Ecology and morphology of mouse lemurs (Microcebus spp.) in a hotspot of microendemism in northeastern Madagascar, with the description of a new species

Delimitation of cryptic species is increasingly based on genetic analyses but the integration of distributional, morphological, behavioral, and ecological data offers unique complementary insights into species diversification. We surveyed communities of nocturnal mouse lemurs (Microcebus spp.) in five different sites of northeastern Madagascar, measuring a variety of morphological parameters and assessing reproductive states for 123 individuals belonging to five different lineages. We documented two different non‐sister lineages occurring in sympatry in two areas. In both cases, sympatric species pairs consisted of a locally restricted (M. macarthurii or M. sp. #3) and a more widespread lineage (M. mittermeieri or M. lehilahytsara). Estimated Extents of Occurrence (EOO) of these lineages differed remarkably with 560 and 1,500 km2 versus 9,250 and 50,700 km2, respectively. Morphometric analyses distinguished unambiguously between sympatric species and detected more subtle but significant differences among sister lineages. Tail length and body size were most informative in this regard. Reproductive schedules were highly variable among lineages, most likely impacted by phylogenetic relatedness and environmental variables. While sympatric species pairs differed in their reproductive timing (M. sp. #3/M. lehilahytsara and M. macarthurii/M. mittermeieri), warmer lowland rainforests were associated with a less seasonal reproductive schedule for M. mittermeieri and M. lehilahytsara compared with populations occurring in montane forests. Distributional, morphological, and ecological data gathered in this study support the results of genomic species delimitation analyses conducted in a companion study, which identified one lineage, M. sp. #3, as meriting formal description as a new species. Consequently, a formal species description is included. Worryingly, our data also show that geographically restricted populations of M. sp. #3 and its sister species (M. macarthurii) are at high risk of local and perhaps permanent extinction from both deforestation and habitat fragmentation.


| INTRODUCTION
Madagascar is one of the world's prime biodiversity hotspots and its endemic group of primates, the lemurs (Primates; Lemuriformes), are flagships for species conservation (Myers, Mittermeier, Mittermeier, Da Fonseca, & Kent, 2000). More than 100 species of lemurs are recognized today making up about one-fifth of all living primate species on earth (Estrada et al., 2017). However, the full extent of lemur species diversity is not yet fully known as several regions in Madagascar are still poorly studied. Intensified biological inventories during recent years have indeed resulted in a considerable rise in lemur species numbers. One example of increased taxonomic recognition is the genus of mouse lemurs (Microcebus). These smallbodied and nocturnal primates can be found in all regions of Madagascar that offer forested habitats, while partially deforested areas appear to offer at least dispersal opportunities (Knoop, Chikhi & Salmona, 2018;Miller et al., 2018;Schüßler, Radespiel, Ratsimbazafy, & Mantilla-Contreras, 2018).
Although rather widespread across the island, mouse lemurs suffer from habitat loss due to ongoing deforestation (Vieilledent et al., 2018). According to the 2020 IUCN assessment, ten species are listed as Endangered, four species are Critically Endangered, while seven are Vulnerable, one is Data Deficient and only two species are categorized as of Least Concern (https://www.iucnredlist.org).
Recent studies indicate that some regions appear to be hotspots of microendemism. One of these is located in northeastern Madagascar where M. lehilahytsara (Kappeler et al., 2005), M. mittermeieri, and M. simmonsi (Louis et al., 2006) are known to occur. Radespiel et al. (2008) surveyed the forests of the Makira region (Anjiahely, Figure 1) and found evidence for three divergent lineages occurring in sympatry, a phenomenon previously undocumented for mouse lemurs. One of these was identified as M. mittermeieri, while the second was newly described as M. macarthurii. The third lineage, named M. sp. #3, was hypothesized to be a new species based on mitochondrial sequence data but could not be formally described given that only a single individual was found.
We conducted additional sampling in northeastern Madagascar to fill the gap between the known distribution of M. simmonsi (Zahamena NP, Betampona SNR, Tampolo; Louis et al., 2006) and the sympatric species pair M. macarthurii and M. mittermeieri at Anjiahely (Radespiel et al., 2008; Figure 1). The presence of the M. sp. #3 lineage was indeed confirmed but only for three study sites south of Anjiahely by genomic data in a companion study (Poelstra et al., 2020). Based on a comprehensive data set generated from restriction-site associated DNA sequencing (RADseq) using a total of 63 mouse lemurs from the entire region (Marojejy NP to Betampona SNR, excluding Ile St. Marie, rejection of a simple isolation-by-distance pattern, formal species F I G U R E 1 Map depicting the study region with confirmed species occurrences (Hotaling et al., 2016;Kappeler et al., 2005;Louis et al., 2006;Radespiel et al., 2008Radespiel et al., , 2012Weisrock et al., 2010). New sampling locations for this study are indicated with "*" and forest cover in 2017/2018 was derived from Vieilledent et al. (2018) and Schüßler et al. (2020). NP, National Park; SNR, Special Nature Reserve; SR, Special Reserve delimitation using SNAPP Bayes factors, BPP and gdi), M. mittermeieri and M. lehilahytsara did not fall into two separate monophyletic clades (Poelstra et al., 2020;Figure 2). Instead, these two latter species exhibited a single isolation-by-distance pattern and high levels of interspecific gene flow, suggesting that separate species status may not be justified (Figure 2; discussed in detail in Poelstra et al., 2020). For the purpose of this study, however, we will still treat these two lineages as separate taxa to be able to test their distinctiveness in other domains.
F I G U R E 2 Maximum likelihood tree illustrating the phylogenetic relationships between Microcebus spp. in northeastern Madagascar as inferred by RAxML (based on nuclear sequence data). Sampling locations are indicated at the tips of the branches. Illustration adapted from Poelstra et al. (2020). NP, National Park; SNR, Special Nature Reserve; SR, Special Reserve Here, we complement the molecular results presented by Poelstra et al. (2020)  2017) and tropical rainforests as primary vegetation (Kottek, Grieser, Beck, Rudolf, & Rubel, 2006). Forest cover has been steadily declining for decades, with lowland rainforests being particularly prone to deforestation (Schüßler, Mantilla-Contreras, Stadtmann, Ratsimbazafy, & Radespiel, 2020;Vieilledent et al., 2018). By 2018, about half of the remaining forested areas were under protection by governmental institutions or nongovernmental organizations . The study region is subdivided by more than seven large rivers that flow from the highlands of the central plateau (west of the study region; Figure 1) eastwards into the Indian Ocean. Large rivers have been considered potential biogeographic boundaries for mouse lemurs (e.g., Martin, 1972;Olivieri et al., 2007).

