CHAPTER 4: EVIDENCE OF SELECTION FROM GENOMIC DATA 144
4.4 DISCUSSION 183
4.4.2 ECOLOGICAL CONCLUSIONS FROM SELECTIONFINDER 186
SelectionFinder analysis revealed an over-‐representation of haplotypes that contain genes associated with responses to biotic and abiotic factors, indicating the predictive power of the application for investigating adaptation of populations to local environmental conditions, and most importantly, for informing the design of field experiments to test predictions. For example, the signatures of selection at known R genes and other defense-‐related genes indicate that A. thaliana populations in the UK have been affected by selection pressure from microbial activity.
More than sixty genes containing a LRR domain were found as potential candidates for selection. Of those, at least 12 have previously been reported to be in some way involved in the regulation of induced cell death (the hypersensitive response) and/or other responses to microbial infection. While some of the genes in this class showed evidence of sweeps across several habitat types, the majority showed evidence of sweeps in a single habitat type. This may suggest that local adaptation to specific habitat types occurs across the UK, and in turn suggests significant differences in both the set of pathogenic
species and the environmental conditions encountered across different habitats.
A number of the LRR genes found to possess signatures of selection control aspects of the phenotype relating to the abiotic conditions of the environment. For example, 3 genes -‐ AT1G05700, AT1G17610 and AT3G20600 (see (Kreps et al. 2002; Wang et al. 2013)) – have previously been reported to confer tolerance to cold conditions; intriguingly, within each habitat type, only one of these three genes exhibits signatures of selection. This may constitute an example of local adaptation.
A signature of selection corresponding to the gene AT2G26290 was found in the population in the wall/rocky outcrop habitat. The kinase produced by this gene becomes activated in conditions of dehydration-‐ or salt-‐stress. Since wall/outcrop habitats are likely to retain less water than other habitats (due to the relative lack of soil), it is reasonable to propose that selection pressures more strongly favour alleles conferring an improved tolerance to dehydration in the wall/outcrop habitat than in other habitats.
Several signatures of selection were discovered at loci associated with flowering time, particularly within the garden habitat type (see Appendix 6.2), and also at a number of LRR loci associated with growth and development – for example, AT1G75820 (associated with root development (Stahl et al. 2013)), AT3G49670 (associated with anther and meristem development (DeYoung et al. 2006; Hord et al. 2006)) and AT4G20270 (also associated with meristem development and cold tolerance (DeYoung et al. 2006)). The Ka/Ks ratios at all but one of the flowering time-‐associated loci indicate strong positive selection. In conjunction with the conclusions regarding optimum points on the r/K spectrum discussed below, this may suggest the presence of unique selection pressures for variation in life history traits within this habitat type.
A fourth group of selection signatures corresponds to a set of several genes of the DAR family -‐ AT5G66610, AT5G66620, AT5G66630 and AT5G66640 -‐ which play a role in the control of organ and seed size (Li et al. 2008). The gene
AT2G39830 also exhibits a similar signature of selection, and is also a determinant of seed size (Li et al. 2008). This signature of selection does not appear across all habitats. Larger seeds require parents to dedicate a greater amount of resources towards the growth of their offspring, which also determines the number of offspring any given individual is ultimately capable of producing. This may, therefore, indicate that different habitats favour different optimal points in the trade-‐off between investment of resources in offspring and in growth, and consequently different points along the r/K strategy continuum. The degree of habitat variability is known to affect the optimal point of this trade-‐off; experimentation has shown that a greater degree of environmental variability favours a strategy of higher fecundity, necessitating a shift in investment of resources away from the growth and longevity of the parent and towards investment in larger numbers of offspring (Rose & Charlesworth 1980).
Interestingly, the signature of selection for the DAR genes was detected in garden habitats (which undergo a substantial degree of human-‐caused disturbance) but not in low-‐disturbance ‘wall/outcrop’ habitats. A hypothetical explanation may be that A. thaliana populations sampled from garden habitats have been undergoing selection at loci affecting seed production in response to factors associated with habitat disturbance (e.g., release of nutrients from cultivation), whereas populations that were sampled from wall/outcrop habitats are fully adapted for survival in the harsh, low nutrient conditions and consequently genetically uniform at the same loci.
This variation in selection for r/K strategies may explain the apparent migratory history of the UK population inferred from PCA clustering of genotypes in Chapter
smaller numbers of seeds than the population-‐wide average (see Chapter 4.4.1). Since human disturbance of the environment at these sites is minimal, a population that establishes at such sites could have adapted to become highly specialised to the habitat type represented by walls and rocky outcrops (and, thus, also somewhat genotypically differentiated from mainland populations). As humans also created other, more disturbed habitats – represented by gardens – in more recent times, there would therefore exist selection pressures in these habitats towards an r strategy, but the populations in less disturbed habitats would remain unaffected, and would therefore not show any signatures of selection due to their already high degree of adaptation to those habitats. This plausible scenario may be tested by common garden experiments (see below. If supported by further evidence, this demonstrates that A. thaliana is a suitable case study for the long-‐ term effects of human actions on the selection pressures exerted on wild populations by their environments.
Researchers wishing to further investigate instances of local adaptation should design common garden experiments using pools of samples drawn from all relevant habitat types. For example, an experiment to further investigate the potential multiple instances of local adaptation in favour of cold tolerance could involve sowing seeds collected from representative populations within each habitat type, at a set of common gardens across all 3 habitat types. Genotypic assays of alleles at the 3 listed loci taken over the course of several generations would then reveal which alleles are favoured in each circumstance, and would also more clearly show the type of selection occurring (i.e., balancing, directional, stabilising, etc.); further hypotheses may then be proposed as to why certain alleles are favoured in a given situation.
4.4.3 PLANT-‐PATHOGEN INTERACTION CONCLUSIONS FROM