The effects of locus number, genetic divergence, and genotyping error on the utility of dominant markers for hybrid identification

Abstract

In surveys of hybrid zones, dominant genetic markers are often used to identify individuals of hybrid origin and assign these individuals to one of several potential hybrid classes. Quantitative analyses that address the statistical power of dominant markers in such inference are scarce. In this study, dominant genotype data were simulated to evaluate the effects of, first, the number of loci analyzed, second, the magnitude of differentiation between the markers scored in the groups that are hybridizing, and third, the level of genotyping error associated with the data when assigning individuals to various parental and hybrid categories. The overall performance of the assignment methods was relatively modest at the lowest level of divergence examined (F st ˜ 0.4), but improved substantially at higher levels of differentiation (F st ˜ 0.67 or 0.8). The effect of genotyping error was dependent on the level of divergence between parental taxa, with larger divergences tempering the effects of genotyping error. These results highlight the importance of considering the effects of each of the variables when assigning individuals to various parental and hybrid categories, and can help guide decisions regarding the number of loci employed in future hybridization studies to achieve the power and level of resolution desired.

Keywords: AFLP; Dominant Markers; Genotyping Error; Hybridization; RAPD; Simulation.

Figure 1

Average efficiency (circles), accuracy (triangles), and performance (squares) values for assignments of individuals into parental or hybrid categories. Assignments are based on a range of locus numbers (25–125) generated across three divergence levels, and incorporating three levels of genotyping error. A value of 0.95 is represented by the dashed line.

Figure 2

Average efficiency (circles), accuracy (triangles), and performance (squares) values for assignments of individuals into each of six categories (P1, P2, F₁, F₂, B × 1, B × 2). Charts labeled “Parental” and “Backcross” represent values averaged across the P1 and P2 and B × 1 and B × 2 categories, respectively. Assignments are based on a range of locus numbers (25–125), simulated at the lowest level of divergence analyzed in this study (F_st = 0.43). A value of 0.95 is represented by the dashed line.

Figure 3

Average efficiency (circles), accuracy (triangles), and performance (squares) values for assignments of individuals into each of six categories (P1, P2, F₁, F₂, B × 1, B × 2). Charts labeled “Parental” and “Backcross” represent values averaged across the P1 and P2 and B × 1 and B × 2 categories, respectively. Assignments are based on a range of locus numbers (25–125), simulated at a divergence level of F_st = 0.55. A value of 0.95 is represented by the dashed line.

Figure 4

Average efficiency (circles), accuracy (triangles), and performance (squares) values for assignments of individuals into each of six categories (P1, P2, F₁, F₂, B × 1, B × 2). Charts labeled “Parental” and “Backcross” represent values averaged across the P1 and P2 and B × 1 and B × 2 categories, respectively. Assignments are based on a range of locus numbers (25–125), simulated at a divergence level of F_st = 0.67. A value of 0.95 is represented by the dashed line.

Figure 5

Average efficiency (circles), accuracy (triangles), and performance (squares) values for assignments of individuals into each of six categories (P1, P2, F₁, F₂, B × 1, B × 2). Charts labeled “Parental” and “Backcross” represent values averaged across the P1 and P2 and B × 1 and B × 2 categories, respectively. Assignments are based on a range of locus numbers (25–125), simulated at the highest level of divergence analyzed in this study (F_st = 0.81). A value of 0.95 is represented by the dashed line.

Figure 6

Proportions of misassigned individuals assigned to each incorrect category for each of the four classes (parental, F₁, F₂, and backcross). The category “NA” includes instances in which an individual was not assigned to any category with a probability of 0.5 or greater. Data for each category are based on the average proportion of misassignments to each of the incorrect categories across all divergence levels and locus numbers. The percentages at the top of each plot indicate the average proportion of individuals in that category that were misassigned in the total dataset.