International Journal of Primatology

, Volume 37, Issue 1, pp 69–88 | Cite as

Niche Divergence in a Brown Lemur (Eulemur spp.) Hybrid Zone: Using Ecological Niche Models to Test Models of Stability

  • Steig E. JohnsonEmail author
  • Kira E. Delmore
  • Kerry A. Brown
  • Tracy M. Wyman
  • Edward E. LouisJr.


Endogenous selection is often implicated in the maintenance of stability of natural hybrid zones. Environmental conditions often vary across these zones, suggesting that local adaptation to ecological conditions could also play a role in this process. We used niche modeling to investigate these alternatives in a hybrid zone between two species of brown lemur (Eulemur rufifrons and E. cinereiceps) in southeastern Madagascar. We produced ecological niche models (ENMs) for parental and hybrid populations and compared values of niche overlap to null expectations using identity and background tests. All three taxonomic groups had nonequivalent ENMs with limited spatial overlap, supporting a role for niche divergence and local adaptation in the maintenance of this zone. However, values of niche overlap between ENMs were not greater than null expectations controlling for background environmental differences. These results could suggest that taxa in this hybrid zone inhabit portions of their environments that are more similar to their backgrounds, i.e., niche conservatism. Nevertheless, we did find evidence of niche divergence when using background tests that examined environmental variables separately. Although we could not rule out models indicating selection against hybrids, most lines of evidence were consistent with predictions for the bounded superiority model of hybrid zone stability. This study thus provides support that exogenous, environmental selection may be responsible for maintaining the hybrid zone, and may be implicated in the evolutionary divergence of these taxa.


Bounded superiority Ecological niche models Geographical selection-gradient Hybrid zone Tension zone 


The potential role of ecological adaptations in speciation —with natural selection driving divergence between populations occupying different habitats— has garnered increased attention (Schluter 2001; Wiens 2004). Hybrid zones, where distinct lineages overlap and reproductive isolation is incomplete, offer a unique perspective on such dynamic evolutionary processes (Arnold 1992). Many hybrid zones are stable over long periods, with hybridizing taxa exchanging genes but remaining distinct (Barton and Hewitt 1985; Hewitt 1988). As environmental conditions may vary across zones of overlap and into allopatric portions of the parental species’ ranges, adaptation to local conditions may reinforce premating barriers and ultimately the evolutionary separation among lineages (Schluter 2001).

Several models have been proposed to explain hybrid zone stability (Barton and Hewitt 1985; Moore 1977). Two models —the tension zone and geographical selection-gradient models— argue that selection acts against hybrids but that continual dispersal of parental forms into hybrid zones maintains them over time (Barton and Hewitt 1985; Bigelow 1965; Endler 1977; Key 1968; Moore and Price 1993). The primary difference between these models is that in tension zones, selection against hybrids is endogenous (heterozygote breakdown), whereas under the geographical selection-gradient model, selection is exogenous, i.e., potentially driven by niche divergence in the parental species. The bounded superiority model was proposed in opposition to these models, arguing that selection could actually favor hybrids within transitional habitats (Anderson 1949; Moore 1977). Similar to the geographical selection-gradient model, selection is exogenous under the bounded superiority model but in this case favors hybrids. Empirical work has provided support for the tension zone and geographical selection-gradient models, whereas far less evidence for the bounded superiority model has been documented (e.g., Alexandrino et al. 2005; Barton and Hewitt 1985; Gligor et al. 2009; Kawakami et al. 2008).

Niche modeling provides a valuable framework for investigating questions of ecological adaptation across landscapes (Kozak et al. 2008; Swenson 2008), which can be used to corroborate findings from genetic, morphological, and behavioral research. These methods use environmental data from known localities to predict the potential geographic distribution of taxonomic groups (Franklin 2009; Peterson et al. 2011). These models are termed ecological niche models (ENMs), and comparisons with known taxonomic boundaries can be used to make inferences regarding ecological separation in hybrids and parental species; in the context of ENMs, niche divergence refers to nonidentical or significantly different predicted spatial distributions based on environmental (often climate) predictors. For example, Swenson (2006) modeled the ecological niches of four avian hybrid zones, each consisting of eastern and western species in North America. The ENMs of eastern species matched their known taxonomic boundaries, suggesting that exogenous selection limits these species. By contrast, the ENMs of western species expanded well into the east, suggesting that biotic factors, e.g., competition, may be limiting their ranges.

We employ ENMs here to test models of stability in a hybrid zone between two species of brown lemur, red-fronted (Eulemur rufifrons) and gray-headed lemurs (E. cinereiceps). These species are part of the brown lemur species complex, which includes seven species found throughout Madagascar (Johnson 2007; Markolf and Kappeler 2013; Mittermeier et al. 2008). Most species within this complex are arboreal and frugivorous (Johnson 2007), and several confirmed or suspected cases of hybridization have been documented (Lehman and Wright 2000; Mittermeier et al. 2006; Pastorini et al. 2009; Wyner et al. 2002).

Eulemur rufifrons and E. cinereiceps hybridize in the Andringitra region of southeastern Madagascar. Data collected from this hybrid zone suggest that it is stable: populations at the center of the zone are at Hardy–Weinberg equilibrium (Wyner et al. 2002), private sites specific to the hybrid population have been identified (Wyner et al. 2002), and the zone is likely too narrow to be explained purely by neutral diffusion (Delmore et al. 2013). Further evidence suggests that it may conform to the bounded superiority model: hybrids are apparently equally as fit as parental forms (Delmore et al. 2011), clines constructed using phenotypic and genetic data are variable in shape and noncoincident in position, and there is no elevation of linkage disequilibrium at the center of the hybrid zone (Delmore et al. 2013). A preliminary analysis of environmental data further suggests that hybrids occupy a transitional or novel habitat characterized by greater seasonality in precipitation, as well as lower seasonality in temperature (Delmore et al. 2013). Furthermore, divergence in the feeding ecology of hybrid and parental forms has been reported, including marked differences in diet during seasons with low food abundance (Johnson 2002, 2007).

We constructed ENMs for hybrid and parental forms in the Andringitra brown lemur hybrid zone and quantified niche overlap using both identity and background tests (McCormack et al. 2010; Warren et al. 2008). We predicted that if this zone conforms to the tension zone model, ENMs should not coincide closely with observed geographic boundaries across parental species and the hybrids because environmental differences are not expected to play a role in species divergence. Furthermore, niche divergence should not be observed (although tension zones can settle in density troughs, which may themselves be associated with environmental shifts; Hewitt 1988); thus, either niche conservatism or null models would be supported under the tension zone model. If the Andringitra hybrid zone conforms to the geographical selection-gradient model, we predicted that there would be overlap between the ENMs of parental forms at the known location of the hybrid zone. Moreover, the ENM of hybrids would be located in this region but overlap with ENMs of parental forms. We further predicted that we should observe niche divergence between parental forms, but comparisons between hybrids and each parental form should demonstrate niche conservatism, as hybrids do not occupy separate ecological niches under this model. Instead, they may possess characteristics from both parental forms that are maladaptive when combined in hybrids. Finally, if this zone conformed to the bounded superiority model, we predicted that the ENM of hybrids should line up with the known location of the hybrid zone and show limited overlap with ENMs of parental forms. Niche divergence should also be documented between hybrids and both parental forms.


