Reproductive isolation and genetic differentiation in North American ...

134
Figure 5. Frequency histograms of telson spine counts in Triops
longicaudatus (solid bars) and T. newberryi (stippled bars).
Figure 4. Frequency distributions of meristic character values in
Triops longicaudatus (solid bars) and T. newberryi (stippled bars).
A. Legless body rings. B. Total body rings.
or fluctuating selection. The latter two have been proposed
as alternate hypotheses to explain large genetic
differences between nearby pools in some freshwater
crustaceans (e.g., Hebert, 1974; Lynch, 1987; Hairston
& Dillon, 1989; Boileau et al., 1992).
Morphological differentiation between species and
among populations
The genetic differentiation between the two North
American Triops species (Figure 2) is accompanied
by significant, but subtle, morphological differentiation
in the numbers of legless body rings and the total
body ring counts (Figure 4). The two species clearly
differ in these two characters, but the nature of the difference
is complex. For both variables, the distribution
of character states for T. newberryi lies between the
two modes of a bimodal distribution representing the
character states for T. longicaudatus. The two species
also differ in the number of spines in the central row
on the dorsum of the telson (Figure 5). Rank-order
comparsions of these two distributions indicate a highly
significant difference, despite extensive overlap in
values. Analysis of carapace ratio did not provide any
clear pattern of differentiation between the two species.
These four morphological characters were also
used to examine patterns of morphological differentation
between populations within each species. Mean
values and standard deviations for the four characters
are given in Tables 3 and 4 for populations of T.
newberryi and T. longicaudatus, respectively. Data are
summarized only for populations in which at least five
individuals were measured. These data were examined
by ANOVA to investigate intraspecific heterogeneity.
Parametric analyses were followed by pairwise multiple
comparisons using the Student-Newman-Keuls
test, and non-parametric (Kruskal-Wallis rank order)
tests were followed by multiple comparisons using
Dunn’s test. Since population variances were heterogeneous
in all comparisons (except for telson spination
inT. longicaudatus males), and the general pattern of
differences was comparable between parametric and
non-parametric analyses, only the latter are presented
here.
In T. newberryi, sample sizes were too small to
allow statistical comparisons among males from different
populations. Comparisons among females, however,
generally indicated that sexual populations do
not differ substantially from each other, nor do sexual
populations consistently differ from unisexual popu-