The maintenance of both spatial and genetic connectivity is paramount to the long-term persistence of small, isolated populations living in environments with extreme climates. We aim to identify the distribution of genetic diversity and assess population sub-structuring and dispersal across dwarfed desert populations of Crocodylus suchus, which occur in isolated groups, usually less than five individuals, along the mountains of Mauritania (West Africa). We used both invasive and non-invasive sampling methods and a combination of mitochondrial DNA (12 S and ND4) and microsatellite markers (32 loci and a subset of 12 loci). Our results showed high genetic differentiation and geographic structure in Mauritanian populations of C. suchus. We identified a metapopulation system acting within four river sub-basins (high gene flow and absence of genetic structure) and considerable genetic differentiation between sub-basins (FST range: 0.12–0.24) with rare dispersal events. Effective population sizes tend to be low within sub-basins while genetic diversity is maintained. Our study suggests that hydrographic networks (temporal connections along seasonal rivers during rainy periods) allow C. suchus to disperse and maintain metapopulation dynamics within sub-basins, which attenuate the loss of genetic diversity and the risk of extinction. We highlight the need of hydrographic conservation to protect vulnerable crocodiles isolated in small water bodies. We propose C. suchus as an umbrella species in Mauritania based on ecological affinities shared with other water-dependent species in desert environments.
Citation: Velo-Antón G, Godinho R, Campos JC, Brito JC (2014) Should I Stay or Should I Go? Dispersal and Population Structure in Small, Isolated Desert Populations of West African Crocodiles. PLoS ONE 9(4): e94626. https://doi.org/10.1371/journal.pone.0094626
Editor: Ulrich Joger, State Natural History Museum, Germany
Received: September 20, 2013; Accepted: March 19, 2014; Published: April 16, 2014
Copyright: © 2014 Velo-Antón, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study was partially supported by National Geographic Society (Grant 8412-08) and by Fundação para a Ciência e a Tecnologia (FCT: PTDC/BIA BEC/099934/2008) through EU Programme COMPETE and the project “Biodiversity, Ecology and Global Change” co-financed by North Portugal Regional Operational Programme 2007/2013 (ON.2 - O Novo Norte), under the National Strategic Reference Framework (NSRF), through the European Regional Development Fund (ERDF). GV-A, RG, JCB, and JCC are supported by FCT (SFRH/BPD/74834/2010, IF/564/2012, SFRH/BD/87885/2012 and IF/00459/2013 respectively). Logistic support for fieldwork was given by Pedro Santos Lda (Trimble GPS), Off Road Power Shop, P.N. Banc d'Arguin (Mauritania), and Ministére Délégué auprès du Premier Ministre Chargé de l'Environnement et du Développement durable of Mauritania. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Habitat connectivity is important for populations living in climate-extreme environments, which are naturally patchy and have temporally variable resources . In particular, water-dependent species in desert environments are examples for dispersal limitations owing to scarce and unpredictable availability of water sources, primary production and shelters . The migration movements of such species are usually constrained by extensive barriers, thus they disperse mostly through a limited number of natural corridors such as hydrographic basins. These basins represent important passageways for the dispersal of numerous relictual aquatic species characterized by small and fragmented distributions . In the Sahara, waterways played a crucial role in the historical dispersal of organisms and still may function as contemporary dispersal routes for several taxa with distinct biogeographical origins (e.g. Nile river) , . The current status of relict Crocodylus suchus  populations across the Sahara-Sahel exemplifies the threats associated with loss of landscape connectivity for aquatic vertebrates with low population densities and restricted dispersal abilities . This species was historically widespread over North Africa but was extirpated throughout much of its original distribution. Increasing aridification in the Sahara-Sahel since the Mid-Holocene (7000 yr) gradually dried the savannah-like ecosystem and most aquatic habitats, which induced local extinctions of aquatic/amphibious taxa , . Further, anthropogenic pressures in the first half of the last century left only a few relict populations in Chad and Mauritania , .
In Mauritania, crocodile populations are known from 81 small rocky pools (locally known as gueltas), located upstream of mountain valleys, and floodplains (locally known as tâmoûrts), located at the foothills of the mountains , . The small area of gueltas (0.001–1.0 ha) restricts their carrying capacity, thus they usually harbour less than five adult individuals . Crocodiles inhabiting these features are highly vulnerable to demographic and environmental variability, which in turn may reduce their genetic diversity and increase risk to their adaptive capacity due to stochastic changes , . However, during the rainy season (July to October), dispersal corridors are formed along raging streams that usually flow to vast plains adjacent to the gueltas and tâmoûrts , . These periodical and short-timed events can seasonally connect isolated habitats, eventually allowing the rescue effect of small populations . Field surveys across Mauritanian mountains have detected possible individual dispersal between water bodies (usually less than 4 km) during the rainy season , suggesting that populations of C. suchus may constitute a metapopulation. Analyses of molecular markers are needed to corroborate field observations and interpretations about the potential connectivity, possibly enabling gene flow between crocodile populations and confirming the existence of a dynamical metapopulation. Assessing genetic structure and gene flow among populations, as well as identifying landscape features that have an impact on gene flow are key to identify dispersal corridors as targets for conservation efforts. The genetic patterns of these relict crocodile populations may help understand possible metapopulations dynamics in desert landscapes and contribute to the prioritization of conservation actions on a species that could serve as a model for connectivity conservation, i.e. strategies for protecting and restoring dispersal corridors , .
Using a combination of mitochondrial DNA (mtDNA) and microsatellite markers, we aim to: 1) assess population sub-structuring across isolated populations of C. suchus in Mauritania; 2) detect the occurrence of contemporary gene flow among gueltas/tâmoûrts and between sub-basins; and 3) quantify genetic diversity within each sub-basin. Taking into account isolated populations and small group size within each locality; we predict genetic structuring, reduced gene flow, low levels of genetic diversity and signs of inbreeding. However, the dispersal ability of crocodiles during the rainy season might counteract the negative effects of their isolation.
