In this study, we mapped 45 newly developed SNP markers, and eight pre-existing microsatellite markers onto the genome of T. californicus. The map includes markers on all 12 T. californicus chromosomes and has an estimated genomic coverage of 75%. This is the first linkage map to be generated for the Copepoda, one of the most abundant animal taxa on the planet , and one of few currently available for crustaceans. All SNP markers are within coding loci, which will facilitate future investigations of chromosome synteny between T. californicus and other taxa.
We used this map to investigate patterns of segregation distortion across the genome of F2 hybrids between two T. californicus populations. We observed some segregation distortion in newly hatched larvae, at fewer than 10% of markers at p < 0.05, which may be caused by pre-zygotic effects or pre-hatching mortality, and was not maintained until adulthood. Segregation distortion, however, was strikingly higher in adult males, affecting 45% of marker loci at p < 0.05. This is comparable to the amount of segregation distortion that has been observed within interpopulation or inter-species crosses in other taxa (e.g. Nasonia spp, 29% of markers in adult males ; Mimulus
guttatus 48% of markers in mature plants ; Arabidopsis lyrata, 50% of markers ; Daphnia magna, 33% of markers ; Daphnia pulex, 21% of markers ; Lepomis spp. 36.8% of markers in fry ). As seen in other studies, distorted markers were not distributed randomly across the genome but clustered together on particular linkage groups. In T. californicus, this clustering may have been exacerbated by the lack of recombination in females, which increases the likelihood that markers physically linked to regions selected against in hybrids will be inherited together.
Our observation of high segregation distortion in F2 hybrid adults but not in F2 nauplii mirrors the pattern seen for several coding loci in different T. californicus interpopulation crosses (ME1, ME2 ; CYC, RISP, CYC1 [30, 41]). This suggests that, in general, observed deviations from Hardy-Weinberg equilibrium in interpopulation T. californicus crosses are the result of selection against genotypes between hatching and adulthood, rather than being due to meiotic drive or differential gametic fertilization success. Although segregation distortion in inter-species and interpopulation hybrids is a common finding, our study is one of rather few to explicitly demonstrate a role of postzygotic selection in generating this phenomenon. Launey and Hedgecock  showed that segregation distortion in the oyster Crassostrea gigas is caused by post-hatching mortality of individuals homozygous for deleterious recessives. Martin and colleagues  identified hybrid genetic combinations affecting survivorship in Iris. Rogers and Bernatchez  found evidence for selection against particular genotypic combinations acting between fertilization and hatching in backcross hybrids between lake whitefish (Coregonus clupeaformis) ecotypes. Niehuis and colleagues  found evidence that cytonuclear co-adaptation caused genotypic-specific mortality between hatching and adulthood in F2 interspecific hybrids in the wasp Nasonia.
While we have data for only one of the two reciprocal crosses, our T. californicus results show very little evidence for cytonuclear coadaptation. All F2 individuals had an SD mitochondrial background: while we did observe an overall excess of SD alleles in nauplii, this had become an overall deficiency of SD alleles in adulthood. In adults we found only two markers (on Chromosome B) where the homozygote matching the SD homozygote was favored, and eight markers where the pattern of segregation distortion was opposite to that which would be expected if the nuclear and mitochondrial genomes within the two populations were co-adapted. For example, we observed dramatic segregation distortion in adult F2 males throughout Chromosome 10, with the direction of segregation distortion indicating strong selection against the nuclear genotype, SDSD, that matched the mitochondrial background. We also observed a large excess of SCSC homozygotes at Chromosome 10 in the non-recombinant backcross, suggesting that this genotype is more fit than the alternative, SDSC. Harrison and Edmands  correspondingly observed a deficiency of SDSD Chromosome 10 homozygotes in males (but not in females) in backcrosses between SD and another population, RP (Royal Palms). Taken together, these results suggest that part or all of Chromosome 10 derived from the SD population has a deleterious effect on viability in males of this interpopulation cross, that appears to act in an incompletely dominant manner. There are several reasons why such an apparently deleterious portion of the genome may be maintained in the SD population. First, this deleterious effect may be only expressed in a hybrid nuclear genetic background. Our study did not detect epistatic interactions involving Chromosome 10, although this may be due to limited power. Second, as we only examined males, it is possible that aspects of SD Chromosome 10 may be advantageous in females. Alternatively, genes on SC Chromosome 10 may cause masculinization; we note, however, that while F2 offspring in this study did not deviate from a 1:1 sex ratio, the concurrently generated nonrecombinant backcross was significantly female biased despite all individuals containing at least one copy of the SC chromosome (V.L. Pritchard and coworkers, unpublished data). Third, the deleterious aspect of Chromosome 10 may not be expressed in the natural environment. It may be