DETECTING AND VERIFYING MICROSATELLITE MARKERS ASSOCIATED WITH DISEASE RESISTANCE IN CULTURED POPULATIONS OF EASTERN OYSTERS, CRASSOSTREA VIRGINICA

The eastern oyster, Crassostrea virginica, is an economically important aquaculture species in the USA, but several diseases, such as Dermo and MSX, have an impact on production. Efforts have been made to develop disease-resistant oyster lines using selective breeding techniques based on phenotype; however, achievement of the desired trait is hindered by the inability to maintain consistent and intense selection pressure in field trials. Marker-Assisted Selection (MAS) offers an effective alternative to traditional breeding techniques by selecting based on genetic markers associated with disease resistance, which can be done even in the absence of diseaserelated selection pressure. The greatest challenge in applying MAS is to identify markers that are consistently associated with the disease-resistant phenotype. In this study, 20 previously published microsatellite markers, including 8 located in regions previously associated with disease resistance were used to genotype and compare oyster populations of the same stock (NEH-RI) deployed in 2012 at two sites (York River, Virginia and Cape Shore, New Jersey) before and after a disease-caused mortality period (March – November 2013). Two markers located in disease-resistant QTLs (Cv02i23, Cvi1g3, and RUCV 97) exhibited significant post-mortality allelic distribution shifts in one site but not the other. Significant differences in allelic distribution before and after selection were detected in 3 markers with no prior evidence of association with disease resistance at either both sites (RUCV270 and RUCV 68) or one site (RUCV 27). These results strengthen the evidence associating markers Cv02i23, Cvi1g3, and RUCV97 with the disease-resistant phenotype and suggest that markers RUCV270, RUCV68 and RUCV 27 warrant further investigation. Additional genetic and functional genomic analyses are required to determine whether these markers are suited for MAS.

in an oyster population Guo 2006, Burreson et al. 2000). While a high parasite load of Dermo or MSX can lead to mass mortality, a lighter burden of the parasites has negative sub-lethal effects on the oysters, including lowered condition index and reduced reproductive output (Ford and Smolowitzv 2007, Dittman et al. 2001, Ford and Figueras 1988. Dermo and MSX are caused by the protozoan parasites Perkinsus marinus and Haplosporidium nelsoni respectively (Ewart and Ford 1993). They affect oyster populations on the Atlantic coast of the United States from Maine to Florida, but Delaware Bay and Chesapeake Bay areas are where they hit the hardest (Burreson et al. 2000, Ford andHaskin 1982). With the increase in sea surface temperature due to climate change, MSX and Dermo have been observed to extend their range to the northeastern US coast and to cause mortalities in this region (Cook et al. 1997, Ford andSmolowitz 2007).
Oyster farmers utilize several strategies to avoid Dermo/MSX-related mortality events. Neither Dermo nor MSX proliferate in low temperature (below 18 ℃ ) and low salinity (below 10 psu) environments (Ewart andFord 1993, Cook et al. 1998). It has been found that altering the timing of seeding could lower mortality caused by MSX.
Oysters deployed late in the season when temperature is falling generally has a lower level of disease intensity and mortality would not start until the following spring (Ewart and Ford 1993). Another approach is to take the advantage of MSX and Dermo pathogens' poor tolerance of low salinity and deploy oysters at sites of low salinity (Paynter and Burreson 1991). However, this approach sacrifices growth since oysters do not grow as well in low temperature or salinity sites (Ewart andFord 1993, Paynter andBurreson 1991). Although MSX and Dermo prevalence and intensity of infection can be reduced with cold winters (both diseases) or low salinity (e.g. caused by freshwater influx after storms, mainly for MSX), disease prevalence and intensity increase as soon as environmental conditions become favorable to both parasite and oyster growth during summer and fall seasons (Ewart and Ford 1993). Farming practices alone are insufficient to prevent economic losses to disease.
Artificial selection for disease resistance is another approach being employed to mediate the negative impacts of disease outbreaks on oyster production. Several lines of evidence indicate that disease resistance is a heritable trait, at least for MSX. For example, wild oyster populations growing in locations with a history of high MSX prevalence in the Chesapeake and Delaware Bays developed quantifiable levels of resistance to the disease in response to natural disease outbreaks (Haskin and Ford 1979). Selective breeding programs were established to build upon the naturally acquired MSX resistance in oysters, and within a small number of successive generations, strains with high resistance were established (Ford and Haskin 1987 (Ewart and Ford 1993). Currently, several oyster lines are available that have been selected for fast growth and disease resistance to a variety of diseases in various environmental conditions (Table 1). Despite the aforementioned genetic improvements gained via selection, traditional methods of selective breeding have some limitations. First, these methods select for phenotypes (e.g. survival to disease outbreaks) instead of genotypes. However, survival is affected by many factors besides disease, such as temperature, salinity, and predation. Thus, the phenotype selected for might not indicate the presence of disease resistance. Secondly, the occurrence of disease is sporadic over space and time. Since it is very important for selective breeding to keep a constant selection pressure on the oyster populations, years with no disease pressure will have a negative impact on the selection process (Guo et al. 2008). Although traditional selective breeding methods through either mass or family selection are a powerful method for producing lines for the industry with certain desired phenotypes, a faster and more accurate way of selecting for disease resistance in oysters is needed.
Marker-assisted selection, whereby selection is imposed on genotype rather than phenotype, could enhance and accelerate the development of disease-resistant oyster lines (Lande and Thompson 1990). It identifies oysters with disease resistance based on genotypes by detecting markers associated with disease resistance. This approach allows the accurate identification of disease resistance in a species in the absence of disease pressure (Collard et al. 2005). In order to perform marker-assisted selection, breeders need genetic markers that have been confirmed to be associated with disease resistance in that species.