| Microcebus sampling
Mouse lemurs were sampled between 2008 and 2017 at five lowland rainforest sites ranging in altitude between 42 and 462 m a.s.l.
Mouse lemurs were captured using Sherman Live traps (H. B. Sherman Traps ® ) or by hand during nocturnal surveys (e.g., Radespiel et al., 2008). Morphometric measurements were taken for each individual (see below) and additional descriptors such as fur coloration were noted and photographed. Ear biopsies (~2 mm 2 ) were collected to provide DNA samples, and all animals were released unharmed within 24 hr at their exact location of capture.
GPS coordinates and the altitude of capture locations were collected to estimate the altitudinal range and Extent of Occurrence (EOO) of mouse lemur species included in this study. The latter measure follows the definition of the IUCN (2012) in which the "shortest continuous imaginary boundary which can be drawn to encompass all the known, inferred or projected sites of present occurrence of a taxon" is used to derive the possible distribution of a certain species.

| Morphometric characterization and reproduction
Captured mouse lemurs were measured for 13 different morphometric variables (ear length, ear width, head length, head width, snout length, inter-and intraorbital distance, lower leg length, hindfoot length, third toe length, tail length, body length, and body mass) following Hafen, Neveu, Rumpler, Wilden, and Zimmermann (1998) and Zimmermann et al. (1998). Mouse lemurs were assigned to two age categories based on their body mass and reproductive state: (a) adult in contrast to (b) young mouse lemurs (<1-year-old) that had a relatively low body mass ( To supplement the comparative data set, we also included published morphometric data from 42 M. lehilahytsara individuals from Mantadia NP (holotype locality; Randrianambinina, 2001) and data from 22 M. mittermeieri individuals that had previously been caught near Anjiahely (Radespiel et al., 2008). The morphometric data set is provided in the supplementary material (Table S1).
Total testes width for the regressed category ranged from 0.0 to 5.3 mm and from 10.2 to 26.2 mm for the enlarged category across all species.
One limitation to the morphometric analyses is that measurements across the five different lineages were taken by five researchers, thus potentially introducing interobserver error. It is worth reporting, however, that two researchers contributed data points to more than one species (D. S. and D. W. R,), and that D. S.
was trained by D. W. R. Furthermore, there was a strict selection of measurements that fully agreed with collection standards before assembling the data set. Finally, the data set was carefully scanned for outliers and inconsistencies within and across species, and a total of 21 measurements was excluded for this reason before data analyses. Reproductive data can also be found in Table S1.

