4 | APPLICATION OFQSTFST COMPARISONS TO PLANT BIOLOGY
Many Q STF ST comparisons falling roughly into seven categories (viz . local adaptation, sexual selection, evolutionary stasis, human-induced evolution, artificial selection, biological invasions, and management and/or conservation) can be applied to plant biology, have been conducted to infer ecological and evolutionary processes. Perhaps the most commonly studied issue is to identify natural selection as a cause of broad-scale clinal variation in morphological and life-history traits (local adaptation, e.g., in Campanulastrum americanum [Prendeville et al., 2013], in Helianthus maximiliani [Kawakami et al., 2011], in two subspecies of Antirrhinum majus [Marin et al., 2020] or various tree species [Savolainen et al., 2007]). As an example of sexual selection, Yu et al. (2011) detected sex-specific selection as the cause of the evolution of sexual dimorphism inSilene latifolia . Using Pinus pinaster as a study species, Lamy et al. (2011) identified selective constraints explaining phenotypic uniformity across species distributions (evolutionary stasis, i.e., canalization or uniform selection).
Other examples are the demonstration of how human-induced habitat changes can either cause or impair adaptation (human-induced evolution, e.g., in Thlaspi caerulescens [Jiménez-Ambriz et al., 2007] and Arabidopsis halleri [Meyer et al., 2010]). In addition, studies on how selective breeding shapes diversification and population structuring of crop species (artificial selection) have been conducted in Oryza sativa (Sreejayan et al., 2011) and Zea mays(Pressoir & Berthaud, 2004). By performingQ STF ST comparisons between the invasive species’ native and invasive ranges (biological invasions), several researchers provided information on the evolution of invasiveness and the adaptive potential of invasive plant species such as Hypericum canariense (Dlugosch & Parker, 2007),Ambrosia artemisiifolia (Chun et al., 2011), Lythrum salicaria (Chun et al., 2009), and Geranium carolinianum (Shirk & Hamrick, 2014).
For management purposes, Gravuer et al. (2005) identified units or populations suitable for translocation in Liatris scariosa . Furthermore, some authors demonstrated that setting conservation priorities should not be based only on neutral marker diversity and thatQ STF ST comparisons can be used to identify populations that are suitable for translocation inArabis fecunda (McKay et al., 2001) and Araucaria araucana(Bekessy et al., 2003). The last two issues, i.e., conservation and management, will be the focus of the next section.
5 | INSIGHTS INTO CONSERVATION AND RESTORATION DERIVED FROMQSTFST COMPARISONS
Because F ST estimates are significantly lower in trees than in most herbaceous perennials and annuals, Chung et al. (2020) recommended that conservation genetic strategies be designed differently for tree species versus other types of plant species. That is, seeds of most tree species (which generally show low values ofF ST) could be sourced from a few populations distributed across the species’ range, whereas seeds of rare herbaceous species (often with high F ST values) should be taken from many populations to capture the highly localized genetic diversity. Based on a small body of available data on seed plant species (data from Lamy et al., 2012; De Kort et al., 2013; Leinonen et al., 2013), on average, Q ST is higher thanF ST in common forest tree species, indicating that their quantitative traits have been subject to diversifying selection and local adaptation (Kremer et al., 1997; Savolainen et al., 2007). It has been suggested that more populations would be needed to preserve enough AGV for adaptively significant quantitative traits than for NGV, particularly in trees (McKay et al., 2001; Hamrick et al., 2006; Chung et al., 2020). More specifically, conservation practitioners may need information about how to capture most AGV and NGV based on known levels of NGV and AGV from population or conservation genetic studies.
