knegt1, Rina C. Grant2, Nicky M. Knox3, Edward M. Kohi1, Emmanuel Mwakiwa1, Bruce R. Page4, Mike J. S. Peel5, Andrew K. Skidmore3, Rob Slotow4, Sipke E. van Wieren1 and Herbert H. T. Prins1 1 Resource Ecology Group, Wageningen University, Droevendaalsesteeg 3a, 6708 PB Wageningen, The Netherlands; Scientific Services, Kruger National Park, Private Bag X402, Skukuza 1350, South Africa; 3International Institute for Geo-Information Science and Earth Observation, PO Box 6, 7500 AA, Enschede, The Netherlands; 4Amarula Elephant Research Programme, School of Biological and Conservation Sciences, University of KwaZulu-Natal, Durban 4041, South Africa; and 5Agricultural Research Council- Range and Forage Institute, PO Box 13054, Nelspruit 1200, South Africa 2 Summary 1. Elucidation of the mechanism determining the spatial scale of patch selection by herbivores has been complicated by the way in which resource availability at a specific scale is measured and by vigilance behaviour of the herbivores themselves. To reduce these complications, we studied patch selection by an animal with negligible predation risk, the African elephant. 2. We introduce the concept of nutrient load as the product of patch size, number of patches and local patch nutrient concentration. Nutrient load provides a novel spatially explicit expression of the total available nutrients a herbivore can select from. 3. We hypothesized that elephant would select nutrient-rich patches, based on the nutrient load per 2500 m2 down to the individual plant scale, and that this selection will depend on the nitrogen and phosphorous contents of plants. 4. We predicted that elephant would cause more adverse impact to trees of lower value to them in order to reach plant parts with higher nutrient concentrations such as bark and root. However, elephant should maintain nutrient-rich trees by inducing coppicing of trees through re-utilization of leaves. 5. Elephant patch selection was measured in a homogenous tree species stand by manipulating the spatial distribution of soil nutrients in a large field experiment using NPK fertilizer. 6. Elephant were able to select nutrient-rich patches and utilized Colophospermum mopane trees inside these patches more than outside, at scales ranging from 2500 down to 100 m2. 7. Although both nitrogen and phosphorus contents of leaves from C. mopane trees were higher in fertilized and selected patches, patch choice correlated most strongly with nitrogen content. As predicted, stripping of leaves occurred more in nutrient-rich patches, while adverse impact such as uprooting of trees occurred more in nutrient-poor areas. 8. Our results emphasize the necessity of including scale-dependent selectivity in foraging studies and how elephant foraging behaviour can be used as indicators of change in the availability of nutrients. Key-words: Colophospermum mopane, nitrogen, nutrient load, patch selection, reuse, spatial scale Introduction Classical optimal foraging theory has neglected the issue of scale (Prins & van Langevelde 2008), although it is well recognized that animals select forage resources at a range of temporal and spatial scales (Spalinger & Hobbs 1992; Wallis*Correspondence author. E-mail: [email protected] DeVries, Laca & Demment 1999; Ball, Danell & Sunesson 2000). Some studies have developed a theoretical framework to explain the spatial scaling of foraging (Bailey et al. 1996; Ritchie & Olff 1999), but support from field research in ‘natural’ systems is scarce (Cromsigt & Olff 2006). With this study, we aim to contribute to the knowledge of foraging theory by conducting a field experiment that incorporates the issues of spatial scale and nutrient heterogeneity. 2011 The Authors. Journal of Animal Ecology 2011 British Ecological Society 876 Y. Pretorius et al. A herbivore that is selective and able to discriminate between food of good or poor quality should have a selective advantage (Fryxell 2008). However, herbivore food selection is, unlike a carnivore, much less concentrated, being distributed as a nested hierarchy of aggregated resources varying widely in nutrient composition and mass (Senft et al.1987; Kotliar & Wiens 1990; Bailey et al. 1996; Searle, Hobbs & Shipley 2005; Fryxell 2008). These aggregated resources can be defined as patches that are discrete spatial units differing from their surroundings in composition and ⁄ or appearance and that cause changes in herbivore foraging behaviour (Kotliar & Wiens 1990; Searle, Hobbs & Shipley 2005). Furthermore, patch selection by herbivores depends on the spatial scales at which the environment is perceived, where ‘grain’ is the smallest scale and ‘extent’ the largest scale, of heterogeneity to which a herbivore responds to patch structure (Kotliar & Wiens 1990). Given that boundaries between the subunits within different hierarchical scales should be defined by the animals’ perceptions and foraging responses (Senft et al. 1987), we tested whether a large herbivore is able to select nutrient-rich patches at multiple scales. However, within a specific spatial scale, patches are not simply distributed uniformly but vary in size, number and local nutrient concentration (Bailey & Provenza 2008). Therefore, at each scale, the resource value per grain should be expressed as the total nutrient load, for example, areas with large nutrientrich patches, with many small nutrient-rich patches or with patches with very high local nutrient concentrations. In our study, spatial scales ranged from 2500 m2, where a herbivore selects between areas with different total nutrient loads, to selection between individual neighbouring plants of the same size and species (Fig. 1). Studying the mechanisms behind patch use is often