Introduction

Dietary switching, i.e., feeding on alternative resources, is the most common primate response to spatial and temporal variations in food availability (Hemingway and Bynum 2005). Dietary flexibility has been defined “…as the capacity to adjust digestive strategy according to the chemical and structural quality of the foods available” (Chapman et al. 2002, p. 344). Therefore, such flexibility is determined primarily by anatomical and physiological adaptations, e.g., mechanisms of mastication and digestion (Chivers 1994; Kinzey 1992; Lambert 1998; Rylands 1993). However, even when anatomical specializations are present, many species still show greater variability in diet than what would be expected by their gut adaptations (Chapman and Chapman 1990). Comparative analyses between different taxa with distinct anatomical adaptations have been performed to understand the degree of dietary flexibility of primates (Milton 1998; Simmen and Sabatier 1996; Tsuji et al. 2013; Yamagiwa and Basabose 2006). For example, cercopithecines with simple digestive tracts exhibit greater dietary diversity and more frequent dietary switching than do colobines, which have specialized stomachs for fiber digestion (Lambert 2002).

In Madagascar, climatic seasonality and unpredictability caused by tropical monsoons (Dewar and Richard 2007; Jury 2003) strongly influence phenological patterns and, thus, the food available to lemurs (Bollen and Donati 2005; Ganzhorn et al. 1999; Wright 1999). Moreover, the climate varies dramatically across the island, particularly between the evergreen rain forests of the east and northeast coasts and the dry deciduous and xerophytic forest formations of the west and south (Jury 2003). One of the most successful primate radiations on the island is the true lemur group (Eulemur spp.), as shown by the presence of one or two species in all forested habitats (Tattersall 1982). Previous field studies focused on the temporal variations in feeding patterns of Eulemur spp. at single sites (Andrews and Birkinshaw 1998; Andriamaharoa et al. 2010; Birkinshaw 1995; Curtis 2004; Donati et al. 2007a; Freed 1996; Overdorff 1993; Rasmussen 1999; Sato et al. 2014; Tarnaud 2006; Vasey 2000), whereas only a few studies have explored regional variations among species or populations (Donati et al. 2009; Johnson 2006; Ossi and Kamilar 2006; Tattersall and Sussman 1998). Based on these studies, the Eulemur complex is identified as semifrugivorous with relatively greater dietary flexibility than the mainly frugivorous Varecia spp., which are found only in the eastern humid forests (Balko and Underwood 2005; Britt 2000; Vasey 2000). Generally, diets of Eulemur tend to be higher in fruit in the eastern region (Donati et al. 2007a; Overdorff 1993), whereas a higher degree of folivory has been recorded in some western populations (Colquhoun 1997; Curtis et al. 1999; Sussman 1977). Thus, overall this genus is considered very flexible, and its capacity for dietary switching is often mentioned as one of the main traits that contributed to the success of the Eulemur radiation (Ossi and Kamilar 2006).

However, during periods of fruit scarcity, Eulemur adopts several behavioral strategies to pursue fruit, an approach that dietary shifters do not usually rely on (Colquhoun 1993; Donati et al. 2007a; Overdorff 1993). Among the strategies observed have been prolonged feeding activities over a 24-h period without shifting food categories (Donati et al. 2007a, 2009); increased ranging efforts (Overdorff 1993; Sato 2013a; Volampeno et al. 2011); reduced cohesion, or fission, of large groups to mitigate scramble competition (Colquhoun 1993; Donati et al. 2011; Freed 1996; Overdorff and Johnson 2003); and, in several Eulemur species, a seasonal shift to an “energy-minimizing strategy” in which animals modify behavioral patterns to minimize energy expenditure (sensu Albert et al. 2013; Campera et al. 2014) by prolonging resting (E. macaco: Colquhoun 1993, 1997) or decreasing daily traveling (E. collaris: Campera et al. 2014). In primates, such flexible and dynamic behavioral strategies often compensate for limited dietary flexibility (limited physio-anatomical flexibility). Thus, there is an apparent inconsistency between traditional knowledge about the dietary flexibility of Eulemur and their observed behavioral flexibility in pursuing primary food resources.

