Impact of Amplified Fragment Length Polymorphism Size Homoplasy on the Estimation of Population Genetic Diversity and the Detection of Selective Loci

Abstract

AFLP markers are becoming one of the most popular tools for genetic analysis in the fields of evolutionary genetics and ecology and conservation of genetic resources. The technique combines a high-information content and fidelity with the possibility of carrying out genomewide scans. However, a potential problem with this technique is the lack of homology of bands with the same electrophoretic mobility, what is known as fragment-size homoplasy. We carried out a theoretical analysis aimed at quantifying the impact of AFLP homoplasy on the estimation of within-and between-neutral population genetic diversity in a model of a structured finite population with migration among subpopulations. We also investigated the performance of a currently used method (DFDIST software) to detect selective loci from the comparison between genetic differentiation and heterozygosis of dominant molecular markers, as well as the impact of AFLP homoplasy on its effectiveness. The results indicate that the biases produced by homoplasy are: (1) an overestimation of the frequency of the allele determining the presence of the band, (2) an underestimation of the degree of differentiation between subpopulations, and (3) an overestimation or underestimation of the heterozygosis, depending on the allele frequency of the markers. The impact of homoplasy is quickly diminished by reducing the number of fragments analyzed per primer combination. However, substantial biases on the expected heterozygosity (up to 15-25%) may occur with $50-100 fragments per primer combination. The performance of the DFDIST software to detect selective loci from dominant markers is highly dependent on the number of selective loci in the genome and their average effects, the estimate of genetic differentiation chosen to be used in the analysis, and the critical bound probability used to detect outliers. Overall, the results indicate that the software should be used with caution. AFLP homoplasy can produce a reduction of up to 15% in the power to detect selective loci. et al. 1995) is becoming one of the most popular methods in the fields of conservation and evolutionary genetics and ecology (Mueller and Wolfenbarger 1999;Bensch and Akesson 2005;Bonin et al. 2007;Meudt and Clarke 2007), as it combines a high reproducibility and information content with the possibility of making genomewide screenings. Because of the anonymous nature of the fragments generated by the AFLP technique, however, one major concern is the incidence of size homoplasy due to the lack of homology of comigrating fragments. This implies that fragments of a given size migrating in a band may involve more than one locus of the genome and, therefore, the inferences obtained from the band can produce misleading conclusions. T HE amplified fragment length polymorphism (AFLP) technique (VosSeveral empirical approaches have been used to estimate levels of homoplasy in AFLP data sets. A few studies have demonstrated the presence of homoplasy in AFLP data by sequencing comigrating fragments within t...

Identifying and reducing AFLP genotyping error: an example of tradeoffs when comparing population structure in broadcast spawning versus brooding oysters

“…For unknown reasons band-present frequency distributions transitioned abruptly between fragments o300 versus 4300 bp in C. virginica (Figure 2). The Caballero et al (2008) predictions for homoplasy effects were seen much more strongly in loci o300 versus 4300 bp in 4% error C. virginica data; the loci o300 bp had p five times larger and H S three times larger than loci 4300 bp. In contrast to expectations, however, among-locus average F ST was twice as high for short fragments (0.0409 vs 0.0212, both significantly different than zero at Po0.001).…”

“…Simulated metapopulations with p ¼ 0:1 studied by Caballero et al (2008) showed homoplasy effects to include a downward bias on F ST and an upward bias on both H e and p, each bias monotonically increasing with shorter fragment size. In this study, a negative correlation between fragment size and band-present phenotype frequency indicated some homoplasy effects in all fragment size classes.…”

“…Homoplasy is a well-documented artifact of scoring dense fragment profiles (Vekemans et al, 2002) with stronger impacts expected on smaller fragments (Althoff et al, 2007;Caballero et al, 2008). Simulated metapopulations with p ¼ 0:1 studied by Caballero et al (2008) showed homoplasy effects to include a downward bias on F ST and an upward bias on both H e and p, each bias monotonically increasing with shorter fragment size.…”

“…Effects on estimation of genetic diversity and population structure have mostly been hypothesized (Bonin et al, 2004), demonstrated with simulations (Caballero et al, 2008), and only recently explored empirically (Arrigo et al, 2009;Herrmann et al, 2010;Crawford et al, 2011). Bonin et al (2004) hypothesized that 0.1% average genotyping error among loci may provide little improvement for population genetic inferences compared with 2%.…”

Genome methylation patterns across castes and generations in a parasitoid wasp

“…This reduces the potential biases due to homoplasy (Caballero et al. 2008). The collected data were entered into a binary data matrix that indicated the banding pattern generated by the isoschizomer endonucleases Hpa II and Msp I.…”