Effects of forest fragmentation on the dietary ecology and activity of a nocturnal lemur community in North West Madagascar

Deforestation and habitat fragmentation is the primary threat to primate populations. The primates that live within degraded and anthropogenically disturbed habitats typical of fragmented landscapes have to cope with lower availability of resources in comparison to primates in continuous, undisturbed forests. While some species are sensitive to forest fragmentation, some evidence exists to suggest that primates can alter their behavior and adapt to such changes, which enables their survival in suboptimal habitat. In this study, we assessed how forest fragmentation and its associated edge‐effects impact the feeding ecology and activity levels of a nocturnal primate community in the Sahamalaza‐Iles Radama National Park, North West Madagascar. From March 06, 2019 to May 17, 2022, we collected data on tree and invertebrate phenology at our study site, and feeding ecology and activity for 159 lemur individuals from four species. Fruit and flower availability varied significantly between continuous and fragmented forest, and between forest core and edge areas, with continuous forest exhibiting higher continuous fruit and flower availability. Lemur feeding ecology varied significantly too, as the feeding niches of all four species were significantly different between continuous and fragmented forest and between core and edge areas. However, lemur activity levels were mostly consistent among all forest areas. The results of this study suggest that nocturnal lemurs are able to adapt their dietary ecology in response to the available food sources within their habitat. Due to this flexible ecology and dietary plasticity, the lemurs do not need to significantly alter their behavior in different environments to fulfill their dietary needs. While nocturnal lemurs demonstrate adaptability and flexibility to degraded habitat, it is unclear how far this plasticity will stretch considering that Madagascar's forests are still being cleared at an alarming rate. Urgent conservation action is therefore needed to ensure the future of lemur habitat.

Urgent conservation action is therefore needed to ensure the future of lemur habitat.

K E Y W O R D S
behavioral ecology, Cheirogaleidae, dietary plasticity, edge-effects, ethology, feeding niches, Sahamalaza