One way to clarify the dietary flexibility of Eulemur is to compare this group with another lemur group that successfully radiated in Madagascar, Propithecus, which is traditionally considered more specialized in terms of its physio-anatomical adaptation for digestion (Campbell et al. 2000; Hill 1953; Richard 1977). The genus Propithecus has extensive molar crests for fracturing leaf material and masticating seeds (Yamashita 1998) and a specialized gut with high fiber digestibility via microbial fermentation (Campbell et al. 2000, 2004b; Hill 1953). Eulemur and Propithecus have been thought to fill similar dietary niches as those filled by cercopithecines and colobines in Africa and Asia (Hemingway and Bynum 2005). Lambert (2002) clarified the distinctive feeding strategies of cercopithecines as frugivore generalists that often consume alternative food and those of colobines as folivore specialists that exhibit a lower frequency of dietary switching. However, there is a paucity of such comparisons between Eulemur and Propithecus, and those that are available are often limited to a descriptive level that does not control for both phylogenetic and ecological factors.

In addition, dietary composition and flexibility may be linked to body mass as suggested by two hypotheses proposed in previous studies (Lehman et al. 2005; Ravosa et al. 1993). The “resource quality hypothesis” predicts that food quality will negatively scale with adult body size in mammals (Ravosa et al. 1993), because larger bodied species have lower energy requirements per unit weight and can feed on low-quality foods (Janson and Chapman 1999; Kay 1984). The “resource seasonality hypothesis” predicts that high seasonal fluctuations of food resource availability will produce strong selective pressures for smaller body size (Lehman et al. 2005; Terborgh and van Schaik 1987).

In this article, we examined the differences between Eulemur and Propithecus in terms of dietary composition and dietary flexibility by assessing the role of environmental condition and body mass, while controlling for phylogenetic relatedness. Based on the dietary niches of cercopithecines and colobines (Lambert 2002), and the two hypotheses on the relationship between diet and body mass (Lehman et al. 2005: Ravosa et al. 1993), we predicted the following:

  1. 1)

    Because Eulemur consumes primarily fruit that is often only seasonally available, the dietary composition of Eulemur should vary more dramatically across seasons than does that of Propithecus.

  2. 2)

    Based on the resource quality hypothesis, populations of Eulemur with higher body mass should exhibit lower proportions of fruit-eating (our proxy of resource quality) than smaller bodied populations. However, such correlations may not be found in Propithecus because we assume that they can use fibrous leaves as an energy source. Alternatively, the opposite prediction may hold: that Propithecus populations with higher body mass eat higher proportions of fruit because large-bodied species tend to be found in habitats that are rich in food (Albrecht et al. 1990).

  3. 3)

    Based on the resource seasonality hypothesis, populations of Eulemur with smaller body mass should exhibit higher seasonal variation in fruit-eating (our proxy of resource seasonality). However, the relationship between body mass and seasonal variation in fruit-eating may not be found clearly in Propithecus because we assume that the diet of Propithecus is mainly and stably composed of leaves across seasons. Alternatively, the opposite prediction should hold: that populations with higher body mass show large seasonal variation in diet because they may be more able to cope with periods of shortage (Lindstedt and Boyce 1985).

  4. 4)

    In general, the diets of Eulemur tend to be higher in leaves in the dry habitats than in the humid habitats (Ossi and Kamilar 2006). This tendency suggests that Eulemur living in drier habitats switches the diet more often from fruit to leaves, and therefore, they will show more dietary variation. Moreover, if this tendency is caused by the geographical variation of fruit abundance in the habitats, the diets of Propithecus may also be higher in fruit in wetter habitats and higher in leaves in drier habitats.

Methods

Dataset on Dietary Composition

To collect data on the diets and feeding behaviors of Eulemur and Propithecus, we searched the published literature found by Google Scholar (http://scholar.google.co.jp/) using the key words diet, feeding, food, foraging, lemur, Eulemur, and Propithecus. We also added data from seven dissertations and three unpublished datasets (E. collaris in Mandena, Campera; E. collaris × E. rufifrons in Berenty, Donati; and P. candidus in Marojejy, Patel). We followed the current taxonomy of lemurs (Markolf and Kappeler 2013; Mittermeier et al. 2008; Rumpler et al. 2011).

