Estimating rates of population shifts is a perennial question in plant ecology, made urgent by rapid environmental change. Forecasting plant population persistence and transformation in places like the Southeast US, requires good estimates of one of two things: (a) a confident mechanistic understanding of how demography and dispersal combine to produce population shifts over many generations; or (b) sufficiently detailed spatiotemporal reconstructions of population shifts in the past. For 20 years, we’ve had a set of tools and data that could conceptually supply this information (paleoecological, phylogegraphical, and demography/dispersal data; climate reconstructions; SDMs, etc), but we still don’t have good estimates. The problem persists because two variables need to be measured with high accuracy: long distance dispersal (LDD), which sets population spread rate, and the distribution of small populations in the past, which could be cryptic refugia. Both of these variables have a pernicious feature: the robust trends in data provide little information about the metric of interest. Here, I use existing theory to estimate the resolution of data that would be required to confidently estimate the important metrics of LDD and small populations. Then, I survey existing data to assess the status and prospects for resolving these challenges.
Results/Conclusions
Both problems stubbornly persist. LDD: No one has measured the vanguard of farthest dispersing propagules with sufficient detail to constrain migration rates. The details of spatial genetic structure provide a promising avenue for improving estimates of dispersal and establishment, but these have yet to be leveraged into testable predictions of population expansion. Small populations: Advances in statistical modeling have allowed improved estimates of changing historical range limits in cases where dense networks of paleodata exist. LDD & Small populations: Taphonomic uncertainty and sparse data still do not allow us to identify cryptic refugia with paleo data alone, but joint inference from paleodata and spatial genetic data could be sufficient to constrain the problem. The approaches are complementary. Paleodata constrain the robust features of changing mean population size while spatial genetic data identify the signal of dispersing small populations. This combined approach leaves a frustrating gap in places like the Southeast, typical of many of the most diverse parts of the Earth, where paleodata do not exist in sufficient density to estimate the historical distribution of small populations.