confounded by the effects of predation risk. For example, the marginal value theorem (MVT) predicts that foragers will depart from a single patch when their instantaneous intake rate drops below the average rate of intake attainable in all patches (Charnov 1976). However, increased predation risk should affect MVT in that time in the patch should decrease (Brown 1999). Therefore, as a study animal, we chose the largest terrestrial herbivore, the African elephant (Loxodonta africana), for which predation risk in our study area is negligible. As elephant are able to select vegetation growing on termite mounds at very small scales (Holdo & McDowell 2004), we predicted that elephant will be able to select nutrient-rich areas at scales ranging from 2500 m2 down to the individual plant scale. Many studies show that savannas are either limited by N or P or co-limited by both (du Toit, Louw & Malan 1940; Weir 1969; Ludwig et al. 2001; Snyman 2002; Augustine, McNaughton & Frank 2003; Cech et al. 2008). Because nonruminants such as the elephant cannot make use of microbial protein and have high food passage rates, they can be expected to incur higher losses of N in the faeces (Foose 1982). Selection for N and P by elephant has been described by Jachmann & Bell (1985), and from diet studies on elephant at our study site in South Africa (Pretorius 2009) and else- where (Woolley et al. 2009), we know that elephant maximize N and P intake depending on the time of the year. Therefore, we expect elephant to select patches and plants of the same species depending on their N and P contents. The impact African elephant have on trees has been a topic of great controversy (Wiseman, Page & O’Connor 2004; De Beer et al. 2006; Lawes & Chapman 2006; O’Connor, Goodman & Clegg 2007; Scholes & Mennell 2008; Chafota & Owen-Smith 2009). Elephant have been implicated as one of the key factors maintaining low tree– grass ratios in savannas (Van de Koppel & Prins 1998). However, few studies have investigated the causal mechanisms of elephant impact, especially the interactions with nutrient availability and distribution (Skarpe et al. 2004). Adverse impact on trees where bark and roots are consumed is not unique to elephant and has been described in voles (Microtus spp.) (e.g. Sullivan et al. 2004), Sika deer (Cervus nippon) (e.g. Yokoyama et al. 2001) or red deer (Cervus elaphus) (e.g. Verheyden et al. 2006). However, the consumption of bark and roots has mostly been attributed to the lack of good quality alternative food sources, especially during winter, rather than to the high nutritional quality of these plant parts (Servello 1984; Bucyanayandi et al. 1992; Verheyden et al. 2006). Similarly, elephant impact on trees is especially prevalent during the dry season when elephant switch from a diet dominated by grass to browse because grass quality decreases to below the animal’s maintenance requirements during this time (Barnes 1982; Beekman & Prins 1989; Pretorius 2009) and an increased consumption of woody material indicates nutritional stress (O’Connor, Goodman & Clegg 2007). We predict that elephant will have a larger impact on trees that represent low-quality food through, for example, increased utilization of roots, whereas impact on trees growing in nutrient-rich patches will be less, with leaves being utilized more. To test these predictions, we experimentally manipulated the spatial distribution of nutrients in a homogenous stand of plant species dominated by Colophospermum mopane trees. Materials and methods In December 2004, we fertilized thirty 50 · 50 m plots in a 35-ha area in the Timbavati Private Nature Reserve, South Africa (24̊14¢11¢¢S; 31̊22¢32¢¢E, at 380 m elevation), with NPK (3 : 2 : 1) fertilizer. The area was re-fertilized 2 years later in an identical fashion. The experimental study site was dominated by a homogeneous layer of C. mopane shrub veld on granitic soils. The herbaceous layer was dominated by Urochloa mosambicensis and Bothriochloa spp. (for a full description of the vegetation, see Van der Waal 2010). The warm, rainy season stretched from October to March, and during the study period, the mean annual rainfall was 420 mm. Data were collected from October 2005 to March 2008. Surface water was not constraining plot use, as it was available 0Æ5 km to the north and south of the closest experimental plot. The experimental layout followed a randomized block design, including three replicates of nine different treatments and nine controls. The fertilizer was applied at each 50 · 50 m plot in one of the three different spatial configurations: one large 50 · 50 m patch, five 10 · 10 m 2011 The Authors. Journal of Animal Ecology 2011 British Ecological Society, Journal of Animal Ecology, 80, 875–883 Soil nutrients & elephant tree utilization 877 Fig. 1. A schematic representation of the study set-up illustrating the spatial scales at which the experiment was conducted. To each spatial configuration, 50 · 50 m control plots were allocated, and at scales smaller than 2500 m2, unfertilized areas neighbouring fertilized patches were used as controls (white blocks and white areas in blocks are representative of unfertilized plots and patches). patches or 25 patches of 2 · 2 m, and using one of the three different nitrogen concentrations (30Æ0, 6Æ0 and 1Æ2 g m)2). One 50 · 50 m control plot was allotted to each of the three spatial configurations, and this was replicated three times (Fig. 1). The lightest and heaviest fertilizer concentrations were excluded, as these were expected to generate too faint a signal or a toxic effect. The result was seven combinations that also differed in their total nutrient load per 50 · 50 m plot (0Æ6, 3, 15 kg N). Each