| INTRODUCTION
Deforestation is one of the primary threats to species survival globally, as more than half of the world's original tropical forest, an area of approximately 9 million km², has already been cleared (Giam, 2017;Whitmore, 1997).Much of the tropical forest that remains is also now severely fragmented, and large expanses of open, anthropogenic grasslands, savannahs, and pasture geographically separate many of the isolated forest fragments (Gascon et al., 1999;Laurance et al., 2011;Taubert et al., 2018).
Further, many of the remaining forest fragments are small and highly degraded with poor quality habitats and low ecological integrity (Coe et al., 2013;Corlett, 2000;Taubert et al., 2018).
Despite ongoing deforestation and fragmentation, tropical forests are still vital habitats for approximately half the world's terrestrial animal species (Bryant et al., 1997) and many animals have had to adapt their behavioral ecology to changing habitats to survive (Lima & Zollner, 1996;Norris et al., 2010).Forest fragmentation is a particular threat for primates (Schwitzer et al., 2017), as their occurrence often solely depends on forest connectivity and availability (Chapman et al., 2003;Estrada & Coates-Estrada, 1996).Additionally, primates often live in multispecies communities and sympatric species often respond and adapt differently to fragmentation (Link et al., 2010;Steffens & Lehman, 2018;Tutin et al., 1997).This makes primates ideal subjects in which to study the capability of animals to adapt their behavior to changes in their habitat structure (Onderdonk & Chapman, 2000), information that is critical to conserve and protect their remaining populations (Schwitzer et al., 2013(Schwitzer et al., , 2017)).
Due to the extent of deforestation within the tropics and the conservation concern for global primate populations, the behavioral and ecological responses of primates to forest fragmentation are already well-studied (Schwitzer et al., 2011).Responses to forest fragmentation are often species-specific among primates (de Almeida-Rocha et al., 2017;Steffens & Lehman, 2018), and some species demonstrate remarkable behavioral adaptability and plasticity to the fragmentation and degradation of their forest habitat (Strier, 2015).For example, some howler monkeys have been able to migrate between isolated forest fragments and quickly adapt to novel food items in new fragments (Richard-Hansen et al., 2000;Silver & Marsh, 2003).Many ecological generalist species with broad feeding niches are also not negatively impacted by forest fragmentation; howler monkey activity patterns in forest fragments mirror those in continuous forests (Cristóbal-Azkarate & Arroyo-Rodríguez, 2007), while bale monkeys dedicate less time to foraging in fragmented forest (Mekonnen et al., 2017).Positive ecological responses to fragmentation have also been reported in capuchins (Cunha et al., 2006) and mouse lemurs (Lehman et al., 2006) which can have higher abundances within fragmented forest and edge habitat.However, fragmentation-induced changes in habitat are often problematic for species with more-specialized ecologies and narrower feeding niches (Wolfheim, 1983).For example, significant changes in the diets of primarily-frugivorous primates, including howler monkeys, mangabeys and colobus monkeys have been observed between continuous and fragmented forest (Cristóbal-Azkarate & Arroyo-Rodríguez, 2007;Onderdonk & Chapman, 2000;Tutin, 1999).This finding has also been reported for folivorous species of lemurs that have subsequently diversified their diet in fragmented habitats (Dröscher & Kappeler, 2014;Irwin, 2008).Some howler monkey populations (Pozo-Montuy & Serio-Silva, 2007) and multiple macaque species (Menon & Poirier, 1996;Ruppert et al., 2018) have to increase their foraging effort in response to habitat fragmentation, while mantled howler monkeys (Clarke et al., 2002) and white-headed langurs (Huang et al., 2017) have exhibited reduced social activity in forest fragments.Further, lemurs of the Eulemur genus have responded to fragmentation-induced habitat degradation by increasing their nocturnal activity (Donati et al., 2016), while howler monkeys increase calling activity (Bolt et al., 2018).However, for many species, an understanding of their ability to adapt to fragmented habitat is not known.
The island nation of Madagascar is well known for both its highly impressive biodiversity and endemism (Ganzhorn et al., 2001;Goodman & Benstead, 2005), and also for its high rates of deforestation and forest fragmentation (Goodman et al., 2018;Harper et al., 2007;Vieilledent et al., 2018).Forest fragmentation in Madagascar is now a serious threat for the island's flagship group of animals, the lemurs, as the small body size of many species prevents them from crossing the large open spaces between isolated forest fragments (Craul et al., 2009;Olivieri et al., 2008).Further, more than 95% of lemur species are now classified as threatened on the International Union for the Conservation of Nature Red List (International Union for the Conservation of Nature, 2020; Schwitzer et al., 2013).In addition to sharp population declines and geographic range contractions (Ganzhorn et al., 2000;Hending, Sgarlata, et al., 2020), forest fragmentation has been observed to cause changes in the behavioral ecology of many lemur species (Seiler, 2012).As with other primate groups, the responses of lemurs to fragmentation are species-specific (Steffens & Lehman, 2018) and some species have had to increase foraging effort (Dinsmore et al., 2016;Gabriel, 2013), alter their activity patterns (Donati et al., 2016;Schwitzer, Kaumanns, et al., 2007) and stray from their regular diet (Donati et al., 2020;Seiler et al., 2014) to survive within fragmented landscapes.Conversely, some species have demonstrated high adaptability to forest fragmentation and degradation (Hending, 2021), and behavioral flexibility and plasticity of both diet (Donati et al., 2011;Eppley et al., 2017;Irwin, 2006) and activity (Cameron & Gould, 2013;Seiler et al., 2014) have been observed in multiple lemur species.While this is encouraging for the conservation of these threatened taxa, the way in which many lemurs survive in fragmented forests is still little known.
In this study, we investigated the effects of forest fragmentation and habitat degradation on the behavioral ecology of four nocturnal lemur species in the Sahamalaza-Iles Radama National Park (hereafter referred to as SIRNP), North West Madagascar.The studyspecies of this investigation were the northern giant mouse lemur (Mirza zaza), the Sahamalaza sportive lemur, (Lepilemur sahamalaza), the Sambirano mouse lemur (Microcebus sambiranensis), and the fattailed dwarf lemur (Cheirogaleus medius).We chose to study these four species as they all live sympatrically in SIRNP, they are all littlestudied and highly threatened by forest fragmentation (Blanco, Dolch, Ganzhorn, Greene, Le Pors, Lewis, Louis, et al., 2020;Blanco, Dolch, Ganzhorn, Greene, Le Pors, Lewis, Rafalinirina, et al., 2020;Randriatahina et al., 2020;Reuter & Schwitzer, 2020), and a detailed knowledge of how they respond to habitat fragmentation, degradation, and edge habitat is urgently needed for their conservation (Schwitzer et al., 2013).Our specific objectives were to investigate how the: (1) Feeding ecology of each species varies between continuous and fragmented forest, and between forest edge and core areas.
(2) Ethology (activity budget) of each species varies between continuous and fragmented forest, and between forest edge and core areas.
Based on the findings of studies focusing on lemurs (Seiler et al., 2014;Steffens & Lehman, 2018), we hypothesized that the behavioral responses of our four study-taxa to forest fragmentation would be species-specific, and we predicted that feeding ecology and activity would vary between continuous and fragmented forest, and between areas of forest edge and forest core.Due to the narrow feeding niche of sportive lemurs (Ganzhorn, 1993), we also predicted that differences in feeding ecology and activity would be most profound for L. sahamalaza between continuous and fragmented forest and between edge and core habitat.However, as lemurs of the Cheirogaleidae family are often highly resilient to deforestation and tolerant of degraded habitat (Hending, 2021;Kappeler & Rasoloarison, 2003;Knoop et al., 2018;Lehman et al., 2016), we predicted that these differences would be much less prominent in M. zaza, M. sambiranensis, and C. medius.