Study System

Occurrence locations for Eulemur cinereiceps, E. rufifrons, and hybrids were recorded during previous research in the study region (e.g., Delmore et al. 2011; Delmore et al. 2013; Ingraldi 2010; Johnson 2002; Table I). For simplicity, we refer hereafter to all three populations (including hybrids) as taxa. The distribution of E. rufifrons includes a portion of western Madagascar south of the Tsiribihina River, as well as a long expanse of eastern forests from the Mangoro and Onive Rivers in the north to the Andringitra Massif in the south (Mittermeier et al. 2008). We include only the eastern portion of this species’ range in our analyses, as marked climatic differences between the now isolated eastern and western distributions would introduce a high degree of environmental heterogeneity in our analyses; western dry forest habitats are not present in the eastern populations of E. rufifrons that border and contribute to the hybrid zone and inclusion of climatic variables from the more arid western range may bias results. Eulemur cinereiceps occupies a narrow range in eastern Madagascar, from the Andringitra Massif to the Mananara River (Johnson et al. 2008; Mittermeier et al. 2008), though with an isolated population described south of this river (H. Andriamaharoa, pers. comm.). The hybrids are distributed in the mountainous Andringitra region, which serves as a biogeographical crossroads between eastern and central Madagascar (Goodman and Lewis 1996). Based on analysis of morphological, microsatellite, and mtDNA markers, the hybrid zone extends up to 70 km in width, from Ambondro in the north to Ankorabe in the south (Delmore et al. 2011, 2013; Fig. 1).
Table I

Performance of Maxent models for Eulemur cinereiceps, E. rufifrons, and their hybrids in southeastern Madagascar, number of presence locations used to build ecological niche models (ENMs), and data sources for presence locations


N a


Omission errorc

Fold 1

Fold 2

Fold 3

Fold 4

Fold 5

E. cinereiceps 1


0.990 (0.01)

0.07, P < 0.0001

0, P < 0.0001

0.07, P < 0.0001

0, P < 0.0001

0.07, P < 0.0001

E. rufifrons 2


0.842 (0.02)

0.02, P < 0.0001

0.27, P < 0.0001

0.34, P = 0.002

0.36, P = 0.0003

0.21, P < 0.0001



0.967 (0.01)

0, P < 0.0001

0.18, P < 0.0001

0.09, P < 0.0001

0.27, P < 0.0001

0, P < 0.0001

AUC [area under the receiver-operating characteristic (ROC) curve] values show average and standard deviation (SD) of fivefold cross-validation and the test omission rate and P-value of the one tail binomial test for each fold.

References: 1Delmore et al. 2011, 2013; Ingraldi 2010; Johnson 2002; 2Delmore et al. 2011, 2013; 3Delmore et al. 2011, 2013; Johnson 2002.

aNumber of presence locations used to build models.

bArea under the ROC curve.

cOptimum threshold determined where sensitivity equals specificity.

Fig. 1

Locations of background environments for Eulemur rufifrons, hybrids, and E. cinereiceps. Points representing the centroid of each pixel (30 arc-seconds) within these ranges were selected for background tests. Limits of the ranges correspond to known river barriers, forest within the eastern escarpment and lowland fragments where these taxa have been recorded, and midpoints between the most proximate parental and hybrid sampling localities.

ENM Construction

Presence locations for ENMs were drawn from the entirety of the range of Eulemur cinereiceps, the hybrid zone, and the eastern range of E. rufifrons. The final set of presence locations was refined in order to avoid spatial autocorrelation and to match the spatial resolution of environmental variables (see later). Systematic sampling was conducted by creating a 1-km grid encompassing all of the presence locations and randomly sampling one occurrence per species per grid cell (Fourcade et al. 2014). This resulted in a dataset of 83 presence locations (21 E. cinereiceps, 25 hybrid, and 37 E. rufifrons; Table I).

We developed ENMs using Maxent v3.3.3k (Phillips et al. 2006). Maxent relies on presence-only records to estimate the probability of occurrence for a species. It finds the probability distribution that is most spread out, or closest to uniform, i.e., maximum entropy, and then constrains that distribution by the values of environmental variables at locations where the species is known to occur (Phillips et al. 2006). Using the mean of environmental values to constrain the distribution minimizes overfitting. Maxent performs well in comparison to other approaches (Elith et al. 2006) and with small numbers of presence locations (Pearson et al. 2007). With the exception of cross-validation (see later), we relied on the default settings for all model parameters.

We developed Maxent models using fivefold cross-validation and model performance was assessed on the held-out, i.e., test folds (Elith et al. 2011). We constructed receiver-operating characteristic (ROC) curves for each fold and used the area under the curve (AUC) to compare model performance (Fielding and Bell 1997). We calculated the test omission rate for each fold through binary predictions using equal sensitivity and specificity; we used a one-tailed binomial test to investigate whether the observed omission rate was better than expected compared to a random prediction (Anderson et al. 2002). The AUC is a threshold-independent test statistic that measures the ability of a model to discriminate between sites where a species is present and those where it is absent, which indicates the efficacy of the model for prioritizing areas in terms of their relative importance as habitat for a species. However, Maxent is a presence-only algorithm; therefore we used the approach suggested by Phillips et al. (2006), applying randomly selected pseudo-absences instead of observed absences to AUC; we drew background selections from within the known ranges of the three taxa in eastern Madagascar (Fig. 1). The AUC ranges from 0 to 1, where 1 indicates perfect discrimination, 0.5 suggests predictive discrimination is no better than random, and values below 0.5 imply performance worse than random. We adopted the interpretation offered by Hosmer and Lemeshow (2000) whereby an AUC value of 0.7–0.8 is considered an acceptable prediction; 0.8–0.9 is excellent and >0.9 is outstanding. We ran the models with increasing regularization multiplier settings (2 and 3) but the default multiplier (1) showed the highest model performance (highest AUC) and the lowest variability; furthermore, increasing regularization did not improve model overfitting in two of the three taxa (results not shown). We did not use spatially independent calibration data, which may have inflated performance estimates (Radosavljevic and Anderson 2014); nonetheless, our background tests using Maxent generated ENMs (see later) do not suggest systematic biases leading to overfitting (see Results).

We constructed the ENMs using six environmental variables: four climate variables, a measure of forest cover, and elevation. The climate data were downloaded from the WorldClim database (Hijmans et al. 2005). We reduced the number of WorldClim climate variables to four (from 19), based on a combination of expert knowledge about the physiological and life history requirements of lemurs, as well as correlation analysis. For the latter, we removed collinear variables with Pearson correlation >0.90 (Syfert et al. 2013). One of the four retained climate variables represented temperature trends (mean annual temperature); another represented temperature seasonality (standard deviation of temperature), while the others represented precipitation trends (mean annual precipitation) and drought incidence (precipitation of driest month). We derived elevation from a digital elevation model (DEM; USGS 2012) and obtained forest cover from layers classified by Conservation International using 2005 satellite imagery (Harper et al. 2007). We upscaled the forest layer to match the resolution of the WorldClim data as it was based on Landsat imagery (30 m resolution). The upscaling process also involved converting the categorical forest variable (forest/nonforest) to a gradient representation (percent forest cover). This was accomplished by the use of a 3 × 3 moving window averaging all of the forest pixels that fell within in each final 30 arc-second pixel. We clipped all variables using the forest layer as the mask so as to include only areas that had greater than zero percent forest cover. We completed all analyses using a Mercator projection in ArcGIS v10.1 (ESRI 2012).