We restricted our data collection to studies conducted for a minimum of 10 mo to account for seasonal effects and to avoid overestimation of temporal tendencies (Tsuji et al. 2013). To standardize the inconsistent categorization of dietary items in the literature, we reclassified food items into the following four categories:

  1. 1)

    Fruits, including ripe fruits, unripe fruits, ripe seeds, and unripe seeds

  2. 2)

    Leaves, including mature leaves, immature leaves, and petioles

  3. 3)

    Flowers, including flowers, nectar, and flower buds

  4. 4)

    Other, including bark, stems, lichens, fungi, animal materials, soil, and unknown matter

We then calculated the percentages of feeding time spent by each population on the four food categories during each month. We acknowledge that time spent feeding is not the best measure to examine food intake (Chivers 1998; Kurland and Gaulin 1987). However, because long-term data on food intake are not commonly available in these two genera (cf. Curtis 2004; Irwin et al. 2014; Meyers 1993; Powzyk and Mowry 2003; Yamashita 2008), we analyzed the data on feeding time to maximize the number of populations as done in previous meta-analyses (Hemingway and Bynum 2005; Ossi and Kamilar 2006; Tsuji et al. 2013). If several independent datasets from multiple groups were available for a single site, we calculated the mean value from all groups, excluding groups with a large portion of “unknown/unidentified materials.”

For the populations for which monthly data on dietary composition were available, we calculated the coefficients of variation (CV) relative to the monthly percentages of a food category and multiplied them by 100. We used this value as an index of dietary flexibility (Hemingway and Bynum 2005). Because we defined dietary flexibility as the ability to shift a diet from the major food category to other food categories, CVs were calculated for the primary major food category of each population. The primary major food was defined as the category with the highest mean monthly feeding time percentage.

Predictors of Dietary Composition and Dietary Flexibility

To examine the factors that may affect the composition of lemur diets, we gathered environmental and morphological data from the original publications or from other publications conducted at the same site. For the environmental variables, we collected data on annual rainfall (mm); the number of dry months, i.e., those months with <100 mm of rainfall over the year; and the mean annual temperature (°C). When data for mean temperature were not available in the original publication, we extracted the annual mean temperature based on data from 1950 to 2000 obtained from the database WorldClim (http://www.worldclim.org/) using the coordinates of the study site.

We collected data regarding the body mass of each population from the original publications when available, or from the database All the World’s Primates (Rowe and Myers 2011). Because significant sexual dimorphism is lacking in Eulemur and Propithecus (Jenkins and Albrecht 1991; Kappeler 1990), we used mean body mass, including both males and females.

Statistical Analysis

We examined the effects of environmental variables, body mass, and their differences between genera on 1) the percentages of feeding time spent on each food category (fruit, leaves, and flowers); 2) the dietary flexibility (CV) of the primary food category; and 3) the dietary flexibility of each food category. Because related species should not be assumed to be independent data points (Ives and Zhu 2006), we performed phylogenetic generalized least squares (PGLS) analysis to test these associations in a phylogenetic context (Grafen 1989). The PGLS method explicitly incorporates the expected covariance among species into a statistical model fit by generalized least squares. We downloaded the consensus phylogenetic tree from the 10kTrees phylogenies project (version 3; http://10ktrees.nunnlab.org/Primates/downloadTrees.php) and modified it to include multiple populations of the same species as polytomies by assuming a branch length of 0.01 (negligible compared with the branch lengths among other species in the tree: interquartile range = 1.03–3.70). To the best of our knowledge, there is no existing phylogeny including Propithecus candidus as a separate species; therefore, we considered this taxon as the sister species of P. diadema based on the molecular biological evidence in Mayor et al. (2004). The hybrid species E. collaris × E. rufifrons in Berenty was considered as E. rufifrons based on a morphological criterion because this population does not differ from individuals of the latter species in terms of body measurements (Donati, pers. obs.). Because all identified independent variables were possible predictors of diet composition and dietary flexibility, we used a minimum adequate model approach to select a limited number of significant independent variables in our models (Crawley 2012). We started from a model including all independent variables plus their interactions with the genera and followed a stepwise procedure that involved individually excluding the variables with the highest P-values, starting from higher order terms. Once we obtained a set containing only significant predictors (P < 0.05), we tested whether the excluded variables became significant in the final model and, if so, retained them. The PGLS was performed using the R package caper (Orme et al. 2013) in R 3.0.3 (R Development Core Team 2014).

