Feeding ecology
Introduction
Colobines have traditionally been referred to as “leaf-eating monkeys”, but this moniker has, over the course of time, been shown to be less accurate than originally thought, as the diversity of feeding strategies the subfamily displays have become more thoroughly understood (Kirkpatrick, 2007). Colobines do of course have morphological adaptations that enable them to consume large quantities of leafy material, adaptations which separate them from the Cercopithecinae. These are most notably dental adaptations (Lucas and Teaford, 1994), large salivary glands (Kay and Davies, 1994) and the presence of a complex stomach and other physiological adaptations of the gastrointestinal tract which, together with symbiotic fermenting bacteria, aid in the effective digestion of plant material (Chivers, 1994). This ability to process plant materials and counter digestion inhibitors and toxins allows colobines access to plant resources not readily available to monogastric primates. It also probably facilitated an adaptive radiation of the Colobinae (Delson, 1994) and an avoidance of competition today with modern day Cercopithecinae with whom they are frequently sympatric. The diversity of colobine diets is now well documented (see Table 7-7 for a summary of Asian colobine diets). Some species have been shown to have high proportions of leaves in their diet such as Rhinopithecus brelichi (Bleisch et al., 1993, Bleisch and Xie Jiahua, 1998), Presbytis hosei (Rodman, 1978) and Trachypithecus leucocephalus (Zhaoyuan Li et al., 2003). Young leaves are often preferred, despite the fact that they are less abundant than mature leaves, as they are easier to digest (Kirkpatrick, 2007). Some other colobines are better characterised as seed predators, having large proportions of their diets made up of seeds, the classic examples being Presbytis rubicunda, whose diet consists of 80% seeds in some seasons (Davies, 1991) and, amongst the African colobines, Colobus satanas with seed consumption levels of up to 60% (Harrison, 1986). Seeds provide high energy relative to young or mature leaves and fruit (Dasilva, 1992) and colobine digestive physiology seems well adapted to maximise capture of seed carbohydrates relative to other food sources (Kay and Davies, 1994), showing that seed predation can be a viable strategy for colobines under the right conditions.
In many species, however, this distinction of seed predator versus folivore is not clear cut. Reliance on particular plant organs can be determined by their seasonal or local availability as has been shown for the best studied species. For example, Nasalis larvatus show considerable variation in levels of leaf and fruit consumption between studies (Bennett and Sebastian, 1988, Yeager, 1989, Boonratana, 1993, Matsuda et al., 2008). Where some colobine species might reasonably be described as leaf-eaters this does not mean they eat any leaves that are available. They are selective. For example, Trachypithecus leucocephalus was found to have a diet of 88% leaves, but over 60% of feeding observations came from only 10 species of plant (Zhaoyuan Li et al., 2003). In general, colobines are “picky eaters”, being quite selective in what species, and what parts of those species, they choose to feed on (Kirkpatrick, 2007).
Data on feeding ecology in the genus Pygathrix has to date been based largely on incidental field observations or short-term studies of captive animals. The seminal work of Lippold (1977) on red-shanked doucs (P. nemaeus) suggested initially that doucs have quite varied diets, with trees that are fruiting being a determining factor in group movement. Later work (Lippold, 1998) suggested that these doucs were one of the most folivorous of colobines, with a diet consisting of 82% leaf matter. Data from comparative morphology of the gastro-intestinal tract of P. nemaeus likewise suggested a largely folivorous diet (Chivers, 1994). This was contradicted somewhat by Pham Nhat (1994), based on his assessment of stomach contents of five red-shanked doucs, which suggested a diet with a considerable fruit component (37%), in addition to a significant leaf element (63%). It should be borne in mind that, unlike most observational field-based feeding ecology studies, stomach content analysis looks at volume as opposed to frequency of feeding records and that standard sampling protocols may underestimate volume of fruit consumption by a factor of five (Chivers and Hladik, 1980).
The somewhat fragmentary nature and methodological differences of these early and important works still left key questions about Pygathrix feeding ecology unanswered: how selective are doucs, what are the key species they utilise and how do seasonal changes alter the use of food resources? These questions remained unaddressed until the commencement of the current study in 2002. Since that time several other studies addressing aspects of feeding ecology in doucs have been conducted.
