A. K. Hundsdoerfer Museum of Zoology




старонка4/9
Дата канвертавання24.04.2016
Памер1.85 Mb.
1   2   3   4   5   6   7   8   9



S.Iberia 17 0.6 (28.9)a 2 (8)a S.Iberia 18 0.5 (28.3)a 2 (5)a S.Iberia 3 0.4 (79.1)a 3 (13)a S.Iberia 24 0.0 (8.6)a 1 (3)a


populations from Costa Vicentina (SW Portugal) and the two populations from south central Algarve, as we found no haplotypes shared among these regions.

Despite the overall high level of populations with shared haplotypes, we did not observe any population with a co- occurrence of haplotypes belonging to different main phylogenetic lineages. The most divergent 12S haplotypes observed to co-occur were ‘S.Iberian’ haplotypes 3 and 24 (uncorrected p-distance 0.9, representing 82% of the maximum intra-lineage divergence observed) and ‘Cádiz’ haplotypes 2 and 5 (uncorrected p-distance 1.1, represent- ing 85% of the maximum intra-lineage divergence).

The differentiation of populations from the south central Algarve and Costa Vicentina as observed for the ‘Portugal’ lineage was also confirmed by AMOVA (Table 5), indicat- ed by a high proportion of the total (within-lineage) genetic variation observed among populations (80.65%). The

‘Cádiz’ lineage had a similarly high proportion of the

variation observed among populations (83.59%), and for both, the ‘Portugal’ and the ‘Cádiz’ lineage, these values were considerably higher than in the ‘S.Iberian’ lineage (within the same type of habitat: 56.26% and 66.84%, see Table 5; from the ‘Gitanilla’ lineage not enough samples were available to perform AMOVA).

For comparison, AMDD values for two haplotypes common in unisexual populations of T. c. cancriformis are shown (Table 4), demonstrating the possible effect of reproductive mode on dispersal success (Korn et al.

2006). Clearly, despite lower sampling efforts, observed AMDD values are higher in this species than in the gonochoric T. mauritanicus.
Evidence for recent gene flow
Evidence for recent gene flow was found on a smaller geographical scale than evidence for dispersal. 38% of the populations for which 12S sequences were available from a minimum of six specimens each, showed associations of geographically spread haplotypes (Table A1; 32% if all populations were considered for which we investigated at least 2 individuals) and thus were indicative of gene flow between populations. However, these populations (i.e. the ones showing associations of geographically spread hap- lotypes) were only observed within the Guadalquivir delta and a small area in southern Cádiz province (between Tahivilla and Benalup).
Genetic diversity among habitat types
Gene diversity of Triops populations differed markedly among all three sampled habitat types within the Guadal- quivir delta (i.e. within a distance of 25 km to the marismas of Doñana National Park; ANOVA, p <0.001, Tukey post-hoc test, p < 0.001 for all pairwise comparisons; Fig. 4a). Triops populations at greater distances to the Guadalquivir delta (>75 km distance to marismas of

Table 5 Results from analysis of molecular variance (AMOVA) based on 12S haplotype frequencies of main lineages within Triops mauritanicus, broken down by habitat types




Lineage

Habitat type

Source of variation

Degrees of freedom

Percent of variation

Probability

Fixation index

‘S.Iberia’

Open, fara

Among populations

6

66.84

<0.001

0.67







Within populations

35

33.16







‘S.Iberia’

Open, closeb

Among populations

9

56.26

<0.001

0.56







Within populations

51

43.74







‘S.Iberia’

Forest/shrubs

Among populations

10

95.50

<0.001

0.95







Within populations

55

4.50







‘S.Iberia’

Marismas

Among populations

8

9.94

<0.001

0.10







Within populations

57

90.06







‘Cadiz’

Open

Among populations

5

83.59

<0.001

0.84







Within populations

36

16.41







‘Portugal’

Open

Among populations

3

80.65

<0.001

0.81







Within populations

20

19.35







a Ponds in >75 km distance to the marismas (natural temporary marshes) of Doñana

b Ponds located 25 km from the marismas



Doñana) were all found in open habitat but had a lower gene diversity than populations found in open habitat within the delta region (Tukey post-hoc test, p <0.05; Fig. 4a). These geographically distant populations showed similarly low gene diversity to populations in closed woodland or shrub habitat within the delta (Tukey post- hoc test, not significant). Gene diversity of Triops populations was positively correlated with estimates of waterbird abundances (regression analysis, R =0.70, p <

0.001; Fig. 4b). Conversely, we found no clear evidence for a possible effect of pasture land use on gene diversity of Triops (ANOVA, p = 0.096), and a statistical model considering only abiotic factors (conductivity, pH, log10

surface area, log10 distance to marismas) could also not sufficiently explain the variability observed among pop- ulations (multiple regression; with distance measurements referring to the present extension of marismas: R =0.49, p =0.20, F = 1.6; with distance measurements considering the borderlines of marisma habitat around year 1900: R =

0.55, p =0.12, F =2.12).