Fieldwork was developed with permission from the Ministére Délégué auprès du Premier Ministre Chargé de l'Environnement, Nouakchott (Permit: 460/MDE/PNBA). This permit was valid for the entire country and no specific permissions were required for any specific locality. Analyses were done at a CITES registered laboratory: 13PT0065/S. Field collection and handling practices were approved by the Committee of Animal Experimentation of the University of Porto (Portugal) under the Directive 2010/63/EU of the European Parliament. No animal was sacrificed and there were no animal husbandry, experimentation and care/welfare concerns.
Study area and sample collection
The study area encompasses three southern Mauritanian mountains: Tagant, Assaba and Afollé (Figure 1). These mountains are associated with six important seasonal hydrographic sub-basins : 1) the endorheic Gabbou that flows within the Tagant, and is isolated and composed by several permanent and inter-connected gueltas; 2) the Gorgol el Abiod that results from water draining from the southern Tagant and northern Assaba; 3) the Gorgol el Akhdar and 4) Garfa sub-basins that are formed through water run-offs from the western Assaba; 5) the Karakoro that flows between the Assaba and the Afollé, although it is mostly fed by run-offs from the later mountain, and is characterized by low water availability; and 6) the Kolimbiné that flows east-ward of the Afollé. With the exception of the Gabbou, the other five sub-basins flow to the Senegal River and are composed of gueltas in mountain slopes and seasonal tâmoûrts at foothills of the mountains . Climate is characterized by a dry and cool season from November to February and a dry, hot season from March to June . Annual rains are scarce and seasonal, occurring in a single wet period from July to October, with most precipitation in August and September . The area is remote and devoid of anthropogenic infrastructure and was recently affected by regional conflict caused by political instability. These factors sometimes hamper field surveys and trans-border research and conservation planning .
On the left: Distribution of Crocodylus suchus localities along major river sub-basins (Gabbou, Gorgol el Abiod, Gorgol el Akhdar-Garfa and Karakoro-Kolimbiné) of southern Mauritania mountains (Tagant, Assaba and Afollé). White dots represent sampled localities for the current genetic study, whereas black dots represent other localities where crocodiles are currently present (Adapted from Brito et al. 2011a). On the right: Haplotype network based on mtDNA (12 s+ND4 markers). White circles represent C. suchus individuals while black squares indicate C. niloticus individuals. In bold, haplotypes that represent individuals sampled in this study. Numbers indicate the number of mutational steps between independent networks.
We collected a total of 94 samples of C. suchus from fieldwork expeditions from 2007 to 2011 , . Samples include 16 fresh tissues, nine near fresh tissues (from specimens found dead), four skin fragments from carcasses, and 65 faecal remains (Fig. 1). The samples collected represent ca. 25% and 5% of the total population of crocodiles in the Tagant/Assaba and Affolé mountains respectively. We captured live crocodiles by hand or using hand nets, and then immediately released them at the point of capture after clipping a 5 mm piece of tail tissue for genetic analysis. Team members obtained the tissue samples following ethical guidelines for use of live reptiles. We used samples obtained from carcasses of C. suchus from Guinea-Bissau and C. niloticus from Malawi as outgroups in the analyses (see Table S1).
DNA extraction, sequencing and genotyping
We performed DNA extractions for fresh and near-fresh tissues using the EasySpin Genomic DNA Tissue Kit or the QIAamp DNA Micro Kit (QIAGEN). We followed Frantz et al.,  protocol after the GuSCN/silica method  for DNA extractions of faecal remains, while used QIAamp DNA Micro Kit (QIAGEN) for skins. We performed all pre-PCR procedures of non-invasive samples in a dedicated laboratory used only for the manipulation of low quality DNA under sterile conditions and quantified DNA through fluorimetry, excluding samples with a DNA concentration lower than 0.4 µg/ml.
We amplified two mtDNA fragments of 421 bp and 878 bp from 12 S and ND4, respectively, for fresh and near fresh samples in 10 µl reaction volumes, containing 5 µl of master mix (Multiplex PCR Kit, QIAGEN), 0.5 µM of each primer and approximately 1 µl of genomic DNA (see Table S2 for primers and PCR conditions). We performed cycle sequencing for both strands using PCR primers and the ABI PRISM BigDye Terminator kit (AB Applied Biosystems) in a My Cycler BioRad Thermal Cycler and sequenced PCR products on an ABI 3130xl Genetic Analyzer (AB Applied Biosystems).
We genotyped invasive samples for a total of 32 microsatellite loci previously developed in crocodilian species ,  and used a subset of 12 loci to genotype non-invasive samples due to their Probability of Identity (PID) power and small fragment sizes (see Table S2 for description of markers, primers, allele range and PCR conditions of each multiplex reaction). The final genotype database comprises a total of 34 genotyped samples: 21 individuals genotyped for 32 microsatellites plus 13 individuals genotyped for 12 microsatellites. We performed PCR amplifications twice and four times for high quality and low quality DNA, respectively, on MyCycler BioRad Thermal Cyclers. Total reaction volume was 10 µl, including 5 µl of the QIAGEN PCR Master Mix, 1 mL of primer mix, and 2 µl of DNA. We screened amplification products on an ABI 3130xl Genetic Analyzer (Applied Biosystems).
We checked and edited electropherograms by eye and aligned all sequences using GENEIOUS R6. Then, we pooled mtDNA sequences with previously published sequences from C. niloticus and C. suchus . Since our goal at mtDNA level was to ensure that all samples belong to C. suchus and to check for historical genetic differentiation by identifying mtDNA haplotypes, we created an haplotype network to infer genetic relationships using TCS 1.21 .