The first was conducted in Cuc Phuong National Park, Vietnam in a four hectare (ha) semi-wild enclosure (Otto, 2005). Although the study was conducted for only 10 days, it showed that a high number of plant species (43) were being utilised by the group of red-shanked doucs. Degree of folivory was lower than had previously been recorded, although the duration of the study meant that considerable sampling error could be expected. While of considerable importance to the management of captive populations of doucs at the Endangered Primate Rescue Center (EPRC), the study’s value for understanding wild douc feeding ecology was limited by its short duration.
The most comprehensive feeding ecology research conducted to date is that by Hoang Minh Duc et al. (2009), who presented a fascinating description of the feeding ecology of P. nigripes in two National Parks in Vietnam, Nui Chua and Phuoc Binh. These parks are radically different from each other in terms of environment and botanical composition (BirdLife International Vietnam Programme and The Forestry Inventory and Planning Institute, 2001), and as such make an ideal study into the degree of dietary flexibility that the species is capable of expressing. While the study was only seven months in duration, it showed a stark difference in approaches between the two douc populations in terms of plant part selectivity, and also documented some seasonal changes in diet, both of which will be explored in more detail in this chapter as they relate to the current study.
While researchers were slow to begin study of the genus Pygathrix, the genus is today under ever increasing scrutiny. At the time of writing there are many research projects at many different sites underway which look at the feeding ecology of all taxa of douc (Dinh Thi Phuong Anh et al., 2008, Ha Thang Long et al., 2008, Lippold et al., 2008, O'Brien et al., 2008, Phaivanh Phiapalath and Pongthep Suwanwaree, 2008). It is expected that in the next few years we will finally have a comprehensive understanding of the feeding ecology of the genus and how it varies between species and sites and through seasons. This current chapter is a contribution to this growing understanding and addresses several key questions relating to feeding ecology of the black-shanked douc in Seima Biodiversity Conservation Area (SBCA), Cambodia. I attempt to determine the key components of their diets, how selective they are in relation to plant species exploitation, what are the key resources for them, how diet changes temporally between seasons as well as throughout the day and how all these factors affect issues relating to conservation such as carrying capacity and population protection.
Methods
In order to determine relative frequencies of exploitation of different plant parts by P. nigripes, I used a scan sampling protocol as detailed in Chapter 3. In addition to plant part consumed, the species of tree being utilised was also noted during observations. Approaches to plant species identifications are detailed in Chapter 3. It should be noted that a tree species was only recorded as a food tree if doucs were observed actually eating from it. Information provided by local guides, and fallen fruit and leaves found on the forest floor, were not included in the analysis, as this information can be misleading. Additionally, tree species identification can be problematic in Cambodia due to the lack of a herbarium for comparative assessment and the fact that in many instances there is simply no systematic structure to assign specimens to species in some groups (D. Middleton, in litt.). This is particularly so for the Lauraceae and to a lesser extent the Moraceae and Rubiaceae.
In order to determine how resource use by the doucs tracked resource availability I calculated an availability index for fruits (including seeds), flowers and young leaves for each month for which phenological data were available. The availability index for each plant part was calculated using the following equation:
∑
∑
= k i k i t ba AI Equation 7.1Where bis the basal area of transect tree i…k which has the plant part in question for that month, a is the abundance score of that plant part on tree i on a scale of 0-3, determined from monthly phenology walks and ti…k is the basal areas (the cross-
sectional area of trees stems at breast height) of all transect trees. This provides an availability index of between 0 and 3.
While many studies use only stem counts to determine resource availability (e.g. Peres, 1994, and see Chapter 4 for this kind of analysis), it has been shown that DBH (diameter at breast height) and associated basal area are good indicators of crown volume and thus abundance (Chapman et al., 1992, Chapman et al., 1994, Phillips, 1995, Heiduck, 2002, Miller and Dietz, 2004) and therefore provide a more accurate estimate of abundance than stem number considered alone. This general approach has been used by several authors (Chapman et al., 1994, Dasilva, 1994, Fashing, 2001). As a simple test of whether these approaches actually measure different variables, I
checked to see how well abundance estimates based on simple stem counts correlated with abundance estimates from Equation 7.1, using Pearson’s Correlation Coefficient. Abundance indices from Equation 7.1 were then tested for correlation with monthly data on consumption of different plant parts (fruit including seed, flowers, and young leaves) taken from behavioural samples using the Spearman Rank Correlation Coefficient, as appropriate for indices. The frequency of a plant part being consumed for a particular month was simply calculated as the number of records for each plant part divided by the total number of feeding observations where a positive identification of plant part was made. A second more restricted statistical test was then performed, including only those plant species that contributed more than 1% of the annual diet, had positive selection ratios, and occurred on transects, similar to Fashing (2001). In this second analysis, fruits and seeds were considered separately to better understand how seed predation at the site tracked seed availability.