| Study site
SIRNP is a 26,000 ha protected area located between the latitudes of 14°04′S -14°37′S and the longitudes of 47°52′E -48°04′E in Madagascar's Sofia Region (Volampeno et al., 2011) (Figure 1).In addition to its designation as a National Park, SIRNP is also classed as a United Nations Educational, Scientific and Cultural Organization Biosphere Reserve (Rode et al., 2013) and, despite its protected status, it has undergone heavy habitat destruction and fragmentation in recent years (Seiler et al., 2014).
Although some isolated forests and matrices of gallery and scrub vegetation remain within the National Park, large areas of SIRNP are now characterized by anthropogenic savannah and grasslands (Volampeno et al., 2011).SIRNP is located in the Sambirano Domain, an area in the north west of Madagascar characterized by transitional subhumid forests, and SIRNP's vegetation is comprised of both deciduous and evergreen plants, of which many species are regionally endemic (Du Puy & Moat, 1998;Koechlin, 1972).The climate of SIRNP is hot, subhumid, and highly seasonal, with a humid wet season (November-April) and a distinctly cooler dry season (May-October) (Mandl et al., 2018).The mean temperature range of SIRNP is 20.6-32.0°C,with an extreme temperature range of 13.2-39.1°C(Hending, Holderied, et al., 2017;Hending, McCabe, et al., 2017b) and a mean annual precipitation of ~1600 mm (Schwitzer, Randriatahina, et al., 2007).
SIRNP is a well-suited location to further investigate the effect of forest fragmentation and edge-effects on the behavioral ecology of lemurs.This is first because SIRNP contains two remaining forests: the 1169 Ha Anabohazo forest, a continuous forest block that has undergone very little anthropogenic disturbance (Randriatahina et al., 2014), and the 976 Ha Ankarafa forest, a highly fragmented forest encompassed of many fragments of varying size, levels of degradation, and anthropogenic disturbance (Seiler, Holderied, et al., 2013a).Second, analyses of tree species diversity, tree size, and structural diversity (Hending, Randrianarison, Andriamavosoloarisoa, Ranohatra-Hending, Holderied, et al., 2023), and transects measuring the variation of microclimatic and abiotic variables (temperature, humidity, and light intensity) from the forest edge into the forest core (Mandl, 2018, unpublished thesis;Mandl et al., 2022) have already established the edge-core gradient of SIRNP's forest as 165 m.The Anabohazo and Ankarafa forests are separated by only ~25 km of grassland, and this provides an opportunity to study the behavioral responses of a lemur community to forest fragmentation on a localized scale, where environmental and habitat-related covariates can be easily controlled for, in comparison to other studies that have attempted to answer similar research questions over a much larger study area (e.g., Chaves et al., 2011;Cristóbal-Azkarate & Arroyo-Rodríguez, 2007).

