- Research article
- Open Access
True lemurs…true species - species delimitation using multiple data sources in the brown lemur complex
© Markolf et al.; licensee BioMed Central Ltd. 2013
- Received: 25 June 2013
- Accepted: 16 October 2013
- Published: 26 October 2013
Species are the fundamental units in evolutionary biology. However, defining them as evolutionary independent lineages requires integration of several independent sources of information in order to develop robust hypotheses for taxonomic classification. Here, we exemplarily propose an integrative framework for species delimitation in the “brown lemur complex” (BLC) of Madagascar, which consists of seven allopatric populations of the genus Eulemur (Primates: Lemuridae), which were sampled extensively across northern, eastern and western Madagascar to collect fecal samples for DNA extraction as well as recordings of vocalizations. Our data base was extended by including museum specimens with reliable identification and locality information for skull shape and pelage color analysis.
Between-group analyses of principal components revealed significant heterogeneity in skull shape, pelage color variation and loud calls across all seven populations. Furthermore, post-hoc statistical tests between pairs of populations revealed considerable discordance among different data sets for different dyads. Despite a high degree of incomplete lineage sorting among nuclear loci, significant exclusive ancestry was found for all populations, except for E. cinereiceps, based on one mitochondrial and three nuclear genetic loci.
Using several independent lines of evidence, our results confirm the species status of the members of the BLC under the general lineage concept of species. More generally, the present analyses demonstrate the importance and value of integrating different kinds of data in delimiting recently evolved radiations.
- Species delimitation
- Taxonomic inflation
- Integrative taxonomy
Species are the fundamental units in biology [1–3]. In fact, species are the fundamental units of comparisons in all fields of biology, including anatomy, behavior, ecology, molecular biology or physiology, underlining the importance of taxonomic studies for all biological disciplines [1, 4–6]. Furthermore, species are also the currency for biodiversity classification and define regions of conservation priority, so-called biological hotspots [7, 8]. Despite their fundamental importance and widespread application, identifying, defining and delimiting species is still one of the most disputed and controversial tasks in evolutionary biology .
Dozens of species concepts have been formulated, but none of them seems to be operational for every individual taxon (see [9–13]). De Queiroz therefore proposed a definition of species that is in agreement with all modern species concepts. Under this so-called general (metapopulation) lineage concept (GLC), the conceptualization of the notion of species and the operational criteria necessary to delimit them became separated [1, 14]. Instead of using a single operational criterion, such as monophyly or interbreeding, seeing species as separately evolving metapopulation lineages through time offers and highlights the importance of using multiple lines of evidence for their delimitation  because different criteria can come to fixation at different times during the divergence process of two populations. In fact, different criteria can lead to important biases in estimates of biodiversity, especially in macroevolutionary and conservation studies depending on species lists [7, 16], and are expected to give incongruent results for the boundaries of recently evolved radiations [15, 17]. However, evaluating multiple lines of evidence not only increases our capacity to detect recently diverged populations, but also can provide stronger evidence of lineage separation when different operational criteria are in concordance [18, 19].
The fauna of Madagascar has enjoyed a constant increase in species numbers in recent years. Descriptions of newly discovered species from all vertebrate groups were based on various criteria for species delimitation, however [20–28]. In this context, an almost threefold increase in the number of endemic primate species (Lemuriformes) over the last three decades has been questioned by several authors [29–31]. For example, newly described lemur species have been delimited solely based on minor variation in mitochondrial DNA (summarized in ). Moreover, sampling per “species” was often limited to one locality encompassed by a pair of Madagascar’s larger rivers. Thus, we have limited information on intraspecific genetic variation across a species’ geographic range, so that the documented extent of mtDNA divergence might just be a result of local population structure. Other taxa have been subject to taxonomic revision without new data and were raised to species level  solely based on the application of the phylogenetic species concept (PSC) in favor of the the biological species concept (BSC). These taxonomic revisions, especially in the genus Eulemur, were based on little evidence , as we outline in the following.
Based on behavioral, anatomical and cytogenetic evidence, Simons and Rumpler  erected and defined the genus Eulemur by splitting the former genus Lemur into two taxa, one containing only Lemur catta and the other containing the “true lemurs”, Eulemur coronatus, E. mongoz, E. rubriventer, E. macaco, E. fulvus fulvus, E. f. albifrons, E. f. collaris, E. f. albocollaris, E. f. rufus and E. f. sanfordi. A further subspecies, E. f. cinereiceps, was resurrected by Groves  based on a drawing by Milne-Edwards from 1890. More recent investigations revealed that this taxon is identical to E. albocollaris and thus the older name E. cinereiceps was adopted for this taxon .
Although hybridization occurs between wild E. f. rufus and E. mongoz, lineage separation of E. coronatus, E. macaco, E. mongoz and E. rubriventer from each other and from the E. fulvus group is considered to be significant by most authors [31, 36] due to frequent sympatry, smaller social units and greater phenotypic differences. The remaining Eulemur taxa were treated as subspecies of the common brown lemur (Eulemur fulvus) and grouped into the polytypic fulvus group , also referred to as the “brown lemur complex” (BLC) . Species status for E. f. albocollaris (cinereiceps) and E. f. collaris was later proposed by Wyner et al. , although both taxa hybridize with E. f. rufifrons[39, 40]. In fact, hybrids of E. cinereiceps and E. collaris are not able to produce fertile offspring, but both taxa can produce fertile offspring with other members of the BLC. Although a number of studies tried to resolve the phylogeny among Eulemur taxa using morphology [41–43], loud calls , hair banding patterns , chromosomal banding patterns [46, 47] or molecular genetics [48–54], phylogenetic relationships among Eulemur taxa, especially among the members of the BLC remain unresolved. Nevertheless, Groves  elevated all members of the BLC to species status without new evidence or new data.
Groves ( pp. 74-75) justified his decision to split E. fulvusinto 7 species as follows:
“What one can insist on is full species status for what are currently regarded as subspecies of E. fulvus. These species are not only sharply distinct externally, but they also appear to differ consistently in craniodental characters. Two of them, collaris and albocollaris (cinereiceps), have unique DNA sequences and are already acknowledged as diagnosably distinct entities. There is no evidence of overlap in phenotypic character states among members of the group, so they qualify as species under the PSC; there is little or no evidence that they form a genetic continuum in the wild, so they also qualify under the BSC.”
However, Tattersall & Schwartz (, p. 17) stated: "…so little of that 'craniodental’ variation can be made pertinent to relationships within the group. Clearly we are dealing with a high degree of homoplasy.'' Thus, apparently homoplastic characters have been used to delimit species under the PSC. Moreover hybridization sensu “a genetic continuum” is not only likely between members of the BLC, but has also been suggested for E. rufifrons and E. fulvus at Betsakafandrika  and E. albifrons with either E. fulvus along the Mananara-Zahamena corridor or with E. sanfordi north of the Bemarivo . Thus, it appears that there is more evidence that species of the BLC form a genetic continuum in the wild than not, and explicit tests of overlap in phenotypic character states are still lacking. While all taxa may be said to represent potential new species, because of remarkable phenotypic differences of males, none of them can yet be shown to have speciated .
Considering the poorly justified decision to split the subspecies of the BLC into seven different species, the main aim of this study was to test this taxonomic hypothesis with new data, and to critically appraise the conceptual and empirical approaches used in delineating these and other lemur species using an approach for species delimitation that covers intraspecific variation of hypothesized lineages for multiple independent data sets. With the present paper we aim to contribute to the topic of species delimitation in recently diverged populations in general, while clarifying the taxonomy of the BLC using several lines of evidence. The usefulness of each type of data for delimiting populations of the BLC can be characterized as follows:
Several studies have investigated the phylogenetic relationships of the members of the Lemuridae[47, 52] without completely resolving the relationships within the BLC. Moreover, these studies used either only mitochondrial DNA  or included not all taxa or only one specimen from captivity [49, 54, 57] in their analyses, which limits their usefulness for delimitation of natural taxa. Therefore, we analyzed one mitochondrial and three nuclear introns to infer species boundaries of natural populations, using phylo- and population genetic methods.
Several authors, including Groves & Eaglen , Tattersall & Schwartz  and Groves & Trueman , investigated cranidodental features of the Lemuridae without resolving relationships between members of the BLC. Later, Viguier  claimed that skull disparity is more controlled by geography than by phylogeny, confirming the homoplasy found in previous studies. Because sample size for taxa of the BLC was quite small in the latter study, we revisit the morphology of lemur skulls, using a geometric morphometric approach.