replicate of these seven treatments, together with their control plots made up a block, and the blocks represented different but contiguous topographical positions in a gently undulating landscape. Large herbivore species commonly found in the area include elephant (L. africana) at a mean density of 0Æ3 elephants km)2, buffalo (Syncerus caffer), impala (Aepyceros melampus), warthog (Phacochoerus aethiopicus), duiker (Sylvicapra grimmia), zebra (Equus burchellii) and steenbok (Raphicerus campestris). C. mopane was the main food plant for elephant in the area (Kos et al. 2007; Pretorius 2009). Data collection on the type of nutrients selected by elephant and the scale of nutrient selection consisted of annual measurements during the wet seasons from 2006 to 2008. We measured leaf nutrient content and accumulated signs of elephant utilization on 600 marked C. mopane trees within the field fertilization experiment and made direct observations of elephant selecting between two C. mopane trees of the same size within C. mopane-dominated areas surrounding the experiment. Data on elephant use of trees, and the occurrence and utilization of coppiced C. mopane trees in and outside fertilized 2011 The Authors. Journal of Animal Ecology 2011 British Ecological Society, Journal of Animal Ecology, 80, 875–883 878 Y. Pretorius et al. patches, were obtained from an elephant utilization assessment on all trees larger than 1Æ5 m within the experimental plots at the end of the study. ANNUAL MOPANE TREE MEASUREMENTS Tree responses were monitored using 600 tagged C. mopane trees. In the controls and in whole-plot fertilizer treatments, C. mopane trees (> 1 m height) closest to twenty randomly generated points were selected. In heterogeneous treatments (patch size either 10 · 10 or 2 · 2 m), ten trees with stems within 2-m distance of fertilized patches were randomly selected, together with ten trees in the unfertilized plot area (> 2-m distance from fertilized patches). Five fully expanded leaf samples were randomly collected from the canopies of marked C. mopane trees during the 2006, 2007 and 2008 growing seasons. In the homogeneous treatments, two pooled samples were analysed per plot while in the heterogeneous treatments, samples were pooled for leaves collected from the plants inside and outside of the fertilized patches. This resulted in 60 samples per year. Prior to milling (through a 1-mm sieve), C. mopane leaves were dried to constant weight at 60 C and weighed. All marked C. mopane trees were assessed for elephant impact (percentage canopy volume reduction) using the scale of Anderson & Walker (1974). For 2006, the height and diameter of marked trees were calculated from digital photographs. At the end of the experiment in 2008, a selection of trees were remeasured to determine tree height changes in relation to visual elephant impact scores. For analysis, as ten trees were measured per patch, elephant utilization was also expressed as the proportion of trees utilized per patch. DIRECT OBSERVATIONS All elephant observations made in areas dominated by C. mopane surrounding the experiment were of randomly encountered solitary adult bulls or bulls in small bachelor groups. All data were collected during the wet season of 2006 and were spatially and temporally independent as only one paired sample was collected at each sighting of a particular elephant, or group of elephant, on a particular day. Each elephant was located using the extensive road network on the reserve and observed from a vehicle for between 5 and 10 min (or until the animal moved out of site) using binoculars. For an observation to be accepted, an elephant had to be foraging in a homogeneous area of C. mopane trees. Only once an elephant moved from the tree, it was first observed feeding on to feed on the next tree, was the next tree used for measurement. Measured trees were all C. mopane, between 3 and 6 m, and the elephant had to have taken at least five bites from the canopy before the tree was sampled. For each of the measured trees, a discarded paired control tree was selected. This discarded tree had to be a C. mopane of similar height and canopy size, which the elephant had walked past (within 2 m) and which was not selected while moving towards tree it fed on. About 50 g of leaf material was collected from various branches on each tree at the same height range at which the elephant fed from. Leaves from both the utilized tree and the discarded tree were stored in separate paper bags, dried at 60 C for 24 h and ground through a 1-mm sieve for later analysis. CHEMICAL ANALYSIS OF LEAF SAMPLES All dried leaf samples were analysed for dry matter, ash, neutral detergent fibre, calcium, sodium, potassium, magnesium, nitrogen and phosphorus contents at the laboratory of the Resource Ecology Group, Wageningen University (the Netherlands). Total Ca, Na, Mg, K, N and P contents were measured with a Skalar San-plus auto analyzer, after destruction with a mixture of H2SO4, selenium and salicylic acid. (Novozamsky et al. 1983). Neutral detergent fibre of dry leaf was determined using the ANKOM filter bag procedure (ANKOM Technology, Macedon, NY, USA) with omission of the sodium sulphite and the heat-resistant a-amylase. A neutral detergent solution was prepared following Goering & van Soest (1970). Condensed tannins were measured with the proanthocyanidin method and total polyphenol with the Folin–Ciocalteu method (Waterman & Mole 1994). ELEPHANT UTILIZATION ASSESSMENT During March 2008, we recorded the position within a 2 · 2 m grid, of all trees taller than 1Æ5 m within 50 · 50 m plots containing five fertilized patches of 10 · 10 m. Fertilized patches were marked at the beginning of the study with