Two out of 35 samples that we successfully analyzed exhibited identical genotypes (a fresh sample and a fecal sample from guelta Legleyta), thus we excluded one from the dataset. We performed four PCR replicates for each sample (discarding samples that failed to have a good amplification rate) after a maximum likelihood evaluation of error rates (allelic dropout and false alleles) using Pedant 1.0 . We obtained the consensus genotype after comparison of four replicates using Gimlet v 1.3.3 , which we used for computing final error rates for each marker and sample. We tested for significant deviations from Hardy–Weinberg equilibrium and presence of linkage disequilibrium within each of the four inferred genetic demes (corresponding to four river sub-basins systems; see results) at each locus for two genetic databases (32 and 12 loci) using ARLEQUIN 188.8.131.52 . We tested for significance using a Markov chain method with 10,000 dememorization steps and 1,000 batches of 10,000 iterations per batch and sequential Bonferroni correction  for multiple comparisons. We used ARLEQUIN to assess indices of population genetic diversity (mean number of alleles per locus (Na), and observed (Ho) and expected (He) heterozygosity values) within each geographic groups (sub-basins) identified in population structure analyses (see below) and using the 12 loci dataset to maximize the number of samples. We calculated allelic richness per sub-basin using the rarefaction algorithm implemented in ADZE 1.0  and standardized to the smallest sample size (N = 5). We used GenAlEx 6.5  to estimate the number of private alleles (Npriv) and the Queller and Goodnight  statistic rxy implemented in IDENTIX 1.1  to calculate pairwise relatedness between all individuals. We also estimated the mean and the variance of rxy for each river sub-basin and compared to their expected distribution, generated by 1,000 permutations of genotypes among individuals, under the null hypothesis of no relatedness.
We used the Bayesian genotype clustering method implemented in STRUCTURE 2.3.4  to infer the number of genetic demes present within both datasets (32 loci including samples with missing data and 12 loci). We used 10 runs for each value of K (number of genetic demes) ranging from 1 to 7, a burn-in period of 1×106 and a run length of 5×106 iterations. We implemented an admixture model with uncorrelated allele frequencies and without prior information on sample population membership. We used Structure Harvester 0.6.93  to estimate the most likely number of K by the highest value of log probability of data L(K) , and the Δ(K) , and CLUMPP 1.1.2  to align cluster membership coefficients from the 10 replicate cluster analyses from each K-value. We calculated pairwise FST values  in ARLEQUIN to estimate genetic differentiation among sub-basins and tested significance of pairwise comparisons using an exact test with 1,000 iterations, adjusting P - values with the sequential Bonferroni correction.
To identify spatial genetic patterns we also performed multivariate analyses using the spatial Principal Component Analyses (sPCA) , and the Discriminant Analysis of Principal Components (DAPC) , in the 12 markers dataset. These two methods do not require data to meet Hardy–Weinberg expectations or linkage equilibrium to exist between loci, and are implemented inside the adegenet package  for the R software (R Development Core Team, 2008). sPCA investigates cryptic spatial patterns of genetic variability using georeferenced multilocus genotypes. We performed two statistical tests in sPCA to detect the existence of global (patches and clines) and local structure (strong genetic differences between neighbors). Then, we applied DAPC to identify and describe clusters of genetically related individuals. This multivariate method maximizes variation between groups and minimizes the variation within groups. To define the number of K groups, we first used K-means clustering of principal components (ranging from 1 to 7) and a Bayesian Information Criterion (BIC). We specified the actual number of clusters and performed the DAPC function retaining 20 principal components of PCA.
We used BIMR  to estimate migration rates of recent gene flow (m) between sub-basins. BIMR uses assignment tests to estimate the proportion of genes derived from migrants within the last generation and assuming linkage equilibrium and allowing for deviation from Hardy-Weinberg equilibrium. We used a burn-in period of 1×105 and a run length of 5×105 iterations for each of the 10 replicate runs, and then averaged pairwise migration rates between and within sub-basins across runs.
Historical genetic structure
The final mtDNA dataset resulted in a concatenated alignment of 421 bp of the 12 S gene and 878 bp of the ND4 gene for 39 individuals (25 from this study and 14 Genbank sequences). Haplotype network reconstruction using statistical parsimony in TCS resulted in three haplotype networks not connected under the 95% connection probability criteria (Fig. 1). One network contained all C. suchus haplotypes and diverged by 53 mutational steps from the second network containing all C. niloticus haplotypes, whereas a third network represented by a single haplotype from Malawi differed in 19 mutational steps from the C. niloticus main network. Mauritanian C. suchus exhibited low genetic diversity (only three haplotypes), with a common haplotype (Mauritania A) widespread across the whole study area (three mountains and main river sub-basins) and a second haplotype (Mauritania B; differing from Mauritania A by a single point mutation) that is restricted to four samples from the Karakoro-Kolimbiné sub-basin (Fig. 1). A third Mauritanian haplotype (Mauritania C, also present in Burkina-Faso), collected 60 km east of our study area, was reported in a previous study  and differs by 5 mutational steps from the most widespread Mauritanian haplotype.
Genetic diversity and effective population size
All loci showed low to moderate polymorphism when considering the entire dataset. The number of alleles per locus ranged from three (CUD68) to seven (CpP302). One locus deviated from Hardy–Weinberg equilibrium at Karakoro-Kolimbiné sub-basin (CpP1409; P<0.05 after Bonferroni correction) for the 12 loci dataset. As this deviation occurred only at Karakoro-Kolimbiné and was not observed for the 32 loci dataset, all loci were retained in the analyses. No evidence of linkage disequilibrium was apparent in any pair of loci/population after adjusting the significance level for multiple comparisons. Mean values of allelic dropout and false alleles amplification across loci (12.5% and 0.0%, respectively) and across samples (7.7% and 0.0%) were low, supporting the robustness of the selected loci.
Gabbou exhibited the lowest genetic diversity values with the exception of the number of private alleles (Table 1). The highest values of genetic diversity were observed in Gorgol el Akhdar-Garfa for most of the nuclear genetic diversity measures (A–R, Ho and He; Table 1) and in the Karakoro-Kolimbiné sub-basin for the mean number of alleles per locus and number of private alleles (Table 1). Relatedness values were similar among sub-basins (mean rxy = 0.21–0.25) with the exception of the Gorgol el Akhdar-Garfa sub-basin that showed the lowest level of inbreeding (Table S3). Individuals exhibiting full sibling relationships (rxy≈0.5) were found within three sub-basins at isolated gueltas (Fig. 2; Table S3), although no significant values were achieved for the mean pairwise relatedness within sub-basins.