Vocalizations in non-human primates are predominantly innate  and may thus provide an additional trait for species delimitation. Loud or long distance calls represent the most distinctive calls in the vocal repertoire and are common in most primates . They typically have a species-specific acoustic structure and have therefore been used to infer phylogenetic relationships [61–67]. Macedonia & Stanger  investigated the phylogeny of the Lemuridae based on loud calls which often, but not always, consist of an introducing series of short explosive elements (chucks), followed by a long lasting scream (croak). These authors found considerable variation in what they called “disturbance advertisement calls” between members of the BLC, but they lumped all of them together for practical purposes so that variation among members of the BLC remains unknown.
Based on genetic data and pelage coloration of a single type specimen of E. f. rufus, this taxon was split into two species: E. rufus occurring north of the Tsiribihina river and E. rufifrons south of it . There are indeed phenotypic differences in pelage coloration among the members of the BLC, but a quantitative comparison of variation within and between populations has not been conducted so far.
Using new data from the field in combination with museum specimens, we examined variation in all four traits among the members of the BLC in order to assess the validity of all species assignments as well as to evaluate the usefulness and consistency of these four data sets in delineating species.
Subsequent pairwise comparisons significantly differentiated males of E. albifrons from all other taxa (Additional file 1: Table S8). Female E. albifrons, however, were not different from E. cinereiceps and E. collaris, but were different from the geographically adjacent E. fulvus and E. sanfordi. In contrast to Groves , who postulated female color differences between E. rufus and E. rufifrons, the present analysis revealed massive overlap and no significant differences between females, but between males. Eulemur cinereiceps was also significantly different from its neighbors, i.e. E. collaris and E. rufifrons.
Summary of next generation sequencing data
NGS sequencing data
Coverage per individual alleles
# of alleles
The phylogenetic tree including museum samples revealed the same pattern as the Bayesian phylogenetic tree without museum samples. Most individuals were found in the expected clade based on their museum labels. Museum samples of E. albifrons, E. sanfordi and E. fulvus confirmed the polyphyletic pattern described above (Additional file 2: Figure S5).
Genealogical sorting index (gsi) and p- values based on 10.000 permutations for the Bayesian consensus trees of all 4 loci and the combined statistic gsiT over all loci
Integration of all analyses
For this reason, morphological shape and acoustic parameter analyses found also the smallest number of significant differences among species in pairwise comparisons. In contrast, variation in pelage coloration, especially in males, could be explained to a high degree by between-group variation, and consequently revealed significant differences between almost all species pairs. All species, except E. cinereiceps showed significant exclusive ancestry for the cytb locus, but also after inclusion of the three nuclear genetic loci. Monophyly of the species of the BLC for the cytb locus, however, is only evident for E. collaris, E. rufus and E. rufifrons (excluding the hybrids from Andringitra). Overall, results of the genetic analyses indicate a substantial amount of incomplete lineage sorting within the BLC, especially for the nuclear loci. This is shown independently by discordance among the Bayesian clustering results of STRUCTURE and the DAPC as well as in the nuclear gene trees and networks. Morphological (see Additional file 2: Figure S4) and genetic divergence of E. coronatus, E. mongoz, E. rubriventer, E. macaco and E. flavifrons is much more pronounced than among the members of the BLC.
As geographic and phylogenetic relationships between taxa of the BLC are crucial for a taxonomic decision, we briefly summarize results for geographically adjacent populations.
E. albifrons and E. fulvus have adjacent geographical populations at the high plateau of Tsaratanana in central northern Madagascar and along the east coast between the National Parks Mananara Nord and Zahamena (see also Figure 7). Eulemur sanfordi is supposed to be separated by the Maevarano du Nord river from western E. fulvus populations and by the Bemarivo river from southern populations of E. albifrons. All three can potentially meet at the headwaters of the Tasaratanana massif and/or crossing rivers. Individuals seen at Tsaratanana resemble phenotypically E. fulvus, but had a mixed genetic composition (ID 496). All three are significantly different in male and female coloration. Additionally, E. albifrons differs significantly from E. fulvus in shape. Eulemur fulvus and E. sanfordi, and E. fulvus and E. albifrons seem to differ also in shape, although not significantly so (p=0.05 and p=0.068). Additionally, E. sanfordi had high gsi values for the cytb and eno loci, suggesting independent evolution for this lineage.
Eulemur rufifrons is geographically adjacent to E. rufus in western Madagascar and to E. fulvus and E. cinereiceps in eastern Madagascar. Furthermore, E. collaris and E. rufifrons are supposed to hybridize at Berenty. Excluding hybrids from Andringitra, E. cinereiceps is different in mtDNA from E. collaris and E. rufifrons, and differs from both in acoustic loud calls (chucks). Differences in female and male pelage coloration of E. cinereiceps and E. rufifrons were also significant. Eulemur collaris and E. rufifrons showed significant differences in all 4 datasets and E. rufifrons and E. rufus differed significantly in pelage coloration, genetics and acoustic parameters. Finally, E. rufus and E. fulvus differed significantly in mtDNA, female and male coloration and loud calls (chucks).
In this study, we investigated the ability of an integrative approach for the delimitation of species of a recently evolved radiation in order to falsify hypothesized lineages, in this case of the Eulemur fulvus[32, 68]. Results clearly indicate the difficulties and discordances that can arise among and within different criteria that are frequently used to delineate taxa. Although we cannot assume that we have covered complete intraspecific variation for all taxa of this study our results also highlight the necessity for a detailed and geographically broad sampling in order to effectively compare intra- and inter- specific variation of hypothesized lineages. In the following, we discuss our results in relation to the taxonomy of the BLC, as well as the significance of the discordances among data sets and their consequences for species delineation in this and other taxonomic groups.
How many species of true lemurs are there?
Lineage divergence occurs when populations accumulate contingent properties, such as reciprocal monophyly for different genes, distinctive ecological or morphological characters, reproductive isolation or adaptive behavioral traits . As speciation is a temporal process, these different contingent properties may not begin to accumulate at the same time during the lineage separation process. In fact, different contingent properties often yield conflicting results, especially in recent or adaptive radiations [17, 71]. Using different contingent properties to delimit species, however, can lead to more robust evidence of lineage separation when they are concordant [18, 72]. In this study we combined multiple lines of evidence for the delimitation of seven allopatric populations of the BLC across the island of Madagascar. This evidence comprised data from mitochondrial and nuclear DNA as well as comparisons of phenotypes in skull shape, pelage coloration and call structure.
Under the general lineage concept of species, we found evidence for the lineage divergence of all seven taxa formerly considered as subspecies of Eulemur fulvus. These lineages seem to have diverged very recently in allopatry, probably triggered by climatic shifts during the late Pleistocene (Markolf & Kappeler, subm). As eulemurs are ecologically highly flexible and occupy most biogeographic regions of Madagascar , it can be assumed that genetic drift is the main mechanism generating the observed divergence of those lineages and that ecological selective processes presumably played a less important role . Therefore, we cannot assume that lineages that are separated by hundreds of kilometers, such as E. collaris and E. albifrons, but occupy similar ecological niches necessarily accumulate strong differences in skull morphology or call structure (see below). Hence, it seems reasonable to make taxonomic decisions based on lineage divergence of geographically adjacent and phylogenetically closer related lineages (see also Markolf & Kappeler, subm). Following this approach, with the exception of E. cinereiceps, E. albifrons and E. sanfordi, we found evidence from three independent types of data supporting the delimitation of the taxa of the BLC as separate species.
However, E. albifrons and E. sanfordi were not only significantly different in male pelage coloration, but also in female coloration, a pattern not expected considering the fact that females of these two species can be hardly distinguished externally. Both species had significant gsi test statistics, indicating lineages divergence. Moreover, E. sanfordi had very high gsi values for the cytb and the eno loci, and DAPC could assign most E. sanfordi individuals with high probability to the respective cluster, suggesting exclusive ancestry for this taxon. A very recent split between these two taxa along with several past migration events (Markolf & Kappeler, subm.) seem to be responsible for a high degree of incomplete lineage sorting and less divergence in other traits analyzed here. Individual 491, treated as E. albifrons in our analyses, was assigned with high probability to E. sanfordi. In fact, we lack phenotypic information for this sample, and it may well represent E. sanfordi as it was sampled north of the Bemarivo. Unfortunately, security issues did not allow us to sample the area north of the Bemarivo more extensively. Thus, it remains unresolved whether E. sanfordi is distributed south up to the Bemarivo river, but species status is warranted. At least the museum sample from Vohemar clusters with E. sanfordi, indicating that this taxon had a much larger distribution than assumed today.