masked in the wild SD population by the presence of a more dominant allele that was lost both from our isofemale lines and the SD parental line used in . Even if this is not the case, previous studies with T. californicus have shown varying experimental conditions to alter the viability of different hybrid genotypes [26, 27, 54]. Additionally, the outcome of replicated experimental hybridizations may vary even under apparently identical conditions , suggesting that even apparently minor environmental changes can have a large influence on the fitness of different genotypes. Finally, T. californicus populations in the wild experience repeated population bottlenecks, which are expected to affect the outcome of selection. Hence even if the deleterious aspect of Chromosome 10 is expressed in the natural environment it may persist in the wild SD population due to drift. Indeed, there is evidence that many T. californicus populations carry such a genetic load . For the SD population in particular, previous studies have indicated a selective disadvantage to SD homozygotes for the coding loci ME2 and CYC, even on the SD mitochondrial background [29, 41]. In contrast, SD homozygotes were favored on a mismatching mitochondrial background, for the coding loci ME1 and RISP [29, 41]. We note, however, that these results vary by sex and study, and none of these four coding loci are on linkage groups exhibiting significant segregation distortion in the current cross (ME1, Chromosome C; ME2, Chromosome A; CYC, Chromosome D; RISP, Chromosome 8, Rose and Edmands, unpublished data). Studies of taxa other than T. californicus have also shown that homozygotes mismatching the cytoplasmic background can be favored in hybrids. Fishman and colleagues , for example, examining segregation distortion in Mimulus F2 hybrids, found a strong excess of M. guttatus homozygotes on a M. nasutus cytoplasmic background. Similarly Martin and colleagues , in a backcross study using Iris brevicaulis and I. fulva, found that, at three QTLs, presence of I. fulva homozygotes decreased long-term survivorship despite a matching cytoplasmic background.
In contrast to the pattern observed for Chromosome 10, we observe heterozygote excess, with no apparent selection against SDSD homozygotes, throughout most of Chromosome 2 and Chromosome 7. We also observe an excess of heterozygotes for Chromosome 7 in the nonrecombinant backcross. This chromosome contains the locus coding for mtRPOL, which has been the focus of recent studies of cytonuclear coadaptation in T. californicus. In comparison to our results, Ellison and Burton , looking at allelic frequencies in F4 hybrid adults, found evidence for selection against the SD mtRPOL genotype in crosses with both SC and another population, AB. They also observed that, unlike for other crosses, recombinant inbred lines with matching SD mtRPOL and SD mtDNA did not demonstrate the same OXPHOS transcriptional profile under conditions of hypo-osmotic stress as SD parentals, suggesting that an epistatically interacting nuclear locus is involved in mitochondrial transcription in SD. They suggested the transcription factor TFAM as a possible candidate. In this context, it is interesting that our results are suggestive of an epistatic interaction between marker TC167 and Chromosome 7; recent work (Rose and Edmands, unpublished) has revealed marker TC167 to be closely linked to TFAM. Nevertheless, the genotypic association patterns between mtRPOL and TC167 are not what would be expected if there is simple co-adaptation between SD mtRPOL and SD TFAM; individuals homozygous for SD mtRPOL exhibit a deficiency of SD homozygotes, and a slight excess of SC homozygotes, at TC167.
Higher divergence between parental lines is expected to result in increased frequencies of distorted loci , particularly due to heterozygote deficits . For example, parental divergence is cited as the reason why crosses between mildly divergent D. pulex populations result in 21% of markers showing segregation distortion, largely due to homozygote deficits , while crosses between highly divergent Daphnia magna populations result in 33% of markers showing transmission ratio distortion, largely due to heterozygote deficits . This pattern is consistent with the prediction that overdominance between alleles of closely related taxa may yield to underdominance between alleles in more distantly-related taxa . Alternatively, divergence may increase the ratio of epistatic interactions involving heterozygous loci. In the current study hybridization between highly differentiated populations (over 20% mitochondrial divergence [15, 16]) led to a high frequency of marker distortion in adults (45%), but no significant heterozygote deficits, indicating relatively slow accumulation of underdominance and/or epistasis involving heterozygotes.
As has been seen in other Tigriopus studies , and in other taxa  it is clear that nuclear loci can interact in a complex way to influence fitness; unfortunately we lack the power to investigate such interactions in more depth in the current study. Additionally, we did not consider epigenetic effects, which have previously been suggested to alter gene transcription in interpopulation hybrids of T. californicus . Overall, our results suggest many intriguing avenues for further investigation into the genetic basis of reduced fitness in interpopulation hybrids of T. californicus. These studies will be greatly facilitated by the recent transcriptome assembly for both the SD and SC T. californicus populations , and by continuing advances in crustacean genomics .