Contemporary genetic structure and gene flow
STRUCTURE analyses identified four genetic demes (K = 4) as the most probable number of groups within our dataset (estimated at both L(K) and Δ(K) methods). These genetic groups correspond to independent hydrographic sub-basins, Gabbou and Gorgol el Abiod in the Tagant mountain, and to combinations of neighbor sub-basins within mountains, Gorgol el Akhdar-Garfa in Assaba and Karakoro-Kolimbiné in Afollé (Fig. 3). The first genetic split (K = 2) clearly shows a partition between sub-basins in the Tagant and Assaba mountains (Gabbou, Gorgol el Abiod and Gorgol el Akhdar-Garfa) from sub-basins in the Karakoro-Kolimbiné in the Afollé mountains (Fig. 3). Genetic assignments were similar in both datasets (12 and 32 loci) for each number of imposed clusters (K), suggesting that increasing the number of markers slightly improve the resolution of population structure (Fig. 3). One exception occurred with one sample in the Gorgol el Abiod that was assigned to the genetic deme occurring at Gorgol el Akhdar-Garfa. When three clusters are imposed (K = 3), sub-basins within Tagant and Assaba mountains are split (results not shown).
The four genetic demes inferred in STRUCTURE (K = 4) are represented with different colors while the two first clusters (K = 2) are represented in black and white. Individual assignment to each genetic deme (K = 2 and K = 4) is shown for each sub-basin (right panel) for 12 and 32 loci. Asterisks indicate non-invasive samples (only 12 microsatellites genotyped) and letters (A and B) represent recent dispersal and historical admixture of the samples 2642 and 2475, respectively.
The barplot of the eigenvalues of the sPCA, and the global randomized Monte–Carlo test indicated global spatial structure of the sampled genotypes (P = 0.009, 9999 permutations), but no local structure (P = 0.25, 9999 permutations) has been detected (Fig. 4). DAPC performed essentially as STRUCTURE, showing K = 2 as the most likely number of clusters (Tagant and Assaba mountains vs. Afollé mountains). To investigate deeper genetic structuring in these two groups, we selected the second best number of clusters (K = 5) in which DAPC recovered the four clusters obtained in STRUCTURE and further subdivided the Karakoro-Kolimbiné in two clusters (Fig. 4).
Dots represent individuals and ellipses roughly group them in five clusters. Color codes identify the same four genetic clusters obtained with STRUCTURE (Fig. 3).
Pairwise FST values yielded moderate to high genetic differentiation among all sub-basins, ranging from 0.12 (between Gabbou and Gorgol el Akhdar-Garfa) to 0.24 (between Gorgol el Abiod and Karakoro-Kolimbiné) (Table 2). Estimates of recent gene flow using BIMr were low between all basin pairs (<0.02%, Table 2) and consistent across runs.
Historical vs. contemporary genetic structure
The presence of mtDNA haplotypes of C. suchus in Mauritania is consistent with the recent findings of Hekkala et al.  that identify western African crocodiles as C. suchus. However, relict populations or individuals of C. niloticus present in West Africa cannot be excluded since the sample sizes and geographic areas surveyed are still limited , . The lack of genetic structure and low genetic diversity (only two mtDNA haplotypes) reflect historical connectivity among extant populations and high levels of panmixia. As in other water-dependent species occurring across the Sahara, a likely hypothesis to explain the absence of historical genetic structure involves the colonization of Mauritanian mountains from southern ranges during a humid phase of the Sahara . Then, with the emergence of a posterior arid phase in the Mid-Holocene, species retreat to their original range or become locally extinct, with some remaining populations restricted to small mountain areas. We also note that long-lived animals with long generation times, such as crocodiles, are prone to exhibit low levels of mtDNA genetic structure , , , , which is partially explained by their low metabolic rates , low mutation rates during DNA replication , and long longevity .
The use of fast evolving markers, which are useful to study fine-scale population structure and contemporary evolutionary forces, revealed high levels of genetic structuring (assignment method) and genetic differentiation (FST) across relict populations in Mauritania, suggesting that geographic isolation might have started since the beginning of the arid phase. In addition to recent geographic isolation, we also note that demographic factors and drift , might also contribute to increase genetic structure among crocodile populations across the Mauritanian mountains in more recent times, as the extreme low population sizes in this biological system could facilitate genetic differentiation by drift. The high FST values (0–15–0.24) between sub-basins of Afollé and sub-basins from both Tagant and Assaba together with the first split of these geographic groups identified with STRUCTURE, suggest that Afollé population may have experienced a longer isolation from the remaining populations. Genetic differentiation and genetic structure of these populations also reflect ecological constraints on dispersal in C. suchus. Pairwise FST values between sub-basins are much higher in comparison to those found in other studies of crocodilians –. This unusual genetic differentiation of desert populations in C. suchus can be explained by a combination of geographic isolation, low population sizes and genetic drift. In addition, genetic demes correspond mainly to sub-basins (Gabbou, Gorgol el Abiod, Gorgol el Akhdar-Garfa and Karakoro-Kolimbiné) supporting previously hypothesized metapopulation dynamics in the region , . We suggest the presence of four metapopulations that mostly correspond to major sub-basins. Seasonal rivers and brooks that are sparsely distributed in temporal river beds within each sub-basin (see Fig. 1) may connect isolated individuals in gueltas and tâmoûrts during the rainy seasons, while these water bodies act as refugia to relict populations during the dry seasons when they remain isolated by sandy areas. However, individuals from relatively close localities but different sub-basins (less than 50 km between gueltas; see Fig. 1) would be largely maintained geographically and genetically disconnected.