Quantitative Analyses of Niche Divergence

We quantified potential niche divergence using three methods. First, we used the identity test implemented in ENMtools (Warren et al. 2008). This test uses a randomization procedure to evaluate whether the ENMs of taxa are more different than expected by chance. We generated pseudoreplicate datasets by pooling occurrence points, randomizing species identities, and creating two new samples of the same sizes as the original samples. We created ENMs for these replicates and estimated niche divergence using two measures of niche overlap: Schoener’s D and Hellinger’s I (Warren et al. 2008). These measures compare estimates of habitat suitability for each taxonomic group in the full study area; values range from 0 (no overlap) to 1 (complete overlap). D assumes that habitat suitability scores are proportional to species abundance while I treats scores as probability distributions. We repeated this randomization procedure 100 times to generate a null distribution and used a one-tailed test to compare observed values of niche overlap to the null distribution.

Second, we used the background test implemented in ENMtools (Warren et al. 2008). Similar to the identity test, this test uses ENMs and a randomization procedure; it aims to determine if the ENM of one taxon predicts that of a second better than expected by chance. Briefly, we compared the ENM of one taxon (A) to another ENM created by choosing a random set of points from the background of the opposing taxon (B). We created background environments by selecting the centroid of every pixel within the ranges of the three taxa within eastern Madagascar (Fig. 1). Where ranges overlapped in the hybrid zone, we created boundaries between Eulemur rufifrons and the hybrid background and E. cinereiceps and the hybrid background with a horizontal line (along the corresponding parallel of latitude) dividing them at the midpoint between the two most proximal presence points, i.e., between E. rufifrons and hybrids at the northern edge of the hybrid zone and between E. cinereiceps and hybrids at the southern limit of the zone (Fig. 1). The number of random points drawn from taxon B’s background is equivalent to the number of occurrence points for taxon B. We compared these ENMs using Schoener’s D and Hellinger’s I. We repeated this procedure 100 times and in both directions to generate two null distributions. We used two-tailed tests to compare observed values of niche overlap between taxon A and B. We considered the niches of taxa diverged if the observed value fell below the 95% CI of the null distribution; i.e., measures of niche overlap are lower. Following Blair et al. (2013), we used a conservative approach when interpreting these results; if the test was significant in one direction but not the other, we failed to reject the null. We took the same approach when results were in opposite directions for the two distributions.

Finally, we used the background test described by McCormack et al. (2010) using custom scripts in R 3.0.3 (R Development Core Team 2014) available in the Electronic Supplementary Material. This is a multivariate method that does not rely on ENMs. Instead, we extracted environmental variables from occurrence points and a random set of background points. We calculated differences for each variable between taxonomic groups and compared them to a null distribution (generated by calculating the difference between background points using a bootstrapping approach and 1000 resamples). We used a two-tailed test, the same environmental variables used to construct ENMs, and the same background used in the previous test to assess niche overlap among taxa. We considered the niches of taxa diverged if the observed value fell above the 95% CI of the null distribution; i.e., differences between background points were less than differences between the occurrence points. The advantage of this background test vs. Warren et al. (2008) is that environmental data are not summarized into a single value, allowing us to examine each variable separately. For simplicity, we term these two analyses the Warren background test and McCormack background test. We note that comparisons among nonsister taxa, i.e., Eulemur cinereiceps and E. rufifrons, may be problematic on theoretical grounds, as we cannot deduce whether niche divergence might have occurred between these species and a possibly large number of more closely related, now extinct lineages since these species originally diverged evolutionarily (Losos 2008; Losos and Glor 2003). However, the parental species are very closely related (Markolf and Kappeler 2013), and there is no evidence for niche divergence in more recent splits in the brown lemur complex (Blair et al. 2013). These comparisons nonetheless should be interpreted with caution.

Ethical Note

We obtained location points for study taxa in previous research and reanalyzed them here (see earlier; Table I). Original sources provide details regarding animal handling procedures. U.S. Fish and Wildlife Services, Institutional Animal Care and Use Committee (Omaha’s Henry Doorly Zoo and Aquarium) and Animal Care Committee (University of Calgary) approved all procedures. The research adhered to the legal requirements of the government of Madagascar.


The Maxent models showed strong discrimination on held out folds, with a mean cross-validated AUC of 0.989 (SD = 0.01) for Eulemur cinereiceps, 0.967 (SD = 0.01) for hybrids, and 0.836 (SD = 0.02) for E. rufifrons (Table I). These AUC values suggest excellent to outstanding discrimination (Hosmer and Lemeshow 2000; Table I) and were further supported by significant binomial tests (omission error, 0–0.36; all folds significant at P < 0.0001, except fold 3 for E. rufifrons with P = 0.002; Table I).

Visual inspection of ENMs indicated that areas with a high probability of occurrence in models for each taxon closely aligned with their known distributions (Figs. 1 and 2). Observed values of niche overlap (Schoener’s D and Hellinger’s I) are provided in Table II for each comparison. Congruent with visual inspection of the ENMs, we rejected the null hypothesis of niche equivalency in all cases; values of niche overlap were lower than expected by chance (Table II; Fig. 3).
Fig. 2

Ecological niche models (ENMs) showing the probability of species presence (logistic output) ranging from 0 to 1 for (A) Eulemur cinereiceps, (B) E. rufifrons, and (C) hybrids in Madagascar. Darker colors represent high probability of species presence. Each distribution is based on models developed using fivefold cross-validation.

Table II

Observed values of niche overlap and results from identity and background tests for Eulemur cinereiceps, E. rufifrons, and hybrids in southeastern Madagascar


E. rufifrons vs. E. cinereiceps

E. rufifrons vs. hybrid

Hybrid vs. E. cinereiceps

Observed niche overlap











P < 0.01

P < 0.01

P < 0.01


P < 0.01

P < 0.01

P < 0.01





Warren background


P < 0.01, P < 0.01

P = 0.08, P = 0.01

P = 0.38, P = 0.34


P < 0.01, P < 0.01

P = 0.04, P = 0.01

P = 0.26, P = 0.24





McCormack background

Forest cover

4.03 C (10.52, 22.74)

13.7 C (25.02, 36.11)

9.76 C (40.05, 53.48)

MA rainfall

363 D (147.5, 238.8)

427 C (485.2, 542.4)

790 D (668.7, 752.0)

MA temperature

3.4 D (2.39, 2.78)

0.66 D (0.012, 0.33)

2.74 (2.60, 2.94)

SD temperature

1.2 D (0.78, 1.02)

0.23 D (0.032, 0.15)

0.97 D (0.70, 0.90)

Driest month

2.87 D (1.20, 1.92)

2.08 C (2.83, 3.23)

4.95 D (4.31, 4.89)


692.5 D (459.5, 553.6)

45.5 C (100.1, 168.3)

647 (602.9, 679.2)

Identical vs. nonidentical may be more appropriate terms for the inferences from identity tests; however, we use convergence and divergence here for simplicity of comparison to other tests. Statistics for the Warren background tests are provided for each pairwise comparison, i.e., the ENM of taxon A in the background of taxon B and vice versa. Values of niche overlap (Hellinger’s I and Schoener’s D) are used in these tests, and inferences are provided below statistics. Note that a convergence or divergence inference would be indicated only if 1) both comparisons are significantly different from null and 2) both comparisons indicate either convergence or divergence, i.e., the same direction. Each environmental variable is examined separately in the McCormack background test (MA = mean annual). Variables showing significant divergence (D) or conservatism (C) are shown in bold. Confidence intervals (95%) for the null distribution of background divergence are shown in parentheses.

