The authors have declared that no competing interests exist.
Conceived and designed the experiments: SKS SPG. Performed the experiments: SKS SM PP. Analyzed the data: SKS JA LK. Contributed reagents/materials/analysis tools: SPG PN. Wrote the paper: SKS RS JA LK SPG.
‡ These authors are shared first authors on this work.
The Sundarbans tiger inhabits a unique mangrove habitat and are morphologically distinct from the recognized tiger subspecies in terms of skull morphometrics and body size. Thus, there is an urgent need to assess their ecological and genetic distinctiveness and determine if Sundarbans tigers should be defined and managed as separate conservation unit. We utilized nine microsatellites and 3 kb from four mitochondrial DNA (mtDNA) genes to estimate genetic variability, population structure, demographic parameters and visualize historic and contemporary connectivity among tiger populations from Sundarbans and mainland India. We also evaluated the traits that determine exchangeability or adaptive differences among tiger populations. Data from both markers suggest that Sundarbans tiger is not a separate tiger subspecies and should be regarded as Bengal tiger (
Tiger (
On the basis of tiger distribution and the potential for connectivity, India is divided into six tiger landscape complexes [
Identification of population units within species is a crucial step for guiding management authorities and policy makers in management practices [
Previous genetic studies on mainland Bengal tigers have evaluated tiger populations from different landscape complexes (or geographic regions) in India, such as North, northwest, central, South and northeast, and found moderate to high levels of genetic diversity [
Our ultimate goal was to use a multi-population comparative approach to: (i) determine whether the tigers in Sundarbans have reduced levels of genetic variability relative to other populations of mainland tigers, as might be expected given their small contemporary population size and insular distribution, (ii) estimate the degree of genetic differentiation, historical and recent migration, (iii) determine time since separation from the mainland and effective population sizes of Sundarbans and mainland tiger populations, (iv) evaluate the traits to determine the exchangeability or adaptive differences among tiger populations, and (v) discuss whether Sundarbans tiger population should be considered as a separate ESU and MU, and implications for their conservation or management.
All tiger samples (blood and tissue) used in this study were provided by forest department of Sundarbans tiger reserve during the routine monitoring, translocations and radio collaring of the conflicted animal. Hence, sample collection did not require any extra handling of the animals. All necessary permissions to get the samples to the National wildlife reference sample repository at Wildlife Institute of India (WII) were obtained from the Field Director of the respective forest reserve (letters no. 2509/WL/2W-296,2(4)/SBR/C-208/09, 401(4)/SBR/C-172/10, 1906/FD/2M-96/06 and 4439/WL/2W/567/06).
In order to adequately assess genetic status of Sundarbans tigers, it is critical to obtain as many biological samples as possible. However, because most of the area in this tiger habitat is inaccessible swamp, it is difficult to non-invasively collect samples, such as feces or hair. We overcome this problem by using opportunistically collected samples by the forest officials. We obtained sixteen tiger samples: blood (
Genomic DNA from blood spots stored on FTA Classic cards (WhatmanTM) was extracted following Smith and Burgoyne [
In order to check for the existence of new haplotypes in the tiger samples from Sundarbans, we amplified and sequenced four mtDNA fragments comprising
Nine microsatellite loci, of which four loci were originally developed for Bengal tiger (PttA2, PttA4, PttC6, PttE5) [
In order to examine the phylogenetic relationships among haplotypes, all published tiger mtDNA sequences (including all extant tiger subspecies) were retrieved from GenBank and compared with CLUSTALW in BioEdit, and then manually edited to achieve the best alignment [
The median-joining networks (MJ Network) were constructed with Network 4.6.1.1 [
The program ModelGenerator was used to determine the most appropriate model of substitution [
