Population specific patterns

Energetic and nutrient requirements within life stages are assumed to be constant across locations as they are based on elephant life history and physiology, and therefore any observed differences in body condition must be attributable to some site-specific variables affecting condition.

The observed lower condition estimates in the AMC (Figs. 4.1, 4.3 – 4.8) for all seasons and both energy classes (with the exception of non-energy stressed individuals in winter) were correlated with an exceptionally high density compared to the other reserves, as well as low rainfall during the study period (see Table 4.1). This supports my prediction that poorer condition would be associated with areas of high elephant density and low rainfall (i.e. productivity). Comparisons with dietary quality measures revealed that, with the exception of summer, these lower body condition scores were also associated with generally lower faecal protein and higher faecal NDF values as compared to other reserves (Tables 4.2 & 4.6), supporting my prediction that lower conditions estimates would be associated with areas of lower dietary quality. If the lower estimate of protein for maintenance requirements (6%, Malpas 1977) is used, this data suggests that elephants in the AMC are on a sub-maintenance diet for nine months of the year as compared to three months in the NCA, Asante Sana and Blaauwbosch, with elephants in Shamwari and Kariega receiving sufficient

protein throughout the year. Although the estimate by Malpas (1977) refers to the dietary protein requirements, and is thus as such not directly comparable to the faecal protein values obtained, the inclusion of a high proportion of browse in the diet would result in an over- and not underestimation of dietary protein, due to the higher levels of protein-precipitating secondary compounds present in browse (Mould & Robbins 1981; Irwin et al. 1993). Elephants in the AMC thus appear to be relying on short flush periods after good rains to boost their protein intake and condition, which then deteriorates during the rest of the year. This suggests that they would be particularly vulnerable to becoming protein limited if an extended period of drought were to occur.

As predicted, the slightly better condition exhibited by energy stressed elephants in Asante Sana in spring (Fig. 4.3) was associated with a much higher than average rainfall for the study period (Table 4.1), as well as consistently good rains recorded from February to June 2007 (Appendix 7). This was not reflected in the dietary quality measures. These higher body conditions scores may, however, be related to the relative foraging effort at that time. Even if the increased rainfall did not affect the quality of the available forage, the abundance of vegetation would increase (Coe et al. 1976), thereby providing the elephants with more food. Additionally, increased rainfall may lead to the formation of smaller puddles and pools, thereby eliminating the need to travel larger distances to and from water sources. This decrease in daily activity would contribute to a decrease energy demands, thus allowing more energy to be located to storage, and ultimately contributing to higher body condition scores.

Although the better condition exhibited by energy stressed individuals in summer (Fig. 4.5) was associated with a large increase in protein and a drop in NDF values at that time (Tables 4.2 & 4.6), the slightly better conditions exhibited in Blaauwbosch by non-energy stressed individuals in spring (Fig. 4.4) was not supported by dietary quality measures. The better condition exhibited during the spring/summer period was also contrary to my expectations based on rainfall.

Based on the low rainfall in this area, it would be expected that these elephants would exhibit lower condition estimates than some of the other reserves. This is especially true for the spring/summer sampling period, as it was preceded by no rains from June to September 2007 (Appendix 7). Although the improved condition of elephants in spring remains unexplained, the increase in faecal protein and decrease in faecal NDF during the summer period is associated with the fruiting of the prickly pear, Opuntia ficus-indica (Brutsch & Zimmermann 1993), an invasive that can be found in large numbers throughout the reserve (pers. obs.). Elephants were seen to feed extensively on these fruits (pers. obs.), thus accounting for the higher dietary quality values and better body conditions observed.

The slightly lower body condition of energy stressed individuals in the NCA in summer (Fig. 4.5) and autumn (Fig. 4.7) were supported by higher values of faecal NDF (Table 4.6) and lower values of faecal protein (Table 4.2) than other reserves during this time, although faecal protein values were only at sub- maintenance level (6%; Malpas 1977) during autumn. The slightly lower body condition estimates obtained for the NCA elephants are also supported by the much higher elephant density than all but the AMC population, again highlighting this as a driver of the observed body condition scores.

These results support those obtained from the regression model, both for dietary quality and for the site specific drivers (Appendix 4), which highlighted elephant density, rainfall during the study period, faecal protein and faecal NDF as drivers of the observed elephant body conditions. Although not all relationships were significant, there was a general trend of populations with large proportions of elephants with higher body conditions (BC 5 and 6) associated with lower elephant densities, higher rainfall during the study period, higher faecal protein and lower faecal NDF (Figs. 4.15 – 4.18). The reverse trend was found for the lower body conditions (BC 3 and 4), with populations with larger proportions of elephants with body condition scores of 3 or 4 being associated with higher

elephant densities, lower rainfall during the study period, lower faecal protein and higher faecal nitrogen. It is important to note that the proportion of elephants within the four body condition scores (BC 3, 4, 5 & 6) within a population add up to one, and that the reverse pattern observed between the two highest and two lowest body condition scores are thus due to auto-correlation of the data, and are both illustrating the same trend.

