Genetic variation in fitness and evolutionary potential
In asexual, clonal populations, fitness can be directly measured as the population’s exponential rate of increase (Bell 2008). Like phenotype, fitness also consists of environmental and genetic components that can be separated in a common garden assay. There is strong evidence for a large amount of genetic variation in fitness among different genotypes of L. minor , and in some cases, even greater variation among genotypes of the same species of Lemnaceae than among closely related species (Ziegler et al. 2015). However, how this variation maps onto the landscape remains unclear (Xu et al. 2015). Although many studies have reported considerable among-site genotype diversity (Vasseur et al. 1993, Cole and Voskuil 1996, Xue et al. 2012, Xu et al. 2015, Ho 2018), it is sometimes thought that L. minor possess poor levels of within site genetic diversity (William C . Jordan 1996, Xu et al. 2015). To our surprise, we found that there was twice as much genetic variation in fitness within sites (among microsites) than among sites (Table 4). This is consistent with studies quantifying intraspecific genetic variation in L. minor using allozymes (Vasseur et al. 1993, Cole and Voskuil 1996, El-Kholy et al. 2015) and amplified fragment length polymorphisms (Bog et al. 2022) that have reported between 4-20 genotypes per site. The source of this genetic variation remains unclear given the low estimates of gene flow (Cole and Voskuil 1996), mutation rates (Sandler et al. 2020), and frequency of sexual reproduction (Hillman 1961, Landolt 1986, Vasseur et al. 1993, Ho 2018) in L. minor .
Genetic variation in fitness is arguably the most important parameter in evolutionary biology since it is what natural selection acts upon, and is therefore directly related to the adaptive potential of a population (Burt 1995). Fisher formalised this relationship in his 1930 fundamental theorem of natural selection (Fisher 1930, Crow 2002) by equating the standardized additive genetic variance in fitness (SVA) with the per generation change in ln mean fitness,\(\overset{\overline{}}{w}\) (Equation 1).
                                                                          \(\text{SV}_{A}=\frac{var(w)}{{\overset{\overline{}}{w}}^{2}}=\ln\left(\overset{\overline{}}{w}\right)\)                                                                    (1)
In a constant environment, all populations experience genetic degradation due to deleterious mutations (Lynch and Gabriel 1990), maladaptive gene flow (Lenormand 2002), and genetic drift (Barton and Partridge 2000). The amount of genetic variation in fitness then represents the population’s ability to counteract these processes and predicts the per-generation increase in mean fitness expected to result from natural selection (Fisher 1930) Likewise, this rate of evolution of fitness, represents the evolutionary potential of a population to respond to maladaptation caused by environmental change. Despite nearly 100 years since Fisher first recognized the crucial importance of this relationship, how much genetic variation in fitness exists in natural populations is a question that still sees considerable debate (Burt 1995, Shaw and Shaw 2014, Hendry et al. 2018).
Although genetic variation in fitness is the result of dominance and epistatic variance in addition to additive variance (Burt 1995, Matsui et al. 2022), this course measure can be used to approximate the upper limit of SVA and therefore rates of evolutionary change. Taking the microsite variance component from the common garden analysis of variance (Table 4), and standardizing it by dividing it by the square of mean fitness, we estimate SVA as 0.0094, or about 1%. This means that fitness is degraded by up to 1% each generation by mutation and immigration, and then restored via purifying selection. Empirical estimates of SVA in wild populations are exceeding scarce. From the 30 estimates in the literature, including just five on plants, SVA seems to range from 1-10% (Burt 1995, Hendry et al. 2018), which is consistent with our findings.