iron stakes, to indicate whether a tree occurred within a fertilized patch or not. We recorded tree species, treatment within the 50 · 50 m plot, tree height and canopy width, coppicing as a result of visible previous impact by elephant, percentage of each type of utilization, total percentage impact and estimation of the age of impact. The area on a branch impacted on by elephant becomes grey after a year through a rainy season (Ben-Shahar & Macdonald 2002). Therefore, the age of impact on a tree was classified as fresh when scars had not yet turned grey. The type of utilization was categorized into five classes: leaf stripping, bark stripping, impact on branches, uprooting and breaking of the main stem. The percentage of leaf stripping, bark stripping and impact on branches was estimated in relation to the availability of the plant part across the entire tree. Uprooting and breaking of the main stem were recorded as 100% impact (i.e. the tree could die). DATA ANALYSIS All data were first tested for normality using the Kolmogorov–Smirnov test, and where proportions were used, data were first arcsine transformed. For the fertilization experiment, block effects were assumed to be negligible as each of the three blocks was a replication containing a set of all treatments and controls. Two types of analyses were used to test elephant selection for specific nutrients. First, we used a one-way anova to test whether N, P, K, Mg, Ca and Na were higher in leaves from trees in fertilized patches compared to unfertilized patches and whether elephant utilized these trees in fertilized patches more than unfertilized patches. Second, we conducted a multiple linear regression analysis to test for effects of the concentrations of different nutrients as independent variables on the proportion of trees utilized by elephant per patch as dependent variable. Annual elephant utilization measurements on marked C. mopane trees were used in an anova to test whether elephant were able to select fertilized patches at a 50 · 50, 10 · 10 and 2 · 2 m scale. Because of the set-up of the experiment, and to compare the relevant unfertilized controls with neighbouring fertilized areas, a separate model was used for each of the three scales. For analysis at the plant scale between neighbouring trees, paired sample t-tests were used to compare nutrient levels in leaves between eaten and uneaten trees. Because leaf stripping (%) and signs of elephant utilization on branches (%) were not normally distributed, the nonparametric Mann–Whitney test was used to detect differences between fertilized and unfertilized patches. Because the number of coppicing C. mopane trees and C. mopane trees killed by elephant are rare events resulting in many zero values in the data, a generalized 2011 The Authors. Journal of Animal Ecology 2011 British Ecological Society, Journal of Animal Ecology, 80, 875–883 Soil nutrients & elephant tree utilization 879 Table 1. Chemical analysis of Colophospermum mopane leaves inside and outside patches fertilized with NPK at three different scales NPK fertilizer Scale (m2) N (%) P (%) Ca (%) K (%) Mg (%) Na (%) Tannin (mg g)1) Absent Present Absent Present Absent Present 4 4 100 100 2500 2500 1Æ92 2Æ04 1Æ95 2Æ22 1Æ93 2Æ11 0Æ12 0Æ13 0Æ12 0Æ14 0Æ12 0Æ13 1Æ49 1Æ51 1Æ36 1Æ25 1Æ55 1Æ51 1Æ02 1Æ08 1Æ03 1Æ05 1Æ06 1Æ05 0Æ22 0Æ23 0Æ22 0Æ23 0Æ23 0Æ22 0Æ00 0Æ00 0Æ01 0Æ01 0Æ01 0Æ00 1084Æ55 1064Æ12 1109Æ65 1021Æ57 1102Æ23 1030Æ03 linear model with a negative binomial error distribution was used to test for differences of these variables inside and outside fertilized patches (White 1996). Results The application of the NPK fertilizer significantly changed the chemical composition of leaves of mopane trees at the experimental site. Both nitrogen (F1,58 = 35Æ566, P < 0Æ001) and phosphorus (F1,58 = 28Æ226, P < 0Æ001) contents of tree leaves were significantly higher in fertilized patches compared to unfertilized patches, while the tannin content was significantly lower in fertilized patches (F1,58 = 13Æ628, P < 0Æ001). However, there was no significant difference for K (even though the fertilizer contained K) or for Ca, Mg, Na or NDF, concentration in leaves (Table 1). The proportion of mopane trees utilized by elephant per patch was significantly higher in fertilized patches vs. unfertilized patches, irrespective of patch size, (F1,58 = 10Æ95, P = 0Æ002). The proportion of trees utilized by elephant per patch increased significantly with increasing nitrogen concentration in leaves, but was not significantly affected by other nutrients (linear regression model: R2 = 0Æ462, F7,52 = 6Æ368, P < 0Æ001) (Fig. 2). Irrespective of patch size and local nutrient concentration, elephant were able to select plots with the highest nutrient loads at a 2500-m2 scale (F3,56 = 4Æ564, P = 0Æ006) (Fig. 3). Fig. 2. The relationship between the proportion Colophospermum mopane trees utilized by elephant per patch and the percentage nitrogen (on a dry matter basis) in tree leaves (n = 60). Fig. 3. Proportion of Colophospermum mopane trees used by elephant in unfertilized patches and fertilized patches with increasing nutrient loads at 100-m2 scale (error bars show 95% confidence interval of the mean, and different letters in the figures denote significant difference). Although the nitrogen content of the leaves was significantly higher in fertilized patches than unfertilized patches at the 100-m2 scale (F1,16 = 18Æ232, P = 0Æ001), this was not the case for the 4-m2 scale (F1,10 = 3Æ727, P = 0Æ082). Hence, the utilization of mopane trees was significantly higher in fertilized patches than in unfertilized patches at the 100-m2 scale (F1,16 = 4Æ733, P = 0Æ045) but not at the 4-m2 scale (F1,10 = 0Æ024, P = 