To assess error rates for genotypes derived from fecal and tissue samples, we randomly selected 10 fecal samples from the field (from Kanha tiger Reserve) and 10 tissue (from Corbett tiger reserve) samples and reanalyzed them (three times for tissue samples and four times with fecal samples). This was done to calculate allelic drop out (ADO) and false allele (FA) error rates using PEDANT 1.0 involving 10,000 search steps for enumeration of per allele error rates [
Number of alleles (
We used SPAGeDi [
To identify possible distinct genetic clusters and to assign individuals to these clusters, we utilized the Bayesian clustering approach implemented in software STRUCTURE 2.3.3. [
We also used a spatially explicit clustering method to identify possible genetic clusters as implemented in the program TESS 2.3 [
We estimated short-term effective population size (
We estimated the number of migrants between all pairs of sample sites using GENECLASS 2.0 [
BayesAss1.3 programme [
In order to infer whether dispersal between tiger populations (Peninsular, northern, and Sundarbans) is sex-biased, we estimated the relative amount of male and female gene flow following the approach of Hedrick et al. [
Demographic analyses was performed to calculate the divergence time and population history of Sundarbans tiger by using different molecular markers (mitochondrial and microsatellite). Bayesian skyline plots [
In order to investigate if Sundarbans tigers have characteristics that distinguish them from other mainland tiger populations, we examined a large sample of the published literature in order to find evidence of adaptive differences or traits, such as morphology, habitat type, size of prey, competition with other predators, tiger density [
Large sized fragments of the mtDNA for at least one fragment of the four mitochondrial genes (
Out of all samples, tissue samples did not show allelic drop out (ADO) and false alleles (FA), however, the fecal-extracted DNA showed that ADO ranged between 0 and 11% and FA between 0 and 4%. Genotyping error rate also varied among loci (
Scat (N = 10) | Tissue (N = 10) | Sundarbans (n = 8) scat & hairs | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Loci | Product size range (bp) | Na | ADO (%) | FA (%) | Product size range (bp) | Na | ADO (%) | FA (%) | Product size range (bp) | Na | ADO (%) | FA (%) |
PttA2 | 186–190 | 3 | 5 | 0 | 188–190 | 2 | 0 | 0 | 188–196 | 3 | 7 | 0 |
PttA4 | 139–145 | 4 | 7 | 0 | 145–153 | 4 | 0 | 0 | 143–151 | 3 | 3 | 1 |
PttC6 | 172–176 | 3 | 0 | 2 | 174–178 | 3 | 0 | 0 | 174–178 | 3 | 0 | 0 |
PttE5 | 183–187 | 2 | 4 | 0 | 174–190 | 4 | 0 | 0 | 182–192 | 3 | 8 | 2 |
FCA304 | 122–142 | 3 | 0 | 0 | 122–128 | 4 | 0 | 0 | 122–126 | 3 | 1 | 0 |
FCA272 | 115–121 | 4 | 11 | 0 | 119–127 | 5 | 0 | 0 | 117–123 | 4 | 1 | 0 |
F41 | 171–187 | 4 | 8 | 0 | 171–179 | 3 | 0 | 0 | 171–191 | 4 | 5 | 0 |
PUN327 | 88–96 | 5 | 0 | 4 | 84–90 | 3 | 0 | 0 | 84–96 | 4 | 1 | 0 |
PUN82 | 111–115 | 3 | 0 | 4 | 101–119 | 5 | 0 | 0 | 114–118 | 3 | 5 | 2 |
Comparison of 2600 bp mitochondrial sequences of the Sundarbans tigers with the sequences from six other tiger subspecies revealed that Sundarbans tigers shared three nucleotide substitutions that were identified as specific (or diagnostic) to Bengal tiger subspecies. The mtDNA amplification target includes 13 single nucleotide polymorphisms (SNPs), of which 3 are diagnostic (fixed differences) for
(a) haplotypes found in Sundarbans tigers (in black) and all other six tiger subspecies (in yellow and green color, from Luo et al. 2004) [
Further comparison of Sundarbans tiger haplotypes with the other Bengal tiger sequences obtained from northern and Peninsular tiger populations revealed that these haplotypes are specific to Sundarbans samples and there were no other shared haplotypes between the samples. Interestingly, one of the two Sundarbans haplotypes observed in the present study corresponds to the unique haplotype (TIG29) as previously reported by Mondol et al. [
Molecular diversity indices such as haplotype (
Tiger population | Microsatellites | Mitochondrial DNA | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
N | MNA | AR | Ho | He | FIS | Ne (95% CIs) | N | S | h | π | |
13 | 3.33 | 3.24 | 0.491 | 0.587 | 0.109 | -242.6 (102Infinite) | 8 | 3 | 0.679 | 0.001 | |
73 | 7.33 | 4.80 | 0.492 | 0.707 | 0.300 | 73.3 (48.4–128.4) | 15 | 19 | 1.00 | 0.002 | |
62 | 4.88 | 4.18 | 0.401 | 0.674 | 0.394 | 47.9 (29.5–92.3) | 8 | 3 | 0.82 | 0.001 |
All nine microsatellites were polymorphic, with between 3 and 4 alleles per locus in the Sundarbans tiger samples. The mean number of alleles (
Two loci in Sundarbans, and six loci each in the northern and Peninsular tiger populations were not in HWE.