A clear taxonomic decision based on our data for E. cinereiceps is difficult. The sample from Manombo (271) clustered as a sister group to E. collaris in the mtDNA gene tree. The rest of our samples were collected from the hybrid population of Andringitra  and had mitochondrial haplotypes introgressed from E. rufifrons. Thus, genetically we have only one sample of “pure” E. cinereiceps from one locality and demarcation of this taxon based on genetics is difficult. Additionally, sample size was also very small for the museum samples and could be one explanation why E. cinereiceps was not found to be significantly different from any of the other members of the BLC in skull shape. However, E. cinereiceps differed in the acoustic structure of their chucks from adjacent E. rufifrons and E. collaris, and from E. rufifrons additionally in pelage coloration. Furthermore, E. cinereiceps and E. collaris have different chromosome numbers. They can therefore not produce fertile offspring  and would consequently qualify as species under the BSC. Further genetic investigations of the hybrid zone at Andringitra, which might shed additional light on the pattern of lineage divergence of E. cinereiceps in relation to E. rufifrons are under way (Johnson, pers. comm.).
Discordance among data sets
We found considerable differences in the ability of different datasets to delimit among members of the BLC. None of the four data sets alone could provide enough evidence for lineage separation of all species. Genetic analyses and pelage coloration could discriminate between most members of the BLC, followed by morphological shape analysis and acoustic analyses of loud calls.
The weak discriminatory ability and low interspecific variation of the acoustic data set might be due to the structure of the calls. Most studies that used acoustic signals for species discrimination in primates analyzed calls with several syllables or even songs [63, 64, 75]. Those signals show necessarily more variation due to the inherent structure of the call. Furthermore, as allopatric populations normally never meet, selective pressure on calls, even those used during intergroup encounters, is probably very low. In fact, acoustic group distances and genetic group distances estimated for the cytb (data not shown) were positively correlated, indicating that genetic drift might be mostly responsible for the small divergence in acoustic parameters. Furthermore, loud calls of E. coronatus and E. rubriventer are much more different from loud calls of the members of the BLC (data not shown), suggesting that in sympatric species selective pressure on call diversity is higher than in allopatric species. Future studies in areas of overlap should employ playbackexperiments to explore this topic in more depth.
The same can be assumed for the divergence of morphological shape, as allopatric populations occupy similar ecological niches. The large overlap of the members of the BLC in the bgPCA including the three smaller eulemurs (Additional file 2: Figure S4) confirms the extensive homoplasy found in previous studies [43, 58].
It can be argued that variation in pelage coloration might be influenced by environmental factors  and storing or preparation conditions of skins sampled in different museums. The same might be the case for acoustic variables that can be highly influenced by the environment and the distance to the animal during recordings . To control for these potential errors, we used only mean values and those acoustic parameters that should be less influenced by the distance to the animal during recording . And, prior to bgPCA, we run general linear models for both data types and included habitat (western dry forest and eastern humid forest) as well as museum for the color analysis as factor in the model. None of them had significant effects on the variables (data not shown). In general, data acquisition and analyses were conservative, and we aimed to cover as much intraspecific variation as possible. Therefore, we included only 17 landmarks for the analysis of shape that could be easily reproduced and placed on all available specimens. Because facial and ventral areas of museum skins were often in bad shape, areas for color measurements were chosen only on the dorsal view of the skins in order to avoid non-homologous placement of the measurement area and to cover variation of as many specimen as possible. Hence, color differences of males are definitely underestimated. As such, however, the method can be easily reproduced by other researchers even for different species.
One obvious drawback of our approach is that all four kinds of data could not be collected for the same individuals. Therefore, direct comparison or even combined analysis of morphological and genetic data such as offered in the software Geneland  could not be conducted. On the other hand we showed that species delimitation using several kinds of data is possible even with a completely non- invasive sampling. Especially the amount of samples for genetic analyses could not have been collected with an invasive approach.
To the best of our knowledge, this is the first study that uses Next Generation Sequencing Technology to sequence multiple independent genetic loci from feces to infer species boundaries in endangered or critically endangered primates. Following the conservative approach above, we intentionally used a high threshold to sort out potential genotyping errors. Under the assumption that sequence variants with errors occur less frequently in the dataset than sequence variants without errors, and that false alleles occur less frequently in individuals than true alleles , our filtering approach and a mean coverage per allele per individual ranging from 107- 355 among the three loci is unlikely to have produced false genotypes. In fact, after discarding sequence reads without both MIDs and unmatched target primers, most sequences were already filtered out. Among the remaining sequences most sequence errors turned out to be chimeras of the two most abundant sequences for an individual. Finally, that the nuclear dataset is unlikely to be influenced by genotyping errors is simply evident because of biological reasons. Although members of the BLC show a substantial mixed nuclear composition, the remaining Eulemur taxa have distinct haplotypes. This pattern was not necessarily expected, but confirms phylogenetic results of previous studies [51, 54] and underlines the validity of our genotyping results.
Although we had known hybrids in the data and these species can hybridize in the wild and in captivity, the mixed nuclear composition of members of the BLC is more likely a consequence of incomplete lineage sorting. With the exception of the individuals from the Andringitra hybrid zone there is no indication of any geographic locality with more admixed individuals as would be expected, if hybridization was the primary cause for admixed ancestry . Nevertheless, the structure results of K=3 revealed mixed ancestry for E. albifrons-E. sanfordi, E. collaris-E. cinereiceps-E. rufifrons and E. fulvus-E. rufus. However, whether this pattern is due to incomplete lineage sorting among phylogenetically closer related species or ongoing gene flow is beyond the scope of this article (but see, Markolf & Kappeler, subm).
Delimiting species with multiple data sources
Using multiple lines of evidence, we showed that delimitation of members of recent radiations can be particularly challenging. Because different datasets can come to different conclusions about the status of species, the use of several independent data is highly recommended in order to avoid false positives. Because taxonomic classification can be treated as a hypothesis that can be modified as new evidence accumulates , several independent data sets allow much stronger tests of a given hypothesis.
Species delimitation in lemurs, however, has been recently criticized for relying too strongly on evidence from mtDNA alone or for using different secondary species concepts (sensu) [30, 31]. It is obvious that species delimitation based on pelage coloration or morphology alone will not be very promising in cryptic species. Nevertheless, there are other methods one could think of to falsify taxonomic hypothesis in cryptic species. Although not intended to clarify species boundaries, delBarco-Trillo et al.  recently showed that chemical composition in scent marks between some eulemurs are significantly different from each other. Integrating this approach into the methods for species delimitation in lemurs would be particularly useful for many of the cryptic species, as scent marks may play a role in species recognition . The same applies to visual and acoustic signals, whose meaning and function to the animals in this context can be tested experimentally (e.g. ).
Lemurs are not the only group of mammals that has been subjected to a substantial increase in species numbers. The number of primates in general more than tripled during the last two decades . In fact, the order primates has been completely revisited following the PSC , resulting of the elevation of many taxa from subspecies to species level without new data. A similar trend can be observed in many other mammalian orders [7, 85], where similar biases have been introduced by the use of the PSC, as e.g. in ungulates . Although a discussion of species concepts is way beyond the scope of this article, the PSC, which was also used to give species status to the members of the BLC, has several shortcomings that make its application inappropriate for theoretical and practical reasons. Although there are many versions of the PSC, they all emphasize a common descent, mostly referred to as monophyly, in conjunction with diagnosability, such as “A species is the smallest diagnosable cluster of individual organisms within which there is a parental pattern of ancestry and descent” . Diagnosability, however, can be achieved even for the smallest possible units that might well represent demes, populations or even family groups due to limited dispersal and reproduction among geographically close individuals of the same species . Therefore, the PSC is very prone to overestimating species diversity based on local genetic structure, as has recently been demonstrated with genetic data from wild mouse lemurs . Cracraft , for example, applied the PSC and proposed species status for the Sumatran tiger based on three diagnostic characters of the cytochrome b unique to tigers from Sumatra and different from all tigers from the mainland. Our three samples of E. rufifrons from Ambadira have three sites diagnosably distinct from sequences of the cytochrome b of E. rufifrons ~20 km to the south along continuous forest. Do they qualify as distinct species? They could under the PSC, but they definitively do not, if we consider that haplotypes of the cytb are shared among individuals from Kirindy and Ranomafana, which is more than 200 km apart and separated by Madagascar’s deforested central highlands.
