The natural cycles hypothesis

The hypothesis that probably aroused the greatest curiosity was the "stable limit cycle" between forests and elephants proposed in 1976 by Graeme Caughley. Whereas several of the earlier descriptions assumed that elephants and woodlands could exist in equilibrium under natural conditions, Caughley began with a radically different view. He argued that there was no natural equilibrium between elephants and woodlands in parts of semiarid Africa. Rather, he visualized a cyclical relationship in which elephants increase and thin out the forests, only to decline later when the forests become sparse, thus allowing the forest to regenerate. This in turn allows the elephants to increase, and the cycle repeats. This constitutes a stable limit cycle in which the trends in tree and elephant densities resemble sine waves with a time lag (fig. 6.7).

To make estimates of the period of the cycle, Caughley went by evidence provided by the trees themselves in the Luangwa Valley. Patterns of forking of stems in mopane trees, indicative of breaking by elephants, showed low frequencies at intermediate sizes compared to small or large sizes, which are suggestive of cycles in elephant usage. There was no basis for fixing ages to mopane trees. On the other hand, size measurements of baobab trees and their transformation to ages showed a peak in abundance (and hence recruitment) at sizes corresponding to 140-year-old trees. Since baobabs seem to regenerate most abundantly in grasslands, this indicated a low forest cover and a low elephant density prior to this time. If the peak in baobab regeneration corresponded to the trough in tree density, which occurred about a quarter period

Figure 6.7

Caughley's stable limit cycle (A) between elephants and trees (from Caughley 1976; reproduced with the permission of Blackwell Science Ltd., U.K.) extended (B) to a gradient of climatic/vegetational types. (From Sukumar 1989a.)

Figure 6.7

Caughley's stable limit cycle (A) between elephants and trees (from Caughley 1976; reproduced with the permission of Blackwell Science Ltd., U.K.) extended (B) to a gradient of climatic/vegetational types. (From Sukumar 1989a.)

or 50 years before the trough in elephant density, then a period of nearly 200 years between successive peaks or troughs in tree or elephant density was indicated.

Caughley recognized that human interference through hunting, fire, and agriculture in the natural system could arrest the cycle and impose an artificial equilibrium. He further reasoned that the period and amplitude of a stable limit cycle could change along a climatic gradient, contracting to a stable equilibrium eventually, but he did not elaborate further.

The Caughley hypothesis requires proximate mechanisms that could drive the cycles between elephants and trees (or, more broadly, woody vegetation). Many have questioned whether any such mechanisms exist in real populations. For a stable limit cycle to operate, there should be a coupling between elephant demography and woody plant biomass. Once woody plants begin to decline in abundance, could the elephant population also follow this trend after only a lag of a few decades? After all, if browse declines in the diet, elephants could switch over easily to grass. They also seem to thrive well on a diet of grass.

Actually, more than one study in East Africa has shown that elephants do not necessarily do too well on a diet in which tall coarse grasses predominate. Richard Laws's extensive studies in Uganda did show that fecundity rates fell considerably in elephant populations living at high densities, in habitats transformed from woodland to grassland, and when presumably their diets had changed from predominantly higher-quality browse to lower-quality grass. The slowing in reproduction was the consequence of both delayed age in first calving by females and longer intercalving intervals. The Murchison South population was clearly a declining one as a result of this demographic shift.

Sylvia Sikes also presented evidence that African elephants could suffer from cardiovascular diseases due to stress caused by loss of tree cover. She even argued that a diet of low-quality grass could make elephants more prone to clogging of arteries, much as a high-fat diet could do in humans.

The main question was whether these conditions were sufficient to produce the regular elephant-tree cycles that Caughley envisaged. This could be modeled much like predator-prey dynamics, but although Caughley provided the basic mathematical formulation in his 1976 article, this was attempted only two decades later. The empirical evidence for the Caughley model is weak. The pattern of age-size distribution of baobab trees (or other trees, for that matter), showing cycles in the form of peaks and troughs, could result in ways other than long-term variations in recruitment. Richard Barnes showed through modeling that such a pattern could arise from a growing elephant population feeding preferentially on young baobabs. My observations of Acacia leucophloea and Acacia suma show that bimodality in size distributions can arise without any significant change in elephant numbers. Nevertheless, I believe that the Caughley model represented a conceptual jump in attempts to understand the complex relationship between elephants and their habitats.