Fig. 3

Results from identity tests evaluating whether ENMs of Eulemur cinereiceps, E. rufifrons, and their hybrids in southeastern Madagascar are more different than expected by chance. Observed niche overlap values are shown with red arrows and null distributions generated using a randomization procedure are shown in black. Table II includes raw values for niche overlap and P-values.

Results were less clear when incorporating background divergence into the null hypothesis. Using ENMs and randomization tests in the Warren background test, we found that the niches of Eulemur rufifrons and E. cinereiceps were more similar than expected by chance, i.e., the observed value of niche overlap fell above the 95% CI (Table II; Fig. 4a and d). We were unable to reject the null hypothesis consistently (and in the same direction) in the remaining two comparisons (Fig. 4b and e, c and f); thus the most conservative interpretation is that the observed divergence in niches viewed in ENMs (Fig. 2), and nonidentical niches shown in the identity test (Fig. 3) may be explained primarily by differences in the environmental background of taxa. However, results for the comparison between E. rufifrons and hybrids may indicate more nuanced patterns: niche conservatism for hybrids and niche divergence for E. rufifrons (although there was a trend for divergence only when quantifying niche overlap using Hellinger’s I; Table II; Fig. 4b and e). This result suggests that hybrids occupy regions within their environment more similar to the background of E. rufifrons, while E. rufifrons inhabits regions within its environment that are more different from the background of hybrids than would be expected by chance (Table II; Fig. 4b and e).
Fig. 4

Results from background tests to determine if the ENM of one taxon predicts that of a second better than expected by chance. Taxa include Eulemur cinereiceps, E. rufifrons, and their hybrids in southeastern Madagascar. Schoener’s D is indicated in the top panels and Hellinger’s I in the bottom panels. Observed niche overlap values are shown with black arrows; null distributions generated using a randomization procedure are shown. Colors correspond to the focal species in each comparison; e.g., a and d show comparison of E. rufifrons distributions (blue) in the E. cinereiceps background (red). C (conservatism) and D (divergence) denote significant differences from null; see Table II for raw values and P-values.

Using the multivariate McCormack background test, we documented significant differences from null in all six of the environmental variables compared between Eulemur rufifrons and E. cinereiceps. We found niche divergence for five of the six environmental variables, i.e., the observed difference was greater than the 95% CI; the remaining variable exhibited niche conservatism; i.e., the observed difference was less than the 95% CI (Table II). When considering E. rufifrons and hybrids, all six environmental variables showed significant differences; we recorded niche divergence for two of these variables and niche conservatism in the remaining four variables (Table II). Four of six environmental variables exhibited significant differences between hybrids and E. cinereiceps. Three of these variables showed niche divergence and one showed conservatism (Table II). All three comparisons exhibited niche conservatism in forest cover, but niche divergence in the standard deviation (SD) of temperature (Table II).


Testing Models for the Maintenance of the Andringitra Hybrid Zone

Our results indicate variable support for each of the three models for hybrid zone stability: the tension zone, the geographical selection-gradient, and the bounded superiority models (Table III). The tension zone and the geographical selection-gradient models both invoke selection against hybrids, with the former indicating endogenous selection, such as reproductive impairment, and the latter exogenous (environmental) selection (Barton and Hewitt 1985; Endler 1977; Moore and Price 1993); the bounded superiority model also involves exogenous selection, but in this case favoring hybrids within particular ecotones (Good et al. 2000; Moore 1977). Under the tension zone model, we predicted substantial overlap among ENMs across parentals and hybrids (or niche conservatism), because adaptation to local environmental conditions was not expected to delimit boundaries across populations (Barton and Hewitt 1985). However, visual inspection of the ENMs indicated well-demarcated ranges across these brown lemur populations, at least in terms of the areas with the highest predicted suitability (Fig. 2). Furthermore, identity tests, which compare the overlap of ENMs through randomization procedures (Warren et al. 2008), indicated that all three taxa were significantly diverged in their niches, i.e., had nonidentical niches. However, under the Warren background tests, we found convergence between the two parental species, while the hybrid-parental comparisons showed no difference from null. Using a conservative interpretation wherein a lack of clear divergence suggests niche conservatism (Blair et al. 2013), these results support the tension zone model. On the other hand, the McCormack background tests (McCormack et al. 2010) suggested niche conservatism for forest cover, divergence for mean annual temperature, and conflicting patterns across the remaining variables (Table II). In all, our results offer only mixed support for the tension zone model, the most commonly cited for hybrid zone stability across many animal species (e.g., Alexandrino et al. 2005; Bronson et al. 2003; Kawakami et al. 2008).
Table III

Predictions for models of hybrid zone stability from previously published work (a–d) and the present study (e) for Eulemur cinereiceps, E. rufifrons (parentals), and their hybrids in southeastern Madagascar


Tension zone

Geographical selection-gradient

Bounded superiority


Population density in hybrid zone



≥ Density in parental range


Hybrid zone width

Narrow relative to dispersal

Narrow relative to dispersal

Width of ecological correlate

b, c

Hybrid zone composition

Parentals and F1 hybrids

Parentals and F1 hybrids

Later generation hybrids

b, c, d

Hybrid fitness



≥ Parental fitness


Cline shape and coincidence

Sigmoid, coincident

Sigmoid, coincident

Variable, noncoincident


ENM vs. geographic range

No correspondence

ENMs of parentals should overlap at hybrid zone



Niche conservatism vs. divergence

Niche conservatism

Niche divergence between parentals; niche conservatism between hybrids and parentals

Niche divergence


When evidence supports a single prediction, bold text is used. Conflicting results for the same prediction are indicated in italics (see text).

a = Irwin et al. 2005; b = Delmore et al. 2011 (morphological and pelage data); c = Delmore et al. 2013 (genetic data); d = Wyner et al. 2002 (genetic data); e = current study.

*Identity test and McCormack background test support divergence in this category; Warren background test supports conservatism.

Under the geographical selection-gradient model, we predicted that the ENMs of parental forms would meet at the hybrid zone and overlap substantially with the ENM of hybrids (Table III). According to visual inspection of the modeled niches, this was not the case: the areas of high suitability in the ENMs of parental forms did not meet, and there was limited overlap between the ENMs of hybrids and both parental forms (Fig. 2). Indeed, the transitions between parental forms and hybrids coincided well with boundaries of the hybrid zone documented through genetic and morphological evidence (Delmore et al. 2013). We also predicted that niche divergence would be documented between parental forms under this model, but not between hybrid and parental forms. Identity test results support the prediction of nonidentical niches between parentals, but they did not support the prediction that hybrids and parentals would be identical. Meanwhile, the Warren background tests showed no support for the former prediction, but support for the latter (Tables II and III). Finally, with the McCormack background tests, the finding of divergence between parentals and convergence (or no difference from null) between hybrids and each parental in elevation does fit the predictions of the geographical selection-gradient model. Nonetheless, the weight of evidence does not provide clear support for this model.

