FIGURE LEGENDS
Figure 1. Pedigree illustration of parental lines (P1 and P2) and initial filial generations, where F1 results from crossing the two parental lines, and F2 results from crossing F1s. Backcrosses, B1 and B2, result from crossing F1 with the parental lines P1 and P2, respectively.
Figure 2. Expectations for the fitness of hybrid F1 offspring depend on ecological and genetic distances between parental lineages. To the left, outcrossing of inbred lines commonly leads to high fitness due to increased heterozygosity (known as ‘genetic rescue’ in conservation genetics). As ecological distance between parental lines increases, benefits of outcrossing are outweighed by negative effects of ecological incompatibility. To the right, hybrid F1 from interspecific crosses commonly show low fitness due to genetic incompatibilities between highly divergent genomes. In between these extremes, indicated by the vertical dotted lines, the relative balance between positive and negative outcomes of both genetic and ecological divergence generates a rich spectrum of possible fitness consequences to the hybrid offspring of matings between natural immigrants and residents arising within metapopulation systems.
Figure 3. Different relative magnitudes of genetic effects can lead to positive or negative increments on trait or fitness values across filial generations, as demonstrated by imputing into the line-cross theory equation (A) arbitrary values for the magnitudes of genetic effects (C-E, top-right corner). The average trait value between parental lines (Pmid) indicates the expected trait value for the F1 given purely additive (α1) genetic effects, as in C. With additive and positive dominance (δ1) effects (as in D), trait values for all filial generations are larger than Pm, therefore showing positive mid- and/or best-parent heterosis. Positive heterosis is also observed for F1 and B1when additive-by-dominance epistasis (α1δ1) is present in addition to additive and dominance effects (as in E). In this case, however, F2 and B2 show negative heterosis due to the loss of positive epistatic benefits, as indicated by the coefficients for source and hybridity indexes (B) for these filial generations (see Box 1 for details). Trait value expectations are indicated for in environment 1. Note that the y-axes in plots (C-E) are in the same scale, since the F2 (square) is taken as reference (μ0).
Figure 4. Non-linear genotype-phenotype maps can lead to heterosis. A) even if genetic effects are additive, i.e. gene product (x-axis) for the heterozygote (Aa) equals the mean of the recessive (aa) and dominant (AA) homozygotes, the phenotypic trait value (Y) for the heterozygote can deviate from the mean expectation (\(\overline{Y}\)). This conclusion can be extended to the case of concave adaptive landscapes where fitness is a non-linear function of the genotype/phenotype. In the context of local adaptation in spatially-structured populations (B-C), individuals in populations P1 (circle) and P2 (square) (presenting either 100% P1 = 0% P2 alleles or 0% P1 = 100% P2 alleles) are at the optima for the different adaptive landscapes corresponding to their respective local environments (yellow for P1 and blue for P2). When P1 and P2interbreed (50% P1 alleles), the fitness for the resulting F1 deviates from the mean fitness of parental populations (Pmid) in either environment, as a consequence of the non-linearity of the fitness landscape. In both examples, populations match “home-vs-away” and “local-vs-foreign” criteria for patterns of local adaptation, but fitness decreases more abruptly in the adaptive landscapes of B than of C, representing more contrasting selective pressures between environments as the optima are further apart. Consequently, while scenario C leads to positive heterosis, hybrid offspring in B show outbreeding depression.
Figure 5 . Representation of literature review work pipeline. The flow diagram (left) depicts each stage of the process with the total number of studies advancing to the next stage displayed over the respective box. The Venn diagrams (right) show the numbers of papers found per keywords searched (colour codes) for two different stages of the process: total papers triaged (upper; N=12,862), and studies selected as containing fitness or fitness-related estimates of offspring from intra- and inter-population crosses (lower; N=111). Although a large total number of studies was found, the Venn diagrams indicate little overlap between results produced by keywords chosen to find ecological studies versus keywords representing more commonly used terms within traditional heterosis literature. See text and table 1 for details.
Figure 6. Proportion of studies on major plant and animal groups for which estimates of fitness or fitness-related traits for both within and between population crosses were found in the literature review. In the outer pie, percentages are shown for groups that appeared in 5% (i.e. N > 5) or more studies out of the total number of studies using animals (N=32) or plants (N=79).
Figure 7. The literature review indicates a paucity of estimates of fitness consequences of interbreeding between natives and immigrants of populations connected by dispersal in natural environmental conditions. Studies found included population crosses across different levels of connectivity, with some (total number of studies presented above the bars) containing crosses between populations connected (“Connected” bars) or not connected (“Not connected”) by natural dispersal, some including crosses at both connectivity levels (“Both”), and some for which connectivity was not possible to be categorized (“Unclear”). For each connectivity category, left bar indicates approach used to obtain crosses between populations. “Introduction” indicates the translocation of individuals from a different population, or [re-]introduction of individuals from different populations into an uninhabited site. Right bars indicate the type of environment in which at least one fitness component or fitness-related trait was estimated for filial generations obtained via experimental crosses: “parental” indicates at least one of the parental environments, “semi-natural” indicates mesocosms, common gardens, or environments not occupied by the parental populations, and “artificial” indicates laboratory or greenhouse. Percentages shown are in reference to the total number of studies within each connectivity category using experimental crosses (left) or using experimental crosses and estimating fitness exclusively in artificial or semi-natural environments (right).