Dietary flexibility of Bale monkeys (Chlorocebus djamdjamensis) in southern Ethiopia: effects of habitat degradation and life in fragments

Background Understanding the effects of habitat modification on the feeding strategies of threatened species is essential to designing effective conservation management plans. Bale monkeys (Chlorocebus djamdjamensis) are endemic to the rapidly shrinking montane forests of the southern Ethiopian Highlands. Most populations inhabit continuous bamboo forest subsisting largely on the young leaves and shoots of a single species of bamboo. Because of habitat disturbance in recent decades, however, there are now also several dozen small populations inhabiting isolated forest fragments where bamboo has been degraded. During 12-months, we assessed Bale monkey responses to habitat degradation by comparing habitat composition, phenological patterns, and feeding ecology in a largely undisturbed continuous forest (Continuous groups A and B) and in two fragments (Patchy and Hilltop groups). Results We found that habitat quality and food availability were much lower in fragments than in continuous forest. In response to the relative scarcity of bamboo in fragments, Bale monkeys spent significantly less time feeding on the young leaves and shoots of bamboo and significantly more time feeding on non-bamboo young leaves, fruits, seeds, stems, petioles, and insects in fragments than in continuous forest. Groups in fragments also broadened their diets to incorporate many more plant species (Patchy: ≥ 47 and Hilltop: ≥ 35 species)—including several forbs, graminoids and cultivated crops—than groups in continuous forest (Continuous A: 12 and Continuous B: 8 species). Nevertheless, bamboo was still the top food species for Patchy group (30% of diet) as well as for both continuous forest groups (mean = 81%). However, in Hilltop group, for which bamboo was especially scarce, Bothriochloa radicans (Poaceae), a grass, was the top dietary species (15% of diet) and bamboo ranked 10th (2%). Conclusions We demonstrate that Bale monkeys are more dietarily flexible than previously thought and able to cope with some degradation of their primary bamboo forest habitat. However, crop raiding and other terrestrial foraging habits more common among fragment groups may place them at greater risk of hunting by humans. Thus, longitudinal monitoring is necessary to evaluate the long-term viability of Bale monkey populations in fragmented habitats. Electronic supplementary material The online version of this article (10.1186/s12898-018-0161-4) contains supplementary material, which is available to authorized users.