| Statistical analyses of morphometrics
We performed a principal component analysis ( (Fox & Weisberg, 2011). M. lehilahytsara individuals from Ambavala were excluded from the ANOVA due to small sample size. For LDA and ANOVA, mouse lemurs were a priori assigned to their respective taxon based on the results of the parallel phylogenomic study ( Figure 2; Poelstra et al., 2020). For PCA, species assignment was done a posteriori to investigate clustering under naïve conditions in which the distance between sample points reflects their distance along the major axes of variation in the data set. Accordingly, points that cluster closely together are more similar to each other than points that do not (Abdi & Williams, 2010). In contrast to that, the LDA aims to minimize distances within pre-defined clusters while maximizing distances among clusters (Balakrishnama & Ganapathiraju, 1998). The PCA was followed by a permutational multivariate analysis of variances (PERMANOVA) as implemented in the "vegan" R package (Oksanen et al., 2019), which tests the null hypothesis of no differences in the position of cluster centroids (Anderson, 2017

| Morphometric distinction between lineages
All morphometric parameters differed significantly among lineages (ANOVA; p < .001; Table S3) and Tukey post hoc tests revealed many pairwise differences ( Figure 3; Table 3; Figure S1). M. sp. #3 can be statistically differentiated from its closest relative, M. macarthurii, by 5 out of 13 parameters. M. macarthurii has smaller body size and longer tail length, and subtle differences were found in head-associated parameters (i.e., ear width, head length and width). By comparison, M. lehilahytsara (from Mantadia NP) and M. mittermeieri (from Anjiahely) differed in 7 out of 13 variables. Major differences were found in snout and tail length, while other differences were more subtle but statistically significant ( Figure 3; Figure S1; Table 3). Both M. lehilahytsara and M.
mittermeieri were significantly smaller than M. sp. #3 and M. macarthurii, which was mainly reflected in the parameters body mass and length, tail length, lower leg, hindfoot, and third toe length ( Figure 3 and Figure S1;  Table 3).
T A B L E 2 Occurrence locations, altitudinal range (m a.s.l.), and estimated Extent of Occurrence (EOO in km 2 as defined by the IUCN, 2012; see Figure 1) of Microcebus spp. in northeastern Madagascar Note  (Table S4). The first two discriminant functions explained together 86.9% of the variation between the groups.

| Reproductive status
At Anjiahely, all male M. macarthurii that were captured from late All males (N = 5) had enlarged testes, while females (N = 3) were not reproductively active ( Figure 6).

| DISCUSSION
We studied the distribution, morphology, and reproductive state of five mouse lemur lineages occurring in a complex spatial pattern

| Distribution of Microcebus spp. in northeastern Madagascar
Our study, in conjunction with the companion study by Poelstra et al.   et al., 2006). We can now confirm its occurrence 75 km and four inter-river systems (IRSs) further north (Figures 1 and 2) which expands its EOO by almost fivefold. The northern range limit for M. simmonsi appears to be the Anove River, which separates it from M. sp. #3. These two species have (so far) only been found in allopatry (despite intensive sampling north of the river), and we consider two alternative hypotheses responsible for this pattern: (a) competitive exclusion at the geographic limits of the respective species ranges (Beaudrot et al., 2013;Hardin, 1960)  sp. #3 that may have a higher competitive potential than the smaller M. simmonsi (Table 3; Thorén, Linnenbrink, & Radespiel, 2011). In the case of M. sp. #3, the subpopulations on both sides of the large Mananara River (Figure 1) were shown to belong to two separate population clusters evolving largely independently from each other ( Figure 2; Poelstra et al., 2020). This suggests that the Mananara River poses a significant barrier to gene flow within this species (Poelstra et al., 2020). This moderate sensitivity to altitude may have

