Detecting and quantifying introgression in hybridized populations: simplifying assumptions yield overconfidence and uncertainty

A growing threat to the conservation of many native species worldwide is genetic introgression from non-native species. Although improved molecular genetic techniques are increasing the availability of species-diagnostic markers for many species, efficient field sampling design and reliable data interpretation require accurate estimates of uncertainty associated with the detection of non-native alleles and the quantification of introgression in native populations. Using fish populations as examples, we developed a simulation model of an age-structured population that tracks the introduction and inheritance of non-native alleles across generations by simulating stochastic mating and survival of individual fish and the resulting transmission of diagnostic markers. To simulate detection and quantification of introgression, we sampled varying combinations of n fish and m diagnostic markers to detect and quantify introgression from thousands of virtual, independent fish populations for a wide range of hybridization scenarios. Using the results of simulated sampling, we quantified the extent to which common simplifying assumptions regarding population structure and inheritance mechanisms can lead to the following: (i) overconfidence in our ability to detect non-native alleles and (ii) unrealistically narrow confidence intervals for estimates of the proportion of non-native alleles present. Under many circumstances, commonly used simplifying assumptions underestimate the probability of failing to detect ongoing introgression and the uncertainty associated with estimates of introgression by orders of magnitude. Such overconfidence in our ability to detect and quantify introgression can affect critical conservation and management decisions regarding native species undergoing or at risk of introgression from non-native species.