Background
Habitat loss and degradation by humans are the major threats to biodiversity worldwide [1,2]. Widespread disturbance to formerly intact forests, particularly in the tropics, is resulting in increasing fragmentation of habitats and biological populations [3]. Given that the global human population is expected to continue to increase in the coming decades, resulting habitat alterations may cause the extinction of thousands of species, including many mammals [4][5][6]. Habitat degradation modifies vegetation composition and structure, consequently reducing habitat quality and food availability for species inhabiting an area [6][7][8][9][10]. This decrease in food availability, in turn, lowers the carrying capacity of populations, and, in extreme cases, results in extirpation or extinction [7,8,11].
Currently, many populations are restricted to small isolated forest patches surrounded by human-dominated landscapes [12][13][14]. The persistence of these populations, therefore, depends on their ability to cope with change and the minimum size and quality of fragments required to sustain them [15][16][17]. One of the central challenges that must be overcome by populations in fragments is meeting their dietary needs in habitats in which the diversity and abundance of plant species has been substantially altered [7,11,18].
Understanding the dietary responses of individual species to habitat degradation and life in fragments is therefore crucial to designing and implementing appropriate species-based management strategies [51,52], especially for dietary specialists which are expected to be less flexible at coping with degradation of their habitats than generalist species [24,53]. For example, until now, no research has yet been conducted to assess the effects of habitat degradation and life in fragments on the feeding strategies of the Bale monkey (Chlorocebus djamdjamensis), an arboreal dietary specialist endemic to the montane forests of the southern Ethiopian Highlands. The Bale monkey is unusual among primates and other mammals for its intense specialization on a single species of bamboo (Arundinaria alpina), which accounts for 77% of its diet in continuous forest [54,55]. The Bale monkey is thought to be at high risk of extirpation because of its specialized niche, small geographic distribution, and the ongoing deforestation occurring across much of its range [54,[56][57][58]. As a result, the species is currently classified as Vulnerable by the International Union for Conservation of Nature (IUCN) [56].
In its high degree of specialization, the Bale monkey appears to provide a striking contrast to its five sister species: vervet monkeys (Chlorocebus pygerythrus), grivet monkeys (C. aethiops), green monkeys (C. sabaeus), Malbrouck monkeys (C. cynosuros) and tantalus monkeys (C. tantalus). Two of these sister species, vervets and grivets, are also native-though not endemic-to Ethiopia and are parapatric to Bale monkeys [59,60]. All members of the genus Chlorocebus, except Bale monkeys, are terrestrial generalists that consume varied omnivorous diets and inhabit a wide range of savanna woodland and grassland habitats over large geographic ranges in equatorial or southern Africa [61][62][63]. Incidentally, an analogous situation exists among monkeys in the genus Cercopithecus where one taxon, the golden monkey (Cercopithecus mitis kandti), is a bamboo specialist while other taxa, including other C. mitis subspecies, tend to be dietary and habitat generalists [63,64].
Intriguingly, the recent discovery of Bale monkey populations during surveys in a few dozen heavily-degraded forest fragments, some with little bamboo left [57], suggested the species might be of greater ecological flexibility than previously believed [54][55][56]65]. This unexpected discovery created the need to evaluate the strategies the monkeys employ in response to habitat degradation and life in fragments by comparing groups inhabiting fragmented habitats with those in continuous forest. We therefore undertook a study comparing the activity, ranging, and dietary patterns of Bale monkeys in fragmented and continuous forests. We recently published evidence that Bale monkeys in fragmented habitats adopt an energy minimization strategy-moving less, feeding less, resting more, and traveling over shorter distances per hour than conspecifics in continuous forest [66]. Along with examining energetic responses to degradation, we sought to determine the dietary strategies Bale monkeys use to cope with the relative scarcity of bamboo in fragments.
The specific aims of the study described here were thus to assess the effects of habitat degradation and life in fragments on (1) habitat quality and temporal patterns of food availability and (2) Bale monkey dietary composition, diversity and selectivity by comparing the feeding ecology between populations in continuous and fragmented forests. We also sought to (3) compare the patterns of dietary flexibility exhibited by Bale monkeys in our study with those of their five sister Chlorocebus species [63], as well as with those of other bamboo-eating mammals, including several other primates (e.g., Cercopithecus mitis kandti [67], Macaca assamensis [68], Prolemur simus [26], Hapalemur spp. [26]) and red and giant pandas [14,34,69]. We hypothesized that any reduction in habitat quality in forest fragments would strongly influence the feeding strategies of Bale monkeys. In particular, we predicted that the anticipated lower abundance of bamboo in fragments [57] would lead Bale monkeys there to consume a greater diversity of food items, plant species and growth forms, including human foods on nearby farms, than conspecifics in continuous forest. We also predicted that Bale monkeys in continuous forest would be bamboo specialists [54], but that conspecifics in fragments would exploit diets more similar to those of other more generalized Chlorocebus species [61,70].

Study site and habitat characteristics
We carried out our study in the continuous Odobullu Forest (06°50′-6°56′N and 40°06′-40°12′E) and two forest fragments (6°44′-06°45′N and 38°48′-38°51′E) in the southern Ethiopian Highlands [66]. Odobullu Forest (hereafter continuous forest) is a large forest within which bamboo is abundant. It covers 141 km 2 (14,100 ha) at elevations ranging from 1500 m to 3300 m asl and lies east of Bale Mountains National Park [54]. The continuous forest consists of four habitat types: mostly bamboo forest and tree-dominated forest but also shrubland and occasional grasslands [55]. It is partially protected by a privately-owned hunting company, Ethiopian Rift Valley Safaris, and disturbance in the home range of our study groups is uncommon due to the steep terrain and remoteness of the area.
Kokosa forest fragment (hereafter Patchy fragment) consists of degraded bamboo with large trees set amidst a matrix of human settlement, cultivated land, shrubland and grazing land. It covers an area of 162 ha and ranges in elevation from 2534 m to 2780 m asl. Most of Patchy fragment is privately owned by local people, though a portion is owned by the community collectively [66]. Selective logging of bamboo is common today.
Afursa forest fragment (hereafter Hilltop fragment) is set upon a hilltop and consists of a mix of secondary forest, shrubland, and Eucalyptus plantation with graminoid and forb cover underneath. Bamboo has been nearly extirpated. Hilltop fragment covers an area of 34 ha at elevations ranging from 2582 m to 2790 m asl and is surrounded by an anthropogenic matrix of cultivated lands, pastures and human settlements. Currently, the district government forbids cutting of trees and use of the fragment for grazing. The edge of the fragment, especially the Eucalyptus plantation, is still used illegally for grazing. Both the Patchy and Hilltop fragments were dominated by bamboo forest only three decades ago [57]. The distance between Hilltop and Patchy fragments is 9 km and they have been separated from one another by human settlement, grazing land and agriculture for many decades [57]. The forest fragments are separated from the continuous forest by ~ 160 km [66].