| Morphometric differences among mouse lemurs
Mouse lemurs are typically regarded as cryptic species exhibiting only subtle interspecific morphological differences (Zimmermann & Radespiel, 2014). Although our measurements of 13 external body parameters generally confirm their cryptic nature, some differences could be detected that can help to distinguish different species.
All analyses confirmed a noticeable divide between the two larger taxa M. sp. #3/M. macarthurii and the two smaller-bodied lineages M. lehilahytsara and M. mittermeieri (along PC1, Figure 4 and SCHÜßLER ET AL.   Radespiel et al., 2012) and can be measured with high accuracy. The more subtle differences in headassociated parameters must be interpreted more carefully, particularly because measurements were not always made by the same person. Nevertheless, it has been suggested that skull parameters may also vary with feeding habits for lemurs and strepsirrhine primates in general (e.g., omnivorous, folivorous or frugivorous etc.; Fabre et al., 2018;Meloro et al., 2015;Viguier, 2004). If validated by future studies, such differences may indicate dietary or even cognitive differentiation between closely related taxa (Zimmermann & Radespiel, 2014).
Conversely, M. lehilahytsara and M. mittermeieri differed in tail length, body length, third toe length, and four head-associated parameters ( Figure 3 and Table 3). This comparison was, however,  (Table 3). If considered as one species, these differences could indicate morphological adaptations to different environmental conditions (highland vs. lowland rainforest) or a morphological gradient across its entire range. These hypotheses, however, will require further testing.  (Table S2), another high-altitude forest, M. lehilahytsara females were observed gestating or lactating in late November, December, and early January (Blanco, 2010). From these sites there was also evidence of rebound polyestry, that is, females undergo renewed estrus after the loss of offspring or early abortions. The capture of juvenile mouse lemurs (2-3 months old) in early February, however, suggested a main birth season at this site in early mid-December. In sum, observations of tail fattening at Tsinjoarivo and Ambatovy in early March suggest that, at least for a portion of the mouse lemur population, reproductive season is over by this time of the year.
Seasonal climatic fluctuations are more pronounced in montane rainforests than in the lowland rainforests of eastern Madagascar.  The studied taxa were found in a variety of habitat types, ranging from nearly undisturbed to selectively logged forest, from shrubby secondary regrowth vegetation to areas dominated by perennial plants (i.e., Aframomum spp.; Miller et al., 2018;Radespiel et al., 2008;Schüßler et al., 2018). However, species differed most substantially in their altitudinal range and the inferred EOO. M. lehilahytsara and M.

| Conclusion and implications for conservation
mittermeieri were found both in montane and lowland forests, but M.
sp. #3 and M. macarthurii occurred only in lowland forests. Locally restricted so-called "lowland specialists" have been found in at least three other cases along the Malagasy east coast, that is, M. gerpi (Radespiel et al., 2012), M. marohita (Rasoloarison et al., 2013) and M.
jollyae (Louis et al., 2006). These taxa share a narrow altitudinal range and a small estimated EOO. Alarmingly, lowland rainforest habitats have disappeared from most of the east coast and our study region is no exception Vieilledent et al., 2018). Under these circumstances, population declines are unavoidable, and ongoing anthropogenic land-use change and forest cover loss SCHÜßLER ET AL.
| 13 of 18   It is planned that one physical specimen will be obtained as a further paratype soon and that this specimen will then be deposited in the Museum of the Zoology Department of the University of Antananarivo, Madagascar. Although not being a standard procedure, this method is most appropriate for endangered primates that should not be prematurely sacrificed if the taxonomic assignment is not yet clear. The same procedure was used for the scientific description of M. gerpi (Radespiel et al., 2012) for which a paratype individual had been collected during a subsequent field mission and was then deposited at the University of Antananarivo.

| Description
Microcebus jonahi is a large-bodied, reddish-brown, and small-eared mouse lemur (Figure 7). This species has short and dense fur. The jonahi were sighted in treeless secondary vegetation except for dense Aframomum angustifolium habitats. It currently inhabits one protected area (Mananara-Nord NP) and a community managed forest area around the village of Ambavala (Schüßler et al., 2018).

| Diagnosis
M. jonahi can be distinguished from other taxa in northeastern Madagascar by morphometric features and genomic distinctiveness.
Compared with its closest relative, M. macarthurii, M. jonahi is longer, has a shorter tail, wider ears, a larger head width and a shorter head length. In addition, M. jonahi can be differentiated from M. macarthurii by its ventral coloration which is rather whitish (Figure 6), but distinctly yellowish orange in M. macarthurii (Radespiel et al., 2008;Radespiel & Raveloson, unpublished data).
Moreover, it can be easily distinguished from the sympatric, small-bodied M. lehilahytsara (at Ambavala) by its higher body mass, larger body size, and longer tail length. Finally, M. jonahi can be differentiated from its southern geographical neighbor, M. simmonsi, by its shorter ear length and its larger inter-and intraorbital distances.
M. jonahi could be unambiguously distinguished from the other four taxa in this study across all analyses of nuclear RADseq data (Poelstra et al., 2020). However, it may not be reliably distinguished from M. macarthurii based solely on mitochondrial sequences, likely due to some introgression from M. jonahi into M. macarthurii in the past (Poelstra et al., 2020).