Detection of hybrids in nature: application to oaks (Quercus suber and Q. ilex)

Abstract

Powerful and accurate detection of first-generation (F1) hybrids and backcrosses in nature is needed to achieve a better understanding of the function and dynamics of introgression. To document the frequency of ongoing interspecific gene exchange between two Mediterranean evergreen oaks, the cork oak (Quercus suber) and the holm oak (Q. ilex), we analyzed 1487 individuals originating from across the range of the two species using eight microsatellite loci and two Bayesian clustering approaches (implemented in the programs STRUCTURE and NEWHY-BRIDS). Simulated data were used to assess the differences between the two clustering methods and to back up the choice of the threshold value for the posterior probability to discriminate admixed from pure individuals. We found that the use of STRUCTURE resulted in the highest power to detect hybrids, whereas NEWHYBRIDS provided the highest accuracy. Irrespective of the approach, the two species were clearly distinguished as independent genetic entities without any prior information. In contrast with previous reports, we found no evidence for unidirectional introgression. The overall hybridization rate was very low (o2% of introgressed individuals). Only two individuals were identified as F1 hybrids and five as early backcrosses. This work shows that the combined application of the two complementary Bayesian approaches and their systematic validation with simulations, fit for the case at hand, helps gain resolution in the identification of admixed individuals.

A nuclear DNA barcode for eastern North American oaks and application to a study of hybridization in an Arboretum setting

“…Oak species with similar reproductive strategies and overlapping geographic regions often hybridize in natural stands (Chybicki & Burczyk, 2010; Curtu, Gailing, & Finkeldey, 2007; Dodd & Afzal‐Rafii, 2004; Dumolin‐Lapègue, Démesure, Fineschi, Le Corre, & Petit, 1997; Dumolin‐Lapègue, Kremer, & Petit, 1999; Efrain Tovar‐Sanchez, 2004; Gerber et al., 2014; Hipp & Weber, 2008; Lexer, Kremer, & Petit, 2006; Moran, Willis, & Clark, 2012; Petit, 1993; Petit et al., 2004; Whittemore & Schaal, 1991). Morphological intermediacy, however, is an imperfect predictor of genetic admixture in oaks, making hybrids difficult to identify (Burgarella et al., 2009; Song, Deng, Hipp, & Li, 2015; Wei, Li, Zhang, & Liao, 2015). Moreover, multispecies hybrid zones are common in oaks and particularly difficult to study due to sampling issues and selection of adequate numbers of informative loci (Craft & Ashley, 2006; Dodd & Afzal‐Rafii, 2004; Peñaloza‐Ramírez et al., 2010; Sullivan, Owusu, Weber, Hipp, & Gailing, 2016).…”

SimHyb: a simulation software for the study of the evolution of hybridizing populations. Application to Quercus ilex and Q. suber suggests hybridization could be underestimated

“…1 includes the expected values of q, the probability of an individual to belong to the Q. suber cluster, provided by STRUCTURE. This parameter can be used to detect hybrid individuals (Burgarella et al 2009) and is frequently considered a proxy of the contribution of one species to the genome. Burgarella et al (2009) reported a good performance of STRUCTURE and this set of markers for the detection of first generation hybrids and first backcrosses, in terms of efficiency and accuracy (sensu Vähä & Primmer 2006).…”

“…Nuclear molecular markers combined with Bayesian approaches have been used for the identification of hybrid individuals in populations, according to the posterior probability of each individual to belong to any of the parental species genetic clusters. This approach was applied by Burgarella et al (2009) in sympatric Q. ilex -Q. suber populations, estimating a current introgression rate of <2%. Later on, this procedure has been applied in studies for many other species, including animals (Neaves et al 2010, Bogdanowicz et al 2012, Malde et al 2017.…”

“…The output of the simulations was downstreamed to STRUCTURE 2.3.4 (Pritchard et al 2000), and the individual q values of the offspring, and the ratio between the offspring and the maternal q were calculated to infer the detection power of the aforementioned microsatellites in identifying hybrids and advanced introgression. For STRUCTURE analyses we used the admixture model assuming independent allele frequencies, a burn-in of 50,000 steps followed by 100,000 iterations, as in Burgarella et al (2009). Results do not vary significantly across multiple runs and with longer burn-in/iteration cycles.…”

“…We used the allele frequencies obtained for the pure Q. ilex and Q. suber individuals identified in Burgarella et al (2009), covering the whole distribution range of Q. suber. We run several simulations with a variable number of input parameters to produce pure and hybrid individuals of known pedigree.…”