Study groups
We selected four Bale monkey groups for this study: two groups within the continuous bamboo forest (hereafter Continuous A and Continuous B) with overlapping home ranges (29% overlap for Continuous A; 47% overlap for Continuous B) [66], one group in the Patchy fragment (Patchy group) and one group in the Hilltop fragment (Hilltop group). The home ranges of continuous forest groups (Continuous A and Continuous B) consisted of exclusively bamboo forest (53.7 and 55.6%) and mixed bamboo forest habitats (46.3 and 44.4%). In contrast, the home ranges of fragment groups consisted of more variable habitat types. Patchy group's range consisted of five habitat classes: grazing land (37.9%), shrubland (29.5%), mixed bamboo forest (17.1%), tree-dominated forest (8.0%) and cultivated land (7.5%) while Hilltop group's range consisted of four habitat classes: shrubland (50.4%), tree-dominated forest (22.7%), Eucalyptus plantation (24.3%) and grazing land (2.7%) [66]. A.M. and two assistants habituated these groups to human observers for 4 months from March to June 2013 by following each group from dawn to dusk on a near daily basis. We identified 10-15 members of each focal group by their distinctive natural markings (e.g., coat color, facial features, tail shape). Group sizes were: Continuous A, 65 individuals; Continuous B, 38 individuals; Patchy, 28 individuals; and Hilltop, 23 individuals [66].

Climate
We recorded climatic data at the continuous forest (Fly campsite, elevation 2758 m asl; 1.5-2.0 km from the two study groups) and at Patchy fragment (Kokosa campsite, elevation 2634 m asl; 1.5 km from Patchy fragment). We measured daily rainfall using Oregon wireless rain gauges and recorded the daily maximum and minimum temperatures using Taylor digital waterproof maximum/ minimum thermometers. We assumed that the rainfall and temperature patterns are similar in each of the two fragments because they are both small, located only 9 km apart, occur at similar elevations, and are oriented in the same north-south and east-west directions. We calculated the monthly and annual rainfall for the period July 2013 to June 2014. We also used the daily maximum and minimum temperatures to calculate monthly means for these variables and calculated annual means by taking the averages of the monthly means.
Though annual rainfall was higher in the fragments (1676 mm SE ± 20.6) than in the continuous forest (1340 mm SE ± 24.8), this difference was not significant (ANOVA: df = 1; F = 2.31; P = 0.136) (Fig. 1). Both study areas were characterized by bimodal rainfall with a long wet season and a short dry season ( Fig. 1) but rainfall was less strongly seasonal in the forest fragments than in the continuous forest (Fig. 1). Mean annual temperature (16.7 °C SE ± 0.4) was significantly higher in the forest fragments than in the continuous forest (14.7 °C SE ± 0.2) (ANOVA: df = 1; F = 48.71; P < 0.001).