Instead, we found stronger support for the bounded superiority model, which predicts niche separation across both parental species and hybrids (Flockhart and Wiebe 2009; Good et al. 2000; Moore 1977). Both visual inspection of high-suitability areas (Fig. 2) and identity tests (Table II) indicate divergence, i.e., nonidentical niches, in the ENMs of hybrids and parentals. While the Warren background test suggested niche conservatism, we did document significant divergence for mean standard deviation of temperature using McCormack background tests (Table II). As each environmental variable used to construct ENMs was examined separately in this set of analyses, it is possible that only temperature seasonality is relevant to niche divergence and the maintenance of reproductive isolation in this system; the inclusion of additional variables which do not distinguish among parentals and hybrids in the ENMs may have swamped out any signal of divergence along relevant niche axes in the Warren background tests.

We also note that although the overall results from the Warren background tests suggest niche conservatism, we did document some trends that may support niche divergence. Specifically, there was a trend toward niche divergence in the comparisons between hybrids and each parental species (Fig. 3). For example, it appears that while hybrids are occupying a portion of their environment that is similar to the background environment of Eulemur rufifrons, E. rufifrons is occupying a portion of its environment that is different than the hybrid background (Fig. 3b and e). This could be considered niche divergence and would support both the geographical selection-gradient and bounded superiority models.

We thus view our analyses as cautious support for the bounded superiority model, but we cannot rule out the tension zone or geographical selection-gradient models across all lines of evidence. This assessment is largely congruent with findings from previous research in the Andringitra brown lemur hybrid zone (Table III). These investigations have shown no indication of strong selection against hybrid forms, as suggested by body condition and the shape and noncoincidence of genetic and morphological clines across the zone (Delmore et al. 2011, 2013). Furthermore, the parental species and hybrids appear to vary in ecological adaptations, such as overall and scarce-season diets (Johnson 2007). We caution, however, that such ecological differences do not provide definitive evidence of niche divergence. As suggested by the Warren background tests, brown lemurs, regardless of ancestry, might selectively use similar habitats (and perform similar ecological behaviors) when they are available; it could instead be differences in available environmental types that cause these taxa to appear distinct in their modeled distributions (Godsoe 2010; Warren et al. 2008). A recent study comparing ENMs across Eulemur also found niche conservatism in all brown lemur sister species pairs (Blair et al. 2013).

Issues of Scale, Resolution, and Variables for Ecological Niche Models

There are inherent difficulties when attempting to infer small-scale processes —such as difference in how animals interact with their local environments— from large-scale analyses as presented here. Indeed there are real concerns that ENMs are at best indirect, coarse predictors of the ecological niches and spatial distribution of organisms, and debate continues about which niche dimensions are predicted from these models (Warren 2012). One concern for ENMs is the potential for high levels of environmental heterogeneity to influence model results. Specifically, a high degree of variation in the background range can bias results toward niche divergence (Blair et al. 2013; Godsoe 2010). Given the considerably larger range of Eulemur rufifrons, it is likely that the distribution of this species maintains greater environmental variation (which would have been even greater had we included the arid west). Nonetheless, we found no evidence of niche divergence in the Warren background tests, suggesting our results are sufficiently conservative despite potential environmental heterogeneity in background ranges.

While background tests are designed to assess the differential use of habitats within larger ranges (McCormack et al. 2010; Warren et al. 2008, 2010), it may be difficult to determine potential niche divergence at small spatial scales. Although we were able to detect niche divergence according to mean annual temperature in the McCormack background test, other climate variables showed conflicting patterns across comparisons, no difference from null, or niche conservatism (Table II). This suggests that Eulemur cinereiceps, E. rufifrons, and their hybrids would be similar in niche requirements and behavior in identical environments, despite recorded differences in dietary ecology in allopatry (Johnson 2007; Overdorff 1993). Two species of mouse lemurs (Microcebus murinus and M. griseorufus) and their hybrids demonstrate a somewhat contrasting but instructive pattern in southeastern Madagascar. Unlike the brown lemurs, the parentals are broadly similar in allopatry (Rakotondranary and Ganzhorn 2011). However, where they overlap in Andohahela National Park, they occupy different microhabitats according to tree size and differ in temporal occupancy via extended torpor in M. murinus; meanwhile hybrids in these environments are more generalist, overlapping with the niches of both parentals (Rakotondranary and Ganzhorn 2011). In general, differential adaptation to environmental types and gradients may play a role in the extent and direction of hybridization in this system (Hapke et al. 2011). Interestingly, a recent study by Kamilar et al. (2015) indicated higher niche overlap using identity tests between these mouse lemur species (I = 0.779, D = 0.476) than recorded in our study taxa. It is possible that some of the evidence pointing to niche conservatism in our regional-scale analysis might obscure similar patterns of divergence among brown lemurs in shared habitats, where reinforcement of population differences is most critical. Indeed, researchers have long recognized subtle niche partitioning in sympatric Eulemur species (Tattersall and Sussman 1998).

In addition to appropriate spatial scales, the selection of relevant variables is an important step in building ENMs (Elith and Leathwick 2009; Warren 2012). Given the ongoing reduction and fragmentation of Madagascar’s forest ecosystems (Allnutt et al. 2008; Harper et al. 2007) with likely substantial impacts on lemur populations (Schwitzer et al. 2014), it may be prudent to consider human factors when attempting to predict lemur distributions and niches. Kamilar and Tecot (2016, this volume) found that including anthropogenic disturbance variables, e.g., distance to human settlements, in Eulemur ENMs improved their performance over models using climate variables alone. Such predictors may have altered the ENMs of the Andringitra hybrids and parental species in our analyses, including the degree of niche overlap. We note, however, that we included an implicit measure of ecosystem integrity —percentage of forest cover— which did not indicate niche divergence in these taxa (Table II).

Implications for Evolutionary Processes and Conservation

In summary, aspects of our results are consistent with all three models of hybrid zone stability, with the bounded superiority model receiving support from the most lines of evidence. This model underscores the potential importance of ecological selection in the maintenance of reproductive isolation and speciation (Nosil 2012; Schluter 2009) and highlights the creative role hybridization could play in these evolutionary processes. Under the bounded superiority model, hybrids occupy transitional or novel habitats and exist relatively independent of parental forms, which could represent the early stages of hybrid speciation (Mallet 2007). Given the lack of empirical support for the bounded superiority model in other systems, future work to disentangle models of hybrid zone stability in this zone will be important. The collection of behavioral data from individuals of distinct ancestry in shared environments could inform this question by identifying differences in the ecological strategies of hybrid and parental forms. Such research could provide insight into the mechanisms behind the patterns observed in the present study, at the appropriate local spatial scales.