Dispersal capacity and contemporary gene flow
The results obtained from the combination of genetic markers unveiled the development of a metapopulation system of C. suchus within the Mauritanian mountains, which likely began with the beginning of the Sahara drying. While mtDNA indicates historical gene flow and spatial connectivity among these relict populations of C. suchus, our results based on microsatellite markers revealed low levels of crocodile movements between sub-basins. The migration rates obtained between sub-basin pairs are much lower than the threshold value (m = 0.1) considered for demographic independence , , suggesting that overland dispersal between basins is rare and each sub-basin has independent dynamics. The high migration estimates, as well as possible full siblings, only found within sub-basins reinforce the hypothesis of a metapopulation dynamic acting mostly within each sub-basin. Demographics and genetic structure is highly influenced by water-level fluctuations in water-dependent species. The present spatial dynamics of this metapopulation system would resemble the source-sink dynamics described for other aquatic organisms in which upstream populations serve as sources and downstream populations as sinks . Such spatial dynamics makes upstream colonization less probable and reduces the likeliness of rescue effect in upstream populations . This would explain the admixed genotype found at southern Assaba Mountain (sample 2475), which could have mixed ancestry from Assaba and Afollé migrants (see K = 2; Fig. 3). However, our study does not include crocodiles from the Senegal River, which prevented our investigation of the migration dynamics between the Senegal River population (and its tributaries) and isolated mountain populations. Moreover, we are cautious in our interpretations because this species is a long-lived organism, which could delay the detection of a more contemporary gene flow.
The reduction of connectivity among fragmented populations is negatively associated with migration rates and gene flow, decreasing the colonization capacity and the rescue effect that ultimately impact both genetic diversity and structure , . If crocodile populations remain isolated, then they are likely to lose genetic diversity over time and increase their susceptibility to extinction by genetic processes, including inbreeding depression, random fixation of deleterious alleles and loss of adaptive potential . Thus, the persistence of relict populations across the Sahara-Sahel largely relies on individuals' dispersal ability, allowing the so-called rescue effect of reduced groups of isolated individuals. Field surveys have found crocodile carcasses of individuals dispersing between water bodies that likely have occurred during the rainy season when water flows partially connect gueltas , . Some of the few individuals restricted to these gueltas might act as dispersers, colonizing other water bodies by overland dispersals. Our study shows evidence of long dispersal between sub-basins for one individual (sample 2642), which leads us to question why and how do crocodiles disperse. Two main benefits of dispersal are the avoidance of kin competition (to limit inbreeding) and the access to better habitats , . However, the dispersal costs in this region must be high since habitats outside gueltas are unsuitable and individuals would have to move long distances among gueltas, thus risking successful dispersal and eventually their lives. Permanent and temporal rivers are thought to be the main way to disperse for crocodiles, as for other semi-aquatic organisms –, but they can use alternative dispersal pathways over land, connecting water bodies and enlarging these metapopulation systems. Although, we have evidences of attempts of crocodile dispersal between water bodies, further work (e.g. combining genetic analysis and telemetry techniques) is needed to confirm if individuals are able to migrate and breed in the sink population, which is crucial to maintaining viable metapopulations. Generalizations of dispersal patterns are complex as these may vary at the inter- and intra-individual level. Inter-individual variability is usually associated with behavioral traits , , while intra-individual variability can occur due to phenotypic plasticity and ontogenetic shifts during the lifetime of an individual , . Dispersal corridors, landscape features that facilitate the movement between ecologically important areas , can be crucial for these relict populations as they may allow periodical inputs of new alleles from adjacent populations. This process prevents the loss of genetic diversity and increases the species' potential to adapt genetically to environmental challenges .
Overall, our study confirms genetic connectivity among isolated water bodies within main sub-basins for C. suchus, highlighting that individual's dispersal abilities help maintain current populations by diminishing the negative effects associated with stochastic events and genetic drift, and thus preventing local extinctions and increasing the populations' recovery in this harsh and mostly unsuitable environment.
Genetic diversity in desert populations of West African crocodile and conservation implications
The current geographic isolation of individual crocodiles across Mauritanian mountains, where they persist in extremely low population sizes (5 individuals in average), make them vulnerable to demographic and environmental events that may reduce genetic variability. However, our results do not show drastically low genetic diversity for any sub-basin. In fact, heterozygosities in C. suchus genetic demes (He ranging from 0.56 to 0.70, Table 1) are comparable to those recorded in large populations of the alligator A. mississippiensis , black caimans, Melanosuchus niger , the Cuban crocodile, Crocodylus rhombifer  and the American crocodile, Crocodylus acutus, , . Water-level fluctuations may drive metapopulation dynamics in water-dependent organisms and shape their genetic diversity patterns . Such a metapopulation dynamic system could counteract the decline of genetic diversity in crocodiles across Mauritanian mountains and reduce the risk of losing adaptive capacity due to stochastic changes . Moreover, we found a low impact of isolation on the reduction of genetic diversity, which may be explained by the buffered effect of intrinsic biological traits, such as long-generation times that may help to curb the loss of genetic diversity , .
Future efforts should be focus on identifying landscape features shaping genetic structure and distribution of genetic diversity across crocodile populations in the region. Further fieldwork efforts are needed within these mountains along with integrative spatial/genetic analyses, such as those employed in landscape genetics approaches . Identifying both the dispersal patterns and the landscape and environmental constrains to individuals dispersal will help in establishing efficient ecological networks  that are needed to protect these fragile desert populations, which may be undergoing specific adaptive processes linked to warming temperatures and populations fragmentation. On the other hand, further studies on the biology, behavior and ecology of C. suchus in isolated localities as complements to molecular approaches will contribute to understanding population dynamics and dispersal rates. In addition to the use of remote sensing techniques to quantify water availability patterns , GPS-based satellite telemetry are promising techniques that may better estimate dispersal rates and the extent of these events.
To curb local population extinctions it is advisable to increase both the carrying capacity of populations and the facilitation of dispersal among populations. Increasing water quality is needed as many gueltas are presently overexploited by humans, resulting in water shortage during the dry season, faecal contamination, excessive eutrophication, and increased activities for excavating pools or pumping water , . Dispersal between water bodies may be enhanced through the installation of small-scale dams and water-reservoirs. Increasing awareness of local human populations for the value of relict crocodile populations should be encouraged, as dispersing crocodiles are often killed . This study highlights the importance hydrographic networks to crocodile dispersal along river beds, because they reduce the risk of inbreeding and extinctions in these fragile habitats. Such conservation activities have been already proposed to preserve crocodile populations in the Gabbou  but should be extended to the remainder sub-basins where C. suchus occur, as these may be considered independent conservation units. Identifying and protecting the species' refugia in temporary freshwater ecosystems can contribute to conservation plans designated to enhance the resilience of freshwater biodiversity .