Vegetation description and temporal patterns of food availability
To examine whether the diet of Bale monkeys was influenced by resource availability, we sampled the vegetation in the ranges of our study groups using two complementary techniques. First, we enumerated all large trees with diameter at breast height (DBH) ≥ 10 cm in 12-24, 50 m × 10 m vegetation quadrats along randomly selected vegetation transects in the home range of each study group. Within quadrats, we measured and recorded the species name, growth form and DBH (in cm) for each tree. Second, we also randomly selected 50% of the vegetation quadrats in each group's range within which we counted and identified all plants ≥ 2 m tall to species level. This second vegetation enumeration technique enabled us to sample bamboo, shrubs and forbs that the monkeys depend on but that are < 10 cm DBH. For the bamboo sampled with this second technique, we also recorded the DBH of each culm.
In each group's home range, we calculated the stem density for all plant species ≥ 2 m tall and basal area (cm 2 /ha) for all large tree species (DBH ≥ 10 cm) and bamboo. We assessed the degree of stem density overlap between the home ranges of the study groups using the Morisita-Horn similarity index, which takes into account both relative abundance and species richness [71]. We classified plant growth forms into five categories: bamboo, trees, shrubs, lianas (including climbers and epiphytes), and forbs. To estimate the biomass of each large tree species and bamboo, we calculated the basal area (BA) of each tree species from the DBH recorded using the formula (BA = [0.5 × DBH] 2 × π) [72].
To evaluate temporal changes in the availability of potential food resources, we carried out monthly phenological assessments over an annual cycle for selected food plant species found at each of the study sites (see [66] for additional details). These species were selected for monitoring because they had been food species for Bale monkeys in a previous 8-month study in continuous forest at Odobullu [54]. At the start of our study, we marked and identified 10-15 individuals of these food species which included: trees (DBH ≥ 10 cm), bamboo (A. alpina), and shrubs. We assigned every monitored plant a relative abundance score for each of its potential food items (young leaves, mature leaves, flowers, ripe fruits, and shoots) via visual inspection, using binoculars where necessary. Relative abundance score ranged from 0 (item absent from plant) to 8 (plant fully laden with the item) at intervals of 1 [66].
We analyzed phenological data from eight species: five trees (Canthium oligocarpum, Dombeya torrida, Galiniera saxifraga, Hagenia abyssinica, and Ilex mitis), two shrubs (Rubus apetalus and Bothriocline schimperi) and bamboo (A. alpina). Ultimately, these species cumulatively accounted for 92.6% of the diet of Continuous A; 93.4% for Continuous B, 50.9% for Patchy and 44.5% for Hilltop groups. The lower contribution of monitored plants to the diets of fragment groups resulted from these groups consuming much less bamboo as well as a greater variety of food species, including difficult-to-monitor insects, graminoids and forbs (cf., [73]), than continuous forest groups. We calculated the monthly mean phenological scores for young leaves, fruits, flowers, and shoots for each individual plant species. We calculated the monthly food availability index (FAI) for each plant part by multiplying the mean phenology scores of species i with the mean basal area of species i and density of the corresponding species i per ha [72].

Feeding ecology
We collected activity data from July 2013 to June 2014 using instantaneous scan sampling [74] at 15-min intervals for up to 5 min duration, typically from 0700 to 1730 [66]. During scans, when a monkey was observed feeding, we recorded the type of food item, growth form and species. We recorded food items as bamboo young leaves, bamboo mature leaves, non-bamboo young leaves (from all species other than bamboo), non-bamboo mature leaves, bamboo shoots, bamboo branchlets (young and thin stems emerging from branches), roots, flowers, fruits, seeds, stems, petioles, insects or mushrooms. We recorded plant growth form as tree, bamboo, shrub, liana (including climbers and epiphytes), forb, or graminoid (grass or sedge). Although most food species consumed were identified in the field, species that could not be identified were collected for taxonomic identification at the National Herbarium in Addis Ababa. We recorded a food item as insects when the monkey was observed manipulating tree bark, searching through dead leaves or directly consuming insects [54]. We collected 28 [66]. Monthly sampling effort was evenly distributed among groups throughout the year.
We assessed dietary composition for each month by determining the proportion of different food items, growth forms and species consumed in each study group. We then calculated annual consumption of food items, growth forms and species as the means of the 12 monthly values for each category. We combined four food items (mature leaves, branchlets, roots and mushrooms) into the category "other" in our analyses because each individually accounted for < 1% of the overall percentage of feeding records. We also compared the identity and contributions of the top five plant species in the diets of each group. We calculated the relative dietary preference (i.e., food selection ratios) by dividing the proportion of annual percentage of feeding records on a particular species i by the percentage stem density of species i in the study group's home range. A selected food species is consumed more frequently than expected based on its proportional representation in the group's home range [72]. A food selection ratio of 1 indicates no selectivity for that food plant species, < 1 indicates a food species is avoided and > 1 indicates a food species is selected. We were only able to calculate selection ratios for trees, bamboo, shrubs, and lianas because stem density cannot be evaluated using the same methods for graminoids and forbs.
To estimate the annual plant species richness of the diet for each study group, we pooled the data from all sampling months within each group. We calculated withinmonth and annual dietary diversity indices for each group using the Shannon-Wiener index (H′), dominance index (D) and evenness index (J) [71] using the software PAST [75]. To assess differences in inter-month dietary similarity among groups in continuous forest and forest fragments, we calculated the inter-month Morisita-Horn's similarity indices (C H ) of each group [76] using EstimateS [77]. To assess the annual diet overlap among groups in continuous forest and forest fragments, we also calculated between group Morisita-Horn similarity indices. The index (C H ) ranges from 0 (no diet overlap) to 1 (complete diet overlap).