If climatic niche divergence among hybrids and parentals maintains the stability of the zone over time, then encroachment into a neighboring taxon’s range should be limited. Thus, an expanding hybrid zone does not likely pose a risk to “pure” populations of Eulemur cinereiceps, a Critically Endangered species (IUCN 2015). However, although estimates of historic land cover change vary, deforestation remains a persistent threat to Madagascar’s ecosystems (Agarwal et al. 2005; Grinand et al. 2013; Harper et al. 2007; Ingram and Dawson 2005; McConnell and Kull 2014). Habitat loss could have the secondary effect of disrupting hybrid zone stability, potentially increasing hybridization and posing particular risks for species with small ranges and populations (Detwiler et al. 2005) such as E. cinereiceps (Brenneman et al. 2012; Irwin et al. 2005). Therefore, preserving the integrity of the forests surrounding the Andringitra Massif may be crucial for conserving this species, as well as the evolutionary and ecological processes that maintain the unique brown lemur hybrid zone.



The authors express their appreciation to Giuseppe Donati for the invitation to contribute to this special issue. The authors also thank the government of Madagascar for permission to conduct the original research reanalyzed here. The authors are grateful for the funding that contributed to the original field research: Primate Action Fund, Primate Conservation, Inc., American Society of Primatologists, National Science and Engineering Research Council of Canada, Omaha’s Henry Doorly Zoo and Aquarium, and University of Calgary. Omaha’s Henry Doorly Zoo and Aquarium Center for Conservation and Research, Madagascar Biodiversity Partnership and Madagascar Institut pour la Conservation des Ecosystèmes Tropicaux facilitated the fieldwork. The authors also thank Alison Porter for assistance with analyses. Finally, we thank three anonymous reviewers and the editor-in-chief for their helpful suggestions to improve the manuscript.

Supplementary material

10764_2015_9872_MOESM1_ESM.r (2 kb)
ESM 1 (R 2 kb)


