In the second of two posts, Sean Hoban completes his discussion of recent advances in the optimization of collecting strategies for ex situ conservation.
Dr McGlaughlin, a professor at the University of Northern Colorado, works on rare plant species, particularly Sibara filifolia, an annual plant restricted to two small California islands. Sampling seed can help preserve this species in case of loss of wild populations (some populations are located in a naval bombing range — thus there is a very real possibility of loss!). McGlaughlin and colleagues recently tested how to sample on the two islands, whose populations exhibit very different population dynamics: one population has high diversity and high outcrossing, while the other population is highly selfing.
The research shows that very different minimum sampling numbers are needed to get the diversity of the two populations. In fact, more than twice as many samples were needed in the former population than the later! The researchers also mention that the optimal protocol depends on the goal, e.g. to capture all the alleles, or just the ‘non-rare’ alleles. The authors point out that much variation (about one third of the variation on one island, and about half the variation on the other) is composed of rare alleles, which may be important for future evolutionary change or disease resistance. These alleles, though, are not targeted by the Brown & Marshall strategy.
As a postdoc at the University of Tennessee, and soon-to-be Conservation Biologist at the Morton Arboretum, I also test sampling strategies. My work is highly complementary to McGlaughlin’s — I use a model-based approach to test sampling strategies for different species and situations.
First, with Allan Strand, we have shown that the minimum sample size should differ for species with different reproductive biology — for example, species with low dispersal and high rates of self-pollination could need about five times as many seeds as species without these characteristics. Second, when sampling only a small portion of the population, more samples should be taken than when you are able to sample randomly (depending on the species, between two and ten times as much seed). This is important because we often do sample nonrandomly (the Brown & Marshall guideline assumes perfect random sampling), by sampling the nearest access point, or along a road or transect when hard to reach all parts of a population!
Thus far, the message generally is “we need to sample more.” For some species, a lot more! However, the third finding was that, when sampling many populations, the Brown & Marshall guideline may lead to occasional over-sampling. This is because genetic variants are often shared to some degree among populations, and thus when visiting multiple populations there are more chances to catch those variants in the sample — the Brown & Marshall guideline does not consider this possibility. So, in some cases of visiting many sites, 25 samples may be more appropriate. Furthermore, selective choice of far-flung populations can be almost as good as sampling every population (Hoban and Schlarbaum, 2014). My colleagues and I are continuing to test how robust these new recommendations are.
It is of course important to remember that all this work is on the “minimum sample” problem — the lowest sample investment to achieve a genetic variation target. However, many seed are inviable, infested, lost, or used for evaluation tests, and seed lose viability over time even under good storage conditions. The minimum sample recommendations do not account for this. Michael Way estimated that the number of seed to be collected from a population, to allow for monitoring, duplication, distribution and decline in viability, should be at least 10,000. A smaller sample of about 500 seed may be much too few to allow for this ‘active use’ of a collection. Thus, we should also keep in mind seed viability, storage and use when deciding how much seed to sample!
Overall, my work with McGlaughlin, as well as similar work (see papers by Juli Caujapé-Castells, Washington Gapare, Patrick Griffith, and Sandra Namoff), shows that a single sampling guideline for all species is not the best approach. A single guideline will lead to frequent under-sampling and sometimes over-sampling. To achieve the best conservation of genetic diversity, we should design our sampling taking into consideration what we know about a species’ biology (selfing rates, dispersal, distribution), as well as the recent ecology and demography of the population (such as recent declines, or absence of pollinators).
We think our approach — to perform a genetic survey to help us understand genetic patterns, and to use simulation models of genetics and ecology — is a really promising way to improve our sampling and thus the effectiveness of ex situ conservation.