Results

Dietary Variations in Eulemur and Propithecus

We gathered data on the monthly dietary composition of 10 populations of Eulemur (7 species and 1 hybrid) (Table I, Fig. 1) and 7 populations of Propithecus (5 species) (Table I, Fig. 2). Fruit-eating comprised large proportions of the diets of the Eulemur species (mean ± SD, 75.6% ± 10.8%), whereas leaves (15.0% ± 8.6%), flowers (6.7% ± 4.8%), and other (2.9% ± 1.9%) constituted minor parts (Fig. 3a). Propithecus tended to consume fruits (37.3% ± 14.7%) and leaves (49.2% ± 12.1%) in more equal proportions, whereas flowers (10.7% ± 4.4%) and other (2.7% ± 2.7%) comprised small parts of their diet (Fig. 3a).

Table I Environmental variables, body mass, dietary composition, and dietary flexibility in each study population of Eulemur and Propithecus.
Full size table
Fig. 1
figure 1

Monthly dietary composition of the populations of Eulemur included in this study. The location of each population is plotted in the lower right map. Months with no data are left blank. If multiple data are available for the same month, mean values are presented.

Full size image
Fig. 2
figure 2

Monthly dietary composition of the populations of Propithecus included in this study. The location of each population is plotted in the lower right map. Months with no data are left blank. If multiple data are available for the same month, mean values are presented.

Full size image
Fig. 3
figure 3

Dietary composition (a) and logarithm of the coefficients of variation (b) for the main food eaten by Eulemur and Propithecus. Fr = fruit, L = leaves, Fl = flowers, Pr = primary major food.

Full size image

In terms of dietary variations, Eulemur had small CVs for fruits (mean ± SD, 21.6 ± 7.0), whereas those for leaves (80.2 ± 39.9) and flowers (136.8 ± 46.3) were large (Fig. 3b). Propithecus tended to have small CVs for leaves (37.5 ± 10.1) and large CVs for fruits (62.4 ± 32.2) and flowers (103.7 ± 25.4) (Fig. 3b). All populations of Eulemur consumed fruit as the primary major food type, whereas only three of seven populations of Propithecus consumed fruits as their primary major food type; the other four Propithecus populations primarily consumed leaves (Table I). The mean CVs for the primary major food were 21.6 ± 7.0 for Eulemur and 36.5 ± 12.1 for Propithecus (Fig. 3b).

Predictors of Diet and Dietary Flexibility

The model for the proportion of fruit-eating in the diet indicated a strong effect of genus, with Eulemur more frugivorous than Propithecus. In addition, fruit consumption increased with increasing body mass in both genera. Propithecus consumed more leaves than Eulemur, and leaf consumption decreased with increasing body mass. No significant differences between genera or effects of other predictors were detected with regard to the proportion of flower eating (Table II).

Table II Minimum adequate models (phylogenetic generalized least-squares) predicting dietary composition and dietary flexibility of Eulemur and Propithecus
Full size table

The variation in the consumption of the primary major food was significantly higher in Propithecus than in Eulemur. It increased with the number of dry months in Eulemur, though no such pattern was detected in Propithecus. The variation in fruit eating increased in populations living in drier habitats and was significantly higher in Propithecus. In contrast, the CVs for leaf eating were higher in Eulemur than in Propithecus, with the former showing more seasonal variation and the latter consuming leaves more consistently over the year. The variability in flower consumption decreased with temperature in both genera. It also increased with the number of dry months in Eulemur, whereas it did not increase in Propithecus. Finally, flower consumption also scaled differently with body mass in the two genera: it decreased with body mass in Propithecus but not in Eulemur (Table II).