As evolution below and at the species level is shaped by population-level processes, taxonomic decisions require sample sizes that cover the whole intraspecific variation . Furthermore, it has been shown repeatedly that gene trees (although this does also apply for trees build form other kinds of data) can substantially differ from the species tree [90–92]. Considering this and the fact that evolution at the species level is often reticulate, monophyly, especially of single genes, is in general not a good criterion for species delimitation. Using multiple genes to estimate phylogenies and delimit species is becoming popular due to advances in sequencing technology, and several new coalescent-based methods for species delimitation have recently been developed [1, 93, 94]. These methods seem very promising for reliably identifying recently diverged lineages. However, any deviation from the standard coalescent model (e.g. panmixia, no gene flow) is likely to overestimate species diversity, and these methods should therefore also be complemented with standard methods from morphology, ecology or behavior .
As conservation organizations and national governments are relying strongly on the decisions of taxonomists to assess the value of protected areas or the allocation of resources for conservation, describing and raising species based on insufficient data can also be a waste of resources and additionally lead to false decisions concerning captive or natural breeding for conservation .
We conclude that according to the criteria investigated in this study members of the brown lemur complex (formerly Eulemur fulvus ssp.) are at present best classified as species according to the general lineage concept of species. As different contingent properties can arise at different times during the lineage separation process and potentially lead to ambiguous conclusions, we suggest, independent of the species concept, the utility of several independent lines of evidence, coupled with field sampling that covers intraspecific variation of the taxa under study for the delimitation of species.
We collected data from 34 different field sites in Madagascar (Figure 7). Sampling localities were a priori chosen based on published distribution data of Eulemur species. We sampled at least 3 different populations per target taxon to cover intraspecific variation, except for E. cinereiceps. Additional data were collected in 5 national history museums (Additional file 1: Table S1-S3) to further increase sample size for genetic (mtDNA) analyses, and to obtain measurements on skull morphometry and fur coloration. Only museum specimens that could unequivocally be assigned to a taxon based on their phenotype, genetic characteristics or confirmed locality were included in the analyses.
In total, we analyzed 1170 loud calls from 24 Eulemur populations. Loud-calls were elicited by presenting species-specific loud calls given during group encounters via a loudspeaker (Davidactive, Visonik) and a Marantz digital solid state recorder (PMD 660; sampling rate: 44.1 kHz, 16 bit amplitude resolution) hidden in the vegetation. Vocalizations were recorded with a Marantz and a Sennheiser directional microphone (K6 power module and ME66 recording head with MZW66 pro windscreen; Sennheiser, Wedemark, Germany). Vocalizations were digitized using AVISOFT-SASLab pro 5.0.07 (R. Specht, Berlin, Germany). We visually inspected and sampled only calls of good quality and low background noise at a sampling frequency of 44.1 kHz. As loud calls often, but not always, consist of an introducing series of short explosive elements (chucks), followed by a long lasting scream (croak), croaks and chucks were processed and analyzed separately. A spectrogram of a typical loud call is given in Additional file 2: Figure S1.
Single calls were submitted to a fast Fourier transformation (1024-pt FFT; time step: 5 ms; frequency range: 22.05 kHz; frequency resolution: 21 Hz) with AVISOFT-SASLab pro. Frequency-time spectra were analyzed with LMA 9.2, a custom software tool to extract different sets of variables from acoustic signals . We focused on acoustic variables that characterize the general call structure and are comparable with acoustic variables that were measured in other studies characterizing the structure of mammalian vocalizations [96–100]. Also, we briefly describe the acoustic variables that were used for the analysis. We measured the mean duration, the mean frequency range, the mean central frequency (DFA2) and the first and second dominant frequency bands, as well as the percentage of time of the call in which the 3rd dominant frequency could be identified . Acoustic variables entered in the analysis were revealed by Pearson’s correlation analysis. We excluded variables exhibiting a correlation coefficient higher than 0.8; the remaining variables were retained and entered into the analysis.
Due to high variation in the number of calls available for each individual, we used the mean for each individual for further statistical analysis. Between-group analysis of principal components (bgPCA) was used to infer and visualize separation between taxa. BgPCA allows to separate and maximize within-group and between-group variation. This is similar but superior to discriminant function analysis (DFA), because DFA needs more cases than variables to reliably discriminate between groups . Significance of group separation was afterwards tested using a randomization test with 999 randomizations. BgPCA and randomization were conducted with the Ade4 package in R (r-project.org). To identify significant differences between pairs of, we conducted a permutational MANOVA (PERMANOVA) with the program PAST  on the first four principal components of the bgPCA. Significance levels were corrected using the false discovery rate (FDR)  in R.
High resolution (18 Megapixels, RAW format) digital photographs of the ventral view of skulls were taken with a Canon 7d digital camera, a Sigma lens (70-200 mm) and with help of a photographic stand. To avoid distortion, which is higher at the fringe of the lens, photos were taken with a distance of 90 cm between the work space of the photographic stand and the sensor of the camera and with a focal length of 200 mm. Skulls were placed in the centre of the image together with a ruler. Use of modeling paste and a water level assured orientation in the horizontal plane. The program tpsDIG  was used to place 17 homologous landmarks on the ventral view of the skull. Landmarks (Additional file 2: Figure S2) were afterwards subjected to generalized procrustes superimposition in R, using the function procGPA of the shapes package . Generalized procrustes superimposition scales, centers and rotates raw coordinates to reduce size differences between objects. BgPCA and a subsequent randomization test on the superimposed coordinates were applied to decompose intra- and interspecific variation and to test for differences between species. The function testmeanshapes of the shapes package in R was used to test for pairwise difference between taxa with subsequent FDR correction of p-values.
Fur color data
Following the method of Bergman & Beehner , raw digital photographs of the dorsal view of museum skins were taken with the same equipment as mentioned above. Pictures were intentionally underexposed to avoid clipping of color channels . Focal length was reduced to 70 mm, and a color chart (MiniColorChecker, Munsell) was included in each photo to control for differences in ambient light conditions. To determine color variation, each image was opened with the raw converter in Photoshop CS5 and all parameters were set to zero, except for the temperature, which was set to 5100K for all photos. Using the PictoColor plugin (http://www.pictocolor.com), we applied a new color profile to each photo based on the 24 colors of the color checker chart.
We measured three areas of each skin by taking the Red, Green and Blue value (RGB) of an area of 50 × 50 pixels with the help of the rectangular marking tool (Additional file 2: Figure S3). One area was a combined measure of two squares of 50 × 50 pixels of the dorso- lateral torso of each specimen. The second area was located on the meso-dorsal stripe that some taxa possess and the third on the centre of the head. Grids and reference lines were used to control for homologous positions of the rectangles in each specimen. Mean RGB values were noted down in an Excel sheet for each area for further statistical analysis. BgPCA with subsequent permutational MANOVA on the first two principal components was conducted to test for pairwise difference between taxa.
More than 500 individual fecal samples were collected from eulemurs in the field from 2008-2011. Feces were stored on silica gel and/or 90% ethanol. After completion of fieldwork, feces were stored at 4°C until DNA extraction. Genomic DNA from the fecal samples was extracted using the QIAamp DNA Stool kit DNA (Qiagen) with a slightly modified protocol as follows. Samples were run for 24 hours at room temperature on a lab rotator in ASL buffer and only a 1/2 InhibitEx tablet was used for 600 μl supernatant of ASL-Buffer. Additionally, centrifugation steps of Qiagen spin columns were done at 8000 rpm instead of 13000 rpm as suggested in the Qiagen protocol. The same sample was sometimes extracted two or three times, which still resulted in sufficient amount of genomic DNA for PCR. A total of 116 individuals were finally used for the different genetic analyses (see Additional file 1: Table S1).
DNA extraction and subsequent PCR for the museum samples was done at a different institution (Abt. Historische Anthropologie, Universität Göttingen) under strict conditions for contamination prevention following Hummel , such as separation of pre- and post-PCR laboratories and the use of disposable protective clothing, glasses, and disposable gloves. Further, all experiments took place with disposable laboratory ware, such as pipette tips and cups, while workbenches and other laboratory equipment were cleaned with detergents (AlconoxTM Detergent, Aldrich, Germany), bi-distillated water and ethanol before use for each sample. Automatic DNA extraction of these samples was done with the QIAGEN EZ1 robotic station and the QIAGEN EZ1 DNA Tissue Kit.
Whereas the whole (1140 bp) cytochrome B gene was analyzed for fecal samples, only a shorter fragment of 223 bp was analyzed for the museum samples consisting of tissue remnants on skulls or pieces of the skin. Primers, PCR mixtures and annealing temperatures are listed in Additional file 1: Tables S4 and S5. We used Roche High Fidelity Taq Polymerase for amplification of DNA extracted from feces and the Qiagen Multiplex PCR plus Kit for the extractions of ancient DNA from museums.
