Genetic diversity and landscape genetic structure of otter (Lutra lutra) populations in Europe
|
|  Yüklə 1.4 Mb.
|
a 1.5 1.0 0.5  0.0
-0.5 -1.0 -1.5 -2.0 Germany
South Norway Iberia
comparisons between otters sampled in: (1) Austria and Czech Republic, Hungary, Latvia-Bielorussia; and (2) Hungary and Serbia-Montenegro and Latvia-Bielorussia. Population sub-structuring STRUCTURE analyses, performed using all the samples and no prior information on their locations, showed likelihood values that slowly tended to reach a plateau at K [ 10 (Fig. 4a), while showing the maximum DK increases between the initial values, that is, from K = 2 to 8. Increasing the K values led to define the number of dis-  -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
FC I
 b 1.5
  1.0
 0.5
 0.0
-0.5 -1.0 tinct clusters, and to clarify the admixture patterns within clusters. To summarize the complex sub-structuring and admixture patterns, we show the averaged qi plotting (population admixture proportions; Fig. 4b, upper), and the individual admixture proportions (Fig. 4b, lower), in the clusters as defined at K = 11. The south Italian otter population was clearly identified and showed no admix- ture (right end of the plot). The two Iberian populations (Portugal and Spain) joined into a single cluster that showed some admixture signals, particularly among the Spanish individuals (left end of the plot). Other local
-1.5 -2.0 Italy
France and Germany showed evidence of sub-structuring
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 FC I
NW D
IR FIN SK
A
C HU Z DK PC I (26.69%) Fig. 3 Factorial correspondence analysis (FCA) of otter individual (a and b), and principal component analysis (PCA) of otter population (c) multilocus scores computed using GENETIX and PCA-GEN (c) variability was significantly distributed among countries (P = 0.001). This geographic partition of genetic vari- ability generated significant global values of FST = 0.13 (95% confidence interval = 0.10–0.16, computed after 1000 bootstraps over loci), and FIS = 0.12 (95% confi- dence interval = 0.11–0.13). In 136 comparisons among the 17 sampling locations, there were only six non-signif- icant pairwise FST values (P [ 0.05), which were due to and admixture, while otters from Fennoscandia showed strong signals of individual admixture. Otters from Aus- tria and Czech Republic joined the same cluster, which was distinct from the Slovakian and other eastern Euro- pean otter samples. Isolated (and nor well sampled) otter populations from UK and Denmark also joined distinct clusters. Most of these patterns were already evident at lower K values (Fig. 4c). At K = 3 all Iberian otters clustered together, showing little admixture. At K = 3 otters from Germany already showed a sharp splitting into two dis- tinct groups, a result that was also confirmed at higher K values. Genetic heterogeneity among Fennoscandian otters was mainly due to evidences for two distinct subpopula- tions in otters sampled from Norway, which was already clear at K = 3. The individual admixture patterns showed that most of the populations included a number of admixed individuals, with the exceptions of Italy and Denmark. All the other populations showed a number of admixed individuals, mainly in France, Slovakia and Fennoscandia. Some individuals sampled from Germany and Norway were also strongly admixed. These findings highlight evidences of genetic distinctions of otters sam- pled form the southern extremes (southern Iberia and Italy) of the species’ distributions, and call for more detailed analyses that can be performed using landscape genetic approaches.