Statistical analyses
We conducted all statistical tests using R version 3.3.2 [78] with significance level set at P ≤ 0.05 unless otherwise stated. We tested data for normality and homogeneity of variances using the Shapiro-Wilk and Levene tests, respectively. We initially calculated and compared variables for each study group individually and examined the differences using the one-way analysis of variance (ANOVA) model followed by the Tukey honest significant difference (HSD) post hoc test. When the results for both groups within continuous forest and fragments were similar, we combined these groups for data analysis unless otherwise stated.
The completeness of plant species recorded in the diet is dependent on sample size. Therefore, we constructed a sample-based rarefaction curve plotting species richness with sampling effort (number of observation days) using PAleontological STatistics (PAST) software [75] to perform a valid comparison of dietary species richness among groups. To examine differences in monthly Shannon-Wiener dietary diversity indices among groups in continuous forest and forest fragments, we conducted a one-way analysis of variance (ANOVA) using the log transformed monthly values as replicas. To examine differences in monthly dietary dominance and evenness indices between continuous forest and fragment groups, we used a generalized linear model (GLM) with a quasibinomial error distribution and logit link-function as recommended for proportional data [79]. We also used a GLM with a quasibinomial error distribution and logit link-function to test for differences in between-month Morisita-Horn similarity indices among groups. We identified differences among groups by post hoc multiple comparisons using function 'glht' from R package multcomp [80]. We used a one-way ANOVA to test for differences in the percentage consumption of each food item and growth form between continuous forest and fragment groups. We applied logit transformations of proportion data before statistical analysis to normalize the data as recommended by Warton and Hui [81]. We used linear regressions to assess whether the availability of non-bamboo young leaves, bamboo young leaves, fruits, flowers, and bamboo shoots was a good predictor of their consumption in each study group.

Vegetation description and temporal variation in resource availability
The vegetation in the ranges of Bale monkey groups inhabiting forest fragments was more diverse (55 species) than in the ranges of groups in continuous forest (23 species) (Additional file 1). We found 24 tree, 14 shrub, 11 liana, 4 forb, 1 bamboo, and 1 fern species in the home ranges of fragment groups but only 12 tree, 2 shrub, 7 liana, 1 forb and 1 bamboo species in the ranges of continuous forest groups (Additional file 1). The ranges of the two continuous forest groups were much more similar in plant species composition and abundance (19 of 23 species shared, Morisita-Horn similarity index = 0.99) than the ranges of the two fragment groups (28 of 55 species shared, Morisita-Horn similarity index = 0.40).
Bale monkey foods were much more abundant in continuous forest than in fragments. Monthly food availability indices of bamboo young leaves (ANOVA: F = 544.00, df = 1, P < 0.001), non-bamboo young leaves (ANOVA: F = 17.17, df = 1, P < 0.001), and fruits (ANOVA: F = 4.19, df = 1, P = 0.05) were all significantly higher in continuous forest than in forest fragments (Fig. 2). Bamboo young leaves were abundant throughout the year in continuous forest, consistently available at low levels in Patchy fragment, and consistently scarce in Hilltop fragment. However, there was no difference in the availability indices of flowers (ANOVA: F = 1.44, df = 1, P = 0.243) and bamboo shoots (ANOVA: F = 0.88, df = 1, P = 0.357) between continuous forest and fragment groups.