| Tree phenology assessments
We conducted all fieldwork between March 06, 2019 to May 17, 2022, a study-period that covered both the wet and dry seasons in SIRNP (Mandl et al., 2018).To assess the seasonal variation in vegetation food availability, we established four 40 × 20 m (800 m²) vegetation plots (as in Haugaasen & Peres, 2005), two of which were situated within Anabohazo forest and the other two in Ankarafa forest.Within each forest, one plot was established in the forest core (>165 m from the edge) and one in the edge area (<165 m from the edge) of the forest, so that we could compare how food availability varied between continuous and fragmented areas, and between forest edge and core areas.The locations of each plot were assigned at the half-way point of a 1 km transect established within the core and edge areas of each forest, and each plot was established at a distance of 20 m from the transect line itself.We marked out all phenology plots at the beginning of the fieldwork period and we tagged and numbered all trees in the plots with a circumference at breast height (CBH) of ≥16 cm (diameter at breast height [DBH] ≥ 5 cm) with colored, biodegradable tape (which we replaced over time).For each month of the year, we visually inspected all of the marked trees within the plots for the presence of potential lemur food items (fruit, flowers, and leaves) on 1 day of the first week of each month, and we recorded the presence/absence of these food resources for each tree (as per Sato, 2013).

| Invertebrate phenology assessments
We also assessed the availability of invertebrates throughout the study period, as arthropods constitute a portion of the diet of cheirogaleid lemurs (Atsalis, 1999a;Lahann, 2007).To do this, we captured insects on a monthly basis at the four phenology plot sites, in parallel to the vegetation assessments.This was so that we could compare invertebrate availability between continuous and fragmented areas, and between forest edge and core areas.To ensure that we were able to capture specimens to represent the overall diversity of invertebrates at the site, both volant and terrestrial, we used three different sampling techniques (as in Dammhahn & Kappeler, 2008).
Second, we set a light trap that consisted of a 2 × 2 m white insectcapture sheet, suspended between two poles, upon which we shone a high-lumen light (Nitecore) and an ultraviolet lamp (Vanskytek) to attract photophilic insects, primarily of the orders Lepidoptera, Coleoptera, Mantodea, and Hemiptera (as in Gadagkar et al., 1989).
Our third and final method consisted of pitfall traps; we dug two  holes in the ground at each site and placed a plastic bucket (30 × 30 × 40 cm) within each hole to capture terrestrial invertebrates of the Orthoptera and Isoptera orders, and also annelids, arachnids, myriapods, and crustaceans (Southwood & Henderson, 2000).On capture nights, we set up all of the equipment in the afternoon and then left it for a period of 4 h (18:00-22:00) to capture specimens.All specimens were identified to their taxonomic order (where possible) using a reference textbook (Brusca et al., 2016).

| Behavioral observations
To collect data on the feeding ecology of each species, we conducted behavioral observations of lemur individuals three nights per week for the duration of the study period.At dusk, we located one random lemur individual (as in Harcourt, 1991;Nekaris, 2001) and followed this focal individual for a period of 6 h (18:00-00:00).We opted for this noninvasive approach rather than capturing, anesthetizing and marking specific individuals to follow with transponders so that our impact on the lemur's stress and physiological health was minimal.
Further, little evidence exists for sex-specificity in lemur diet and foraging behavior, so we did not need to be able to identify the individual identity or sex of our animals (e.g., Ganzhorn et al., 2004;Hemingway, 1999;but see O'Mara and Hickey, 2014).We spread our survey effort over the entire area of the Anabohazo forest and over the whole area of three fragments of Ankarafa (Ankarafa I [69.7 Ha], Ankarafa II [18.31 Ha],and Meleintena [61.93 Ha]) to minimize the chance of collecting data for multiple nights for individual lemurs.We also spread our survey effort for each species evenly throughout the study-period, as available plant and invertebrate food items would likely vary significantly throughout the year and between seasons.As M. sambiranensis was only confirmed to inhabit Anabohazo forest at the time of our study (Hending, Randrianarison, et al., 2022), we were not able to collect any data for this species from Ankarafa.We conducted our behavioral follows for the first half of the night only, as there is little qualitative evidence of differences in activity level and feeding behavior in nocturnal lemurs between the first and second halves of the night (Dammhahn & Kappeler, 2008;Mandl et al., 2018).
During the focal observations, we continuously observed the individual's feeding activity, and we recorded the time of all instances of feeding and the food item consumed.We categorized the food items as either fruit, leaves, flowers, gum, invertebrates or vertebrates.
We identified the species of any plant items consumed with the expertise of local guides and published literature (Schatz, 2001).We also identified the taxonomic order of any invertebrate prey items (when possible), and we identified any vertebrate prey items (reptiles and amphibians) with a field guide (Glaw & Vences, 2007).
In addition to focal observations of feeding activity, we also conducted interval sampling of activity over the course of the nightly follows to calculate an activity budget for each species.We observed the behavior of the focal individual and we recorded the specific activity at 5 minute intervals.As per Harcourt (1991), we categorized the behaviors at the intervals as either feeding, resting, locomotion, vocalizing, or grooming/social behavior.If we could not observe a specific behavior of an individual due to the lemur being in very dense vegetation, we noted the observation at that respective interval as "out of site."If we lost sight of an individual for a period of at least 30 min and we could not relocate the lemur, we ended the sampling for that night and marked the follow as incomplete.The same team of observers collected the data for all behavioral observations (two researchers and one local guide) to avoid any potential observational bias between different follows at different sites.All participants were equipped with headlamps (Tikka+; Petzl) and high-lumen hand-torches (EC20; Nitecore) to enable them to continuously observe the small-bodied, fast-moving lemurs among the dense vegetation.Over the course of the fieldwork, behavioral observation effort was allocated equally between species and between individuals in both the edge and core areas of the forests.
This was so that we could assess how activity and feeding ecology varied for all four species between continuous and fragmented forest, and between edge and core forest areas.