 a -17000
 -18000
 -19000
K = 11 -20000
-21000 -22000 2 4 6 8 10 12 14 16 18 K 
b K = 11
K = 11
c K = 2
K = 3 Fig. 4 Estimated population structure in otter. a Mean log-likeli- hood values for four replicated STRUCTURE runs with prior K values ranging from 1 to 18. b Population clustering (upper) and individual admixtures (lower) plottings for K = 11. c Individual admixtures (lower) plottings for K = 2 and 3. Each individual is represented as a vertical bar partitioned into one or more segments; the length of segments being proportional to the individual membership values (qi) Landscape genetics of otter populations The existence of large geographical gaps among population distributions and sampling locations, prevented us to per- form a global pan-European landscape genetic analysis. Therefore, we used the main clusters from the STRUCTURE results and geographical proximity of sampling locations to identify six groups of otter samples: (1) Iberia (central- southern Portugal and Spain; excluding 4 samples from northern Spain); (2) France (only samples collected along the Atlantic coast; seven isolated samples from north-east France and Massif Central were excluded); (3) central Europe (including otters from Germany, Slovakia, Czech Republic, Serbia and Montenegro); (4) Germany; (5) Slo- vakia and Czech Republic; (6) Fennoscandia (Norway, Sweden and Finland). These groups were further analyzed using STRUCTURE and GENELAND aiming to infer their spatial genetic structure. Isolated populations (England, Ireland, Denmark, Italy), poorly represented geographical groups (Hungary, Latvia, Belarus) and populations lacking of geographical information (Austria) were excluded from the following analyses. Results showed that: 1) Iberia. Otters sampled from Iberia were split into two sub-populations by both GENELAND (Fig. 5a) and STRUC- TURE (not shown). The sub-population defined by cluster 1 included the otters sampled from Coimbra in the centre of Portugal, and from Extremadura, Castilla-La Mancha and Leo´ n in central Spain. Cluster 2 included the otters sampled from Santare´m, South Portalegre, Lisbon, Evora, Setubal, Beja, Faro in south Portugal, and from South Extremadura and Andalusia in south Spain. As it is shown in Table 4, this splitting is supported by: (a) smaller deviations from HWE (lower FIS values in the sub-populations than in the pooled Iberian population); (b) highly significant FST value between sub-populations 1 and 2; (c) strongly reduced isolation-by-distance (IBD) in sub-population 1 as assessed through the Mantel test; and d) 91% individ- uals that were correctly assigned to the sub-populations 1 or 2 from which they were originally sampled. 2) France. Samples collected in France were split into three sub-populations (Fig. 5b), although an alternative splitting into four sub-populations was also partially supported by both GENELAND and STRUCTURE (not shown). The small sample size (n = 6) of sub-popu- lation 3 prevented us to estimate reliably the optimal K and the value of the population genetic parameters. Deviation from HWE was significantly reduced only in sub-population 1, which did not show any detectable IBD effect and no mis-assigned individuals (Table 4). The average FST was highly significant. 
Fig. 5 Maps of the individual posterior probability to belong to distinct genetic clusters as identified by GENELAND in the following geographic areas: a Iberian Peninsula; b Western France; c Germany; d Fennoscandia (see also Fig. 1). The highest membership values of individuals are identified by white areas Table 4 Subdivision of sampling groups according to Bayesian analyses performed using STRUCTURE
K number of sub-populations identified by STRUCTURE, Subpop individuals for each population or subpopulation, HWE number of loci that showed a departure from HWE, FIS fixation index and its significance (P), FST average FST values among sub-populations and its significance (P), RXY Mantel test of correlation between genetic and geographic distance matrices and its significance (P), Assign Assignment test (Paetkau et al. 2004) as implemented in GENEALEX 3) Otters from central Europe (Germany, Czech Repub- lic, Slovakia, Serbia and Montenegro) are fragmented in a number of genetically distinct clusters. Both GENELAND and STRUCTURE (data not shown) indicate an optimal value of K = 4. This splitting reduced the deviations from HWE both in terms of loci that were not in equilibrium in the pooled population vs. the four sub-populations, and in terms of FIS values. However, significant signals of IBD persisted in sub-populations 2 (corresponding to otters from Slovakia, Serbia and Montenegro) and 4 (otters from Czech Republic). Sub- populations 1 (Germany) and 3 were also not in HWE, and showed a number of mis-assigned individuals (Table 4). These results invited us to further investi- gate the sub-population structure at lower geograph- ical scales in Germany, Czech Republic and Slovakia. 4) Samples collected in Germany were split into two sub- populations by STRUCTURE and GENELAND (Fig. 5c). Sub-population 2 included otters from south Branden- burg and Saxony, while sub-population 1 grouped samples from upper Brandenburg and Mecklen- burg, except for four probably mislabeled individuals (indicated by the arrows in Fig. 5c). The splitting into two subpopulations strongly reduced the number of loci not in HWE and the FIS values, particularly in population 1 that also showed a reduction of IDB. The FST between sub-populations 1 and 2 was significant (Table 4). 5) Otters from Czech Republic and Slovakia were split into two sub-populations by GENELAND and STRUCTURE (data not shown), according to their geographical origins. The two sub-populations were in HWE and there were no signals of IBD; 100% of the samples were correctly assigned (Table 4). 6) The sub-structure of the Fennoscandian samples was not clearly resolved with this data set: three sub- populations were detected by both STRUCTURE and GENELAND (Fig. 5e), although an alternative splitting into four populations was also supported by STRUCTURE (not shown). Sub-population 1 included only samples collected in south-west Norway; sub-population 2 otters collected in north and central Norway and central Sweden; sub-population 3 grouped samples from Finland and north Sweden. The pooled samples were strongly deviating from HWE, and IBD was very significant (Table 4). However, also the three sub- populations were not in HWE; the IDB was reduced in the sub-populations, with the exception of population 2. Nevertheless, 94% of the individuals were correctly assigned. Interestingly, the sub-population located in south-west Norway was assigned to a distinct cluster, suggesting genetic isolation. |