Traits
Association studies are an effective approach to identify genetic markers associated with a certain trait, such as growth, shape, or disease resistance. It is particularly helpful for complex traits that are controlled by multiple genes, such as disease resistance in oysters (Ford and Haskin 1987). Association studies utilize genetic markers and compare allele frequencies of these markers between two groups of organisms, one group with the trait of interest (case group) and one control group without the trait of interest (Cardon and Bell 2001). If a significant difference in the genotype is consistently observed at a certain marker between the two groups, it indicates that the marker is closely located to a nearby genetic variant that is causing the difference or, less likely, is the direct genetic cause of the observed difference in phenotype (Hirschhorn et al. 2001). Such studies have been commonly used to identify causes of diseases in humans (Hirschhorn et al. 2005, Oka et al. 1999) and economically important traits in agriculture, such as increased yield and disease resistance (Bai et al. 1999, Zhou et al. 2002, Brondani et al. 2002, Bermingham et al. 2013, Zila et al. 2013). These studies usually are supported with genetic maps for the targeted species, so that the relative positions of markers on the chromosomes are known. It is worth noting that for the purpose of MAS, an available genetic map is not mandatory. Genetic markers with unknown locations on the chromosome can still be tested for association with the trait of interest, and once validated, can be used in MAS.
But the availability of the genetic map of the targeted organism will be helpful in selecting genetic markers to make sure that they span the whole genome, ensuring high coverage. Furthermore, once a genetic marker detects a signal of difference between the control and case groups, a genetic map facilitates locating the genetic markers that have evidence of association on the chromosome and pin-pointing the functional genes through fine-scale mapping (Cardon andBell 2001, Zhou et al. 2002).
It has also been noted that the results of association studies often cannot be replicated by further studies, and that an association from a single report should not be trusted fully (Hirschhorn et al. 2005). Thus, verification of previous association study results in multiple populations is necessary.