Crocodiles are usually viewed as a conservation flagship species and considered a valuable indicator in ecosystem monitoring and restoration programs , . In contrast to other African regions, villagers from Mauritanian mountains peacefully accept crocodile presence in gueltas , , ,  and even consider their potential for local economy by launching crocodile-watching activities . Moreover, C. suchus in Mauritania constitutes a good candidate umbrella species, since it's conservation is tied to other aquatic or semi-aquatic organisms (e.g. amphibians, Amietophrynus xeros, Hoplobatrachus occipitalis; fishes, Barbus spp., Tilapia spp., Clarias anguillaris; reptiles, Varanus niloticus, Python sebae), which also live in isolated spots across the Sahara-Sahel. The use of genetic tools to study dispersal in umbrella species has been considered as a realistic approach to implement ecological networks that could assist species protection at the regional level . Therefore, our study on C. suchus provides an initial framework to manage and protect threatened desert populations that depend on permanent or seasonal water bodies.
Crocodylus suchus samples analysed in this study. For each sample (ID), we list the country, mountain chain, basin, sub-basin, locality, GPS coordinates of collection sites and their genetic assignment at both mtDNA (haplotype) and microsatellite level (genetic deme).
Loci used on this study with primer information, allele ranges, PCR conditions and references for description of primers used both for mtDNA and microsatellite markers.
Pairwise relatedness (rxy) among all crocodile individuals.
Input file (Structure format) containing genotype information for 32 microsatellite markers.
Input file (Phylip format) containing sequences of the 12 s mtDNA fragment.
We thank F. Martínez-Freiría, Z. Boratyński, D.V. Gonçalves, P. Tarroso, P. Sierra, C.G. Vale, T.L. Silva, N. Sillero, and A.S. Sow for their help during field work, and J. Cavadas, D. Castro and S. Lopes for their assistance during lab work. We thank David Rodriguez for the revision of the language and literature. Logistic support for fieldwork was given by Pedro Santos Lda (Trimble GPS), Off Road Power Shop, P.N. Banc d'Arguin (Mauritania), and Ministére Délégué auprès du Premier Ministre Chargé de l'Environnement et du Développement durable of Mauritania.
Conceived and designed the experiments: JCB RG GV-A. Performed the experiments: GV-A RG JCB JCC. Analyzed the data: GV-A RG. Contributed reagents/materials/analysis tools: GV-A RG JCB JCC. Wrote the paper: GV-A. Discussed the results and contributed to the manuscript: GV-A RG JCB JCC.
- 1. Shkedy Y, Saltz D (2000) Characterizing core and corridor use by Nubian ibex in the Negev desert, Israel. Cons Bio 14: 200–206.
- 2. Fagan WF, Unmack,PJ, Burgess C, Minckley WL (2002) Rarity, fragmentation, and extinction risk in desert fishes. Ecology 83: 3250–3256.
- 3. Murphy NP, Breed MF, Guzik MT, Cooper SJB, Austin AD (2012) Trapped in desert springs: phylogeography of Australian desert spring snails. J Biogeogr 39: 1573–1582.
- 4. Drake NA, Blench RM, Armitage SJ, Bristow CS, White KH (2011) Ancient watercourses and biogeography of the Sahara explain the peopling of the desert. PNAS 108: 458–462.
- 5. Brito JC, Godinho R, Martínez-Freiría F, Pleguezuelos JM, Rebelo H, et al. (2013) Unravelling biodiversity, evolution and threats to conservation in the Sahara-Sahel. Biol Rev 89: 215–231.
- 6. Schmitz A, Mansfeld P, Hekkala E, Shine T, Nickel H, et al. (2003) Molecular evidence for species level divergence in African Nile Crocodiles (Crocodylus niloticus) (Laurenti, 1786). Comptes Rendus Palevol 2: 703–712.
- 7. Trape S (2009) Impact of climate change on the relict tropical fish fauna of Central Sahara: threat for the survival of Adrar mountains fishes, Mauritania. PLoS ONE 4 : doi:10.1371/journal.pone.0004400.
- 8. Brito JC, Martínez-Freiría F, Sierra P, Sillero N, Tarroso P (2011a) Crocodiles in the Sahara Desert: An Update of Distribution, Habitats and Population Status for Conservation Planning in Mauritania. PLoS ONE 6: e14734
- 9. Brito JC, Campos JC, Gonçalves D, Martínez-Freiría F, Sillero N, et al. (2011b) Status of Nile crocodiles in the lower Senegal River basin. Crocodile Specialist Group Newsletter 30: 7–10.
- 10. Madsen T, Shine R, Olsson M, Wittzell H (1999) Conservation biology: restoration of an inbred adder population. Nature 402: 34–35.
- 11. Saccheri I, Kuussaari M, Kankare M, Vikman P, Fortelius W, et al. (1998) Inbreeding and extinction in a butterfly metapopulation. Nature 392: 491–494.
- 12. McRae BH, Hall SA, Beier P, Theobald DM (2012) Where to Restore Ecological Connectivity? Detecting Barriers and Quantifying Restoration Benefits. PLoS ONE 7: e52604
- 13. Crooks KR, Sanjayan M (2006) Connectivity Conservation. Cambridge University Press. Cambridge.
- 14. Campos JC, Sillero N, Brito JC (2012) Normalized Difference Water Indexes have dissimilar performances in detecting seasonal and permanent water in the Sahara-Sahel transition zone. J Hydrol 464–465: 438–446.
- 15. Cooper A, Shine T, McCann T, Tidane TA (2006) An ecological basis for sustainable land use of Eastern Mauritanian wetlands. J Arid Environ 67: 116–141.
- 16. Ewi M (2010) A decade of kidnappings and terrorism in West Africa and the Trans-Sahel region. African Security Review 19: 64–71.