0Æ88). At plant scale, N and P contents from leaves of eaten trees were not significantly different from leaves from neighbouring trees of the same species that were not eaten (N: t14 = )1Æ242, P = 0Æ235, P: t14 = )1Æ61, P = 0Æ13). For the elephant utilization assessment at the 100-m2 scale, utilization of tree leaves via leaf stripping occurred more inside fertilized patches than outside (Z = )4Æ694, n = 786, P < 0Æ001), whereas total impact on branches did not differ between patches (Z = )0Æ946, n = 786, P = 0Æ344) (Fig. 4a,b). Tree killing, through uprooting of trees or breaking main tree trunks, occurred more outside fertilized patches than inside (Wald chi-square = 3Æ818, d.f. = 1, P = 0Æ05) (Fig. 4c). Tree coppicing tended to be more frequent inside fertilized patches than in the surrounding control areas, but this was not significant (Wald chi-square = 3Æ181, d.f. = 1, 2011 The Authors. Journal of Animal Ecology 2011 British Ecological Society, Journal of Animal Ecology, 80, 875–883 880 Y. Pretorius et al. (a) (a) (b) (b) Fig. 5. Differences in the proportion of coppicing trees in the presence or absence of fertilizer (a) and the proportion signs of fresh leaf stripping on coppiced and noncoppiced trees (b) (error bars show 95% confidence interval of the mean). (c) Discussion FORAGING AT DIFFERENT SPATIAL SCALES Fig. 4. Differences in elephant utilization at a 100-m2 scale between fertilized (n = 176) and unfertilized (n = 610) patches, distinguishing different use-categories: (a) leaf stripping, (b) impact on branches and (c) killing of trees, respectively (error bars show 95% confidence interval of the mean). P = 0Æ075). As expected, coppiced trees had significantly more signs of fresh leaf stripping than noncoppiced trees (Wald chi-square = 5Æ398, d.f. = 1, P = 0Æ02) (Fig. 5). Two decades ago, scales as the application of traditional optimal foraging theory were problematic, Senft et al. (1987) introduced a hierarchy into herbivore foraging theory to integrate foraging decisions at different spatiotemporal. Since then, many studies have illustrated how herbivores of various body sizes are able to select nutrient-rich patches at a range of spatial scales (WallisDeVries, Laca & Demment 1999; Durant, Fritz & Duncan 2004; Cromsigt & Olff 2006). We also found that elephant are able to select nutrient-rich patches at scales ranging from 2500 to 100 m2. Contradictory to the theoretical predictions of Ritchie & Olff (1999) that larger species will not be able to detect food patches at fine scales, we found that even the largest terrestrial mammal is able to select nutrient-rich plant parts at fine scales within a homogeneous tree species stand. To our knowledge, this has never been shown for elephant. At a regional scale (100 km), in a study conducted in C. mopane woodlands in Botswana, no relationship could be found between elephant impact on C. mopane trees and leaf N con- 2011 The Authors. Journal of Animal Ecology 2011 British Ecological Society, Journal of Animal Ecology, 80, 875–883 Soil nutrients & elephant tree utilization 881 tent even though soil N levels and leaf N were related, and leaf N differed significantly between sampling sites (BenShahar & Macdonald 2002). However, this latter study was conducted at larger scales than our study, and more importantly, no analysis was carried out on the relationship between the type of elephant impact and the nutrient status of the plants. Soil and leaf samples were collected in the dry season when elephant are more limited by energy (Pretorius 2009) and thus less selective of other nutrients such as N and P. NITROGEN SELECTION Selection of plant species high in N has been described for elephant within their daily foraging ranges (Jachmann & Bell 1985), for barnacle geese (Branta leucopsis) in fertilizer experiments at 250-m2 scale (Ydenberg & Prins 1981), and at even smaller scales, black colobus monkeys (Colobus satanas) in the rain forests of Cameroon have been found to select plant parts in relation to their nitrogen content (McKey et al. 1981). Some authors reason that switching between plant species or plant parts is a behavioural adaptation to cope with declines in N content (Mattson 1980) and that elephant in particular switch between plant species and plant parts to sequester the greatest amount of digestible protein per unit time (O’Connor, Goodman & Clegg 2007). For example, cambium in the bark and roots of plants contains relatively high levels of N (Mattson 1980) and Hiscocks (1999) found that cambium samples from bark and root that elephant utilized were higher in N than cambium from unutilized trees. Therefore, uprooting of trees in nutrient-poor areas, such as was found in our study, could be a result of low leaf N levels, which encourages elephant to utilize other plant parts with higher N content. CONTRIBUTION TO OPTIMAL FORAGING THEORY Irrefutably, the contribution of this study to the current body of knowledge of optimal foraging theory is the concept of patch selection by a herbivore based on the total nutrient load within the grain size of a specific spatial scale. In situ patch size does not simply increase uniformly, but patches vary in size and can be scattered or aggregated depending on the scale of observation. Moreover, the local nutrient concentration between patches can also vary widely (Fryxell 2008). Thus, when studying patch selection by a herbivore at a particular scale, the available nutrients must be expressed as total nutrient load per grain size, which is a product of the number of patches, local nutrient concentration of patches and patch sizes. We found that at a scale of 2500 m2, elephant are able to select the patch combinations resulting in the highest