  1. Agarwal, D. K., Silander, J. A., Gelfand, A. E., Dewar, R. E., & Mickelson, J. G. (2005). Tropical deforestation in Madagascar: analysis using hierarchical spatially explicit, Bayesian regression models. Ecological Modelling, 185, 105–131.CrossRefGoogle Scholar
  2. Alexandrino, J., Baird, S. J. E., Lawson, L., Macey, J. R., Moritz, C., & Wake, D. B. (2005). Strong selection against hybrids at a hybrid zone in the Ensatina ring species complex and its evolutionary implications. Evolution, 59, 1334–1347.Google Scholar
  3. Allnutt, T. F., Ferrier, S., Manion, G., Powell, G. V. N., Ricketts, T. H., Fisher, B. L., Harper, G. J., Irwin, M. E., Kremen, C., Labat, J. N., Lees, D. C., Pearce, T. A., & Rakotondrainibe, F. (2008). A method for quantifying biodiversity loss and its application to a 50-year record of deforestation across Madagascar. Conservation Letters, 1, 173–181.CrossRefGoogle Scholar
  4. Anderson, E. (1949). Introgressive hybridization. New York: John Wiley & Sons.CrossRefGoogle Scholar
  5. Anderson, R. P., Peterson, A. T., & Gomez-Laverde, M. (2002). Using niche-based GIS modeling to test geographic predictions of competitive exclusion and competitive release in South American pocket mice. Oikos, 98, 3–16.Google Scholar
  6. Arnold, M. L. (1992). Natural hybridization as an evolutionary process. Annual Review of Ecology and Systematics, 12, 237–261.CrossRefGoogle Scholar
  7. Barton, N. H., & Hewitt, G. M. (1985). Analysis of hybrid zones. Annual Review of Ecology and Systematics, 16, 113–148.CrossRefGoogle Scholar
  8. Bigelow, R. S. (1965). Hybrid zones and reproductive isolation. Evolution, 19, 449–458.Google Scholar
  9. Blair, M. E., Sterling, E. J., Dusch, M., Raxworthy, C. J., & Pearson, R. G. (2013). Ecological divergence and speciation between lemur (Eulemur) sister species in Madagascar. Journal of Evolutionary Biology, 26, 1790–1801.CrossRefPubMedGoogle Scholar
  10. Brenneman, R. A., Johnson, S. E., Bailey, C. A., Ingraldi, C., Delmore, K. E., Wyman, T. M., Andriamaharoa, H. E., Ralainasolo, F. R., Ratsimbazafy, J. H., & Louis, E. E. (2012). Population genetics and abundance of gray-headed lemurs (Eulemur cinereiceps): potential consequences of fragmentation and cyclones in an endangered primate. Oryx, 462, 298–307.CrossRefGoogle Scholar
  11. Bronson, C. L., Grubb, T. C., & Braun, M. J. (2003). A test of the endogenous and exogenous selection hypotheses for the maintenance of a narrow avian hybrid zone. Evolution, 57, 630–637.CrossRefPubMedGoogle Scholar
  12. Delmore, K. E., Louis, E. E., & Johnson, S. E. (2011). Morphological characterization of a brown lemur hybrid zone (Eulemur rufifrons × E. cinereiceps). American Journal of Physical Anthropology, 145, 55–66.CrossRefPubMedGoogle Scholar
  13. Delmore, K. E., Brenneman, R. A., Lei, R., Bailey, C. A., Brelsford, A., Louis, E. E., & Johnson, S. E. (2013). Clinal variation in a brown lemur (Eulemur spp.) hybrid zone: combining morphological, genetic and climatic data to examine stability. Journal of Evolutionary Biology, 26, 1677–1690.CrossRefPubMedGoogle Scholar
  14. Detwiler, K. M., Burrell, A. S., & Jolly, C. J. (2005). Conservation implications of hybridization in African cercopithecine monkeys. International Journal of Primatology, 26, 661–684.CrossRefGoogle Scholar
  15. Elith, J., & Leathwick, J. R. (2009). Species distribution models: ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics, 40, 677–697.CrossRefGoogle Scholar
  16. Elith, J., Graham, C. H., Anderson, R. P., Dudik, M., Ferrier, S., Guisan, A., Hijmans, R. J., Huettmann, F., Leathwick, J. R., Lehmann, A., Li, J., Lohmann, L. G., Loiselle, B. A., Manion, G., Moritz, C., Nakamura, M., Nakazawa, Y., Overton, J. M., Peterson, A. T., Phillips, S. J., Richardson, K., Scachetti-Pereira, R., Schapire, R. E., Soberon, J., Williams, S., Wisz, M. S., & Zimmermann, N. E. (2006). Novel methods improve prediction of species' distributions from occurrence data. Ecography, 29, 129–151.CrossRefGoogle Scholar
  17. Elith, J., Phillips, S. J., Hastie, T., Dudik, M., Chee, Y. E., & Yates, C. J. (2011). A statistical explanation of MaxEnt for ecologists. Diversity and Distributions, 17, 43–57.CrossRefGoogle Scholar
  18. Endler, J. A. (1977). Geographic variation, speciation and clines, vol. 10. Princeton: Princeton University Press.Google Scholar
  19. ESRI. (2012). ArcGIS desktop: Release 10.1. Redlands: Environmental Systems Research Institute.Google Scholar
  20. Fielding, A. H., & Bell, J. F. (1997). A review of methods for the assessment of prediction errors in conservation presence/absence models. Environmental Conservation, 21, 38–49.CrossRefGoogle Scholar
  21. Flockhart, D. T. T., & Wiebe, K. L. (2009). Absence of reproductive consequences of hybridization in the northern flicker (Colaptes auratus) hybrid zone. The Auk, 126, 351–358.CrossRefGoogle Scholar
  22. Fourcade, Y., Engler, J. O., Rödder, D., & Secondi, J. (2014). Mapping species distributions with MAXENT using a geographically biased sample of presence data: a performance assessment of methods for correcting sampling bias. PLoS ONE, 9, e97122.CrossRefPubMedPubMedCentralGoogle Scholar
  23. Franklin, J. (2009). Mapping species distributions. Cambridge: Cambridge University Press.Google Scholar
  24. Gligor, M., Ganzhorn, J. U., Rakotondravony, D., Ramalijaona, O. R., Razafimahatratra, E., Zischler, H., & Hapke, A. (2009). Hybridization between mouse lemurs in an ecological transition zone. Molecular Ecology, 18, 520–533.CrossRefPubMedGoogle Scholar
  25. Godsoe, W. (2010). Regional variation exaggerates ecological divergence in niche models. Systematic Biology, 59, 298–306.CrossRefPubMedPubMedCentralGoogle Scholar
  26. Good, T. P., Ellis, J. C., Annett, C. A., & Pierotti, R. (2000). Bounded hybrid superiority in an avian hybrid zone: effects of mate, diet, and habitat choice. Evolution, 54, 1774–1783.CrossRefPubMedGoogle Scholar
  27. Goodman, S. M., & Lewis, B. A. (1996). Description of the Réserve Naturelle Intégrale d'Andringitra, Madagascar. Fieldiana: Zoology, 85, 7–19.Google Scholar
  28. Grinand, C., Rakotomalala, F., Gond, V., Vaudry, R., Bernoux, M., & Vieilledent, G. (2013). Estimating deforestation in tropical humid and dry forests in Madagascar from 2000–2010 using multi-date Landsat satellite images and the random forests classifier. Remote Sensing of Environment, 139, 68–80.CrossRefGoogle Scholar
  29. Hapke, A., Gligor, M., Rakotondranary, S. J., Rosenkranz, D., & Zupke, O. (2011). Hybridization of mouse lemurs: different patterns under different ecological conditions. BMC Evolutionary Biology, 11, 297.CrossRefPubMedPubMedCentralGoogle Scholar
  30. Harper, G. J., Steininger, M. K., Tucker, C. J., Juhn, D., & Hawkins, F. (2007). Fifty years of deforestation and forest fragmentation in Madagascar. Environmental Conservation, 34, 1–9.CrossRefGoogle Scholar
  31. Hewitt, G. (1988). Hybrid zones—natural laboratories for evolutionary studies. Trends in Ecology and Evolution, 3, 158–167.CrossRefPubMedGoogle Scholar
  32. Hijmans, R. J., Cameron, S. E., Parra, J. L., Jones, P., & Jarvis, A. (2005). Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology, 25, 1965–1978.CrossRefGoogle Scholar
  33. Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression (2nd ed.). New York: John Wiley & Sons.CrossRefGoogle Scholar
  34. Ingraldi, C. (2010). Forest fragmentation and edge effects on eight sympatric lemur species in southeast Madagascar. Master's thesis, University of Calgary.Google Scholar
  35. Ingram, J. C., & Dawson, T. P. (2005). Inter-annual analysis of deforestation hotspots in Madagascar from high temporal resolution satellite observations. International Journal of Remote Sensing, 26, 1447–1461.CrossRefGoogle Scholar
  36. Irwin, M. T., Johnson, S. E., & Wright, P. C. (2005). The state of lemur conservation in south-eastern Madagascar: population and habitat assessments for diurnal and cathemeral lemurs using surveys, satellite imagery and GIS. Oryx, 39, 204–218.CrossRefGoogle Scholar
  37. IUCN. (2015). The IUCN red list of threatened species. Retrieved from Accessed 6 Jan 2014.
  38. Johnson, S. E. (2002). Ecology and speciation in brown lemurs: White-collared lemurs (Eulemur albocollaris) and hybrids (Eulemur albocollaris × Eulemur fulvus rufus) in southeastern Madagascar. Unpublished Ph.D. dissertation, University of Texas at Austin, Austin, TX.Google Scholar
  39. Johnson, S. E. (2007). Evolutionary divergence in the brown lemur species complex. In L. Gould & M. L. Sauther (Eds.), Lemurs: Ecology and adaptation (pp. 187–210). New York: Springer Science+Business Media.Google Scholar
  40. Johnson, S. E., Lei, R., Martin, S. K., Irwin, M. T., & Louis, E. E. (2008). Does Eulemur cinereiceps exist? Preliminary evidence from genetics and ground surveys in southeastern Madagascar. American Journal of Primatology, 70, 372–385.CrossRefPubMedGoogle Scholar