Dietary species richness, diversity and similarity
Overall, at least 65 plant species (1 bamboo, 12 trees, 5 shrubs, 8 lianas, ≥ 25 forbs and ≥ 14 graminoids) were food sources for Bale monkeys. They also ate one species of mushroom and presumably many unidentified species of insects. Dietary species richness was much higher in groups inhabiting forest fragments (≥ 61 species: Patchy ≥ 47 species; Hilltop ≥ 35 species) than in groups inhabiting continuous forests (12 species: Continuous A = 12 species; Continuous B = 8 species). The rarefaction curves for dietary plant species richness reached a plateau for each of the four study groups, suggesting we sampled intensively enough to obtain robust values for dietary species richness in all groups (Additional file 2).
The mean monthly Shannon-Wiener diversity index (H′) of food species was significantly higher in fragments than in continuous forest (ANOVA: F = 178.60, df = 1, P < 0.001; Fig. 3a). However, mean monthly dietary species evenness (J) was not significantly different between groups inhabiting fragments and those in continuous forest (GLM: F = 0.35, df = 1, P = 0.555; Fig. 3b). Lastly, mean monthly food plant species dominance was significantly higher for groups inhabiting continuous forest than for those in fragments (GLM: F = 163.60, df = 1, P < 0.001; Fig. 3c). Between-month dietary species similarity was significantly greater for groups in continuous forest than for groups in forest fragments (GLM: F = 380.80, df = 1, P < 0.001; Fig. 3d). Annual dietary species overlap was much lower between the two fragment groups (

Consumption of different growth forms
In forest fragments, a total of 10 tree, 1 bamboo, 5 shrub, 7 liana, 24 forb, 14 graminoid and 1 mushroom species were food sources for Bale monkeys whereas in continuous forest only 3 tree, 1 bamboo, 1 shrub, 4 liana, 2 forb, and 1 graminoid species were food sources for the monkeys. Groups in fragments spent less time feeding on bamboo ( (Table 1). Bothriochloa radicans was only a minor (< 1%) dietary species for the other study groups, though 5 other graminoid species were more commonly consumed than B. radicans by the group in Patchy fragment. Galiniera saxifraga, a tree, was the second most frequent food source in continuous forest (Mean = 6.6%) and Hilltop fragment (11.8%) and the third most frequent food source in Patchy fragment group (7.4%).

Dietary preference
The selection ratios of bamboo, tree, shrub, and liana food species accounting for > 0.5% of the annual diets of the study groups are presented in Table 2. Despite its dominance in the diets of the continuous forest groups, bamboo (Arundinaria alpina) had selection ratios of just below 1.00 in continuous forest (Continuous A = 0.94 and Continuous B = 0.95) owing to its extremely high stem density in this forest type. Although they ate much less bamboo, the fragment groups also exhibited comparable selection ratios to Oleaceae Jasminum abyssinicum

Temporal variability in food item availability and consumption
Bamboo young leaf and shoot consumption were significantly correlated with availability over time in Continuous groups A and B and in Patchy fragment group (Table 3).
It is possible that similar relationships between these variables also existed in Hilltop fragment, but we did not track changes in bamboo abundance over time here because of the low density and small sizes of individuals of bamboo at this site. The consumption of fruits and flowers were also significantly correlated with availability for both groups inhabiting continuous forest and fruit consumption was significantly correlated with availability for Hilltop fragment group (Table 3).

Dietary responses to habitat degradation by Bale monkeys compared to other primates
Habitat degradation affects plant species richness, diversity and structure in forest fragments, ultimately reducing the availability of food resources for many primate species [48,82,83]. Specifically, the destruction or degradation of mature continuous forest promotes the growth in light gaps of pioneer species including fast-growing graminoids, forbs, shrubs, lianas and trees [9,44,[84][85][86]. In our study, Bale monkeys in fragments exploited many of these pioneer species (Table 1), broadening their diet Asteraceae   to include a far greater diversity of plant species (indigenous, exotic, and/or cultivated) and growth forms than conspecifics in continuous forest. Primates inhabiting fragments frequently eat a higher percentage of leaves than conspecifics in continuous forest [41,42,46,49]. Bale monkeys, however, ate a much lower percentage of leaves in fragments than in continuous forest largely because of the lower availability of bamboo in the former. In fragments, Bale monkeys modified their diet by increasing consumption of fruits, stems, petioles and insects as well as the leaves of a number of species other than bamboo. Interestingly, the much higher fruit consumption in fragments occurred despite fruit being significantly less available in fragments than in continuous forest.