Of particular interest for studies on primate feeding ecology is the issue of how selective the species in question is in terms of food items it chooses to eat. I determined what species of plant were selected by doucs at high frequencies relative to their availability within the habitat by calculating a selectivity ratio (SR). Calculations followed McKey and Gartlan (1981), except that selectivity ratios were calculated only by tree species, not by plant part within each species, due to low sample sizes for many species. Selectivity ratio was calculated using the following equation:
i i i b f SR = Equation 7.2
Where fi is the percentage of feeding records for species i, and bi is the percentage of total basal area for species i calculated from Transects 1 and 2. Species that are positively selected by the animals have an SR of more than 1.0, those that are avoided have a SR of between 0 and 1.0, and those that are neutral should have an SR of approximately 1.0. In some instances feeding records were made for tree species which were not captured by the transects, which did not allow for the calculation of a selectivity ratio. In these instances the selectivity ratio was merely noted as positive. For additional clarification on selectivity, data on the 15 tree species with the highest summed basal areas calculated from transects were included in the analysis in order to ascertain at what levels these common species in the environment were being utilised.
Daily, seasonal and age and sex class variability in relative frequencies of plant part consumption were assessed using Chi-squared statistical techniques. All statistical tests were performed using SPSS version 12.0.1.
Results
Consumption by plant part
Feeding behaviour considered the consumption of eight different food types. The relative frequencies of consumption of each food type across the entire study duration, from most frequent to least frequent, with all age and sex classes, excluding infants, grouped together (n = 1116), are:
seeds 39.70%, fruit 9.77%, figs 1.61%
young leaves 24.01% flowers 8.78% other 0.18%. unknown leaves 10.04% mature leaves 5.91%
These percentages are shown in Figure 7-1.
Figure 7-1 Frequencies for feeding on different food types, age and sex classes grouped, excluding infants (n = 1116)
Some studies do not differentiate between some of the feeding subcategories used in this study, so for the sake of comparison grosser divisions are also given here. Using only the categories of fruit, flower, leaves and other (what remains), figures are:
Seeds 39.7% Leaves (young) 24.0% Leaves (mature) 5.9% Leaves (unk) 10.0% Fruit 9.8% Flowers 8.8% Figs 1.6% Other 0.2% Seeds Leaves (young) Leaves (mature) Leaves (unk) Fruit Flowers Figs Other
51.08% fruits 8.78% flowers, 39.96% leaves 0.18% other.
The only instance of an individual feeding from a food source categorised as “other” occurred when one individual, who was not identified by sex or age category, was seen to be ripping the bark off twigs of a Dipterocarpus alatus tree, and consuming it. At no time were doucs seen to drink water from any source, nor even to lick moisture from leaves.
Hourly trends
Black-shanked doucs at SBCA showed a highly significant association between the time of day and which food items were selected for consumption (χ2 = 136.21, df = 9, p < 0.01). Most significantly, seeds were heavily exploited at the expense of all other food types in the early morning (see Table 7-1 and Figure 7-2) with levels considerably higher than would be expected from the general patterns of Figure 7-1. This preference changed during the rest period over midday with fruits and seeds then under-represented and leaves and flowers over-represented. There was no clear pattern of targeted exploitation in the evening before the doucs stopped activities for the day.
0% 10% 20% 30% 40% 50% 60%
Seeds Fruits Leaves Flowers
5 am - 8 am 8am - 11am 11am - 2pm 2pm - 5pm
Time Seeds Fruit Leaves Flowers
5 am-8 am 58.27 6.50 30.35 4.88 8 am-11 am 32.61 17.39 38.86 11.14
11 am-2 pm 26.24 0.71 53.90 19.15 2 pm-5 pm 30.08 16.10 48.73 5.08
Table 7-1 Percentage of food items eaten at different times of the day (n = 1114).