I qualitatively extended the Caughley model to span the spectrum of habitat types from the semiarid savanna woodlands to aseasonal tropical rain forests where elephants are found (fig. 6.7, table 6.3). Elephant demography can be properly understood only in the light of the evolutionary history of a population. This would have been important in shaping life history traits such as age at first reproduction or longevity. Similarly, the structure of a forest would depend on how the individual species have evolved in relation to climate, soil, competition, herbivory, and so on. The link between elephant population dynamics and forest dynamics would depend on what the latter has to offer in terms of forage. Rain forests have a high biomass of plants, but few of these are palatable or available to elephants. Savanna woodlands have a low biomass, but a high proportion of this is food for elephants. Thus, Caughley's limit

Table 6.3

Characteristics of elephant-vegetation dynamics in different habitats.

Semiarid Savanna Equatorial

Woodland Rain Forest

Rainfall Total quantity/year Coefficient of variation Vegetation

Density of large trees Total biomass of vegetation Proportional availability of woody plants as food Proportion of total trees that can be pushed over by elephants Quantity of edible woody plants Seasonality in forage resources Carrying capacity K Mean K (elephant /km2) Variation in K Demography

Age at first conception in female elephants Mean age in years Variance Mortality rates Maximum population growth rates per year Elephant-woodland dynamics

Low Low


High High

High High High

Low Low Low*


High range

Up to about 5% Highly fluctuating

Relatively stable

Source: Modified from Sukumar 1989a. *An exception may be fruit availability.

cycles model can be examined in the light of selection pressures on elephants and trees across this climatic gradient.

Selection pressures would obviously have varied considerably across the gradient from semiarid to tropical moist habitats. Rainfall in the semiarid habitat is not only very seasonal annually, but also highly variable from one year to another (in statistical terms, the coefficient of variation in rainfall is inversely related to the quantity of rainfall). High-quality food may be superabundant seasonally or during certain years, while it would be scarce in other periods. Elephants can also feed on a large proportion of plants—herbs, shrubs, and trees—in the structurally simple savanna-woodland. If the availability or quality of one type (say, grasses) declines at any time, they can switch over to another plant type or part (say, bark). On the contrary, rainfall is much less variable in aseasonal rain forests. Food, on the other hand, is more dispersed, much of it being of low quality. The rain forest is structurally complex and rich in species, but most plants are virtually useless as forage. Elephants have to search widely for nutritionally poor palms, lianas, selected fruits, or grasses and weeds growing in tree-fall gaps. The remaining plants have deadly chemi-cals—the so-called plant secondary compounds—that deter feeding. What little is available, however, is more predictable on a seasonal or interannual basis.

Let us consider what life history traits elephants should evolve in these different habitats. I elaborate on this subject in chapter 7, but it would be useful to state briefly the main arguments here. In the highly fluctuating semiarid habitat, it would be advantageous for the elephant to have traits necessary for a rapid rate of increase when favorable conditions return after a population decline. An early age at first reproduction (in females) and a shorter intercalv-ing interval are two traits of fecundity that would be adaptive in response to favorable conditions. Individuals should be able to delay reproduction (late age at maturity or longer calving intervals) when conditions are unfavorable. Thus, one could expect to see greater flexibility in reproductive traits in elephants of semiarid regions. The opposite can be expected in elephants that have evolved in more stable rain forests.

Observed demographic traits are not entirely genetic in origin, but also have strong environmental components. Nutrition would obviously have a major influence on reproductive traits. The relative abundance of high-quality forage in the semiarid savanna woodlands could thus enable elephants to be highly productive. Elephants in the rain forest have to spend energy in searching far and wide for suitable food. During one season, few species or trees might be in fruit. Food is also of lower overall quality. Such populations cannot be as productive as their counterparts in the savanna woodlands. Indeed, given their more stable environment, they do not need to be as productive for their survival and persistence.

The savanna woodlands have a higher carrying capacity for elephants. Densities anywhere up to three elephants/km , or even five elephants/km in localized areas, can be attained here as opposed to not more than one elephant/ 5 km in the aseasonal rain forests. Elephants in the former habitat can increase at a rapid rate of up to 4% per annum or even higher to large populations. Rain forest elephants can possibly grow at a more modest rate of less than 1% per annum.

What would be the fate of trees across the vegetation spectrum in response to elephant foraging? In savanna woodlands, a rapidly increasing elephant population would result in a noticeable decline of canopy trees, most of which are palatable to elephants. Charles Fowler has produced compelling evidence that large mammal populations are most productive when they are close to the carrying capacity of the habitat and not at half the carrying capacity, as called for by the logistic function of population growth. If this were to be true for elephants, the steepest growth in a population would be seen when the decline in woody vegetation has begun and elephant density is near its peak. Density-dependent brakes operate slowly in such a situation. Elephants, by virtue of their life history traits, are resilient to sudden population changes. Their ability to switch from browse to grass is also an insurance against immediate nutritional stress. Trees would continue to decline at a rapid rate. Recovery would be possible only when elephants decline to sufficiently low numbers or, alternatively, move to another area. Elephants and trees would fluctuate in their population numbers or biomass with a high amplitude and a low period.

