IV.4 Medio socioeconómico
IV.4.8 Vías y Medios de comunicación existentes, disponibilidad de servicios
Feeding behaviour of large herbivores is notoriously difficult to study because, firstly, it is a complex process that is influenced by numerous factors and, secondly, many of its determinants vary spatially and temporally making them difficult to quantify over large areas and for long periods. For example, diet selection is influenced by seasonal and inter-annual variation in the phenology of plants, which results in a constantly changing array of available food types, each with its own chemical composition and set of constraints to harvesting, chewing and digestion. Choice of what and where to eat is further complicated by sex and age related differences in the ability to harvest, chew and digest food, and by non-dietary factors such as distance from surface water that constrain use of habitats. Study of foraging behaviour is especially difficult under natural conditions where the researcher has no control over the target animals or the environment in which they live. This is particularly true for elephants because they eat a wide variety of food types (10 were identified in this study) and forage over large areas, which exacerbates the already difficult task of quantifying the choice set.
Given the complexity of the foraging process, it is understandable why most research into elephant feeding behaviour has been limited to descriptions of easily observable features of foraging,
such as seasonal variation in diet or sexual segregation in habitat selection, with little or no attempt to uncover the underlying causes of these phenomena. Although past research has described the characteristic features of elephant feeding behaviour, it has not uncovered the underlying principles and, therefore, little explanatory power has been gained.
To gain greater explanatory power, an attempt was made in this thesis to uncover the driving force behind the feeding response of elephants. This was done by firstly constructing a theoretical framework for the foraging strategy based on what is known about the body size, digestive physiology, and non-dietary requirements of elephants and, secondly, by testing the theory against actual landscape- level responses of elephants using a mechanistic foraging model. As far as the author is aware, this study represents the first attempt to investigate the foraging ecology of elephants using a true landscape-level approach.
The purpose of this section is (1) to highlight the theoretical and practical aspects of the study that were instrumental in its successful implementation, (2) to compare the approach used in this study with that employed in previous studies of herbivore foraging behaviour, and (3) to point out the new insights into the foraging ecology of elephants that were gained.
7.2.1.
A theoretical basis for resource selectionUnlike most studies of resource selection, where the underlying ecological theory is implicit (Guisan & Zimmermann, 2000; Austin, 2002; Austin, 2007), this study began by constructing an explicit conceptual framework for the optimal foraging strategy of elephants that was based on the effects of body size, type of digestive system, and salient non-dietary factors. This was an important first step because (1) it identified a nutritional currency that could potentially be used to rank the profitability of habitats and food types for elephants, (2) it provided a theoretical foundation for the research that could be tested empirically, and (3) it integrated potential distal influences into a single proximal currency which served to reduce the complexity of the problem.
The theoretical analysis predicted that elephants should optimise their nutritional fitness by selecting habitats and diets that maximise their rate of intake of digestible energy and nutrients. The findings were consistent with this hypothesis and provided fresh insight into the foraging ecology of elephants. The benefit of integrating ecological theory with statistical modelling was also demonstrated.
7.2.2.
Detailed quantification of the pattern of resource availability at the landscape levelThe development of GPS and satellite collars has revolutionised the study of resource selection by free- ranging animals at both local and landscape scales (Douglas-Hamilton, 1998; Galanti et al., 2000a; Blake et al., 2001; Johnson et al., 2002a; Johnson et al., 2002b; Nielsen et al., 2003; Frair et al., 2005). However, meaningful interpretation of the fine-scale position data from these collars is only possible when spatial databases describing the environment of the landscape are available at an equally fine scale (Guisan & Zimmermann, 2000). It is the opinion of the author that the success of the landscape-level approach used in this study was largely due to the effort put towards accurately quantifying the spatio- temporal pattern of resource availability over the entire study landscape (see chapters 3 & 4). Considering that a true landscape-level response is driven by the relative profitability of the available habitats rather than by an absolute measure, a study that aims to uncover landscape-level influences must quantify resource availability simultaneously across the entire landscape. The species composition and phenological variation of the herbaceous and woody strata, the structure of the woody layer, the topo- edaphic variables that influence plant phenology, recent fire history, and the spatio-temporal pattern of salient non-dietary factors can be considered the minimum set of landscape-scale variables required to study the foraging ecology of mammalian herbivores at the landscape level.
It should be noted, however, that detailed quantification of resource availability at the landscape scale has only become practicable in recent times on account of manifold advancements in satellite and GIS technology. Despite these technological advancements, a large proportion of the required data still has to be collected using a ground-based approach, which is both time consuming and labour intensive. This makes studies that extend for longer than one season difficult because rapid changes in food availability between seasons are not easily quantified, especially when the vegetation resource is markedly heterogeneous. Future research should be aimed at developing remote sensing techniques that are capable of continuously collecting the necessary data over large areas.
7.2.3.