  41. Kamilar, J. M., & Tecot, S. (2016). Anthropogenic and climatic effects on the distribution of Eulemur species: An ecological niche modeling approach. International Journal of Primatology, 37.Google Scholar
  42. Kamilar, J. M., Blanco, M., & Muldoon, K. M. (2015). Ecological niche modeling of mouse lemurs (Microcebus spp.) and its implications for their species diversity and biogeography. In S. M. Lehman, U. Radespiel, & E. Zimmerman (Eds.), Dwarf and mouse lemurs of Madagascar: Biology, behavior and conservation biogeography of the Cheirogaleidae. Cambridge: Cambridge University Press (in press)Google Scholar
  43. Kawakami, T., Butlin, R. K., Adams, M., Paull, D., & Cooper, S. J. B. (2008). Genetic analysis of a chromosomal hybrid zone in the Australian morabine grasshoppers (Vandiemenella, viatica species group). Evolution, 63, 139–152.CrossRefPubMedGoogle Scholar
  44. Key, K. H. L. (1968). The concept of stasipatric speciation. Systematic Biology, 17, 14–22.CrossRefGoogle Scholar
  45. Kozak, K. H., Graham, C. H., & Wiens, J. J. (2008). Integrating GIS-based environmental data into evolutionary biology. Trends in Ecology & Evolution, 23, 141–148.Google Scholar
  46. Lehman, S. M., & Wright, P. C. (2000). Preliminary study of the conservation status of lemur communities Betsakafandrika region of eastern Madagascar. Lemur News, 5, 23–25.Google Scholar
  47. Losos, J. B. (2008). Phylogenetic niche conservatism, phylogenetic signal and the relationship between phylogenetic relatedness and ecological similarity among species. Ecological Letters, 11, 995–1003.Google Scholar
  48. Losos, J. B., & Glor, R. E. (2003). Phylogenetic comparative methods and the geography of speciation. Trends in Ecology and Evolution, 18, 220–227.CrossRefGoogle Scholar
  49. Mallet, J. (2007). Hybrid speciation. Nature, 446, 279–283.CrossRefPubMedGoogle Scholar
  50. Markolf, M., & Kappeler, P. M. (2013). Phylogeographic analysis of the true lemurs (genus Eulemur) underlines the role of river catchments for the evolution of micro-endemism in Madagascar. Frontiers in Zoology, 10, 70.CrossRefPubMedPubMedCentralGoogle Scholar
  51. McConnell, W. J., & Kull, C. A. (2014). Deforestation in Madagascar: Debates over the island’s forest cover and challenges of measuring forest change. In I. R. Scales (Ed.), Conservation and environmental management in Madagascar (pp. 67–104). London: Routledge.Google Scholar
  52. McCormack, J. E., Zellmer, A. J., & Knowles, L. L. (2010). Does niche divergence accompany allopatric divergence in Aphelocoma jays as predicted under ecological speciation? Insights from tests with niche models. Evolution, 64, 1231–1244.PubMedGoogle Scholar
  53. Mittermeier, R. A., Konstant, W. R., Hawkins, F., Louis, E. E., Langrand, O., Ratsimbazafy, J., Rasoloarison, R., Ganzhorn, J. U., Rajaobelina, S., Tattersall, I., & Meyers, D. M. (2006). Lemurs of Madagascar (2nd ed.). Washington, DC: Conservation International.Google Scholar
  54. Mittermeier, R. A., Ganzhorn, J. U., Konstant, W. R., Glander, K., Tattersall, I., Groves, C. P., Rylands, A. B., Hapke, A., Ratsimbazafy, J., Mayor, M. I., Louis, E. E., Rumpler, Y., Schwitzer, C., & Rasoloarison, R. M. (2008). Lemur diversity in Madagascar. International Journal of Primatology, 29, 1607–1656.CrossRefGoogle Scholar
  55. Moore, W. S. (1977). An evaluation of narrow hybrid zones in vertebrates. The Quarterly Review of Biology, 52, 263–277.CrossRefGoogle Scholar
  56. Moore, W. S., & Price, J. T. (1993). Nature of selection in the northern flicker hybrid zone and its implications for speciation theory. In R. G. Harrison (Ed.), Hybrid zones and the evolutionary process (pp. 196–225). New York: Oxford University Press.Google Scholar
  57. Nosil, P. (2012). Ecological speciation. Oxford: Oxford University Press.CrossRefGoogle Scholar
  58. Overdorff, D. J. (1993). Similarities, differences, and seasonal patterns in the diets of Eulemur rubriventer and Eulemur fulvus rufus in the Ranomafana National Park, Madagascar. International Journal of Primatology, 14, 721–753.CrossRefGoogle Scholar
  59. Pastorini, J., Zaramody, A., Curtis, D. J., Nievergelt, C. M., & Mundy, N. I. (2009). Genetic analysis of hybridization and introgression between wild mongoose and brown lemurs. BMC Evolutionary Biology, 9, 32.CrossRefPubMedPubMedCentralGoogle Scholar
  60. Pearson, R. G., Raxworthy, C. J., Nakamura, M., & Peterson, A. T. (2007). Predicting species distributions from small numbers of occurrence records: a test case using cryptic geckos in Madagascar. Journal of Biogeography, 34, 102–117.CrossRefGoogle Scholar
  61. Peterson, A. T., Soberón, J., Pearson, R. G., Anderson, R. P., Martínez-Meyer, E., Nakamura, M., & Araújo, M. B. (2011). Ecological niches and geographic distribution. Princeton: Princeton University Press.Google Scholar
  62. Phillips, S. J., Anderson, R. P., & Schapire, R. E. (2006). Maximum entropy modeling of species geographic distributions. Ecological Modelling, 190, 231–259.CrossRefGoogle Scholar
  63. R Development Core Team. (2014). R: A language and environment for statistical computing. (Version 3.03). Vienna, Austria.Google Scholar
  64. Radosavljevic, A., & Anderson, R. P. (2014). Making better Maxent models of species distributions: complexity, overfitting and evaluation. Journal of Biogeography, 41, 629–643.Google Scholar
  65. Rakotondranary, S. J., & Ganzhorn, J. U. (2011). Habitat separation of sympatric Microcebus spp. in the dry spiny forest of south-eastern Madagascar. Folia Primatologica, 82, 212–223.CrossRefGoogle Scholar
  66. Schluter, D. (2001). Ecology and the origin of species. Trends in Ecology and Evolution, 18, 372–380.CrossRefGoogle Scholar
  67. Schluter, D. (2009). Evidence for ecological speciation and its alternative. Science, 323, 737–741.CrossRefPubMedGoogle Scholar
  68. Schwitzer, C., Mittermeier, R. A., Johnson, S. E., Donati, G., Irwin, M., Peacock, H., Ratsimbazafy, J., Razafindramanana, J., Louis, E. E., Jr., Chikhi, L., Colquhoun, I. C., Tinsman, J., Dolch, R., LaFleur, M., Nash, S., Patel, E., Randrianambinina, B., Rasolofoharivelo, T., & Wright, P. C. (2014). Averting lemur extinctions amid Madagascar's political crisis. Science, 343, 842–843.CrossRefPubMedGoogle Scholar
  69. Swenson, N. G. (2006). Gis-based niche models reveal unifying climatic mechanisms that maintain the location of avian hybrid zones in a North American suture zone. Journal of Evolutionary Biology, 19, 717–725.Google Scholar
  70. Swenson, N. G. (2008). The past and future influence of geographic information systems on hybrid zone, phylogeographic and speciation research. Journal of Evolutionary Biology, 21, 421–434.Google Scholar
  71. Syfert, M. M., Smith, M. J., & Coomes, D. A. (2013). The effects of sampling bias and model complexity on the predictive performance of MaxEnt species distribution models. PLoS ONE, 8, e55158.CrossRefPubMedPubMedCentralGoogle Scholar
  72. Tattersall, I., Sussman, R., & Sussman, R. (1998). 'Little brown lemurs' of northern Madagascar. Folia Primatologica, 69(Supplement 1), 379–388.CrossRefGoogle Scholar
  73. USGS. (2012). Global 30 Arc-Second Elevation (GTOPO30). Retrieved from Accessed 8 Nov 2014.
  74. Warren, D. L. (2012). In defense of ‘niche modeling.’. Trends in Ecology & Evolution, 27, 497–500.CrossRefGoogle Scholar
  75. Warren, D. L., Glor, R. E., & Turelli, M. (2008). Environmental niche equivalency versus conservatism: quantitative approaches to niche evolution. Evolution, 62, 2868–2883.CrossRefPubMedGoogle Scholar
  76. Warren, D. L., Glor, R. E., & Turelli, M. (2010). ENMTools: a toolbox for comparative studies of environmental niche models. Ecography, 33, 607–611.CrossRefGoogle Scholar
  77. Wiens, J. J. (2004). Speciation and ecology revisited: phylogenetic niche conservatism and the origin of species. Evolution, 58, 193–197.CrossRefPubMedGoogle Scholar
  78. Wyner, Y. M., Johnson, S. E., Stumpf, R. M., & DeSalle, R. (2002). Genetic assessment of a white-collared × red-fronted lemur hybrid zone at Andringitra, Madagascar. American Journal of Primatology, 57, 51–66.CrossRefPubMedGoogle Scholar

Copyright information

© Springer Science+Business Media New York 2015

Authors and Affiliations

  • Steig E. Johnson
    • 1
    Email author
  • Kira E. Delmore
    • 2
  • Kerry A. Brown
    • 3
  • Tracy M. Wyman
    • 1
  • Edward E. LouisJr.
    • 4
  1. 1.Department of Anthropology and ArchaeologyUniversity of CalgaryCalgaryCanada
  2. 2.Department of ZoologyUniversity of British ColumbiaVancouverCanada
  3. 3.School of Geography, Geology and the Environment, Centre for Earth and Environmental Science Research (CEESR)Kingston UniversitySurreyUK
  4. 4.Center for Conservation and ResearchOmaha’s Henry Doorly Zoo and AquariumOmahaUSA

Personalised recommendations