The allele size permutation test [
Using mtDNA, the pairwise
Bengal tiger populations | northern | Peninsular | Sundarbans |
---|---|---|---|
0 | 0.058 |
0.250 |
|
0.030 |
0 | 0.116 |
|
0.070 |
0.069 |
0 |
*
**
Bayesian cluster analysis of microsatellite genotypes in STRUCTURE suggested the existence of five genetic clusters (mean Ln P (X/K) = -2882.64), when data were analysed across a large geographical scale (i.e., northern + Peninsular + Sundarbans). However, delta
Each individual is represented by a vertical bar and indicates the probability of membership in each cluster.
TESS gave results similar to STRUCTURE
We did not detect first generation migrants in Sundarbans population with GENECLASS using a threshold
Arrows show the direction of gene flow.
Estimates of effective population size (
The estimated amount of the genetic differentiation from male gene flow, “
For the mtDNA, the Bayesian skyline plots revealed slight, but steady growth of the mitochondrial effective size during the last tens of thousands of years. However, the median in the skyline plot shows a decrease of the population size beginning about 1000–2000 years ago (
Interestingly, our results from microsatellite data unambiguously point to a scenario similar to the one revealed by mtDNA. The DIYABC analyses suggested that recent divergence explains the observed genetic structure. With the data set divided according to STRUCTURE and TESS, all the explored coalescent scenarios involved estimates of a recent to ancient divergence (
Pop 1 = northern India, Pop 2 = Peninsular India and Pop 3 = Sundarbans.
t1 | t2 | |||||
---|---|---|---|---|---|---|
northern | Peninsular | Sundarbans | Ancient | |||
4 410 (1 020–9 510) | 21 200 (8 020–47 600) | 1 080 (359–8 430) | 7 590 (1 840–19 200) | 125 (33–1 160) | 862 (226–3 770) | |
3 020 (927–9 300) | 7 270 (3 340–45 100) | 1 320 (389–7 940) | 11 700 (2 600–18 800) | 384 (98.4–1 540) | NA | |
3 550 (1 030–9 460 | 19 000 (5 930–47 600) | 1 260 (312–7 980) | NA | 139 (32.5–1 120) | NA |
Posterior probabilities (95% CI) after direct and logistic regression on the 1% simulated data most similar to the observed data, and mode (95% CI) of estimated time since divergence and admixture events (number of generations, t1 and t2).
Further assessment of exchangeability using ecological data, such as differences in morphology, habitat type, size of prey species, and competition with other predators revealed that Sundarbans tiger landscape is distinct from other tiger landscapes in India (
Ecological exchangeability Parameter | Sundarbans tiger landscape | Mainland Bengal tiger landscape |
---|---|---|
Small skull size and body weight of females ~80 kg [ |
Large skull size and body weight of female ~ 160 kg [ |
|
Small size prey (Chital and Wild pig) [ |
Large size prey (Sambar and Nilgai) [ |
|
Mangrove forest [ |
Tropical forest [ |
|
None [ |
Leopard [ |
|
4.3 tiger per 100 km2 [ |
16 tiger per 100 km2 [ |
We observed a statistically positive
The main result of our study is the finding that Sundarbans tiger is not a separate tiger subspecies and should be regarded as Bengal tiger (
Our findings indicate that Sundarbans tiger population contains two new closely related mitochondrial lineages, which had not yet been detected previously. The frequencies of a number of mtDNA lineages in the Sundarbans deviate noticeably from those in mainland tiger populations, suggesting that founder effects and genetic drift may have had a considerable influence on the Sundarbans gene pool, although low sample size may have also biased our results. Analyses of other samples from nearby tiger populations in the mainland, such as northeast India and Brahmaputra flood plains did not reveal Sundarbans haplotype (data not shown). The absence of the haplotype unique to Sundarbans from other tiger populations indicates that practically no emigration of females has taken place from Sundarbans to Peninsular, northern or north east India after geographical isolation.