Functional representation of the plant/elephant interfaceAlthough variation in search time, handling time and bite mass across food types have long been recognised as important determinants of diet selection by herbivores (Cooper & Owen-Smith, 1986; Spalinger & Hobbs, 1992; Parsons et al., 1994; Farnsworth & Illius, 1996; Sauvant et al., 1996; Haschick & Kerley, 1997; Farnsworth & Illius, 1998; Pastor et al., 1999; Wilson & Kerley, 2003b; Baumont et al., 2004), attempts to explain the diet composition of elephants have to a large extent been
based solely on analyses of the chemical composition of food types and plant species (Williamson, 1975; Osborn, 2004). In this study, a mechanistic ingestion model that functionally represented the plant/elephant interface showed that constraints to harvesting and chewing strongly influence diet selection by elephants (see chapter 4). In addition, by recognising that a high rate of food intake is made possible by simultaneous prehension and chewing of food, the mechanistic ingestion model revealed the important role played by the trunk in the nutritional ecology of elephants.
By functionally integrating the effects of seasonal variation in protein density, search time, handling time, and trunkload mass across food types, the mechanistic ingestion model provided a possible explaination for the seasonal change in the diet of elephants. Elephants may eat more grass during the rainy season because, at this time, the high patch density, short handling time and large trunkload mass of grass compensate for its low protein content to the extent that it has a higher rate of short-term protein intake relative to browse (Figure 4.62). With the onset of the dry season, grass begins to dry out and senesce. This has the effect of lowering the protein content, increasing search time, increasing handling time (because grass tufts must first be shaken to remove dry material before being ingested; see Table 4.2 and Table 4.3), and reducing the mass of a trunkload. The net result is a lower short-term rate of protein intake from grass relative to browse during the dry season, and hence the observed shift towards a diet largely composed of browse from woody plants.
Implementation of the ingestion model at the landscape scale also provided potential insight into why it should be necessary for some grazers to switch to a diet of browse at certain times, and why browsers rarely need to consume grass in large quantites (Clauss et al., 2003a). During the wet season, high quality grass is generally abundant in savannas and can be consumed at a higher rate than browse, making it a more profitable food source at this time (Figure 4.62). However, as the dry season progresses both the abundance and quality of the grass resource decline, which results in a lower rate of intake than when feeding on grass during the wet season. The potential for a nutritional deficit caused by the reduction in the rate of intake is further increased by a simultaneous decline in grass quality. As the dry season continues a point is eventually reached when, in most parts of the landscape, it is more profitable for grazers to feed on browse than grass. This is because the properties of the browse resource (quality, bite mass, search and handling times) are relatively constant and are not subject to marked seasonal fluctuations. Therefore, in order to meet their nutritional requirements during the dry season, grazers must either seek out the parts of the landscape where changes to the grass resource, relative to wet season conditions, have been negligible (e.g. swamps, flood plains or bottom lands with high levels of residual soil moisture), or they must begin to include larger portions of browse in their diets. Browsers,
on the other hand, do not have to deal with a marked decline in profitability of their primary food source because savanna landscapes generally provide a relatively constant supply of browse over the annual cycle (Figure 4.62). This is because (1) woody plants are generally deeper rooted than herbaceous plants, and therefore have access to soil water for longer, and (2) variability in the timing of leaf drop and flush across woody plant species means that invariably there is some component of the woody layer with green leaf at any given time (see chapter 4).
Most studies of resource selection by herbivores have not attempted to include both top-down and bottom-up cross-scale interactions in their models of habitat and diet selection (Johnson et al., 2004). The use of a mechanistic intake model to create a functional predictor variable (short-term rate of protein intake) that was the integrated result of foraging decisions made at fine to coarse scales proved to be a simple way of incorporating the effects of cross-scale interactions in the selection process. The technique also served to simplify the habitat selection models and, in so doing, made them easier to interpret.
7.2.4.
Sex related differences in feeding behaviourAlthough sex related differences in diet and habitat selection by elephants have been recorded and potential explanations presented (Stokke, 1999; Stokke & du Toit, 2000; Stokke & du Toit, 2002; Shannon et al., 2006; Smit et al., 2007), a mechanistic justification for these phenomena based on sex related differences in handling time, trunkload mass and the relative importance of dietary and non- dietary factors has not been previously presented. This study showed that sex related differences in trunkload mass and handling time across food types (see Chapter 4) may cause bulls and cows to have different spatio-temporal patterns of the rate of energy and nutrient intake over the landscape and that this may ultimately result in different patterns of habitat and diet selection between the sexes. Sexual segregation in habitat selection may also be explained by the short-term rate of protein intake being a more important explanatory variable than cost distance from water for the spatial distribution of family units, and the converse being the case for the spatial distribution of bulls.
7.2.5.
Importance of a landscape level perspectiveThe study showed that a true landscape-level approach is required to study habitat and diet selection by elephants. Short-term rate of protein intake and cost distance from water were shown to influence elephant distribution at the landscape level. In other words, the attractiveness of a habitat to elephants was not determined by the absolute measure of the factors for the habitat in question, but rather by a
measure that is relative to the other habitats in the landscape. Development of a generic model to predict habitat selection by elephants throughout their range will only be achieved by using a technique that accounts for this conditionality of response.
The results of the study also highlighted the importance of not only testing the congruence of predictor variables across multiple spatial scales, but also testing the predictive consistency of variables over time.