- 17. Frantz AC, Pope LC, Carpenter PJ, Roper TJ, Wilson GJ, et al. (2003) Reliable microsatellite genotyping of the Eurasian badger (Meles meles) using faecal DNA. Mol Ecol 12: 1649–1661.
- 18. Boom R, Sol CJA, Salimans MM, Jansen CL, Wertheimvandillen PME, et al. (1990) Rapid and simple method for purification of nucleic-acids. J Clin Microbiol 28: 495–503.
- 19. FitzSimmons NN, Tanksley S, Forstner MR, Louis EE, Daglish R, et al.. (2001) Microsatellite markers for Crocodylus: new genetic tools for population genetics, mating system studies and forensics. Crocodilian biology and evolution (ed. by G. Grigg, F. Seebacher and C.E. Franklin). Chipping Norton, Australia: Surrey Beatty, pp 51–57.
- 20. Miles LG, Isberg SR, Moran C, Hagen C, Glenn TC (2009) 253 Novel polymorphic microsatellites for the saltwater crocodile (Crocodylus porosus). Conserv Gen 10: 963–980.
- 21. Hekkala E, Shirley MH, Amato G, Austin JD, Charter S, et al. (2011) An ancient icon reveals new mysteries: Mummy DNA resurrects a cryptic species within the Nile crocodile. Mol Ecol 20: 4199–4215.
- 22. Clement M, Posada D, Crandall K (2000) TCS: a computer program to estimate gene genealogies. Mol Ecol 9: 1657–1660.
- 23. Johnson PCD, Haydon DT (2007) Maximum likelihood estimation of allelic dropout and false allele error rates from microsatellite genotypes in the absence of reference data. Genetics 175: 827–842.
- 24. Valière N (2001). Gimlet: a computer program for analysing genetic individual identification data. Molecular Ecology Notes, 2: , 377–379.
- 25. Excoffier L, Lischer HEL (2010) Arlequin suite ver 3.5: A new series of programs to perform population genetics analyses under Linux and Windows. Mol Ecol Resour 10: 564–567.
- 26. Rice WR (1989) Analyzing tables of statistical tests. Evolution 43: 223–225.
- 27. Szpiech ZA, Jakobsson M, Rosenberg NA (2008) ADZE: a rarefaction approach for counting alleles private to combinations of populations. Bioinformatics 24: 2498–2504.
- 28. Peakall R, Smouse PE (2012) GenAlEx 6.5: genetic analysis in Excel. Population genetic software for teaching and research-an update. Bioinformatics 28: 2537–2539.
- 29. Queller DC, Goodnight KF (1989) Estimating relatedness using genetic markers. Evolution 43: 258–275.
- 30. Belkhir K, Castric V, Bonhomme F (2002) IDENTIX, a software to test for relatedness in a population using permutation methods. Mol Ecol Notes 2: 611–614.
- 31. Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics 155: 945–959.
- 32. Earl DA, vonHoldt BM (2012) STRUCTURE HARVESTER: a website and program for visualizing STRUCTURE output and implementing the Evanno method. Conserv Genet Resour 4: 359–361.
- 33. Falush D, Stephens M, Pritchard JK (2003) Inference of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics 164: 1567–1587.
- 34. Evanno G, Regnaut S, Goudet J (2005) Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Mol Ecol 14: 2611–2620.
- 35. Jakobsson M, Rosenberg NA (2007) CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23: 1801–1806.
- 36. Weir BS, Cockerham CC (1984) Estimating F-statistics for the analysis of population structure. Evolution 38: 1358–1370.
- 37. Jombart T (2008) adegenet: a R package for the multivariate analysis of genetic markers. Bioinformatics 24: 1403–1405.
- 38. Jombart T, Devillard S, Balloux F (2010) Discriminant analysis of principal components: a new method for the analysis of genetically structured populations. BMC genetics 11: 94.
- 39. Faubet P, Gaggiotti OE (2008) A new Bayesian method to identify the environmental factors that influence recent migration. Genetics 178: 1491–1504.
- 40. Glenn TC, Staton JL, Vu AT, Davis LM, Bremer JRA, et al. (2002) Low mitochondrial DNA variation among American alligators and a novel non-coding region in crocodilians. J Exp Zool 294: 312–324.
- 41. Cedeño-Vázquez JR, Rodriguez D, Calmé S, Ross JP, Densmore LD, et al. (2008) Hybridization between Crocodylus acutus and Crocodylus moreletii in the Yucatan Peninsula: I. evidence from mitochondrial DNA and morphology. J Exp Zool 309A: 661–673.
- 42. González-Trujillo R, Rodriguez D, González-Romero A, Forstner MRJ, Densmore III LD, et al. (2012) Testing for hybridization and assessing genetic diversity in Morelet's crocodile (Crocodylus moreletii) populations from central Veracruz. Conserv Gen 13: 1677–1683.
- 43. Ray DA, Dever JA, Platt SG, Rainwater TR, Finger AG, et al. (2004) Low levels of nucleotide diversity in Crocodylus moreletii and evidence of hybridization with C. acutus. Conserv Gen 5: 449–462.
- 44. Martin AP, Palumbi SR (1993) Body size, metabolic rate, generation time, and the molecular clock. PNAS 90: 4087–4091.
- 45. Bromham L (2002) Molecular clocks in reptiles: life history influences rate of molecular evolution. Mol Biol Evol 19: 302–309.
- 46. Galtier N, Jobson RW, Nabholz B, Glemin S, Blier PU (2009) Mitochondrial whims: metabolic rate, longevity and the rate of molecular evolution. Bio Lett 5: 413–416.
- 47. Excoffier L, Foll M, Petit RJ (2009) Genetic consequences of range expansions. Annu Rev Ecol Evol Syst 40: 481–501.
- 48. de Thoisy B, Hrbek T, Farias IP, Vasconcelos WR, Lavergne A (2006) Genetic structure, population dynamics, and conservation of black caiman (Melanosuchus niger). Biol Conserv 133: 474–482.
- 49. Davis LM, Glenn TC, Strickland DC, Guillette Jr, LJ, Elsey RM, et al. (2002) Microsatellite DNA analyses support an east-west phylogeographic split of American alligator populations. J Exp Zool 294: 352–372.
