Life histories predict genetic diversity and population structure within three species of octopus targeted by small-scale fisheries in Northwest Mexico

The fishery for octopus in Northwest Mexico has increased to over 2,000 tons annually, but to date the specific composition of the catch has been ignored. With at least three main species targeted by artisanal fisheries in the region with distinct life histories, the lack of basic biological information about the distribution, metapopulation size and structure of each species could impede effective fisheries management to avoid overexploitation. We tested if different life histories of three species of octopus could help predict observed patterns of genetic diversity, population dynamics, structure and connectivity and how this information could be relevant to the sustainable management of the fishery. We sequenced two mitochondrial genes and genotyped seven nuclear microsatellite loci to identify the distribution of each species in 20 locations from the Gulf of California and the west coast of the Baja California peninsula. We tested five hypotheses derived from population genetic theory based on differences in the fecundity and dispersal potential for each species. We discovered that Octopus bimaculoides with low fecundity and direct development (without a planktonic phase) had lower average effective population size and genetic diversity, but higher levels of kinship, population structure, and richness of private alleles, than the other two species. These features indicated limited dispersal and high local recruitment. In contrast, O. bimaculatus and O. hubbsorum with higher fecundity and planktonic phase as paralarvae had higher effective population size and genetic diversity, and overall lower kinship and population structure than O. bimaculoides. These observations supported higher levels of gene flow over a larger geographical scale. O. bimaculatus with the longest planktonic paralarval duration and therefore larger dispersal potential had differences in the calculated parameters possibly associated with increased connectivity. We propose O. bimaculoides is more susceptible to over exploitation of small, isolated populations and could have longer recovery times than the other two species. This species may benefit from distinct fishery management within each local population. O. bimaculatus and O. hubbsorum may benefit from fishery management that takes into account metapopulation structure over larger geographic scales and the directionality and magnitude of larval dispersal driven by ocean currents and population connectivity among individuals of each locality. The distribution of each species and variations in their reproductive phenology is also important to consider when establishing marine reserves or seasonal fishing closures.

229 6.501 (Peakall & Smouse 2012) to estimate the genetic differences observed within and among 230 populations; in other words to estimate genetic structure. We used FreeNA to measure the effect 231 of null alleles on F ST estimates of population structure, taking into account the frequency of null 232 alleles estimated with the expectation maximization method (EM) (Chapuis & Estoup 2007).  (Table 2). Since local retention is expected to decrease with increasing PPD (Byers & 242 Pringle 2006), we expect that genetic relatedness within populations will be lower in the species 243 with the longest known PPD (O. bimaculatus). We used Queller & Goodnight (1989) relatedness 244 metric to calculate pairwise relatedness to describe the number of alleles shared between pairs of 245 individuals and then calculated the average within each population as implemented in GenAlex 246 6.2 (Peakall & Smouse 2012). Statistical significance was assessed by 9,999 permutations and 247 10,000 bootstraps to estimate 95% confidence intervals around the hypothesis of random mating.  (Fig 2 A). 265 Nucleotide divergence between the three species ranged from 3.3 -7.1% for the 16S rRNA gene 266 and from 6.3 -10.4% for the COI gene (Table 3) (Table S3).

282
Before assigning individuals to species in order to test our five a-priori hypotheses we 283 excluded individuals that did not meet our criteria. We excluded 17 samples that showed missing 284 data at two or more microsatellite loci. In our dataset, 92.78 % of individuals assigned to one 285 species using 16S rRNA and COI sequences (Fig. 2 A) were correctly assigned to the same 286 species using microsatellite genotypes (Fig. 2 B). However, we found 20 individuals that did not 287 comply with the 2/3 rule of ancestry to a single species according to the nuclear genome and 288 were excluded from further analyses. These individuals showed a shared ancestry between O. 289 hubbsorum and O. bimaculatus, mainly in the localities of Puerto Peñasco, Puerto Refugio and 290 Puerto Libertad (Table S4). These locations are within the limit of the geographic range between 291 the two species (Table S5). In Puerto Peñasco, two cases were observed in which the mtDNA 292 identified the individuals as O. bimaculatus, whereas their microsatellite ancestry assigned them 293 to O. hubbsorum (Table S5).

294
The STRUCTURE analyses showed a modal frequency that supported the presence of at   318 The seven loci were polymorphic for the three species (Table 4). Results generally 319 supported our prediction about higher allelic diversity and effective size in highly fecund species 322 bimaculoides (N E = 3.62 ± 0.47, R A = 4.50 ± 0.48) than in O. hubbsorum (N E = 5.02 ± 0.53, R A 323 = 4.54 ± 0.12), while results for O. bimaculatus were mixed and showed the largest diversity of 324 effective alleles (N E = 5.64 ± 0.28), and the lowest allelic richness (R A = 4.14 ± 0.07). 325 We observed that the species with direct ontogenetic development (O. bimaculoides) had 326 the largest average frequency of private alleles (R PA = 1.60 ± 0.48), compared to the species with 327 a planktonic paralarval phase (   (Table 5).
334 Genetic structure within species 335 Pooling sampling locations according to species molecular identification (Fig 1), we 336 found that the microsatellite data AMOVA test supported the prediction that O. bimaculoides 337 with direct ontogenetic development had higher levels of genetic structure (F ST = 0.19, P = 338 0.000), compared to species with pelagic paralarvae (Table 6). Also, we accepted the hypothesis  (Table S6).  We analyzed slowly evolving haploid markers (the mitochondrial genes 16S rRNA and 358 COI) and rapidly-evolving, hypervariable, nuclear markers (seven microsatellite loci) to infer the 359 geographic distribution of three molecularly identified species of octopus among 20 fishing 360 localities from Northwest Mexico and corroborated that differences in the fecundity and potential 361 paralarval planktonic drift (or lack thereof) influence genetic diversity and population structure 362 found within each species.

363
A minimum of 3% genetic divergence in the COI gene is considered a threshold to 364 distinguish species in metazoans (Hebert et al. 2003). We found a higher divergence among the 365 three species (6% -10%), suggesting they are reproductively isolated biological taxa. We    Manuscript to be reviewed

Contemporary effective population size
Average and 95% confidence intervals for the contemporary effective population size (N e ) for three species of octopus. Locations were pooled according to the results of the genetic assignment of species (Fig. 2). N e was estimated with two methods, including linkage disequilibrium (LD; lowest allele frequency used 0.05 and 0.02 respectively) and Molecular coancestry (M C ). Manuscript to be reviewed    Samples used for 16S rDNA and COI Overlap zones