Three nuclear introns were sequenced, using 454 amplicon sequencing on a Roche GS Junior 454 Sequencing platform, which allows to directly score both alleles in a diploid individual without extensive cloning procedures. However, prior to sequencing, amplicon libraries have to be constructed and each amplicon requires its own combination of MID tags to assign individuals to the correct sequence after pooling all amplicons for emulsionPCR and subsequent sequencing. A two-step PCR procedure was used to construct amplicon libraries of the three introns nramp (natural resistance macrophage protein), vwf (van willebrand factor) and eno (enolase). Initially, target-specific primers (Additional file 1: Table S4) were designed with help of published sequences from Horvath et al.  and Perelman et al. . These primers were equipped with a universal tail (M13) for the first PCR. After control on an agarose gel, PCR products were purified using magnetic beads (Beckmann and Coulter), and purified products were diluted with Molecular Biology Grade Water to approximately equimolar (5-20 ng/μl) concentrations for the next PCR. Primers for the second PCR included the GS Junior Titanium fusion primer sequences, 1-10 different MIDs for both forward and reverse primers and the template-specific sequence, which in our case were the universal tails of the previous PCR. This approach allowed us to use only 10 different forward and reverse fusion primers to individually tag 10×10=100 individuals for all three introns. The second PCR was run with the same conditions as the first. For the rest of the procedure we followed the GS junior Amplicon Library Preparation Method Manual, the GS Junior emPCR Amplification Method Manual Lib-A and the GS Junior Sequencing Method Manual from Roche.
Genotyping of individuals
After initial quality filtering and processing (i.e. adaptor removal) by the Roche/454 GS Junior software, further preprocessing was carried out by custom Perl scripts. First, sufficiently long reads were selected that perfectly matched a pair of barcode (MID) tags. Target-specific primers were removed that need to be found at the 5′ and (as reverse complement) at the 3′ end. All reads from the same gene locus were moved to a separate file. Then each sequence file was compressed by (a) removing (duplicate) reads with a perfectly identical copy in the same individual, and (b) noting the number of read copies in the FASTA comment, together with the individual identifier (corresponding to the MID tag pair).
All sequences with <10-fold coverage were discarded from the dataset.
Insertions/deletions that occurred only in one non-duplicate sequence in the whole dataset of a gene locus were discarded from the dataset, because they were likely to be a consequence of homopolymers.
Variants of each individual were sorted for coverage and checked for chimeras. If one of the sequences was likely to be a chimera of the sequences with highest coverage, they were discarded.
The two sequences with highest coverage were finally taken as the true alleles for diploid individuals, if more than one sequence was left in the end.
Final alignments for each locus were produced with SeaView and manually inspected by eye. The best fitting substitution models were calculated for each locus with jModeltest2  and chosen based on Akaike’s Information Criterion (AIC). Haplotypes were collapsed using FaBox  and translated into a genotype matrix for population genetics analyses. Input files for different software packages were also created with help of the web server GALAXY  and Microsoft Excel.
For the combined analysis of the cytb of museum and field samples, a simple Neighbor Joining Tree was calculated using the pairwise deletion option in SeaView with 10.000 bootstraps. Phylogenetic trees for the cytb without museum samples and the three nuclear loci were estimated separately using MrBayes 3.2.1. . In all analyses, we used two runs with four Monte Carlo Markov Chains (MCMC), the default temperature of 0.2, 10.000.000 generations and a sampling frequency of 1000. After a burn-in of 25% we retained 15.002 trees. Substitution model parameters were adjusted as before according to the results from jModeltest. The program Tracer and the uncorrected potential scale reduction factor (PSRF, should approach one) in MrBayes were used to check for the adequacy of the burn-in and sufficient convergence of the Markov chains.
We calculated the genealogical sorting index (gsi)  to quantify exclusive ancestry of lineages. The gsi ranges form zero to one, where zero indicates complete lack of divergence and one indicates monophyly. As the significance of the gsi statistic is measured through randomizations of group labels across the tips in a rooted gene tree, hypothesized lineages are tested against the null hypothesis of no divergence. Therefore, significance of the gsi statistic indicates exclusive ancestry of lineages, whereas the value of the gsi measures the degree of lineage divergence. The gsi was calculated separately and combined for all loci using the Bayesian phylogenetic trees.
As phylogenetic trees are often not appropriate to illustrate relationships due to reticulate evolution or incomplete lineage sorting, we calculated statistical parsimony haplotype networks for the three nuclear loci using NETWORK 4.611 (http://www.fluxus-engineering.com) .
We used two population genetic methods to test for population structure with the nuclear genotype matrix. STRUCTURE version 2.2 [117, 118] was used for Bayesian clustering of individuals into populations. To infer the correct number of K (clusters), 20 independent runs of 1.000.000 generations and a burn-in of 250.000 generations was used in an admixture model with correlated allele frequencies from K=1- 20. The number of K was inferred over all runs with STRUCTURE HARVESTER  after the ln likelihood of the data and after the method of Evanno et al. . CLUMPP  was used to permute over all runs for a given K, and assignment probabilities were plotted in R.
Discriminant Analysis of Principal Components (DAPC)  of the adegenet package in R was used to infer the probability of individuals belonging to predefined phenotypic species. DAPC is a multivariate method to infer the genetic structure of populations. The advantage of this method is that it does not assume Hardy Weinberg Equilibrium and linkage disequilibrium as STRUCTURE and other population genetic clustering methods, which is likely to be violated in most natural populations . The alpha score was used to choose the number of retained principal components and subsequent discriminant functions in order to avoid over-fitting of the data by retaining to many principal components as suggested by the manual.
We thank the Ministére d’Eaux et Forêt of Madagascar and the Département Biologie Animale of the University of Madagascar for their continued authorization and support of our field work in Madagascar. We thank Rodin M. Rasoloarison and various Malagasy locals for help during fieldwork and Christina Glaschke for help in the lab. Thanks to the National History Museum London, the American Museum of National History in New York, the Smithsonian Institution in Washington, the Museum of Comparative Zoology in Cambridge and the Naturhistorisches Museum of Berlin for the support of our work. We are grateful to the Deutsche Forschungsgemeinschaft (DFG) for financial support.
