Historical and contemporary gene dispersal in wild carrot (Daucus carota ssp. carota) populations

Abstract

Background and aims: Wild carrot is the ancestor of cultivated carrot and is the most important gene pool for carrot breeding. Transgenic carrot may be released into the environment in the future. The aim of the present study was to determine how far a gene can disperse in wild carrot populations, facilitating risk assessment and management of transgene introgression from cultivated to wild carrots and helping to design sampling strategies for germplasm collections.

Methods: Wild carrots were sampled from Meijendel and Alkmaar in The Netherlands and genotyped with 12 microsatellite markers. Spatial autocorrelation analyses were used to detect spatial genetic structures (SGSs). Historical gene dispersal estimates were based on an isolation by distance model. Mating system and contemporary pollen dispersal were estimated using 437 offspring of 20 mothers with different spatial distances and a correlated paternity analysis in the Meijendel population.

Key results: Significant SGSs are found in both populations and they are not significantly different from each other. Combined SGS analysis indicated significant positive genetic correlations up to 27 m. Historical gene dispersal sigma(g) and neighbourhood size N(b) were estimated to be 4-12 m [95 % confidence interval (CI): 3-25] and 42-73 plants (95 % CI: 28-322) in Meijendel and 10-31 m (95 % CI: 7-infinity) and 57-198 plants (95 % CI: 28-infinity) in Alkmaar with longer gene dispersal in lower density populations. Contemporary pollen dispersal follows a fat-tailed exponential-power distribution, implying pollen of wild carrots could be dispersed by insects over long distance. The estimated outcrossing rate was 96 %.

Conclusions: SGSs in wild carrots may be the result of high outcrossing, restricted seed dispersal and long-distance pollen dispersal. High outcrossing and long-distance pollen dispersal suggest high frequency of transgene flow might occur from cultivated to wild carrots and that they could easily spread within and between populations.

Fig. 1.

Illustration of the sampling from a wild carrot population in Meijendel (WM), South Holland, The Netherlands. Small crosses indicate the positions of samples.

Fig. 2.

Illustration of the sampling from a wild carrot population in Alkmaar (WAL), North Holland, The Netherlands. Small crosses indicate the positions of samples.

Fig. 3.

Correlograms summarizing the spatial genetic structure for the wild carrot population in Meijendel (WM) and Alkmaar (WAL) and the combined spatial structure analysis for both of the populations using GenAlEx 6·2 (Peakall and Smouse, 2006; Smouse et al., 2008). The autocorrelation coefficient r indicates the genetic similarity between pairs of individuals whose geographical separation falls within the specified distance class, e.g. 0–14, 15–27 m, etc. (Peakall and Smouse, 2006). The error bars show the 95 % confidence interval of r estimated with bootstrap re-sampling. The number of pairwise individuals within a distance class is given in parentheses. The dotted lines indicate the 95 % upper and lower confidence limits for the null hypothesis of no spatial genetic structure. ***P < 0·005 indicates significantly positive genetic correlation.

Fig. 4.

Bar chart showing the mean Sp values of plant species with different mating systems and wild carrot. The Sp values of different mating systems are taken from Vekemans and Hardy (2004) based on the Sp statistics analysis of 47 plant species. Numbers in parentheses indicate the numbers of plant species used in different mating systems. Plant mating systems are classified as selfing (average selfing rate s higher than 90 %), mixed-mating (10 % <s < 90 %), outcrossing (s < 10 % but self-compatible) and self-incompatible as indicated in Vekemans and Hardy (2004). The study by Vekemans and Hardy (2004) found the Sp value was significantly related to the mating system. The mean Sp value of wild carrot was estimated using the genetic and geographical distance data of wild carrots collected from Meijendel (WM) and Alkmaar (WAL) according to Vekemans and Hardy (2004) and Hardy et al. (2006) using SPAGeDi 1·2 (Hardy and Vekemans, 2002). The error bars indicate s.e.

Fig. 5.

Correlations of among-sibship correlated paternity (i.e. the probability of two mother plants at a distance apart mating with the same father) and distance between mother plants. The correlated paternity was estimated with the pairwise kinship coefficient, and therefore it became negative when individuals were less related than the average; a threshold distance of 110 m was thus set to define unrelated pollen pools as a reference for calibrating the kinship coefficient estimates (Robledo-Arnuncio et al., 2006, 2007). The r_s value represents Spearman's rank correlation coefficient (n = 190).

Fig. 6.

Relationship between Ψ and distance between mother plants. Pairwise observed Ψ (Ψ_O) was calculated based on the kinship coefficients relative to a proposed unrelated pollen pool (a threshold distance of 110 m) (Robledo-Arnuncio et al., 2006). Expected Ψ (Ψ_e) was calculated using the probability density function of pollen dispersal of Normal, Exponential and Exponential-power (Austerlitz et al., 2004; Robledo-Arnuncio et al., 2006). Parameters of different probability density functions were estimated by the KINDIST program of POLDISP 1·0c (Robledo-Arnuncio et al., 2007).