The graph depicting the relationship between faecal protein and the proportion of the population within a specific body condition (Fig. 4.17) indicates that at approximately 8.5% faecal protein, the point at which the two lower body condition (BC 3 & 4) lines meet the X-axis (i.e. proportion = 0), none of the population falls within the lower body condition scores. This suggests that 8.5% faecal protein could possibly be the lower threshold value of faecal protein needed for the maintenance of elephant body condition within a population. A similar trend can be seen between the proportion of the population and faecal NDF (Fig. 4.18), where none of the population falls within the two lower body condition scores at the point where their lines (BC 3 & 4) converge, at approximately 71% faecal NDF. This suggests that 71% faecal NDF could be the upper threshold above which elephant body condition begins to deteriorate (i.e. some of the population are in poor condition). These observations, however, need to be further tested using a larger dataset, and provide a useful working hypothesis.

All four regression graphs (Figs. 4.15 – 4.18) show two clusters of data, with the AMC clustered away from the other populations. It is thus possible that the correlations between the data could simply be due to Simpson’s paradox. However, there is no evidence that Addo is a separate population of data, and I feel that the AMC is only the extreme of a gradient, and that the processes determining elephant body condition are the same within all the reserves. I am

thus confident that these regressions are valid. However, the model fit R2 was

0.274), and thus only explained 35.4% and 27.4%, respectively, of the variability observed, and it is therefore possible that some other, un-measured parameter could have contributed to the observed body condition distributions.

The observed variation in body condition of elephants across seasons in both energy classes in the AMC (Fig. 4.9) seemed to be roughly correlated with the rainfall recorded three months prior to the sampling dates (Appendix 7). Good body conditions in June were related to the peak in rainfall recorded in March, April and May, and the decrease in condition recorded in September was related to the decrease in rainfall recorded in June, July and August. Rainfall remained fairly constant, with no change recorded in condition in December. However, although good rains were recorded for December/January, body condition decreased from summer to autumn. With the exception of the good body condition scores observed in winter, body conditions were correlated with dietary quality measures (Tables 4.4 & 4.6). This could be seen especially in the large drop in condition from winter to spring, which was associated with a significant increase in faecal NDF and a significant decrease in faecal protein at that time.

Although the changes were not always significant, elephants of the NCA showed a similar trend in body condition (Fig. 4.10) in response to rainfall (Appendix 7), as illustrated by the AMC elephants. However, whereas the AMC elephants decreased in condition from summer to autumn, elephants from the NCA remained in similar condition. Comparisons with dietary quality measures revealed no correlations between the observed conditions and dietary quality.

The lack of variation in body condition scores for any of the energy classes across seasons in Kariega (Fig. 4.1) could be attributed to the high rainfall experienced almost throughout the year (Appendix 7), as well as the relatively large size of the reserve and low elephant density. Although the elephants were not in exceptional condition, their consistent condition throughout the course of the year suggests that they were receiving sufficient nutrients for maintenance.

This is supported by the consistently good faecal protein values obtained in the reserve (Table 4.4).

The drop in body condition score experienced by energy stressed elephants in Asante Sana in summer (Fig. 4.12) could not be attributed to rainfall. Although no rainfall was recorded for September, almost 90 mm of rain was recorded in October (Appendix 7), and rainfall in the reserve was much higher during the study period than the annual average. Dietary quality measures showed a mixed response in summer, with both NDF and protein values increasing (Tables 4.4 & 4.6), and it is therefore uncertain whether the observed changes in body condition could be attributed to changes in dietary quality. However, elephants in Asante Sana maintained good body conditions for most part of the year, possibly due to their very low density.

The lack of variation in body condition of elephants at Blaauwbosch (Fig. 4.13) was contrary to what was expected, not only based on the low average annual rainfall and small reserve size, but also the absence of rainfall from June to September 2007 (as discussed above). With the exception of higher body condition estimates for non-energy stressed individuals being associated with a drop in NDF during spring (Table 4.6), dietary quality measures were not

correlated with the body conditions observed. The fairly consistent conditions

were, however, supported by the low elephant density.

With the exception of energy stressed elephants in summer, Shamwari elephants maintained condition throughout the year (Fig. 4.14). This is in accordance with expectations, as Shamwari has a relatively low population density and rain falls throughout the year (Appendix 7). The decrease in condition experienced in summer could be explained by the two low rainfall months in September and October 2007. Dietary quality remained high throughout the study period (Tables 4.4 & 4.6), thus supporting the good body conditions observed.

Energy stressed and non-energy stressed classes across seasons for Kariega, Asante Sana, Blaauwbosch and Shamwari (Figs. 4.11 – 4.14) showed an inconsistency in their trends, with energy stressed individuals sometimes in better and sometimes in worse condition than non-energy stressed individuals, and this pattern being different for each site. It is important to note that the differences in condition between these groups were fairly small and non- significant, and that these patterns could have been created by small, natural variations in condition of the individual elephants comprising the classes over time.