total nutrient loads, whether it was many small patches of high local nutrient concentration per 2500 m2 or few large patches of lower local nutrient concentration per 2500 m2. These findings may not only have implications for the way patch selection by herbivores is interpreted at different scales but also offer a possible explanation why some studies find, contrary to the MVT, that herbivores select suboptimal patches or leave patches when there is still much food left (Schaefer & Messier 1995). This dependence of MVT predictions on the scale at which a herbivore perceives patches, which is very difficult to determine, has also been noted by Kotliar & Wiens (1990) and Searle, Hobbs & Shipley (2005). IMPACT OF KEYSTONE HERBIVORES Furthermore, our study illustrates how elephant can possibly facilitate other species by maintaining tree ‘islands of fertility’(Ludwig, de Kroon & Prins 2008; Treydte, Heitkönig & Ludwig 2009) through the reuse of trees and dung deposition, which may potentially initiate positive feedbacks into the abiotic environment (van der Waal 2010). Jachmann & Bell (1985) and subsequently Smallie & O’Connor (2000) hypothesized that, by continuously hedging C. mopane trees, elephants can maintain a supply of preferred coppice growth through time. The increased proportion of coppiced mopane trees because of previous breaking of branches by elephant inside fertilized plots and the higher occurrence of leaf stripping on these trees indicate that the fertilizer in our experiment might have triggered a positive feedback response that may persist through time. In a fenced off clipping experiment conducted in northern Botswana, elephant broke into the experimental site 3 years after initiation. The assessment of elephant impact 1 month after the break-in revealed higher utilization of trees subjected to simulated browsing 3 years before compared to control trees (Makhabu & Skarpe 2006). Often, the problem with studying keystone species under natural conditions is that their effects on other organisms only become apparent once they themselves have been removed. This was demonstrated by the removal of impala by disease followed by a subsequent increase in woodland cover in East Africa (Prins & van der Jeugd 1993). The advantage of an experimental approach such as ours is that the mechanisms behind these effects become clearer without the removal of the keystone species. Both soil fertility and elephant have been identified as major determinants of tree– grass ratios in savannas (Scholes 1990; Fritz et al. 2002; O’Connor, Goodman & Clegg 2007; Sankaran, Ratnam & Hanan 2008). However, the interaction between these two factors, and more specifically how soil nutrient status affects the way in which elephant utilize their environments, was until recently unclear (Olff, Ritchie & Prins 2002; Skarpe et al. 2004). This study therefore contributed to a better understanding of this mechanism, but more importantly the results show how future studies should interpret patch selection by herbivores at different scales. By concentrating limiting resources, patchiness may play a critical role in maintaining ecosystem productivity (Aguiar & Sala 1999) and herbivory has been shown to create spatial heterogeneity and induce vegetation patterning through the process of self-facilitation (Groen 2007). Recent research using hyperspectral remote sensing in Kruger National Park (Skidmore et al. 2009) and in saltmarshes (Schmidt & Skid- 2011 The Authors. Journal of Animal Ecology 2011 British Ecological Society, Journal of Animal Ecology, 80, 875–883 882 Y. Pretorius et al. more 2001; Schmidt et al.2004) shows that foliar quality of species within patches can be mapped and discriminated. In other words, it is possible, at varying scales, to understand foliar quality across space. Linking wildlife and livestock observations to these maps may allow us to understand animal behaviour and survival. Conclusion We conclude that more emphasis should be placed on the nutrient status of soils when trying to understand the impact elephant have on trees (Fritz et al. 2002), as changes in elephant foraging behaviour can be used as indicators of changes in the availability of nutrients. Acknowledgements We thank managers and owners of the Associated Private Nature Reserves for allowing us access to the reserve to collect our data, all the co-workers on the TEMBO programme for their support and the field staff for assisting us with data collection in South Africa. We appreciate financial support from the WOTRO fund, Shell-SA and Omnia for helping us realize this study. References Aguiar, M.R. & Sala, O.E. (1999) Patch structure, dynamics and implications for the functioning of arid ecosystems. Tree, 14, 273–277. Anderson, G.D. & Walker, B.H. (1974) Vegetation composition and elephant damage in Sengwa Wildlife Research Area, Rhodesia. Journal of the South African Wildlife Management Association, 4, 1–14. Augustine, D.J., McNaughton, S.J. & Frank, D.A. (2003) Feedbacks between soil nutrients and large herbivores in a managed savanna ecosystem. Ecological Applications, 13, 1325–1337. Bailey, D.W. & Provenza, F.D. (2008) Mechanisms determining large-herbivore distribution. Resource Ecology: Spatial and Temporal Dynamics of Foraging (eds H.H.T. Prins & F. van Langevelde), pp. 7–28. Springer, the Netherlands. Bailey, D.W., Gross, J.E., Laca, E.A., Rittenhouse, L.R., Coughenour, M.B., Swift, D.M. & Sims, P.L. (1996) Mechanisms that result in large herbivore grazing distribution patterns. Journal of Range Management, 49, 386–400. Ball, J.P., Danell, K. & Sunesson, P. (2000) Response of a herbivore community to increased food quality and quantity: an experiment with nitrogen fertilizer