S1). A median-joining network was drawn including all the New Zealand mtDNA control region haplotypes and considering the putative population subdivision (Fig. 2). The network shows a complex structure Atezolizumab with a central core of missing (unsampled or extinct) haplotypes and no clear clustering based on population origin. A total of 84 individuals for the mtDNA data set and 79 for the microsatellite data set was examined to test for possible differentiation among putative Oceanic, Coastal, and Hauraki Gulf populations. The program

STRUCTURE was used to test population structure without a priori sample assignment to a specific population. The most likely number of clusters was found to be two (K = 1, LnP = −4,269.93; K = 2, LnP = −4,261.49; K = 3, LnP = −4,565.97; K = 4, LnP = −4,381.87, K = 5, LnP = −4,449.32, K = 6, LnP = −4,632.84). The application of the Evanno method also identified K = 2 as the most likely number of cluster (K = 2, ΔK= 70.89; K = 3, ΔK = 10.19; K = 4, ΔK = 4.45; K = 5, ΔK = 2.18). However, the clusters do not identify any population as previously assigned based on the sample origins (Oceanic, Coastal or Hauraki Gulf) (Fig. 3). The Principal Component check details Analysis

(PCA) was also performed and the first two principal components

explained MCE 64.42% of the total variation (Fig. 4). Along both components, the Coastal population appears slightly differentiated from the others. FST values, based on the mtDNA data, suggested small but significant genetic differentiation between the putative Hauraki Gulf, and the Coastal and Oceanic populations, although no significant differentiation was detected between the Coastal and Oceanic groups. Φst values did not indicate any significant differentiation (Table 2). The AMOVA analysis indicated 98% of the overall variation due to variance within populations (Vc = 0.49, FST 0.0168, P = 0.00098). In contrast, pairwise comparisons of FST values based on microsatellites showed a small but significant differentiation between Oceanic and Coastal populations, a result supported by Jost’s DEST values (Table 2). However, the result obtained with AMOVA, which gave us an overall estimate of FST, was not significant (FST = 0.00953, P = 0.81916). When testing for sex-biased dispersal, there was no significant difference (P > 0.05) for FIS, FST, HE, and HO between males and females. However, this test is known to have relatively low power, particularly when small sample sizes are used (see Goudet et al. 2002).