| Data analysis
To compare tree species composition and foliage availability between the four study areas (Anabohazo core, Anabohazo edge, Ankarafa core, Ankarafa edge), we used a data set of 9619 trees collected in parallel to this study (2019-2022: see Hending, Randrianarison, Andriamavosoloarisoa, Ranohatra- Hending, Holderied, et al., 2023, for details) and a χ 2 test.We used χ 2 tests to compare the availability of fruiting and flowering trees between the four study areas using the count data obtained from our tree phenology assessments.We additionally used χ 2 tests to assess whether leaf, fruit, flower, and invertebrate availability varied significantly throughout the year at both the core and edge areas of each forest, using monthly count data from our phenological assessments.We also used χ 2 tests to compare how the diet of each lemur species varied between the four study areas (following Dammhahn & Kappeler, 2008;Lahann, 2007;Thorén et al., 2011;Wright et al., 2011).To investigate feeding niche overlap between the four study areas within species, we used the package "EcoSimR" (Gotelli et al., 2015) to calculate Pianka's indices (O jk , Krebs, 1998) via the formula: where p ij is the proportion food item i is of the total food items consumed by species j, and p ik is the proportion food item i is of the total food items consumed by species k.We randomized the observed feeding data over 1000 simulations, and we determined the statistical significance of observed niche overlap values via confidence intervals which we compared with null models.We compared feeding niche breadth between the four study areas for each species by calculating standardized Levin's indices (B A , Feinsinger, et al., 1981), using the formulas: where B is Levin's measure of niche breadth.Finally, we used linear mixed models (LMMs) to compare how the activity budget of each species varied among the four study areas, where the percentage of time of each individual lemur engaged in each behavior category (Table 1) was the response variable, the study area was the fixed effect, and month of the follow was a random effect to control for behavioral differences between seasons and over the course of the year (following Mandl et al., 2018).We did not control for individual ID, as only one follow was conducted on each individual lemur.flowers simultaneously in the core area, while 97.0%(N = 162) of all trees bore leaves, 10.1% (N = 17) bore fruit, and 5.9% (N = 10) bore flowers simultaneously in the edge area.For Ankarafa forest, 97.6%
The availability of leaf-bearing trees varied significantly between the four study areas (χ 2 = 4876.70,df = 672, p < 0.001), as did fruit and flower-bearing trees (χ 2 = 799.12,df = 297, p < 0.001).While leaf availability did not vary throughout the year at all four study areas, fruit and flower availability did vary significantly across the year at some study areas (fruit: Anabohazo core and Ankarafa core; flowers: Anabohazo core and Anabohazo edge Supporting Information: File 2).

| Feeding ecology
We completed a total of 159 follows over the 3-year study period, a total of 48 for M. zaza among the four sites (Figure 2).However, the diet of M. sambiranensis was not significantly different between the core and edge areas of Anabohazo (χ 2 = 62.54, df = 47, p = 0.064) (Figure 2).
For M. zaza, L. sahamalaza, and C. medius, diet did not significantly overlap between any of the four study areas (Table 3).Feeding niche also did not overlap significant between M. sambiranensis populations in both the core and edge areas of Anabohazo forest (Table 3).The feeding niche breadth of M. zaza was highest in the edge of Anabohazo and lowest in the edge of Ankarafa, while L. sahamalaza feeding niche breadth increased in Ankarafa and edge areas in comparison to Anabohazo and core forest areas (Table 4).The feeding niche breadth of C. medius was the opposite of that for L. sahamalaza, while M.
sambiranensis feeding niche breadth was marginally higher in the edge of Anabohazo in comparison to the core (Table 4).