Discussion

Dietary Flexibility

The genus was the strongest predictor of differences in the proportions of frugivory and folivory and of seasonal variation in the populations examined. Populations of Eulemur relied heavily on fruits as their primary food throughout the year. The diet of Propithecus appeared more diverse, although the proportions of the major foods in their diet varied dramatically over the year. Therefore, once ecological variables and interspecific differences were controlled, the dietary flexibility of Propithecus was far greater than that of Eulemur, suggesting that the simple digestive system of Eulemur allows for only limited dietary shifts, while the specialized digestive tract of Propithecus does not constrain dietary shifts. This result is in contrast to our prediction 1, although similar dietary differences between genera have been observed between E. collaris × E. rufifrons and P. verreauxi in Berenty (Simmen et al. 2003). Our data indicate different dietary strategies than what has been observed in cercopithecines and colobines (Lambert 2002), thus suggesting that these two anthropoid subfamilies are not good analogues for lemurids and indriids.

To explain these dietary differences between Eulemur and Propithecus, we must reconsider the anatomical and physiological adaptations of the two genera. Eulemur possesses a simple digestive tract (Campbell et al. 2000; Hill 1953; Schwitzer 2009) that allows for a much shorter digestive time than that of Propithecus. In Eulemur, the time to the first appearance of food in feces after feeding is 1.6–3.3 h (Campbell et al. 2004a; Overdorff and Rasmussen 1995), while in Propithecus it is 24.5 h (Campbell et al. 2004a) and in cercopithecines between 16.6 and 31.5 h (Lambert 1998). As in other frugivorous primates, a nonspecialized digestive system leads these lemurs to maximize their intake of ripe fruit rich in soluble carbohydrates and rapidly eliminate indigestible materials (Lambert 1998; Schwitzer 2009). However, E. fulvus exhibits a tooth morphology similar to that of folivorous lemurs (Boyer 2008) and possesses better masticatory effectiveness and fiber digestibility than Varecia variegata (Campbell et al. 2004b; Overdorff and Rasmussen 1995). The combination of these factors may enable Eulemur to use alternative foods, thus showing more dietary flexibility than Varecia (Vasey 2000) despite their overall adaptations to frugivory.

In contrast to Eulemur, Propithecus has several anatomical characteristics specialized for the digestion of fibrous materials, such as molars with long crests (Yamashita 1998), a large intestine developed for bacterial fermentation, and an elongated small intestine for nutrient absorption (Campbell et al. 2000, 2004b; Hill 1953). For example, P. coquereli exhibits almost twice the fiber digestibility values reported for E. fulvus [digestibility of neutral detergent fiber/acid detergent fiber (NDF/ADF): 42/22% vs. 60/47%; Campbell et al. 2004b]. However, Propithecus spp. are more of an ecological generalist than is Indri indri, a closely related folivore specialist (72%–81% of their overall diet comprises leaves) (Britt et al. 2002; Powzyk and Mowry 2003). Considering the specific anatomical traits of Propithecus, such as their relatively smaller large intestines, longer small intestine, and faster gut passage times than those of Indri indri (Campbell et al. 2000; Hill 1953), Powzyk and Mowry (2003) suggested that Propithecus may adopt a feeding strategy that maximizes the intake of carbohydrates and fats from high-quality foods rather than an alternative strategy that maximizes the efficiency of fiber digestion. Such a digestive strategy also differs from that of colobine monkeys. In fact, foregut-fermenting colobines exhibit greater fiber digestibility (NDF/ADF: 77/80%) (Edwards and Ullrey 1999) but narrower dietary flexibility compared with that of hindgut-fermenting primates (Lambert 2002; Milton 1998). Moreover, the molars of Propithecus, which have specializations for fracturing leaves, are also well suited for seed consumption (Yamashita 1998). Therefore, such anatomical and physiological adaptations may be related to the significant dietary flexibility observed in Propithecus.