Age and sexual variation
As noted in Chapter 5, variation in feeding frequency was detected, with males feeding less frequently than expected and immatures more (χ2 = 14.916, df = 2, p = < 0.01). The data show considerable age and sex variation in consumption of different food types (χ2 = 58.511, df = 10, p = < 0.01) (Figure 7-3). Male diets were dominated by seeds, with a frequency of almost 60%. Female and immature consumption of seeds was still high, but considerably lower than males’, with frequencies of 43.8% and 34.8% respectively. Conversely, male consumption of leaves was relatively low, with levels of 22.1% overall, compared to females who consumed leaves at almost twice the frequency with 40.9%, and immatures at 36.8%. Other food items played a lesser role in the diet of all animals. For both fruits and flowers females had the lowest consumption frequencies followed by males and immatures. Figs were eaten infrequently, but with high intensity.
Sub-category Female n Male n Immature n
Seeds 43.8% 154 59.3% 166 34.8% 71 Leaves (imm) 25.9% 91 9.0% 25 17.7% 36 Leaves (unk) 9.9% 35 8.2% 23 13.2% 27 Fruit 8.8% 31 11.0% 31 14.2% 29 Flowers 4.5% 16 7.5% 21 12.7% 26 Leaves (mat) 5.1% 18 5.0% 14 5.9% 12 Figs 2.0% 7 0.0% 0 1.5% 3
Table 7-2 Frequencies for feeding on different food types, by age and sex class
0% 10% 20% 30% 40% 50% 60% Seeds Leaves (imm) Leaves (unk)
Fruit Flowers Leaves (mat)
Figs
Female Male Immature
Figure 7-3 Frequencies for feeding on different food types, by age and sex class
(nfemale = 352, nmale = 280, nimmature = 204).
30.06% 24.86% 14.45% 6.74% 0.19% 13.49% 10.21% 0.33% 5.18% 3.85% 47.99% 9.87% 9.53% 23.24% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% Seeds Leaves (imm) Leaves (unk)
Fruit Flowers Leaves (mat)
Other
Dry Season Wet Season
Figure 7-4 Frequencies for feeding on different food types, by season (ndry=519, nwet=598).
Seasonal variation
Diet was significantly different in the dry season (November – April) compared to the wet (May – October) (χ2 = 63.045, df = 4, p < 0.01) (Figure 7-4). Most notably, feeding on seeds was considerably under-represented in the dry season (30%) and over- represented in the wet season (48%). This discrepancy was reduced, however, when fruits and seeds were analysed as one food type (data not shown). Conversely, consumption frequencies of flowers were considerably higher in the dry season (14.5%) than in the wet season (3.8%). Consumption of leaves in all categories was similar between seasons, with total frequencies of 38.3% in the wet and 41.8% in the dry. There were no significant differences in consumption of immature or mature leaves by season.
Consumption versus availability
Indices were calculated to determine whether resource use was tracking resource availability. Resource availability of fruit (including seeds), flowers and young leaves was used to calculate a monthly availability index (see Equation 7.1, Methods section). This availability index was then tested for a correlation with abundance estimates derived from the percentage of stems with a plant part confirmed as present for each month, to determine whether these different methods return radically different abundance levels. It was found that for all plant parts on both transects, there was a highly significant correlation between methods (Table 7-3). While both methods appear to track abundance, the availability index, which uses basal area and abundance estimates from phenology data was used for all subsequent analysis based on its assumed higher accuracy.
Percentage of Stems with Plant Part Present
Availability Index Fruit Flower Young Leaf Transect 1 0.722** 0.712** 0.719** Transect 2 0.679** 0.819** 0.920**
Table 7-3 The correlations between calculated availability indices using Equation 7.1 and simple stem counts of phenology trees with a plant part confirmed present (Pearson correlation values and probabilities are given).
** p < 0.01
I also tested whether Transects 1 and 2 were correlated in terms of the calculated relative abundances by month (see Chapter 4 for descriptions of transects). Abundance of fruits (rs = 0.485, n = 17, p = 0.048) and flowers (rs = 0.569, n = 17, p = 0.017) on the
two transects were strongly correlated, this was less so for young leaf abundance (rs = 0.407, n = 17, p = 0.105). Transect 2 showed higher spikes in young leaf
abundance index scores during the end of the dry/start of the wet season transition. Transects data were pooled to create a single availability index for comparison with feeding data. Calculated availability indices for fruit (including seeds), flowers and young leaves across the entire study based on all transect trees are shown in Figure 7-5. Fruits and seeds were not differentiated in this analysis as not all trees were exploited by the doucs and not all trees were identified, and therefore a classification other than broadly “fruit” was not possible.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Janu ary Febr uary Marc h April May June July Augus t Septe mber Octo ber Nov/D ec Janu ary Febr uary Marc h April May June A v a ila b ili