| Activity
We observed some differences in the activity budget of each species between the study areas (Figure 3).For M. zaza, LMM test results revealed no significant difference among the four study areas for time

| Phenology
The forests of SIRNP are seasonal, with high levels of rainfall from November to April and almost no rain between May and October (Hending, Randrianarison, Andriamavosoloarisoa, Ranohatra-Hending, Solofondranohatra, et al., in, 2023;Mandl et al., 2018).As many Malagasy tree species fruit and flower at the onset of the wet season (Dammhahn & Kappeler, 2008), it is therefore unsurprising that flower and fruit availability varied over the course of the year at some of our study areas (Table 2, Supporting Information: File 2).This statement is also true of invertebrates, as many invertebrate orders are only active under certain climatic conditions (Wolda, 1988).For example, cicadas (order Hemiptera) are the most visually and acoustically prominent invertebrates at the end of the dry season in SIRNP, but they are absent for much of the rest of the year (pers.obs.).Tree composition is highly variable among SIRNP's two forests, and among their edge and core areas (Hending, Randrianarison, Andriamavosoloarisoa, Ranohatra-Hending, Holderied, et al., 2023), which is likely why fruit and flower availability varies among these areas.Tree fruiting and flowering phenology is species-specific (König et al., 2018), and trees more abundant in edge and fragmented forest such as shade-intolerant and successional species will have different phenological traits to shadetolerant and old-growth species more-typical of core and continuous forest areas (Metzger, 2000).On the other hand, the transitional forests of SIRNP encompass both evergreen species that retain their leaves year-round and deciduous species that lose leaves in times of low rainfall (Du Puy & Moat, 1998;Koechlin, 1972).Leafing phenology is unlikely to vary among our study areas, as we observed here, as deciduous trees often lose leaves in response to the same climatic cues (Wright & Cornejo, 1990).

| Feeding ecology
The diet of M. zaza, L. sahamalaza, and C. medius varied significantly among the four different study areas (Figure 2), a finding further supported by our analysis of Pianka's indices (Table 3).As discussed in the previous section, tree species composition and fruiting and flowering phenology was highly variable among the study areas, and different food sources were available to resident lemur populations in variable quantities over the duration of this investigation.While our findings can be attributed to phenology to some extent, many tree species and invertebrates were present throughout all study areas (Supporting Information: Files 1 and 3), and only rare or specialized species were restricted to specific areas (Hending, Randrianarison, Andriamavosoloarisoa, Ranohatra-Hending, Holderied, et al., 2023).
This makes it less clear why diet varied so significantly.Although many nocturnal lemurs have broad feeding niches, some species have preference for specific types of food (Dammhahn & Kappeler, 2008;Dröscher & Kappeler, 2014;Fietz & Ganzhorn, 1999;Thorén et al., 2011).The high number of plant species, and for the Cheirogaleid species the high variety of invertebrate groups, that we observed to be consumed suggest that nocturnal lemurs can The percentage (%) of the diet of (a) Mirza zaza, (b) Lepilemur sahamalaza, (c) Cheirogaleus medius, and (d) Microcebus sambiranensis, comprised of leaves, fruit, flowers, gum, insect secretions, invertebrates, and vertebrates, in the core and edge areas of two forests of the Sahamalaza-Iles Radama National Park, North West Madagascar.Each data point used to calculate each percentage value represents an observation of a feeding instance.
adapt their diet depending on the food sources that are available to them.Broad and varied diets are widely reported for the Cheirogaleidae (Atsalis, 1999a;Fietz & Ganzhorn, 1999;Génin, 2008;Radespiel et al., 2006), and our results mirror these observations and lend further support to the theory of Cheirogaleids as resilient, adaptable and generalist primates (Hending, 2021).Our study also suggests dietary flexibility for L. sahamalaza, which has been reported in other sportive lemurs (Ganzhorn, 1993;Mandl et al., 2018;Nash, 1998).In concordance with our original hypothesis, M.
sambiranensis diet did not vary significantly between core and edge areas of Anabohazo forest (Figure 2), although our Pianka's index demonstrated that the dietary niche of M. sambiranensis in the core and edge areas did not overlap.In general, Table 3 would suggest that diet is highly specific to each surveyed area within species.However, it may be that all food species and categories consumed by each species are dietary items at all sites, and we simply were not able to observe it (≤15 follows per study area per species).It must be acknowledged here that increasing sample size of follows, and the number of months and years over which observations take place may reveal dietary overlap among study areas.
The feeding niche breadths of the three Cheirogaleidae species were all high, as reported for other members of this family (Atsalis, 1999a;Fietz & Ganzhorn, 1999;Henke von der Malsburg & Fichtel, 2018;Radespiel et al., 2006).This is exemplified by the wide variety of fruit, flowers, gum, and invertebrates consumed by these lemurs, and the prominence of vertebrates, including a souimanga sunbird, in the diet of M. zaza (Supporting Information: File 3).Interestingly, feeding niche breadth varied among study areas within M. zaza and C. medius (Table 4), although this is again likely due to the differences in tree species composition and phenology between forests and among core and edge areas.C. medius feeding niche breadth was higher in Anabohazo forest in comparison to Ankarafa; the percentage of fruiting trees was higher overall in Anabohazo in comparison to Ankarafa (Table 2), and C. medius are known to rely heavily on sugar-rich fruits during the hyperphagic period before hibernation (Fietz & Ganzhorn, 1999;Lahann, 2007).
As we observed L. sahamalaza to eat food types other than leaves more frequently in Ankarafa, feeding niche breadth was higher in Ankarafa forest than in Anabohazo (Table 4, Supporting Information: File 3).Sportive lemurs are primarily folivorous (Dröscher et al., 2016) but can adapt their diets to incorporate fruits when leaves of sufficient nutritional value are not available in times of environmental hardship (Dinsmore et al., 2016;Mandl et al., 2018;Seiler et al., 2014).
Although our niche breadth indices therefore suggest that Ankarafa may have lower availability of good-quality leaves, the diet of L.
sahamalaza is flexible to enable its survival where its preferred food choice is not readily available.To summarize, our findings suggest that forest fragmentation and its associated edge effects have a measurable impact on the feeding ecology of nocturnal lemurs and  responses to fragmentation may be species-specific, as we originally hypothesized.However, nocturnal lemurs have high dietary plasticity (Dammhahn & Kappeler, 2008;Heck et al., 2016), and they are able to rely on a wide variety of food sources to survive in a range of habitats.