Ecological Factors

The general principle that larger species tend to consume lower energy food is consistent with the differences in body mass and diet between Propithecus and Eulemur, with the former being heavier (range, 2803–6503 g) and more folivorous than the latter (range, 1220–2408 g). Interestingly, our results showed the opposite tendency at the intrageneric level. In accordance with our alternative prediction (prediction 2), populations with higher body mass exhibited higher proportions of fruit eating and lower proportions of leaf eating. Thus, dietary quality in each habitat can be one of the constraints on body size of populations within a narrow range of taxa with similar physio-anatomical adaptations (Albrecht et al. 1990). However, our prediction 3 based on the resource seasonality hypothesis was not supported because the PGLS models did not select body mass as a predictor for CV of each food in both genera. Nevertheless, ecogeographic variation in the size of Propithecus is related more to resource seasonality than to quality; i.e., sifakas in habitats with less resource seasonality have higher body mass (Lehman et al. 2005). Our analysis supports the notion that P. verreauxi living in highly seasonal habitats actually have smaller body mass (Kirindy in Lewis and Kappeler 2005; Beza Mahafaly in Richard et al. 2000) and consume more leaves and less fruit (Kirindy in Lewis and Kappeler 2005; Beza Mahafaly in Yamashita 2008). As in the case of Alouatta spp., which also possess a specialized digestive tract (Espinosa-Gómez et al. 2013), primates using hindgut fermentation reduce their body mass as their level of fruit consumption decreases. However, to fully understand ecogeographic size variation, it is important to consider other possible causative factors, such as Bergmann’s rule, in which body mass increases with decreasing ambient temperature, e.g., Microcebus (Lahann et al. 2006) and Eulemur (Gordon et al. 2015), or the energetic equivalent rule that predicts an inverse relationship with population density, e.g., Indriidae (Lehman 2007).

Eulemur living in habitats with longer dry seasons and higher maximum temperatures have been observed to consume higher proportions of leaves (Ossi and Kamilar 2006), although such a relationship was not detected in our analysis (see also Donati et al. 2015). Some populations of Eulemur inhabiting arid habitats exhibit an increase in folivory during the dry season (Ankarafantsika in Sato et al. 2014; near Kirindy in Sussman 1977). However, because fruit availability does not necessarily diminish during the dry season in the deciduous forests of western Madagascar (Rasmussen 1999; Sato 2013b; Sorg and Rohner 1996), some populations also living in the dry region do not exhibit such extreme seasonality in folivory (Ambato Massif in Colquhoun 1997; Anjamena in Curtis 2004; Ankarafantsika in Rasmussen 1999; Berenty in Simmen et al. 2003). Nevertheless, as we hypothesized in our prediction 4, populations of both genera living in habitats with less rainfall exhibited larger seasonal variation in fruit eating. This tendency is also suggested from the model for CVs of primary food: longer dry seasons increased seasonal variation of Eulemur’s primary food (= fruit). Despite stable phenological availability over the year, fruits in the dry forest may not always be a nutritive resource for Eulemur spp. as, for example, they may contain high levels of fiber (Bollen et al. 2005). During these time windows it might be more advantageous for animals with simple digestive tracts to switch to flowers or young leaves.

Similarly, long dry seasons may affect the timing and production of flowering, and the seasonal availability of flowers to some populations may have caused the highly variable flower consumption in Eulemur. Interestingly, the CVs of flower eating decrease with increasing body mass in Propithecus. This is likely driven by P. diadema in Tsinjoarivo, a small-bodied population living in humid forests; this population depends heavily on mistletoe flowers (Bakerella clavata) as a fallback food when preferred foods are scarce (Irwin 2008).