in a boreal forest. Journal of Applied Ecology, 37, 247–255. Barnes, R.F.W. (1982) Elephant feeding behaviour in Ruaha National Park, Tanzania. African Journal of Ecology, 20, 123–136. Beekman, J.H. & Prins, H.H.T. (1989) Feeding strategies of sedentary large herbivores in East Africa, with emphasis on the African buffalo, Syncerus caffer. African Journal of Ecology, 27, 129–147. Ben-Shahar, R. & Macdonald, D.W. (2002) The role of soil factors and leaf protein in the utilization of mopane plants by elephants in northern Botswana. BMC Ecology, 2, 3. Brown, J.S. (1999) Vigilance, patch use and habitat selection: foraging under predation risk. Evolutionary Ecology Research, 1, 49–71. Bucyanayandi, J.D., Bergeron, J.M., Soucie, J., Thomas, D.W. & Jean, Y. (1992) Differences in nutritional quality between herbaceous plants and bark of conifers as winter food for the vole Microtus pennsylvanicus. Journal of Applied Ecology, 29, 371–377. Cech, P.G., Kuster, T., Edwards, P.J. & Venterink, H.O. (2008) Effects of herbivory, fire and N2-fixation on nutrient limitation in a humid African savanna. Ecosystems, 11, 991–1004. Chafota, J. & Owen-Smith, N. (2009) Episodic severe damage to canopy trees by elephants: interactions with fire, frost and rain. Journal of Tropical Ecology, 25, 341–345. Charnov, E.L. (1976) Optimal foraging, the marginal value theorem. Theoretical Population Biology, 9, 129–136. Cromsigt, J.P.G.M. & Olff, H. (2006) Resource partitioning among savanna grazers mediated by local heterogeneity: an experimental approach. Ecology, 87, 1532–1541. De Beer, Y., Kilian, W., Versfeld, W. & van Aarde, R.J. (2006) Elephants and low rainfall alter woody vegetation in Etosha National Park, Namibia. Journal of Arid Environments, 64, 412–421. Durant, D., Fritz, H. & Duncan, P. (2004) Feeding patch selection by herbivorous Anatidae: the influence of body size, and of plant quantity and quality. Journal of Avian Biology, 35, 144–152. Foose, T.J. (1982) Trophic strategies of ruminant versus nonruminant ungulates. PhD thesis, University of Chicago, Chicago, United States. Fritz, H., Duncan, P., Gordon, I.J. & Illius, A.W. (2002) Megaherbivores influence trophic guilds structure in African ungulate communities. Oecologia, 131, 620–625. Fryxell, J.M. (2008) Predictive modelling of patch use by terrestrial herbivores. Resource Ecology: Spatial and Temporal Dynamics of Foraging (eds H.H.T. Prins & F. van Langevelde), pp. 105–123. Springer, the Netherlands. Goering, H.K. & van Soest, P.J. (1970) Forage Fiber Analyses (Apparatus, Reagents, Procedures and Some Applications). Agricultural Handbook 379. United States Department of Agriculture, Washington DC. Groen, T.A. 2007. Spatial matters: how spatial processes and patterns affect savanna dynamics. PhD Thesis, Wageningen University, the Netherlands. Hiscocks, K. (1999) The impact of an increasing elephant population on the woody vegetation in southern Sabi Sand Wildtuin, South Africa. Koedoe, 42, 47–55. Holdo, R.M. & McDowell, L.R. (2004) Termite mounds as nutriënt-rich food patches for elephants. Biotropica, 36, 231–239. Jachmann, H. & Bell, R.H.V. (1985) Utilization by elephants of the Brachystegia woodlands of the Kasungu National Park, Malawi. African Journal of Ecology, 23, 245–258. Kos, M., Hoetmer, A.J., Pretorius, Y. & de Boer, W.F. (2007) Seasonal diet switch in elephant and impala in a Mopane woodland. 5th Annual Science Network Meeting. South African National Parks, Skukuza. pp. 44. Kotliar, N.B. & Wiens, J.A.. (1990) Multiple scales of patchiness and patch structure: a hierarchical framework for the study of heterogeneity. Oikos, 59, 253–260. Lawes, M.J. & Chapman, C.A. (2006) Does the herb Acanthus pubescens and ⁄ or elephants suppress tree regeneration in disturbed Afrotropical forest? Forest Ecology and Management, 221, 278–284. Ludwig, F., de Kroon, H. & Prins, H.H.T. (2008) Impacts of savanna trees on forage quality for a large African herbivore. Oecologia, 155, 487–496. Ludwig, F., de Kroon, H., Prins, H.H.T. & Berendse, F. (2001) Effects of nutrients and shade on tree-grass interactions in an East African savanna. Journal of Vegetation Science, 12, 579–588. Makhabu, S.W. & Skarpe, C. (2006) Rebrowsing by elephants three years after simulated browsing on five woody plant species in northern Botswana. South African Journal of Wildlife Research, 36, 99–102. Mattson, W.J. (1980) Herbivory in relation to plant nitrogen content. Annual Review of Ecology and Systematics, 11, 119–161. McKey, D.B., Gartlan, J.S., Waterman, P.G., Gillian, F.L.S. & Choo, M. (1981) Food selection by black colobus monkeys (Colobus satanas) in relation to plant chemistry. Biological Journal of the Linnean Society, 16, 115–146. Novozamsky, I., Houba, V.J.G., Vaneck, R. & Vanvark, W. (1983) A novel digestion technique for multi-element plant analysis. Communications in Soil Science and Plant Analysis, 14, 239–248. O’Connor, T.G., Goodman, P.S. & Clegg, B. (2007) A functional hypothesis of the threat of local extirpation of woody plant species by elephant in Africa. Biological Conservation, 136, 329–345. Olff, H., Ritchie, M.E. & Prins, H.H.T. (2002) Global environmental controls of diversity in large herbivores. Nature, 415, 901–904. Pretorius, Y. (2009) Satisfying giant appetites: mechanisms of small scale foraging by large African herbivores. PhD thesis, Wageningen University, Wageningen, the Netherlands. Prins, H.H.T. & van der Jeugd, H.P. (1993) Herbivore population crashes and woodland structure in East Africa. Journal of Ecology, 81, 305–314. Prins, H.H.T. & van Langevelde, F. (2008) Assembling a diet from different places. Resource ecology: spatial and temporal dynamics of foraging (eds H.H.T. Prins & F. van Langevelde), pp. 129–155. Springer, Netherlands. Ritchie, M.E. & Olff, H. (1999) Spatial scaling laws yield a synthetic theory of biodiversity. Nature, 400, 557–560. Sankaran, M., Ratnam, J. & Hanan, N. (2008) Woody cover in African savannas: the role of resources, fire and herbivory. Global Ecology and Biogeography, 17, 236–245. Schaefer, J.A. & Messier, F. (1995) Habitat selection as a hierarchy: the spatial scales of winter foraging by muskoxen. Ecography, 18, 333–344. 