- 50. Milián-García Y, Venegas-Anaya M, Frias-Soler R, Crawford AJ, Ramos-Targarona R, et al. (2011) Evolutionary history of Cuban crocodiles Crocodylus rhombifer and Crocodylus acutus inferred from multilocus markers. J Exp Zool 315: 358–375.
- 51. Rodriguez D, Cedeño-Vázquez JR, Forstner MRJ, Densmore III LD (2008) Hybridization between Crocodylus acutus and Crocodylus moreletii in the Yucatan Peninsula: II. evidence from microsatellites. J Exp Zool 309A: 674–686.
- 52. Rodriguez D, Forstner MRJ, Moler PE, Wasilewski JA, Cherkiss MS, et al. (2011) Effect of human-mediated migration and hybridization on the recovery of the American crocodile in Florida (U.S.A.). Conserv Gen 12: 449–459.
- 53. Hastings A (1993) Complex interactions between dispersal and dynamics – lessons from coupled logistic equations. Ecology 74: 1362–1372.
- 54. Waples RS, Gaggiotti OE (2006) What is a population? An empirical evaluation of some genetic methods for identifying the number of gene pools and their degree of connectivity. Mol Ecol 15: 1419–1439.
- 55. Morrissey MB, de Kerckhove DT (2009) The maintenance of genetic variation due to asymmetric gene flow in dendritic metapopulations. Am Nat 174: 875–889.
- 56. Gotelli NJ, Taylor CM (1999) Testing macroecology models with stream-fish assemblages. Evol Ecol Res 1: 847–858.
- 57. McRae BH, Hall SA, Beier P, Theobald DM (2012) Where to Restore Ecological Connectivity? Detecting Barriers and Quantifying Restoration Benefits. PLoS ONE 7: e52604
- 58. Velo-Antón G, Parra JL, Parra-Olea G, Zamudio KR (2013) Tracking climate change in a dispersal-limited species: reduced spatial and genetic connectivity in a montane salamander. Mol Ecol 22: 3261–3278.
- 59. Shirk J, Wallin DO, Cushman SA, Rice CG, Warheit KI (2010) Inferring landscape effects on gene flow: a new model selection framework. Mol Ecol 19: 3603–3619.
- 60. Tellería JL, Ghaillani HEM, Fernández-Palacios JM, Bartolomé J, Montiano E (2008) Crocodiles Crocodylus niloticus as a focal species for conserving water resources in Mauritanian Sahara. Oryx 42: 292–295.
- 61. Cote J, Clobert J, Brodin T, Fogarty S, Sih A (2010) Personality dependent dispersal: characterization, ontogeny and consequences for spatially structured populations. Philos T Roy Soc B 365: 4065–4076.
- 62. Chapman BB, Hulthén K, Blomqyist DR, Hansson LA, Nilsson JA, et al. (2011) To boldy go: individual differences in boldness influence migratory tendency. Ecol Lett 14: 871–876.
- 63. Bilton DT, Freeland JR, Okamura B (2001) Dispersal in freshwater invertebrates. Annu Rev Ecol Syst 32: 159–181.
- 64. Fortuna MA, Gómez-Rodríguez C, Bascompte J (2006) Spatial network structure and amphibian persistence in stochastic environments. P Roy Soc Lond B Bio 273: 1429–1434.
- 65. Campbell Grant EH, Nichols JD, Lowe WH, Fagan WF (2010) Use of multiple dispersal pathways facilitates amphibian persistence in stream networks. PNAS 107: 6936–6940.
- 66. Piersma T, Drent J (2003) Phenotypic flexibility and the evolution of organismal design. TREE 18: 228–233.
- 67. Dingemanse NJ, Kazem AJN, Réale D, Wright J (2010) Behavioural reaction norms: where animal personality meets individual plasticity. TREE 25: 81–89.
- 68. Massot M, Clobert J, Ferriére R (2008) Climate warming, dispersal inhibition and extinction risk. Glob Change Biol 14: 461–469.
- 69. Nevado B, Mautner S, Sturmbauer C, Verheyen E (2013) Water level fluctuations and metapopulation dynamics as drivers of genetic diversity in populations of three Tanganyikan cichlid fish species. Mol Ecol 22: 3933–48.
- 70. Fagan WF, Holmes EE (2006) Quantifying the extinction vortex. Ecol Lett 9: 51–60.
- 71. Lippé C, Dumont P, Bernatchez L (2006) High genetic diversity and no inbreeding in the endangered copper redhorse, Moxostoma hubbsi Catostomidae, Pisces: the positive sides of a long generation time. Mol Ecol 15: 1769–1780.
- 72. Bishop JM, Leslie AJ, Bourquin SL, O'Ryan C (2009) Reduced effective population size in an overexploited population of the Nile crocodile (Crocodylus niloticus). Biol Conserv 142: 2335–2341.
- 73. Manel S, Holderegger R (2013) Ten years of landscape genetics. TREE 28: 614–621.
- 74. Baguette M, Blanchet S, Legrand D, Stevens VM, Turlure C (2013) Individual dispersal, landscape connectivity and ecological networks. Biol Rev 88: 310–326.
- 75. Hermoso V, Ward DP, Kennard MJ (2013) Prioritizing refugia for freshwater biodiversity conservation in highly seasonal ecosystems. Divers Distrib 19: 1031–1042.
- 76. Ross JP (1998) Crocodiles. Status Survey and Conservation Action Plan. 2nd edition. Available: http://www.flmnh.ufl.edu/natsci/herpetology/act-plan/plan1998a.htm Accessed 2007 Dec 04.
- 77. Shirley MH, Oduro W, Beibro HY (2009) Conservation status of crocodiles in Ghana and Côte-d'Ivoire, West Africa. Oryx 43: 136–145.
- 78. Shine T, Böhme W, Nickel H, Thies DF, Wilms T (2001) Rediscovery of relict populations of the Nile crocodile Crocodylus niloticus in south-eastern Mauritania, with observations on their natural history. Oryx 36: 260–262.