Parameter Estimation
The parameter estimates obtained in our analyses were in units of coalescent generations and for Thuja plicata and Tsuga heterophylla generally fit with most of our expectations based on the history of the bioregion. For both species, the population sizes estimated for the coastal population are slightly larger than those of the ITR (Table 3), consistent with their current distributions. For both species, the median divergence time estimates between the coastal and inland rainforests were approximately 252,000 generations ago (Table 3). The timing of the population bottleneck event for both species was estimated to be between 50 and 90 Kya (Table 3). The time of the population expansion for Thuja plicata was ~1,050 generations ago, whereas the time of population expansion forTsuga heterophylla was nearly twice that value (2,020 generations ago). The magnitude of the Thuja plicata coastal bottleneck was apparently slightly larger than that of the inland bottleneck. The opposite was true for Tsuga heterophylla , where the bottleneck in the inland was more severe than that on the coast (Table 3). Not surprisingly, the populations with the most severe bottleneck also had the largest population expansion rates (Table 3). Note that this expansion rate was modeled backwards in time; a negative rate indicates the population is getting smaller towards the past, thus expanding forward in time. In both species, migration rates from the coast to the ITR were larger than migration rates from the ITR to the coast (Table 3).
In order to interpret our time estimates, which in coalescent analyses are expressed in units of generations, we need to consider the generation length of each species. The generation length is essentially the average amount of time between consecutive generations in a population. For western redcedar, estimates of trees reaching maturity typically range from 20-30 years (Turner 1985), however trees can reach maturity as early as 10 years in some open grown areas (Minore 1990). The same is true for western hemlock, where most estimates suggest maturity is reached between 25-30 years (Owen et al. 1984) but with trees reaching maturity earlier in some cases (Tesky 1992). To be conservative, we assumed a relatively short generation length of 10 yrs / generation and a longer generation length of 25 yrs / generation for both species. In doing this, we can convert our estimates of the timing of the events from generation into years while accounting for uncertainty in generation length amongst the populations (Table 4). From these generation lengths, we calculate median divergence times between inland and coastal populations between 2.5 - 6.3 Mya (Table 4), where each estimate has an associated 95% CI (Table 4). This indicates the divergence between inland and coastal populations, for both species, wasbefore the onset of the Pleistocene. Similarly, given the overlap in 95% CI for the estimates for both species, we cannot reject that these species’ disjunct populations diverged in concordance.