The differences in genetic diversity observed among habitat types were also confirmed by AMOVA. The proportion of genetic variation observed among populations differed markedly among all three habitat types, ranging from 9.9% in marisma habitat to 95.5% in enclosed forest or shrubland (Table 5), indicating a low level of gene flow







Fig. 4 Gene diversity H calculated from 12S haplotype frequencies in populations of the ‘S.Iberian’ lineage of Triops mauritanicus. a H in relation to main habitat types [Forest/s. = forest or shrubland; Open f. (open, far) = ponds at >75 km distance from marismas of Doñana National Park; Open c. (open, close) = ponds ≤25 km from marismas], based on populations 001–039 (see Table A1); error bars indicate

95% confidence intervals, capital letters show statistical classification from Tukey post-hoc test (p < 0.05; ANOVA, p < 0.001). b H in relation to estimates of waterbird abundance (1 = low, 5 = high), based on populations 001–032 from Sevilla and Huelva provinces; regression analysis showed strong correlation (R = 0.70, p < 0.001)






among populations surrounded by forest or shrubland and highest gene flow within marisma habitat. Populations in open habitat were indicated to have intermediate levels of gene flow and, according to the differences in gene diversity observed between the Guadalquivir delta and more distant areas, also showed a difference in the proportion of genetic variation observed among populations (‘open, close’, delta region: 56.26% of total variation;

‘open, far’, >75 km geographic distance to marismas:

66.84%).
Morphological analysis of adult specimens
All three discriminant function analysis models of morpho- logical data yielded significant results (p < 0.001; Table 6). Thus, the multivariate models successfully separated the statistical groups representing phylogenetic lineages of Triops mauritanicus.
Morphological data of males in all presently known lineages
The overall discriminating power of the discriminant function analysis (DFA) model was high, indicated by a low value of Wilk’s lambda (0.009; see Table 6) and a high classification success: 85% of all individuals were classified correctly. The latter value was only slightly lower (82% correct) when a jackknife procedure was applied to the analysis. Classification success was highest for Triops m. simplex and the ‘Portugal’ and ‘Gitanilla’ lineages (Table 7). In contrast, the ‘S.Iberian’ lineage showed the lowest levels of correct classification (Table 7), apparently due to the fact that its morphology is intermediate between that of the

‘Cádiz’ lineage and T. m. mauritanicus (Fig. 5), so that

individuals of the ‘S.Iberian’ lineage with a less typical combination of characters may easily be misclassified to one of those morphologically similar lineages. The mor- phological differentiation of the ‘S.Iberian’ lineage from T. m. mauritanicus is lower than its differentiation from the

‘Cádiz’ lineage (squared Mahalanobis distances 4.4 and

6.1, respectively; Table 8). This morphological proximity of the ‘S.Iberian’ lineage to the nominotypical subspecies is also

Table 6 Results from discriminant function analysis (DFA) for morphological characters
Specification of DFA model Wilks’ lambda F p


Males, all lineages

0.009

16.8

<0.001

Males, Iberian lineages

0.037

13.0

<0.001

Females, Iberian lineages

0.084

8.9

<0.001

The haplotypes were used to pre-define statistical groups tested in each analysis

apparent in the prevailing placement of misclassified obser- vations into that taxon (up to 28% of all individuals of the ‘S. Iberian’ lineage if a jackknife procedure was applied at the population level). When the DFA model was used for identification of Iberian populations, this resulted in a high proportion of populations in the ‘S.Iberian’ lineage that could not clearly be assigned to that lineage (Table 9). Among the

19 populations of the ‘S.Iberian’ lineage classified using a

population-level jackknife procedure, two populations were even misclassified as T. m. mauritanicus.

The NJ tree (Fig. 5) based on squared Mahalanobis

distances between the group centroids of the phylogenetic lineages is in agreement with the phylogenetic reconstruc- tions based on molecular data, with the exception of the position of Triops m. mauritanicus: the ‘Cádiz’ and ‘S. Iberian’ lineages group between the nominotypical subspecies and T. m. simplex, the ‘Portuguese’ and the ‘Gitanilla’ lineage, whereas the latter four form a monophyletic clade in all molecular phylogenetic reconstructions with sufficient resolution (see above).


Morphological data of Iberian males
The reduced dataset including only males from the south- west Iberian Peninsula caused fewer problems in retaining homogeneity of variances, so that in total, one more dependent variable could be included in the model (Table 2). The set of variables in the model resulted in better separation of all four south-west Iberian lineages, as indicated by the consistently higher squared Mahalanobis distances between group centroids of Iberian lineages as compared to distances achieved for the above-described model (Table 8). Consequently, classification success of observations was higher for the morphologically less well differentiated ‘S.Iberian’ and ‘Cádiz’ lineages. For popula- tions, classification success was 89%, both for the ‘S. Iberian’ lineage and for total number of populations (Table 9). None of the populations was misclassified, as the individuals from populations which were not classified correctly could not be assigned clearly to a single lineage.
Morphological data of Iberian females
The results of discriminant function analysis obtained for Iberian females were in agreement with those for males. However, with a value of 0.084 for Wilks’ lambda, discriminating power was lowest for that DFA model (Table 6). Accordingly, distances between groups tended to be lower (Table 8) and classification success of observations was clearly lower than that obtained for males (Table 7). This resulted in almost 20% lower success of population classification (70% correct classification; data not shown) compared to the classification based on males.


Table 7 Classification success of observations (individuals) for discriminant function analysis (DFA) of morphological characters; a priori classification probability set to ‘same for all groups’
Classification based on resubstitution sampling Classification based on jackknife sampling
Males of all lineages

% correct C G P S.I MM MS % correct C G P S.I MM MS

C

80

16

1

0

2

1

0

C

80

16

1

0

2

1

0

G

94

0

16

0

1

0

0

G

94

0

16

0

1

0

0

P

97

0

0

35

1

0

0

P

97

0

0

35

1

0

0

S.I

59

4

1

2

20

7

0

S.I

56

4

1

2

19

8

0

MM

90

0

0

0

3

26

0

MM

79

0

0

0

6

23

0

MS

100

0

0

0

0

0

17

MS

94

0

1

0

0

0

16

S.Ia

63

6

5

3

65

24

0

S.Ia

56

7

5

3

58

30

0
1   2   3   4   5   6   7   8   9


База данных защищена авторским правом ©shkola.of.by 2016
звярнуцца да адміністрацыі

    Галоўная старонка