One would hardly notice any damage to trees in the rain forest. The bark or leaves of most trees are unpalatable to elephants. Larger trees are immune to being pushed over by elephants. Considerable biomass is locked up as wood in these trees and thus is unavailable for elephants. Life history traits that result in a more stable demography in elephants and the lack of any elephant-related declines in woody vegetation would result in a near-stable equilibrium between elephants and vegetation. I am aware that rain forests are not necessarily the highly stable entities depicted in popular writing or even the earlier scientific literature; hurricanes, fires, and droughts can ravage these forests, albeit rarely. Under natural conditions, however, the rain forests of Africa and Asia inhabited by elephants are relatively stable over timescales of decades or centuries compared to savannas.

In regions of medium rainfall, the deciduous forests could be intermediate between the semiarid savannas and the rain forests in the nature of elephant-vegetation dynamics. Elephant life history traits would have medium plasticity. Structurally, the deciduous forests could witness considerable fluxes in the understory trees, but much less in the canopy trees. Elephant-vegetation cycles could have a lower amplitude and longer period here than in the savanna woodlands.

What is the evidence that elephant life histories and vegetation dynamics actually vary in the fashion described above? The earliest ages at first reproduction in female elephants, in the range of 10-13 years, are known only from the semiarid habitats such as Etosha, Hwange, Kruger, Gonarezhou, Mkomazi, and Tsavo (all African sites) and eastern Sri Lanka, while elephants in moist forest habitats such as Budongo (Uganda), Congo, southern India, and My-

anmar (Burma) begin reproducing much later (chapter 7). Intercalving intervals also seem to be slightly lower in such semiarid habitats, although the difference with moist habitats seems less pronounced. Variability in fecundity traits is also high in semiarid regions, as seen from the example of the Kabalega Falls elephant population in Uganda. These patterns broadly support my contention of life history variation across habitats. We need to know more, however, of the rain forest elephant both in Africa and in Asia.

Vegetation change across habitats also seems to follow the expectations outlined above. Tsavo is one example of a semiarid ecosystem in which the vegetation has fluctuated between grassland and woodland during the past century. There is historical evidence that the Tsavo region had few elephants (possibly in part due to hunting for ivory) and a largely grassland habitat (settled by pastoralists) toward the end of the nineteenth century. The region has since seen the establishment of woodland, an increase in the elephant population, decline in woodland, and a crash in elephant numbers during the great drought of 1970-1971 and later by poaching, followed by a resurgence in woodland. In the Luangwa Valley, too, a decline in fecundity was recorded by Dale Lewis during 1980-1981, about a decade after the high fecundity observed there by John Hanks. Lewis's explanation that a skewed sex ratio was responsible for this decline in fecundity is not tenable (a ratio of 1 adult male to 2.5 adult females is hardly skewed in elephant populations; see chapter 7). It is more likely that some poorly understood feature of habitat change caused this decline.

In southern Indian deciduous forests, interestingly, the impact of elephants on the vegetation is felt most acutely in understory trees and shrubs. In vegetation plots my research team has set up at Mudumalai, the greatest impact has been on the understory tree Kydia calycina and a shrub, Helicteres isora. Zizyphus xylopyrus is another small tree pushed over by elephant. Some of the canopy trees, such as Tectona grandis and Grewia tiliaefolia suffer damage to bark by elephants, but death due to elephants is rare in adult trees, especially in the former species. Other common canopy trees, such as Lagerstroemia microcarpa, Terminalia crenulata, and Anogeissus latifolia, are hardly affected by elephants. The deciduous forest thus witnesses considerable flux in the under-story, but is more stable in its canopy layer. I know of no large rain forest where elephants have caused (perceptible) instability in woody vegetation, although elephants could potentially inflict selective damage and change species composition if introduced into a forest where elephants have never existed.

The key question is whether limit cycles as visualized by Caughley can be generated in an elephant-tree system by incorporating realistic parameters of elephant and tree densities or biomass. This has been recently modeled by Kevin Duffy and associates. Using a range of realistic parameters, they showed that a fixed equilibrium between elephants and trees was the eventual outcome, and that an elephant-tree limit cycle was highly unlikely.

Even if the Caughley model itself is only of historical interest, the relationship between elephants and woody plants can still be visualized as going from near stability to a highly unstable dynamics in relation to environmental variability across a rainfall gradient. Human-transformed habitats could further add to this complexity. How would vegetation respond to a rapid compression of elephants to high densities from loss of habitat?

0 0

Post a comment