We assessed changes in genetic composition of seven tsetse populations in southeast Uganda in order to gain insight into the population dynamics of G. f. fuscipes. In general, our results provide evidence for temporal stability of G. f. fuscipes populations over the one to two year period that we examined. With the exception of just one or two populations discussed below, mitochondrial haplotype frequencies and microsatellite allele frequencies exhibited little change over time and effective population sizes were generally large.
Compared to other riverine species of tsetse, estimates of Ne
for G. f. fuscipes
were similar to or larger than estimates for G. palpalis palpalis
in Equatorial Guinea [31
] and 2 to 3 orders of magnitudes larger than estimates for G. p. gambiensis
on islands off the coast of Guinea [5
]. Values of Ne
for G. f. fuscipes
populations were also generally larger than estimates for a savannah species, G. pallidipes
, in Kenya [34
]. The large effective population sizes and overall stability of G. f. fuscipes
populations support the hypothesis [35
] that seasonal variation in tsetse numbers, in which larva develop in utero, should be relatively small, since they do not depend on surface water or moist media for breeding. Nonetheless, the lack of variation in genetic structure over time is surprising given the reduced abundance of G. f. fuscipes
observed during the dry season in Uganda [12
]. To reconcile our results with this observation, which may reflect the low efficiency of traps used for monitoring [36
], we suggest that populations of G. f. fuscipes
in dry season refugia remain large, and that seasonal invasion of marginal wet-season habitat (e.g., at Mukongoro, Bunghazi) must occur in waves of tsetse that are large enough to be representative of the refugia population. Large populations of pupa, which develop in the ground over a period of weeks, may also help to ensure the continuity of tsetse populations and would contribute to reducing the variance in genetic changes over time.
In contrast to the other populations, estimates of Ne were low for populations MS and especially OT, where both moment and likelihood methods produced values of only about 200. These values could be indicative of small populations. Ne may also be influenced by overlapping generations and temporal variance in reproductive success as well as the forces of selection, mutation and migration. In this study, however, the low values of Ne observed in these populations probably reflected small differences in the location of trapping sites used for the two temporal samples. Generation 13 from MS was sampled at a distance of about 4 km from the original site at which generation 0 was sampled. Likewise, generation 11 from OT was sampled at a single site that was 11-20 km from the relatively widely dispersed sites from which generation 0 was sampled. Thus, for these sites, which were the only two sites sampled at different locations across years, fine-scale spatial genetic variation could be responsible for the apparent temporal variation in gene frequencies, thus depressing estimates of Ne.
Given that genetic variation in MS and OT samples can probably be attributed to microgeographic variation, the change in genetic composition of the population at JN likely reflects the only significant temporal change observed in this study. Although microsatellite allele frequencies were largely invariate, mtDNA haplotype frequencies here differed significantly between generation 0 and generation 13. Junda (JN), along with sites BN and MS, lies along a narrow zone of contact between two long-diverged and historically-isolated groups of G. f. fuscipes
]. In 2008, populations at all three of these sites harbored both "southern" and "northern" mtDNA haplotypes. Interestingly, in Junda, individuals with the "southern" haplotypes disappeared from the sample after 13 generations. This could be due to a particularly small population of females and stochastic variation in female reproductive success, although in tsetse, the latter is more likely to be true among males than females [5
]. Mating success can also be influenced by Wolbachia
, a symbiont that may impose mating barriers due to cytoplasmic incompatibility between infected and uninfected tsetse individuals [38
], thereby biasing mating in favor of infected females and potentially producing mitochondrial sweeps [39
]. Given the change in mtDNA observed at Junda, flies here should be examined for Wolbachia
. If present, the zone of contact in Uganda may provide a unique opportunity to monitor symbiont-induced population changes over time.