2011 The Authors. Journal of Animal Ecology 2011 British Ecological Society, Journal of Animal Ecology, 80, 875–883 Soil nutrients & elephant tree utilization 883 Schmidt, K. & Skidmore, A.K. (2001) Exploring spectral discrimination of grass species 809 in African rangelands. International Journal of Remote Sensing, 22, 3241–3434. Schmidt, K.S., Skidmore, A.K., Kloosterman, E.H., Van Oosten, H., Kumar, L.O. & Janssen, J.A.M. (2004) Mapping coastal vegetation using an expert system and hyperspectral imagery. Photogrammetric Engineering and Remote Sensing, 70, 703–715. Scholes, R.J. (1990) The influence of soil fertility on the ecology of southern African dry savannas. Journal of Biogeography, 17, 415–419. Scholes, R.J. & Mennell, K.G. (2008) Elephant Management: A Scientific Assessment for South Africa. Wits University Press, Johannesburg, South Africa. Searle, K.R., Hobbs, N.T. & Shipley, L.A. (2005) Should I stay or should I go? Patch departure decisions by herbivores at multiple scales. Oikos, 111, 414– 424. Senft, R.L., Coughenour, M.B., Bailey, D.W., Rittenhouse, L.R., Sala, O.E. & Switft, D.M. (1987) Large herbivore foraging and ecological hierarchies. BioScience, 37, 789–799. Servello, F. (1984) Pine vole diet quality in relation to apple tree root damage. Journal of Wildlife Management, 48, 450–455. Skarpe, C., Aarrestad, P.A., Andreassen, H.P., Dhillion, S.S., Dimakatso, T., du Toit, J.T., Halley, D.J., Hytteborn, H., Makhabu, S., Mari, M., Marokane, W., Masunga, G., Modise, D., Moe, S.R., Mojaphoko, R., Mosugelo, D., Motsumi, S., Neo-Mahupeleng, G., Ramotadima, M., Rutina, L., Sechele, L., Sejoe, T.B., Stokke, S., Swenson, J.E., Taolo, C., Vandewalle, M. & Wegge, P. (2004) The return of the giants: ecological effects of an increasing elephant population. Ambio, 33, 276–282. Skidmore, A.K., Ferweda, J.G., Mutanga, O., van Wieren, S.E., Peel, M., Grant, R.C., Prins, H.H.T., Balcik, F.B. & Venus, V. (2009) Forage quality of savannas: simultaneously mapping foliar protein and polyphenols for trees and grass using hyperspectral imagery. Remote sensing of environment, 114, 64–72. Smallie, J.J. & O’Connor, T.G. (2000) Elephant utilization of Colophospermum mopane: possible benefits of hedging. African Journal of Ecology, 38, 352– 359. Snyman, H.A. (2002) Short-term response of rangeland botanical composition and productivity to fertilization (N and P) in a semi-arid climate of South Africa. Journal of Arid Environments, 50, 167–183. Spalinger, D.E. & Hobbs, N.T. (1992) Mechanisms of Foraging in Mammalian Herbivores: New Models of Functional Response. The American Naturalist, 140, 325–348. Sullivan, T.P., Sullivan, D.S., Reid, D.G. & Leung, M.C. (2004) Weasels, voles and trees: influence of mustelid semiochemicals on vole populations and feeding damage. Ecological Applications, 14, 999–1015. du Toit, P.J., Louw, J.G. & Malan, A.I. (1940) A study of the mineral content and feeding value of natural pastures in the Union of South Africa. Onderstepoort Journal of Veterinary Science and Animal Industry, 14, 123–127. Treydte, A.C., Heitkonig, I.M.A. & Ludwig, F. (2009) Modelling ungulate dependence on higher quality forage under large trees in African savannahs. Basic and Applied Ecology, 10, 161–169. Van de Koppel, J. & Prins, H.H.T. (1998) The importance of herbivore interactions for the dynamics of African savanna woodlands: an hypothesis. Journal of Tropical Ecology, 14, 565–576. Van der Waal, C. 2010. Nutrients in an African Savanna: the consequences of supply heterogeneity for plants and animals. PhD thesis, Wageningen University, the Netherlands. Verheyden, H., Ballon, P., Bernard, V. & Saint-Andrieux, C. (2006) Variations in bark-stripping by red deer Cervus elaphus across Europe. Mammal Review, 36, 217–234. WallisDeVries, M.F., Laca, E.A. & Demment, M.W. (1999) The importance of scale of patchiness for selectivity in grazing herbivores. Oecologia, 121, 355– 363. Waterman, P.G. & Mole, S. (1994) Analysis of Phenolic Plant Metabolites. Blackwell Scientific, Oxford, UK. Weir, J.S. (1969) Chemical properties and occurrence on Kalahari sand of salt licks created by elephants. Journal of Zoology, 158, 293–310. White, G.C. (1996) Analysis of frequency count data using the binomial distribution. Ecology, 77, 2549–2557. Wiseman, R., Page, B.R. & O’Connor, T.G. (2004) Woody vegetation change in response to browsing in Ithala Game Reserve, South Africa. South African Journal of Wildlife Research, 34, 25–37. Woolley, L.A., Millspaugh, J.J., Woods, R.J., Janse van Rensburg, S., Page, B.R. & Slotow, R. (2009) Intraspecific strategic responses of African elephants to temporal variation in forage quality. Journal of Wildlife Management, 73, 827–835. Ydenberg, R.C. & Prins, H.H.T. (1981) Spring grazing and the manipulation of food quality by Barnacle geese. Journal of Applied Ecology, 18, 443–453. Yokoyama, S., Maeji, I., Ueda, T., Ando, M. & Shibata, E. (2001) Impact of bark stripping by sika deer, Cervus nippon, on subalpine coniferous forests in central Japan. Forest Ecology and Management, 140, 93–99. Received 22 August 2010; accepted 25 January 2011 Handling Editor: Atle Mysterud 2011 The Authors. Journal of Animal Ecology 2011 British Ecological Society, Journal of Animal Ecology, 80, 875–883