Models with a porpoise.

F ontaine et al. provide insight into the structure of a geographically disjunct population on a severe downtrend (1). To understand these data, they turned to a Bayesian statistical approach, providing the mathematical base for changing existing beliefs in light of new evidence. Their hybrid simulation/genomic analysis is one of the latest demonstrations that increasing crossfertilization between bioinformatics and evolutionary biology is as simple as “ABC”— approximate Bayesian computation. At issue is whether small populations per se are dangerous from a genetic perspective, signaling imminent extinction, or if their ecology predisposes them to undergo periodic contractions, followed by expansion from which they recover. To evaluate this question, we must distinguish between ancient and modern population bottlenecks. Although harbor porpoises (Phocoena phocoena) are currently a species of “least concern,” with an estimated several hundred thousand worldwide (2), the Black Sea subspecies (P. phocoena relicta) is protected by the Agreement on the Conservation of Cetaceans in the Black Sea, Aegean, Mediterranean, and Contiguous Atlantic Area. Is this sufficient to repurpose scarce resources to more pressing conservation cases? If protective measures are working, there may be a genetic signal of an expanding population that will be captured by a different statistical approach. After all, 7 billion humans today once started as small, geographically separated groups, with local populations flickering in and out of existence. Debates about our own population history stimulated researchers to develop new analytic tools, and out of population and evolutionary genetics came computational techniques anchored in Bayesian statistics (3, 4). Bayesian statistics can improve the accuracy of genetic risk assessment for families carrying mutations that could increase their probability of developing disease phenotypes vs. the risk frequency in the general population because we know their prior information (5). By combining new data, a better model builds confidence in making future decisions. Mendelian inheritance is, after all, a flawed mathematical model (6). In Bayesian statistics, posterior probability distributions, modeling the distribution of an unknown quantity (a parameter) treated as a random variable becomes conditional on evidence obtained from experiments or surveys. Increasingly, biologists turn to simulation to broadly test model distributions, exposing the need to estimate, fit, and assess biological parameters used to build subsequent models of how populations grow and contract. Parameters that improve model fit can include initial population size, fitness penalties for different genotypes, migration and mutation rates, expansion or contraction rates, and generation times. Fontaine et al. (1) try out various models of population expansion and collapse by using fisheries and genetic data on harbor porpoises in the Black Sea. Based on evidence gathered from a relatively small number of stranded animals, they compare them with samples from the adjacent Aegean Sea. Approximately 7,500 to 8,000 y ago, a rapid sea level rise caused the Mediterranean to overflow into a lowlying fresh water lake, the Black Sea. Early farming communities clustered around this lake drowned in an event so sudden that the Mediterranean spill over the sill at the Bosporus might have taken less than a decade, with linkage to the memory probably becoming Noah’s flood of the Judeo–Christian Bible (7) (Fig. 1). As only a limited number of porpoises from the Mediterranean would have entered the Black Sea, surviving the salinity and temperature shift in the new habitat, most researchers would expect to detect a significant population bottleneck and loss of genetic diversity over time. Despite the physical barriers imposed by the narrow migration corridor to open water, Fontaine et al. discover that the animals in the Aegean and Black Sea are in Hardy– Weinberg equilibrium (1). Although this does not imply a lack of morphological or behavioral differences, it raises the possibility that local extinctions are reversed by immigration, so an exploration of bottleneck biology took a Bayesian turn. Fifteen different alternative demographic scenarios were examined closely, with two gaining very high statistical support. The population bottleneck idea has become a critical feature of conservation biology because it underpins the current extinction crisis. A population grown from fewer genetic ancestors will have less variant phenotypes responsive to selective gradients. As a propagule of a once-large population, it faces random environmental catastrophes. The harbor porpoises of the eastern Mediterranean (P. phocoena relicta) were caught in two catastrophes, the first of which was approximately 7,500 y ago, followed by a second event in the past five decades linked to direct hunting for meat and blubber, indirect bycatch mortality in fishing nets, and massive mortality caused by intense eutrophication of the Black Sea. Fisheries scientists suggest that perhaps 70% of the harbor porpoise population in the Black Sea has already been lost through direct and indirect effects (8). With the concomitant loss of top predators in the Black Sea ecosystem, the short-beaked common dolphin (Delphinus delphis) and the common bottlenose dolphin (Tursiops truncatus), reversing a regime shift now may not even be possible (9, 10). How close are we to the extinction point for the Black Sea harbor porpoise? It depends on how accurate and/or reliable our tests for genetic bottlenecks really are. Detecting the decline of a population under pressure usually entails sampling nondestructively to conserve numbers, and hence the popularity of genome-based methods (11). At root is the idea that a bottleneck test detects the decline in species’ abundance by testing the decline in effective population size (Ne). If you test a small but stable population by incorrectly calling a bottleneck, scarce resources are diverted needlessly. Conversely, if you fail to detect a decline in an endangered or threatened species, the delay likely results in extinction. Like the Mirror of Erised (12) showing Harry Potter the deepest, most desperate desires of his heart and not a true reality, use of a flawed bottleneck model or an invalid test might reflect a biased agenda, used consciously or unconsciously. Prediction of the consequences of global change to carnivores and primates with minimum Fig. 1. A pair of harbor porpoises (P. phocoena) from which a Black Sea relict population is derived.