| Activity
The amount of time dedicated to each behavioral category for all four species did not vary much among the study areas (Figure 3), which reflects our original prediction for the three cheirogaleid species.
However, the resting, locomotion and vocalizing time for M. zaza, and the feeding time of L. sahamalaza did vary significantly among study areas.Further, M. sambiranensis appeared to spend less time feeding and more time engaged in social behavior in the core of Anabohazo in comparison to the edge.Food sources of M. sambiranensis may be more abundant or richer in energy in Anabohazo core in comparison to the edge (Figure 2, Supporting Information: File 3), meaning that this species needs to dedicate less time to feeding and can spend more time engaging in social activity in this habitat type (Figure 3).
The lemurs of the Cheirogaleidae family survive in a range of different forest types, undisturbed and highly degraded areas, and a range of anthropogenic environments (Forbanka, 2018;Hending et al., 2018;Hending, Andrianiaina, et al., 2017;Kappeler & Rasoloarison, 2003;Knoop et al., 2018;Lahann, 2008;Lehman et al., 2016;Webber et al., 2020).As so many different habitat types are suitable for cheirogaleid lemur survival, it is therefore expected that these lemurs do not need to significantly alter most of their behaviors between them.In contrast, we had originally expected L.
sahamalaza activity to vary between the two forests and between core and edge areas, as behavior for this species (Seiler, Holderied, et al., 2013b;Seiler et al., 2014) and other sportive lemur species (Dinsmore et al., 2016;Muir et al., 2019)  feeding appears to be significantly higher in the edge of Ankarafa forest (Figure 3) where less of this species' preferred food source of leaves may be available (Figure 2).However, the edge areas and degraded forest at Ankarafa have already been observed to provide sufficient food sources, sleeping sites, and shelter from predators for L. sahamalaza (Seiler, Holderied, et al., 2013a;Seiler, Schwitzer, et al., 2013;Seiler et al., 2014); the requirements of this species appear to be met within all of our study areas.
Behavioral similarities of each species among study areas are likely due to three different factors.First, our feeding ecology data suggest that sufficient food sources are always available to each species, due to the lemur's broad feeding niches and dietary plasticity (Figure 2, Table 3, Supporting Information: File 3).Lemurs with more-specialized diets often have to spend more time feeding and searching for food in suboptimal habitats where food is less abundant (Cameron & Gould, 2013;Donati et al., 2011;Ratsimbazafy, 2006).As our studyspecies can adapt their feeding behavior to available food sources, they likely did not need to allocate a significantly larger portion of their activity toward feeding or foraging for suitable food (but see results for L. sahamalaza and M. sambiranensis).Second, our study-species are present within both core and edge areas of each forest (but see Hending, Randrianarison, et al., 2022), and densities appear somewhat consistent throughout.Encounter rates with conspecifics is therefore likely to be similar among study areas for each species, resulting in similar levels of social behavior within each forest and within both core and edge areas.In comparison, social activity and group interactions can be reduced in lemur species more sensitive to habitat degradation and fragmentation (Arrigo-Nelson, 2006;Donati et al., 2011;Gabriel, 2013).Finally, vocal activity would be expected to be fairly consistent among study areas as observed here (Figure 3), as loud calls serve important functions related to spacing, sexual advertisement, and reproductive isolation in these cryptic nocturnal species (Hending, Seiler, et al., 2020;Seiler et al., 2019).