- Fujita MK, Leaché AD, Burbrink FT, McGuire JA, Moritz C: Coalescent-based species delimitation in an integrative taxonomy. Trend Ecol Evol. 2012, 27: 480-488. 10.1016/j.tree.2012.04.012.Google Scholar
- de Queiroz K: Ernst Mayr and the modern concept of species. Proc Natl Acad Sci USA. 2005, 102: 6600-6607. 10.1073/pnas.0502030102.PubMed CentralPubMedGoogle Scholar
- Naomi SI: On the integrated frameworks of species concepts: Mayden’s hierarchy of species concepts and de Queiroz’s unified concept of species. J Zool Syst Evol Res. 2011, 49: 177-184. 10.1111/j.1439-0469.2011.00618.x.Google Scholar
- Sites JW, Marshall JC: Operational criteria for delimiting species. Ann Rev Ecol Evol S. 2004, 35: 199-227. 10.1146/annurev.ecolsys.35.112202.130128.Google Scholar
- Tobias JA, Seddon N, Spottiswoode CN, Pilgrim JD, Fishpool LDC, Collar NJ: Quantitative criteria for species delimitation. Ibis. 2010, 152: 724-746. 10.1111/j.1474-919X.2010.01051.x.Google Scholar
- Wiens J, Penkrot T: Delimiting species using DNA and morphological variation and discordant species limits in spiny lizards (Sceloporus). Sys Biol. 2002, 51: 69-91. 10.1080/106351502753475880.Google Scholar
- Agapow P, Bininda-Emonds O, Crandall K, Gittleman J, Mace G, Marshall J, Purvis A: The impact of species concept on biodiversity studies. Quart Rev Biol. 2004, 79: 161-179. 10.1086/383542.PubMedGoogle Scholar
- Balakrishnan R: Species concepts, species boundaries and species identification: a view from the tropics. Sys Biol. 2005, 54: 689-693. 10.1080/10635150590950308.Google Scholar
- Pavlinov IY: The Species Problem, Why Again?. The Species Problem- Ongoing Issues. Edited by: Pavlinov IY. 2013, Rijeka (HR): InTech, 3-37.Google Scholar
- Claridge M, Dawah H, Wilson M: Species: The Units of Biodiversity. 1997, London: Chapmann & Hall LtdGoogle Scholar
- Coyne J, Orr H: Speciation. 2004, Sunderland, MA: Sinauer AssociatesGoogle Scholar
- Mallet J: Species, concepts of. Encyclopedia of Biodiversity. Edited by: Levin SA. 2001, New York: Academic press, 427-440. 5Google Scholar
- Mayden R, Claridge M, Dawah H, Wilson MR: A hierarchy of species concepts: the denouement in the saga of the species problem. Species: The Units of Biodiversity. Edited by: Claridge M, Dawah H, Wilson M. 1997, London: Chapman and Hall, 381-384.Google Scholar
- de Queiroz K: The general lineage concept of species, species criteria, and the process of speciation. Endless forms: Species and speciation. Edited by: Howard D, Berlocher S. 1998, Oxford: Oxford University Press, 57-75.Google Scholar
- Bacon CD, McKenna MJ, Simmons MP: Evaluating multiple criteria for species delimitation: an empirical example using Hawaiian palms (Arecaceae: Pritchardia). BMC Evol Biol. 2012, 12: 23-10.1186/1471-2148-12-23.PubMed CentralPubMedGoogle Scholar
- Isaac N, Purvis A: The “species problem” and testing macroevolutionary hypotheses. Diversity & Distributions. 2004, 10: 275-281. 10.1111/j.1366-9516.2004.00092.x.Google Scholar
- Leaché A, Koo M, Spencer C, Papenfuss T, Fisher R, McGuire J: Quantifying ecological, morphological, and genetic variation to delimit species in the coast horned lizard species complex (Phrynosoma). Proc Natl Acad Sci USA. 2009, 106: 12418-10.1073/pnas.0906380106.PubMed CentralPubMedGoogle Scholar
- Dayrat B: Towards integrative taxonomy. Biol J Lin Soc. 2005, 85: 407-415. 10.1111/j.1095-8312.2005.00503.x.Google Scholar
- de Queiroz K: Species concepts and species delimitation. Sys Biol. 2007, 56: 879-886. 10.1080/10635150701701083.Google Scholar
- Andriaholinirina N, Fausser JL, Roos C, Zinner D, Thalmann U, Rabarivola C, et al: Molecular phylogeny and taxonomic revision of the sportive lemurs (Lepilemur, Primates). BMC Evol Biol. 2006, 6: 17-10.1186/1471-2148-6-17.PubMed CentralPubMedGoogle Scholar
- Craul M, Zimmermann E, Rasoloharijaona S: Unexpected species diversity of Malagasy primates (Lepilemur spp.) in the same biogeographical zone: a morphological and molecular approach with the description of new species. BMC Evol Biol. 2007, 83: 7-Google Scholar
- Gehring PS, Ratsoavina FM, Vences M, Glaw F: Calumma vohibola, a new chameleon species (Squamata: Chamaeleonidae) from the littoral forests of eastern Madagascar. Afr J Herpetol. 2011, 60: 130-154. 10.1080/21564574.2011.628412.Google Scholar
- Goodman SM, Raherilalao MJ, Block NL: Patterns of morphological and genetic variation in the Mentocrex kioloides complex (Aves: Gruiformes: Rallidae) from Madagascar, with the description of a new species. Zootaxa. 2011, 2776: 49-60.Google Scholar
- Groeneveld LF, Weisrock DW, Rasoloarison RM, Yoder AD, Kappeler PM: Species delimitation in lemurs: multiple genetic loci reveal low levels of species diversity in the genus Cheirogaleus. BMC Evol Biol. 2009, 9: 30-10.1186/1471-2148-9-30.PubMed CentralPubMedGoogle Scholar
- Louis E, Engberg S, Lei R, Geng H, Sommer J: Molecular and morphological analyses of the sportive lemurs (family Megaladapidae: genus Lepilemur) reveas 11 previousely unrecognized species. Mus Tex Tech Univ. 2006, 49: 1-49.Google Scholar
- Radespiel U, Olivieri G, Rasolofoson D, Rakotondratsimba G, Rakotonirainy O, Rasoloharijaona S, et al: Exceptional diversity of mouse lemurs (Microcebus spp.) in the Makira region with the description of one new species. Am J Primatol. 2008, 70: 1033-1046. 10.1002/ajp.20592.PubMedGoogle Scholar
- Wollenberg K, Vieites D, van der Meijden A, Glaw F, Cannatella D, Vences M: Patterns of endemism and species richness in Malagasy cophyline frogs support a key role of mountainous areas for speciation. Evolution. 2008, 62: 1890-1907. 10.1111/j.1558-5646.2008.00420.x.PubMedGoogle Scholar
- Weisrock DW, Rasoloarison R, Fiorentino I, Ralison J, Goodman S, Kappeler PM, Yoder A: Delimiting species without nuclear monophyly in Madagascar’s mouse lemurs. PLoS ONE. 2010, 5: e9883-10.1371/journal.pone.0009883.PubMed CentralPubMedGoogle Scholar
- Groeneveld L: PhD Thesis. Species delimitation, phylogeography and population genetics of the endemic Malagasy dwarf lemurs (genus Cheirogaleus). 2008, University of Göttingen, Behavioral Ecology and Sociobiology Unit, German Primate Center (DPZ)Google Scholar
- Markolf M, Brameier M, Kappeler PM: On species delimitation: Yet another lemur species or just genetic variation?. BMC Evol Biol. 2011, 11: 216-10.1186/1471-2148-11-216.PubMed CentralPubMedGoogle Scholar
- Tattersall I: Madagascar’s Lemurs: Cryptic diversity or taxonomic inflation?. Evol Anhropol. 2007, 16: 12-23. 10.1002/evan.20126.Google Scholar
- Groves C: Primate taxonomy. 2001, Washington: Smithsonian Institution PressGoogle Scholar
- Simons EL, Rumpler Y: Eulemur: new generic name for species of Lemur other than Lemur catta. Comptes rendus de l’Académie des sciences. Série 3, Sciences de la vie 1988. 1988, 307: 547-551.Google Scholar
- Johnson S, Lei R, Martin S, Irwin M, Louis E: Does Eulemur cinereiceps exist? Preliminary evidence from genetics and ground surveys in southeastern Madagascar. Am J Primatol. 2007, 70: 372-385.Google Scholar
- Pastorini J, Zaramody A, Curtis DJ, Nievergelt CM, Mundy NI: Genetic analysis of hybridization and introgression between wild mongoose and brown lemurs. BMC Evol Biol. 2009, 9: 32-10.1186/1471-2148-9-32.PubMed CentralPubMedGoogle Scholar
- Johnson SE: Evolutionary divergence of the brown lemur species complex. Lemurs: Ecology and Adaption. Edited by: Gould L, Sauther ML. 2006, Massachusetts: Springer US, 187-210.Google Scholar
- Tattersall I: The Primates of Madagascar. 1982, New York: Columbia University PressGoogle Scholar