Our study revealed relatively low levels of genetic variation in Sundarbans tigers, using both mitochondrial sequence and nuclear genetic markers. The levels of genetic variation in Sundarbans tigers (
A closer look at the population genetic pattern inferred from our mitochondrial and microsatellite data analyses suggest that Sundarbans tigers are the most divergent group of Bengal tigers. Both the applied mtDNA and microsatellites demonstrate genetic differentiation and structure between tigers in the different geographical regions, with the Sundarbans being significantly differentiated from the northern and Peninsular populations. Pairwise
This finding was also supported by the detection of Sundarbans as a separate population by the Bayesian clustering analysis (STRUCTURE and TESS) for microsatellite markers. These results are not surprising, given the field-based knowledge on the current and historical landscape connectivity between Sundarbans and mainland areas [
We observed that the microsatellites have considerable allele sharing among different tiger populations. The lack of population specific autosomal markers could mean the tiger populations are not as evolutionary isolated as the mitochondrial phylogeny suggests. However, the complete reciprocal monophyly, high divergence, and geographically structured lineages for the mitochondrial data would also mean that if ongoing gene flow were responsible for allele sharing, migration would have to be strongly male biased. The absence of first generation migrants, high assignment probabilities to the population of origin, and low long-term migration between Peninsular and Sundarbans populations as estimated by the Bayesian methods may suggest sex-biased dispersal between these populations until very recent times.
Interestingly, the demographic analyses using different markers (mtDNA and microsatellites) showed similar results and suggested a recent rather than an ancient divergence of Sundarbans tiger population. Bayesian skyline analysis provided clear evidence of a recent historical reduction in effective population size. These results are supported by Approximate Bayesian analyses (ABC), which suggest that Sundarbans tiger population might have diverged from the mainland tiger population within last 2000 years. The DIYABC estimates involved wide posterior distributions but unanimously suggested that recent divergence explain the observed genetic structure of Bengal tigers in India. ABC analysis suggests high effective population size of Sundarbans (
It is striking that for both marker types the data point towards a contraction that took place in the last 2000 years and that is difficult to explain by recent anthropogenic activities alone. Several independent lines of evidence indicate that series of climatic changes profoundly influenced the geography and vegetation in many parts of Sundarbans, leading to shifts in the extension and distribution of different habitat types. These biogeographical changes are thought to have had a profound impact on the geographic distribution of the Sundarbans fauna, including tigers [
(1- Rajaji NP, 2- Corbett TR, 3- Dudhwa TR, 4- Buxa TR, 5- Ranthambhore TR, 6- Bandhavgarh TR, 7- Kanha TR, 8- Pench TR, 9- Palamau TR, 10-Simlipal TR, Sundarbans- Sundarbans Tiger Reserve).
The recent deforestation in the last couple of centuries has most certainly affected the Sundarbans tigers, though we may not be able to detect and date such very recent factors. Indeed, it is also possible that all these ancient climatic and recent anthropogenic factors combined have shaped the history of present day Indian tiger populations, including the Sundarbans [
Consequences of the level of ecological exchangeability from other Bengal tigers reinforce the distinctiveness of Sundarbans tiger population, and suggest that morphological and behavioral changes in Sundarbans tigers might be adaptations to new mangrove habitat and availability of small-sized prey (
Altogether, both historical and genetic evidence support the fact that Sundarbans tigers were connected by gene flow with other mainland tigers from peninsular India until very recent. Hence, we finally conclude that isolation from mainland tiger population; subsequent gene flow and local adaptation have jointly shaped the genetic architecture of Sundarbans tiger in this marshy ecosystem.
This study gives the first description of the genetic diversity and population structure of Sundarbans tigers and as such should be used for their conservation management. A closer look at the population genetic pattern inferred from our mitochondrial and microsatellite data analyses suggests that Sundarbans tigers are the most divergent group of Bengal tigers. Hence, it should be managed as a separate conservation unit (CU). According to the definitions of Ryder [
(a) Selection of best possible number of genetic clusters on the basis of DIC criterion for BYM and CAR, detecting 5–6 genetic populations. (b) Individual assignment probabilities of Bengal tiger to genetic clusters using the model-based program of TESS (run K = 5 and 6).
(DOCX)
Demographic analyses methodology.
(DOCX)
Observed Null allele frequency.
(DOCX)
We are thankful to the Director, Dean and Research Coordinator of the Wildlife Institute of India (WII) for their strong support to carry out this study. We would like to express our thanks to Dr. Mukesh of wildlife forensic lab (WII) and Eeva Jansson of the University of Oulu for reviewing the manuscript and their valuable suggestions. We thank the Forest Department of Sundarbans Tiger Reserve for providing samples for this study. We also thank the editor and two anonymous reviewers for their constructive comments, which helped us to improve the manuscript.