Mikaniopsis clematoides
Another common dietary response to habitat degradation among primates is to consume more secondary successional species, including shrubs, forbs, or graminoids [39,[41][42][43][44]87]. The Bale monkeys in our study clearly fit this pattern, obtaining more than half their diet from shrubs, forbs, and graminoids in forest fragments (Additional file 3).
Primates in fragments also exhibit a tendency to consume exotic species and/or human crops from surrounding human matrix [46,47,88], a habitat absent from the ranges of conspecifics in continuous forest. Bale monkeys in both fragments in our study engaged in cropraiding, though the group in Patchy fragment, whose range included more areas of human use [66], had a diet containing a higher overall percentage of crops. Farmer responses to crop raiding by Bale monkeys included throwing stones, hunting with spears, chasing them with dogs, or positioning scarecrows in cultivated areas (Mekonnen, personal observation). In addition to crops, Bale monkeys in fragments also consumed bamboo planted near the homes of local people, triggering additional human-monkey conflict, particularly at Patchy fragment (Mekonnen, personal observation).
Lastly, the species richness of primate diets in fragments often differs from in continuous forests, increasing substantially for some primates (e.g., Alouatta pigra [48]; Cercopithecus mitis boutourlinii [49]), while decreasing for others (e.g., Ateles geoffroyi [41]; Propithecus diadema [42]). Bale monkeys appear to adopt the former approach, consuming many more plant-and probably insect-species in fragments. The strategy of continuous forest Bale monkeys to focus primarily on bamboo is simply not an option for monkeys in fragments where bamboo populations have been degraded or almost eradicated and the monkeys must diversify their diet to survive.

Dietary flexibility in Bale monkeys relative to other Chlorocebus species
Several of the Chlorocebus species are well-studied and eat varied diets with the top food item ranging from fruit in Nigerian (C. tantalus: [89]) and Senegalese (C. sabaeus: [90]) populations to gum or flowers in Kenyan populations (C. pygerythrus: [70,91,92]) (Table 4). Among Chlorocebus, Bale monkeys (C. djamdjamensis) are unique in their heavy reliance on the young leaves and shoots of bamboo in relatively undisturbed continuous forest habitats.
Intriguingly, our study revealed that C. djamdjamensis inhabiting fragments consumed diets more comparable to those of the other less specialized Chlorocebus species than to continuous forest-dwelling C. djamdjamensis populations. For example, percentages of fruit and graminoid consumption by C. djamdjamensis in fragments were similar to those reported for East African C. pygerythrus populations (Table 4). Further, levels of invertebrate consumption by the Hilltop group of C. djamdjamensis mirrored levels of invertebrate consumption by C. sabaeus in West Africa (Table 4). Lastly, C. tantalus's diet in West Africa was 2-3 times more species rich than the diets of C. djamdjamensis in continuous forest though actually somewhat less species rich than the diets of C. djamdjamensis in fragments (Table 4). Though the one dietary commonality among C. djamdjamensis groups in our study was a greater reliance on leaves than in any of the other Chlorocebus spp. (maximum 25% of the diet), consumption of leaves still varied widely among C. djamdjamensis groups.
The remarkable dietary flexibility exhibited by C. djamdjamensis in fragments has at least two possible explanations. First, they may retain some of the ancestral ecological flexibility characteristic of other members of the genus Chlorocebus, only expressing this plasticity when habitat degradation requires them to diversify their diets beyond primarily bamboo and a handful of other species. A second possibility is that genetic introgression (hybridization) between C. djamdjamensis and parapatric C. aethiops and C. pygerythrus in fragmented forest areas [57,60,93] endows some C. djamdjamensis populations with the ability to radically alter their diets in fragments.