- Wyner Y, Absher R, Amato G, Sterling E, Stumpf R, Rumpler Y, Desalle R: Species concepts and the determination of historic gene flow patterns in the Eulemur fulvus (Brown Lemur) complex. Biol J Linn Soc. 1999, 66: 39-56.Google Scholar
- Jekielek J: Hybridization of Brown Lemurs at Berenty Reserve, Madagascar. MSc Thesis. 2004, Alberta: University of Alberta, Department of Biological SciencesGoogle Scholar
- Wyner YM, Johnson SE, Stumpf RM, Desalle R: Genetic assessment of a white collared× red fronted lemur hybrid zone at Andringitra, Madagascar. Am J Primatol. 2002, 57: 51-66. 10.1002/ajp.10033.PubMedGoogle Scholar
- Groves C, Eaglen R: Systematics of the Lemuridae (Primates, Strepsirhini). J Hum Evol. 1998, 17: 513-538.Google Scholar
- Groves C, Trueman JWH: Lemurid systematics revisited. J Hum Evol. 1995, 28: 427-437. 10.1006/jhev.1995.1033.Google Scholar
- Tattersall I, Schwartz J: Phylogeny and nomenclature in the “Lemur-Group” of Malagasy Strepsirrhine primates. Anthropol Pap Am Mus. 1991, 69: 18-Google Scholar
- Macedonia J, Stanger K: Phylogeny of the Lemuridae revisited: evidence from communication signals. Fol Primatol. 1994, 63: 1-43. 10.1159/000156787.Google Scholar
- Macedonia J, Shedd D: Metachromism and Its Phylogenetic Implications for the Genus Eulemur (Prosimii: Lemuridae). Fol Primatol. 1991, 57: 221-231. 10.1159/000156589.Google Scholar
- Djelati R, Brun B, Rumpler Y: Meiotic study of hybrids in the genus Eulemur and taxonomic considerations. Am J Primatol. 1997, 42: 235-245. 10.1002/(SICI)1098-2345(1997)42:3<235::AID-AJP6>3.0.CO;2-Y.Google Scholar
- Rumpler Y, Warter S, Ishak B, Dutrillaux B: Chromosomal evolution in prosimians. Human Evolution. 1989, 4: 157-170. 10.1007/BF02435444.Google Scholar
- Delpero M, Pozzi L, Masters JC: A composite molecular phylogeny of living lemuroid primates. Fol Primatol. 2006, 77: 434-445. 10.1159/000095390.Google Scholar
- Horvath J, Weisrock DW, Embry S, Fiorentino I, Balhoff J, Kappeler PM, et al: Development and application of a phylogenomic toolkit: Resolving the evolutionary history of Madagascar’s lemurs. Genome Res. 2008, 18: 489-499. 10.1101/gr.7265208.PubMed CentralPubMedGoogle Scholar
- Pastorini J, Forstner M, Martin R: Relatiomships among Brown Lemurs (Eulemur fulvus) Based on Mitochondrial DNA Sequences. Mol Phylogenet Evol. 2000, 16: 418-429. 10.1006/mpev.2000.0796.PubMedGoogle Scholar
- Pastorini J, Thalmann U, Martin R: A molecular approach to comparative phylogeography of extant Malagasy lemurs. Proc Natl Acad Sci USA. 2003, 100: 5879-5884. 10.1073/pnas.1031673100.PubMed CentralPubMedGoogle Scholar
- Pozzi L, Gamba M, DelPero M, Giacoma C: Eulemur phylogeny: Evidence from molecular and acoustic data. Int J Primatol. 2006, 27 (Suppl 1): 434-Google Scholar
- Wyner Y, DeSalle R, Absher R: Phylogeny and character behavior in the family Lemuridae. Mol Phylogenet Evol. 2000, 15: 124-134. 10.1006/mpev.1999.0723.PubMedGoogle Scholar
- Yoder A, Yang Z: Divergence dates for Malagasy lemurs estimated from multiple gene loci: geological and evolutionary context. Mol Ecol. 2004, 13: 757-773. 10.1046/j.1365-294X.2004.02106.x.PubMedGoogle Scholar
- Lehmann S, Wright P: Preliminary study of the conservation status of Lemur communities in the Betsakafandrika region of eastern Madagascar. Lemur News. 2000, 5: 23-25.Google Scholar
- Mittermeier R, Konstant W, Hawkins F, Louis E, Langrand O, Ratsimbazafy J, Rasoloarison R, Ganzhorn JU, Rajoabelina S, TAttersall I, Meyers DM: Lemurs of Madagascar. 2006, Washington, DC: Conservation International, 2Google Scholar
- Perelman P, Johnson WE, Roos C, Seuánez HN, Horvath JE, Moreira MAM, et al: A molecular phylogeny of living primates. PLoS Genet. 2011, 7: e1001342-10.1371/journal.pgen.1001342.PubMed CentralPubMedGoogle Scholar
- Viguier B: Is the morphological disparity of lemur skulls (Primates) controlled by phylogeny and/or environmental constraints?. Biol J Linn Soc. 2002, 76: 577-590. 10.1046/j.1095-8312.2002.00081.x.Google Scholar
- Hammerschmidt K, Freudenstein T, Jürgens U: Vocal development in squirrel monkeys. Behaviour. 2001, 138: 1179-1204. 10.1163/156853901753287190.Google Scholar
- Wich SA, Nunn CL: Do male “long-distance calls” function in mate defense? A comparative study of long-distance calls in primates. Behav Ecol Sociobiol. 2002, 52: 474-484. 10.1007/s00265-002-0541-8.Google Scholar
- Konrad R, Geissmann T: Vocal Diversity and Taxonomy of Nomascus in Cambodia. Int J Primatol. 2006, 27: 713-745. 10.1007/s10764-006-9042-3.Google Scholar
- Mendez-Cardenas M, Randrianambinina B: Geographic variation in loud calls of sportive lemurs (Lepilemur ssp.) and their implications for. Am Jo Primatol. 2008, 70: 828-838. 10.1002/ajp.20554.Google Scholar
- Merker S, Driller C, Perwitasari-Farajallah D: Elucidating geological and biological processes underlying the diversification of Sulawesi tarsiers. Proc Natl Acad Sci U S A. 2009, 106: 8459-8464. 10.1073/pnas.0900319106.PubMed CentralPubMedGoogle Scholar
- Nietsch A, Kopp M: Role of vocalization in species differentiation of Sulawesi tarsiers. Fol Primatol. 1998, 69: 371-378. 10.1159/000052725.Google Scholar
- Pozzi L, Gamba M, Giacoma C: The use of Artificial Neural Networks to classify primate vocalizations: a pilot study on black lemurs. Am J Primatol. 2009, 72: 337-348.Google Scholar
- Thinh VN, Hallam C, Roos C, Hammerschmidt K: Concordance between vocal and genetic diversity in crested gibbons. BMC Evol Biol. 2011, 11: 36-10.1186/1471-2148-11-36.PubMed CentralPubMedGoogle Scholar
- Zimmermann E, Vorobieva E, Wrogemann D: Use of vocal fingerprinting for specific discrimination of gray (Microcebus murinus) and rufous mouse lemurs (Microcebus rufus). Int J Primatol. 2000, 2000 (21): 837-852.Google Scholar
- Groves C: Red- fronted lemurs are not the same as red lemurs. Australas Primatol. 2006, 18 (1): 23-Google Scholar
- Delmore KE, Louis EE, Johnson SE: Morphological characterization of a brown lemur hybrid zone (Eulemur rufifrons × E. cinereiceps). Am J Phys Anthropol. 2011, 145: 55-66. 10.1002/ajpa.21466.PubMedGoogle Scholar
- Evanno G, Regnaut S, Goudet J: Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular ecology. 2005, 14: 2611-2620. 10.1111/j.1365-294X.2005.02553.x.PubMedGoogle Scholar
- Wake DB: Problems with species: Patterns and processes of species formation in salamanders. Ann Missouri Bot Gard. 2006, 93: 8-23. 10.3417/0026-6493(2006)93[8:PWSPAP]2.0.CO;2.Google Scholar
- de Queiroz K: Species concepts and species delimitation. Syst Biol. 2007, 56: 879-886. 10.1080/10635150701701083.PubMedGoogle Scholar
- Ossi K, Kamilar J: Environmental and phylogenetic correlates of Eulemur behavior and ecology (Primates: Lemuridae). Behav Ecol Sociobiol. 2006, 61: 53-64. 10.1007/s00265-006-0236-7.Google Scholar
- Dutrillaux B, Rumpler Y: Chromosomal evolution in Malagasy lemurs. II. Meiosis in intra- and interspecific hybrids in the genus Lemur. Cytogenet Cell Genet. 1977, 18: 197-211. 10.1159/000130763.PubMedGoogle Scholar
- Meyer D, Hodges JK, Rinaldi D, Wijaya A, Roos C, Hammerschmidt K: Acoustic structure of male loud-calls support molecular phylogeny of Sumatran and Javanese leaf monkeys (genus Presbytis). BMC Evol Biol. 2012, 12: 16-10.1186/1471-2148-12-16.PubMed CentralPubMedGoogle Scholar
- Santana SE, Lynch Alfaro J, Alfaro ME: Adaptive evolution of facial colour patterns in Neotropical primates. Proc Roy Soc B: Bio Sci. 2012, 279: 2204-2211. 10.1098/rspb.2011.2326.Google Scholar
- Maciej P, Fischer J, Hammerschmidt K: Transmission characteristics of primate vocalizations: implications for acoustic analyses. PLoS ONE. 2011, 6: e23015-10.1371/journal.pone.0023015.PubMed CentralPubMedGoogle Scholar