| Next steps, conclusions and conservation implications
In summary, the feeding ecology of SIRNP's nocturnal lemur community appears to be significantly affected by forest fragmentation and its associated edge-effects, while activity appears to be little-affected.
These findings are primarily due to differences in the availability of food sources among sites, and they highlight dietary flexibility and plasticity in nocturnal lemurs.The adaptability that these lemurs have to resources available to them at different locations means that they do not have to significantly alter their behavior to meet their dietary requirements.
While we have focused on how feeding ecology and activity level of nocturnal lemurs changes in response to forest fragmentation, we note that seasonality has been reported as a notable influencer of lemur feeding ecology and activity in other studies at different locations (Atsalis, 1999b;Campera et al., 2021;Dammhahn & Kappeler, 2008;Mandl et al., 2018;Randrianambinina et al., 2003;Thorén et al., 2011).
Seasonality is something that we controlled for in this study by spreading our survey effort evenly throughout the year, but is something that should be investigated in detail in the future for SIRNP's nocturnal lemur community.
Regardless of whether they can survive in degraded or disturbed habitat, all lemurs require forest habitat for food, shelter, and ultimately their survival (Schwitzer et al., 2013).Further, there will be a limit to the behavioral flexibility that some lemurs demonstrate, beyond which they will not be able to adapt their diet and activity any further, and these adaptations may lead to future extinction debt (Alcocer-Rodríguez et al., 2021).Over 80% of Madagascar's original forest has already been cleared, and this important habitat continues to be threatened by ongoing deforestation and unmitigated climate change (Harper et al., 2007;Hending, Holderied, et al., 2022;Vieilledent et al., 2018).
Urgent conservation actions are now required to safeguard its future, and that of its resident lemur communities (Schwitzer et al., 2013).writing-review and editing (equal).

F
I G U R E 1 The Sahamalaza-Iles Radama National Park and the location of Ankarafa forest (a) and Anabohazo forest (b).
Figure created inArcMap, with a scale of 1:7,000,000 for Madagascar and a scale of 1:350,000 for the zoomed panel.Red lines represent the protected area boundary (inner), and a 3-km buffer (outer).

T
A B L E 3 Pianka's indices of feeding niche overlap for individuals of four species of lemurs within the core and edge areas of the Anabohazo and Ankarafa forests of the Sahamalaza-Iles Radama National Park, North West Madagascar.Pianka's index values are given in the bottom-left half of each species section, and the significance of the overlap (P-values obtained via one-tailed confidence intervals) are given in the top-right half.T A B L E 4 Levin's standardized indices of feeding niche breadth for individuals of four species of lemurs within the core and edge areas of the Anabohazo and Ankarafa forests of the Sahamalaza-Iles Radama National Park, North West Madagascar.
often varies between habitat types and over habitat degradation gradients.Feeding activity of this species was significantly different among study areas, and time F I G U R E 3 The percentage (%) of the activity budget of (a) Mirza zaza, (b) Lepilemur sahamalaza, (c) Cheirogaleus medius, and (d) Microcebus sambiranensis, observed as feeding, resting, locomotion, vocalizing and social/grooming behaviors, in the core and edge areas of two forests of the Sahamalaza-Iles Radama National Park, North West Madagascar.