Population(s) to be protected in situ or to be sampled for seed banking purposes could be estimated using the formulae: PNGV = 1 –F ST (orG ST)N for NGV, where PNGV = proportion of NGV captured by sampling, N = number of populations (Ceska et al., 1997; Hamrick et al., 2006) and PAGV = 1 –Q STN for AGV, where PAGV = proportion of AGV captured by sampling. However, one should be aware that if there are more than two alleles per locus for the neutral markers, then Q ST and G STare on different scales, and the formulae PAGV = 1 –Q STN and PNGV = 1 –F STN cannot be interpreted in the same way (J. D. Nason, pers. comm.). For multi-allelic markers, it depends on µ whether this is problematic. In addition, as Ф ST (a function of the among-population variance component and the within-population component, which is based on genetic distances among alleles for the neutral markers) is conceptually similar to Q ST, it is advisable to use Ф ST rather thanG ST, F ST, or θ(Edelaar et al., 2011). The calculations for 99% capture of AGV and NGV can be key to figuring out ideal sample sizes, especially when resources are limited. Based on the average values of De Kort et al. (2013) forF ST and Q ST (annuals,n = 19, 0.308 versus 0.451 [i.e., Q ST is about 1.5 times greater than F ST]; herbaceous perennials, n = 14, 0.267 versus 0.299 [Q ST is about 1.1 times greater]; woody perennials, n = 18, 0.074 versus 0.269 [Q ST is about 3.6 times greater]; recalculated from De Kort et al., 2013), to capture 99% of NGV and AGV for woody perennials, just two and four populations would be needed using the abovementioned formulae, respectively. On the other hand, on average, four populations of herbaceous perennials would be needed to secure 99% of NGV and AGV, respectively, because the average difference betweenQ ST and F ST is small (0.032). For annuals, on average, four and six populations are needed to secure 99% of NGV and AGV, respectively.
We apply this approach to a real-life example: for the widespread treePopulus balsamifera , Keller et al. (2011) reported a meanФ ST value of 0.067 estimated from 310 nuclear SNP loci and a mean Q ST value of 0.421 (range = 0.127–0.832) obtained from 13 ecophysiological and phenological traits originating from 20 populations across North America. To capture 99% of NGV, two populations of this tree would be needed using the above formula. When we apply the mean Q ST value to the formula, at least six populations would be necessary to capture the same level of AGV. However, the value of Q ST depends on the trait under consideration: for traits with a highQ ST, more populations should be sampled than for traits with a low Q ST. Given this, the prudent thing would be not to use the average Q ST but the maximum Q ST in these calculations. If this logic is applied to P. balsamifera , as for bud setQ ST = 0.832, then up to 25 populations would need to be targeted to maintain enough variation for the trait. This does not mean that NGV is not essential; there is probably a reservoir of genetic variation in every population that is neutral now but may be selectively important if environmental conditions change. Furthermore, NGV can be very informative about the populations’ past demography which is often of interest in conservation biology (Frankham, 2015; Allendorf, 2017; DeWoody et al., 2021; García-Dorado & Caballero, 2021).
The application of the above formulae to plants with different life forms, as well as the example of Populus balsamifera , suggests that conservation and management policies or actions based solely onF ST could potentially be misleading. Again, these findings stress that guidelines and conservation genetic strategies should be designed based on genetic information on both NGV and AGV together for tree and herbaceous (whether perennial or annual) species. In addition, managers or practitioners should design restoration and conservation strategies by knowing that, on average,Q ST is about 3.6, 1.5, and 1.1 times greater thanF ST in woody plants, annuals, and herbaceous perennials, respectively.
To summarize, F ST estimates appear to be more closely related to AGV than within-population genetic diversity metrics (e.g., H e,%P , or AR ) in seed plant species. Thus,F ST should be considered as a more predictable parameter for conservation and restoration purposes. Together with the metrics of H e, %P , or AR , the particular degree of F ST (i.e., low, moderate, or high) is important for prioritizing populations for collection and identifying appropriate sources for reintroductions (Hamrick & Godt, 1996; Ottewell et al., 2016; Chung et al., 2021). Thus, the importance of the proper consideration of F ST information (and Q ST, if available) in conservation management cannot be overstated, particularly when it comes to annuals and herbaceous perennials.