Bamboo consumption across bamboo eating mammals
Adaptation to bamboo-dominated forests and diets appears to have evolved at least six times among the mammals: giant pandas in China [34,94], red pandas in India, Nepal, Bhutan, Myanmar, and China [69], bamboo lemurs (Hapalemur/Prolemur spp.) in Madagascar [26,95], Assamese macaques (Macaca assamensis) in China [68,96], golden monkeys in Uganda and Rwanda [67,97], and Bale monkeys in Ethiopia (this study; Table 5). Most of the primate taxa are members of ecologically-flexible genera (Macaca: [98]; Chlorocebus: [63]) or species (Cercopithecus mitis: [64,99]), while giant and red pandas belong to different more specialized families in the order Carnivora [69].w  Agmen et al. [89]   Tan [26] Macaca assamensis 12 Among the other bamboo-eating primates, the closest phylogenetically and geographically to Chlorocebus djamdjamensis is Cercopithecus mitis kandti (Table 5). Both taxa feed primarily on a single species of African highland bamboo (Arundinaria alpina) though C. mitis kandti rely on it less than C. djamdjamensis populations in continuous forest and more than C. djamdjamensis populations in fragmented forest ( [54,100]; This study). Giant and red pandas are arguably the best known obligate specialist folivores, exploiting diets consisting almost entirely of bamboo [34,94]. Neither species exhibits an ability to cope with intensive habitat degradation [34,94]. Among primates, some bamboo lemurs appear to be the most inclined towards extreme specialization [26]. In particular, the greater bamboo lemur (Prolemur simus) consumes a diet of 95% bamboo [26] and does not appear to exist outside of bamboo forest habitat [101,102]. P. simus also relies heavily on an unusually cyanogenic bamboo species [95] and is probably the only 'obligate specialist' on bamboo among the bamboo-eating primates. Indeed, recent studies of several other bamboo lemurs (Hapalemur spp.) found they can survive in habitats without bamboo, consuming more species-rich diets in these habitats, including a high percentage of graminoids in the cases of H. alaotrensis [103] and H. meridionalis [36]. The increased consumption of graminoids by these Hapalemur spp. provides an interesting parallel to the Bale monkeys in our study, which also consumed more graminoids at fragmented sites where bamboo is scarce. Overall, it appears that, with the exception of Prolemur simus, bamboo eating primates are more dietarily flexible than giant and red pandas. This pattern is consistent with the evidence that the bamboo feeding adaptation in pandas is much older than it is for any of the bamboo feeding primates (e.g., [69,[93][94][95]).

Implications for conservation and management
Our study revealed that, like most other bamboo-eating primates, Bale monkeys have the flexibility to cope with changes in the identity and abundance of foods resulting from habitat degradation and loss of bamboo, at least over the short-term. More intensive longterm studies of Bale monkeys in both fragmented and continuous habitats are, however, needed to examine and address some of the potential drawbacks of life in fragments. The greatest conservation concern raised by our study is that of human-monkey conflict at fragmented sites, especially at Patchy fragment. As in many other sites where primates crop raid [104], humans near fragments in our study sometimes responded to Bale monkey crop raiding in a manner that put Bale monkeys at risk, hunting them with spears and dogs. A more detailed study of this human-monkey conflict and its impact on Bale monkey survivorship in fragments should be a priority along with developing and implementing strategies to mitigate this conflict [105]. Any Bale monkey habitat restoration programs undertaken at fragments should focus on increasing fragment sizes, minimizing edge effects, incorporating matrix habitats into management plans, and mitigating human monkey-conflict (cf., [88,106]). Moreover, the remaining continuous bamboo forest habitat in the southern Ethiopian Highlands should be protected from further deforestation both to best ensure the long-term persistence of Bale monkeys [93] and to prevent the functional homogenization of biodiversity in this important region for conservation [19,107].

Conclusions
Bale monkeys in fragments have smaller group sizes, and experience lower food availability and habitat quality relative to those in continuous forest ([66]; This study). Consequently, they consume more diverse species-rich diets, including more secondary and cultivated food resources. While Bale monkeys are the only specialized members of a genus, Chlorocebus, whose other five species are all ecological generalists, we hypothesize that they have either retained the ancestral Chlorocebus ability to fall back on a generalist diet where necessary or that populations in fragments have reacquired this ability through interbreeding with parapatric grivet (C. aethiops) or vervet (C. pygerythrus) populations. Despite the encouraging dietary flexibility documented among Bale monkeys in our study, the long-term conservation prospects for populations in forest fragments remain unclear and will require long-term population monitoring and conservation actions to ensure their persistence in the southern Ethiopian Highlands.

Additional files
Additional file 1. Stem density of all plant species (≥ 2 m tall) within vegetation quadrats in the home ranges of study groups. Continuous A (n = 9110 stems), Continuous B (n = 5410 stems), Patchy (n = 3388 stems) and Hilltop (n = 2312) groups (* exotic species).
Additional file 2. Sample based rarefaction curves of plant species consumed by Bale monkeys among four study groups. Samples were collected in the continuous forest (Continuous A, N = 52 days (A); Continuous B, N = 54 days (B) and forest fragments (Patchy fragment, N = 62 days (C); Hilltop fragment, N = 67 days (D). The red (rarefaction) curves represent the cumulative number of plant species consumed by the study groups and blue curves represent the 95% confidence intervals.
Additional file 3. The proportion of feeding records devoted to consuming different plant growth forms by the four study groups. Proportions were summarized from N = 12 months, mean ± SE.