- Guillot G, Renaud S, Ledevin R, Michaux J, Claude J: A unifying model for the analysis of phenotypic, genetic, and geographic data. Sys Biol. 2012, 61: 897-911. 10.1093/sysbio/sys038.Google Scholar
- Galan M, Guivier E, Caraux G, Charbonnel N, Cosson JF: A 454 multiplex sequencing method for rapid and reliable genotyping of highly polymorphic genes in large-scale studies. BMC Genom. 2010, 11: 296-10.1186/1471-2164-11-296.Google Scholar
- Hewitt G: Speciation, hybrid zones and phylogeography-or seeing genes in space and time. Mol Ecol. 2001, 10: 537-549.PubMedGoogle Scholar
- Groves C: What, if anything, is taxonomy?. Gorilla J. 2000, 21: 12-15.Google Scholar
- del Barco-Trillo J, Sacha CR, Dubay GR, Drea CM: Eulemur, me lemur: the evolution of scent-signal complexity in a primate clade. Phil Trans Biol Sci. 2012, 367: 1909-1922. 10.1098/rstb.2011.0225.Google Scholar
- Smadja C, Butlin RK: On the scent of speciation: the chemosensory system and its role in premating isolation. Hered. 2009, 102: 77-97. 10.1038/hdy.2008.55.Google Scholar
- Kappeler PM: Behavioral ecology of strepsirrhines and tarsiers. The evolution of primate societies. Edited by: Mitani JC, Call J, Kappeler PM, Palombit R, Silk JB. 2012, Chicago: University of Chicago Press, 17-42.Google Scholar
- Zachos FE, Apollonio M, Bärmann EV, Festa-Bianchet M, Göhlich U, Habel JC, et al: Species inflation and taxonomic artefacts—A critical comment on recent trends in mammalian classification. J Mammal Biol. 2013, 78: 1-6.Google Scholar
- Groves C, Grubb P: Ungulate Taxonomy. 2011, Baltimore: The Johns Hopkins University PressGoogle Scholar
- Cracraft J: Species concepts and speciation analysis. Current Ornithology. 1983, 1: 159-187.Google Scholar
- Avise JC: Phylogeography: The History and Formation of Species. 2000, Cambridge: Havard University PressGoogle Scholar
- Cracraft J, Feinstein J, Vaughn J, Helm-Bychowski K: Sorting out tigers (Panthera tigris): mitochondrial sequences, nuclear inserts, systematics, and conservation genetics. Anim Conserv. 1998, 1: 139-150. 10.1111/j.1469-1795.1998.tb00021.x.Google Scholar
- Camargo A, Sites J: Species Delimitation: A Decade After the Renaissance. The Species Problem - Ongoing Issues. Edited by: Pavlinov YI. 2013, Rijeka (HR): INTECH, 225-247.Google Scholar
- Knowles LL, Kubatko LS: Estimating species trees: Practical and theoretical aspects. 2011, Wiley Blackwell: CanadaGoogle Scholar
- Nichols R: Gene trees and species trees are not the same. Trend Ecol Evol. 2001, 16: 358-364. 10.1016/S0169-5347(01)02203-0.Google Scholar
- Ence DD, Carstens BC: SpedeSTEM: a rapid and accurate method for species delimitation. Mol Ecol Res. 2011, 11: 473-480. 10.1111/j.1755-0998.2010.02947.x.Google Scholar
- Rannala B, Yang Z: Improved reversible jump algorithms for bayesian species delimitation. Genetics. 2013, 194: 245-253. 10.1534/genetics.112.149039.PubMed CentralPubMedGoogle Scholar
- Schrader L, Hammerschmidt K: Computer- aided analysis of acoustic parameters in animal vocalizations: A multi-parametric approach. Bioacoustics. 1997, 7: 247-265. 10.1080/09524622.1997.9753338.Google Scholar
- Fichtel C, Hammerschmidt K: Responses of redfronted lemurs to experimentally modified alarm calls: evidence for urgency based changes in call structure. Ethology. 2002, 108: 763-778. 10.1046/j.1439-0310.2002.00816.x.Google Scholar
- Fichtel C, Perry S, Gros-Louis J: Alarm calls of white-faced capuchin monkeys: an acoustic analysis. Anim Behav. 2005, 70: 165-176. 10.1016/j.anbehav.2004.09.020.Google Scholar
- Fichtel C: Acoustic differences in loud calls of Decken's and Crowned sifakas. Primate Conversation. 2013, in pressGoogle Scholar
- Gros-Louis JJ, Perry SE, Fichtel C, Wikberg E: Vocal repertoire of Cebus capucinus: Acoustic structure, context, and usage. Int J Primatol. 2008, 29: 641-670. 10.1007/s10764-008-9263-8.Google Scholar
- Manser MB: The acoustic structure of suricates’ alarm calls varies with predator type and the level of response urgency. Proc Roy Soc Lond B. 2001, 268: 2315-2324. 10.1098/rspb.2001.1773.Google Scholar
- Mitteroecker P, Bookstein F: Linear discrimination, ordination, and the visualization of selection gradients in modern morphometrics. Evol Biol. 2011, 38: 100-114. 10.1007/s11692-011-9109-8.Google Scholar
- Hammer O, Harper DAT, Ryan PD: PAST: Paleontological statistics software package for education and data analysis. Paleontol Electronica. 2001, 4: 9-Google Scholar
- Benjamini Y, Hochberg Y: Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc B Met. 1995, 57: 289-300.Google Scholar
- Rohlf FJ: TPS Series. 2004, http://life.bio.sunysb.edu/ee/rohlf/software.html,Google Scholar
- Dryden IL: P, Dryden IL: Package 'Shapes (2012)’. http://www.maths.nottingham.ac.uk/~ild/shapes,
- Bergman T, Beehner J: A simple method for measuring colour in wild animals: validation and use on chest patch colour in geladas (Theropithecus gelada). Biol J Linn Soc. 2008, 94: 231-240. 10.1111/j.1095-8312.2008.00981.x.Google Scholar
- Stevens M, Párraga C, Cuthill I, Patridge J, Troscianko T: Using digital photography to study animal coloration. Biol J Linn Soc. 2007, 90: 211-237. 10.1111/j.1095-8312.2007.00725.x.Google Scholar
- Hummel S: Ancient DNA Typing. Methods, Strategies, and Applications. 2003, New York, Berlin, Heidelberg: SpringerGoogle Scholar
- Gouy M, Guindon S, Gascuel O: SeaView version 4: a multiplatform graphical user interface for sequence alignment and phylogenetic tree building. Mol Biol Evol. 2010, 27: 221-224. 10.1093/molbev/msp259.PubMedGoogle Scholar
- Gilles A, Meglécz E, Pech N, Ferreira S, Malausa T, Martin JF: Accuracy and quality assessment of 454 GS-FLX Titanium pyrosequencing. BMC Genomics. 2011, 12: 245-10.1186/1471-2164-12-245.PubMed CentralPubMedGoogle Scholar
- Darriba D, Taboada GL, Doallo R, Posada D: jModelTest 2: more models, new heuristics and parallel computing. Nat Methods. 2012, 9: 772-PubMed CentralPubMedGoogle Scholar
- Villesen P: FaBox: an online toolbox for fasta sequences. Mol Ecol Notes. 2007, 7: 965-968. 10.1111/j.1471-8286.2007.01821.x.Google Scholar
- Goecks J, Nekrutenko A, Taylor J, Team TG: Galaxy: a comprehensive approach for supporting accessible, reproducible, and transparent computational research in the life sciences. Genome Biol. 2010, 11: R86-10.1186/gb-2010-11-8-r86.PubMed CentralPubMedGoogle Scholar
- Ronquist F, Huelsenbeck JP, Mark PVD: MrBayes 3.1 manual. 2010, http://mrbayes.sourceforge.net/manual.php,Google Scholar
- Cummings MP, Neel MC, Shaw KL: A genealogical approach to quantifying lineage divergence. Evolution. 2008, 62: 2411-2422. 10.1111/j.1558-5646.2008.00442.x.PubMedGoogle Scholar
- Polzin T, Daneshmand SV: On Steiner trees and minimum spanning trees in hypergraphs. Oper Res Lett. 2003, 31: 12-20. 10.1016/S0167-6377(02)00185-2.Google Scholar
- Pritchard J, Stephens M, Donnelly P: Inference of population structure using multilocus genotype data. Genetics. 2000, 155: 945-PubMed CentralPubMedGoogle Scholar
- Pritchard J, Wen X, Falush D: Documentation for structure software: Version 2.2. 2007, http://pritch.bsd.uchicago.edu/software,Google Scholar
- Earl DA, von Holdt BM: STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conserv Genet Resources. 2012, 4: 359-361. 10.1007/s12686-011-9548-7.Google Scholar
- Jakobsson M, Rosenberg NA: CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics. 2007, 23: 1801-1806. 10.1093/bioinformatics/btm233.PubMedGoogle Scholar
- Jombart T, Devillard S, Balloux F: Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC Genet. 2010, 11: 94-PubMed CentralPubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.