Behavioral Flexibility and Feeding Strategies

The results of our dietary niche meta-analysis strongly indicate that Eulemur spp. should be considered fully frugivores with low levels of dietary flexibility, whereas Propithecus spp. exhibit the dietary profile of generalists, with high levels of seasonal flexibility. This conclusion is also supported by increasing evidence of the behavioral strategies used by Eulemur to avoid dietary shifts. The Eulemur spp., which are all cathemeral (Donati et al. 2015), have been observed to seasonally expand feeding activities from a 12-h cycle to a 24-h cycle to meet energy requirements when feeding on low-quality food (Engqvist and Richard 1991; E. collaris: Donati et al. 2007a, 2009; E. collaris × E. rufifrons: Donati et al. 2009; E. fulvus: Tarnaud 2006; cf. Curtis et al. 1999). During lean periods when fruits are still available, Eulemur tends to increase its daily path length to visit patchily distributed fruiting trees (E. flavifrons: Volampeno et al. 2011), and they sometimes expand their ranging areas to include food patches very far from the boundaries of their usual home ranges, a behavior known as habitat shifting (E. collaris: Campera et al. 2014; E. rufifrons: Overdorff 1993; E. fulvus: Sato 2013a). In addition, Eulemur spp. have been frequently observed to use a fission–fusion strategy and become less cohesive during periods with low fruit availability (E. macaco: Colquhoun 1993; E. coronatus: Freed 1996; rare case for E. rufifrons: Overdorff and Johnson 2003). This flexible social system is also used in degraded forests when the animals are faced with small, low-quality food patches (E. collaris: Donati et al. 2011). Thus, Eulemur appears to adopt a “power-feeding strategy,” in which animals process a large volume of food per unit time to meet nutritional requirements (Donati et al. 2007a), or an “energy-maximizing strategy,” in which animals modify behavior to maximize energy income during resource scarce periods (sensu Albert et al. 2013; Campera et al. 2014), even increasing energy expenditure for pursuing food (see also “high-cost and high-yield strategy” in Agetsuma and Nakagawa 1998). This strategy enables them to compensate for their nonspecialized physio-anatomical adaptations with highly flexible activity, ranging and grouping patterns that maximize food acquisition. In some cases, Eulemur may adopt an “energy-minimizing strategy” during resource scarce periods. E. collaris in Mandena decreased daily path lengths in the lean seasons (Campera et al. 2014). E. macaco at Ambato Massif reduced diurnal activities without increasing nocturnal activities during the dry season, when they could consume fruits of several introduced plants (Colquhoun 1993, 1997, 1998). Thus, Eulemur probably do not always adopt an energy-maximizing strategy, but they appear to change their feeding strategy in a flexible manner depending on the situation with regard to fruit resources. Such behavioral flexibility adopting energy-maximizing and energy-minimizing strategies is likely to be the key reason behind the expansion of this genus into a wide variety of habitats across Madagascar (Campera et al. 2014; Donati et al. 2011; Ossi and Kamilar 2006).

Unlike Eulemur, Propithecus reduces activity levels to conserve energy during the dry season, when they consume lower quality food (P. verreauxi: Norscia et al. 2006; P. coronatus: Pichon et al. 2010) and/or increase daily feeding time on low-quality fallback food (P. verreauxi, Yamashita 2008). However, although they exhibit seasonal fluctuations in activity, Propithecus remains strictly diurnal (P. verreauxi: Erkert and Kappeler 2004). During periods of ubiquitous, low-quality food consumption focused on such resources as mature leaves, Propithecus often reduces its daily path length and/or ranging area (P. diadema: Irwin 2006; Powzyk 1997; P. tattersalli: Meyers 1993; Meyers and Wright 1993; P. verreauxi: Norscia et al. 2006; cf. P. edwardsi: Meyers and Wright 1993). Even in those cases in which the core areas within home ranges change seasonally, the location of ranging boundaries do not change dramatically throughout the year (Propithecus edwardsi: Gerber et al. 2012). In accordance with this behavioral pattern, Propithecus seems to adopt an “energy-conservation strategy” (sensu Wright 1999) or an “energy-minimizing strategy” (Albert et al. 2013; Campera et al. 2014) in which they rely on their ability to digest fibrous materials and minimize energy expenditure.

In conclusion, explorations of the evolution of feeding strategies should include comprehensive analyses of behavioral flexibility with regard to the pursuit of primary food within the ecological and phylogenetic frameworks of the species considered. Habitat shifting, for example, has been reported only in brown lemurs (Campera et al. 2014; Overdorff 1993; Sato 2013a; Scholz and Kappeler 2004), whereas fission–fusion has been more frequently observed in the E. macaco–E. coronatus clade (Colquhoun 1993; Freed 1996). Because these remarkable examples of flexibility have been reported only sporadically, meta-analyses performed at this stage cannot be conclusive. Considering the variation shown in multiannual phenological patterns of most Malagasy forests (Bollen and Donati 2005; Wright 1999) and the paucity of behavioral data spanning longer than a single year (Erhart and Overdorff 2008), our knowledge of lemur feeding strategies is likely to be incomplete.