Marker Development in Eastern Oysters
Association studies required a large number of polymorphic markers (Liu and Cordes 2004). Recently, an increasing number of various types of genetic markers have been developed for eastern oysters, such as Amplified Fragment Length Polymorphism (AFLP) markers, microsatellites (MS), and Single Nucleotide Polymorphisms (SNPs). Over 200 AFLP markers were used in construction of the first genetic map for eastern oysters (Yu and Guo 2006). AFLP markers are dominant and cannot be transferred among populations, which makes them less favorable than other co-dominant markers like microsatellites and SNPs, but they are still valuable for saturating linkage maps and increasing coverage (Guo et al. 2008). Over 300 microsatellite markers have been generated in eastern oysters (Brown et al. 2000, Wang and Guo 2007, Wang et al. 2009, Reece et al. 2004). Many of the newly developed microsatellite markers were generated from ESTs (Expressed Sequence Tags), meaning that they are part of the coding sequence. More than 100 SNPs markers have been generated from both ESTs and resequencing of functional genes with known or predicted functions. If association is detected using markers developed from coding sequences, there is a higher chance of detecting functional genes underlying the disease resistance trait, compared to an association detected at neutral loci (Liu and Cordes 2004).
Among these markers, microsatellite markers are particularly informative because they are highly polymorphic and are inherited in a co-dominant manner, meaning that heterozygotes can be identified (Wright and Bentzen, 1995). They have been used in association studies to identify markers associated with various traits in humans, plants and animals (Oka et al. 1999, McKnight et al. 2006, Zhou et al. 2002, Vigouroux et al. 2002.

Previous Association Studies on Disease Resistance in Eastern Oysters
Successful association studies have been done to map disease resistance QTLs and to identify markers associated with disease resistance in eastern oysters (Yu and Guo 2006;Guo et al. 2008). Several challenges to association studies for disease resistance in oysters exist. First, it is hard to quantify the phenotype of disease resistance. The easiest evaluation of disease resistance in the field is survival.
However, survival can be affected by factors other than disease, including environmental stress and predation. Disease prevalence and intensity in the presence of disease pressure is a more specific indicator of disease resistance, but, it is expensive and, in some cases, oysters may be able to tolerate a high parasite load without obvious negative effects on survival or performance (Guo et al. 2008).
Secondly, there is a lack of well-characterized highly-inbred susceptible and resistant oyster lines. As a result, the usual control-case association study design cannot be directly implemented, since there are no obvious candidates for the control groups and case groups. To overcome this challenge, another approach has been used in oysters, in which control and case oysters are taken from a single population but at different times regarding the disease outbreaks. The control group consists of oysters randomly sampled from the population before a mortality event caused by diseases, usually at the beginning of the summer. The case group consists of oysters randomly sampled from the population right after a mortality event, as these survivors are likely to be resistant to diseases. The two groups of oysters are then sacrificed and genotypes are then compared. Markers that show significant allele distribution shifts between the two groups are thought to be linked to disease response. If affected markers are clustered close on the genetic map and have shifts in the same direction, the region containing these markers is identified as a QTL (quantitative trait loci) for disease resistance. In order to facilitate mapping these QTLs, oyster families (instead of oyster populations) were used in these association studies (reviewed in Guo et al. 2008).
The results from these family-based association studies in oysters are summarized in Figure 1. The first map (Figure 1a) summarizes the results from several association studies for resistance to Dermo disease, as evaluated through allele frequency shifts after a Dermo disease-related mortality event in the field (reviewed in Guo et al. 2008).
This map includes 313 markers (249 AFLPs, 47 MS, and 17 SNPs), resulting in a total genetic length of 729.9 cM. The average interval between markers is 2.3 cM. The map describes the location of 47 functional genes or expressed sequences. Based on results from previous association studies, 26 AFLP markers, 1 SNP, and 6 microsatellite markers had significant after-mortality frequency shifts. After mapping these markers on a linkage map, regions containing 2 or more markers with frequency shifts in the same direction were used to identify 8 QTL regions potentially associated with disease resistance (Figure 1a; Guo et al. 2008). The 6 microsatellite markers showing significant post-mortality shifts and clustered in QTL regions were RUCV 97, RUCV 58, Cv02i23, Cvi2j24, Cvi2i4, and Cvi1g4.
In an unpublished study performed in 2010, Zhang, Guo, and Gomez-Chiarri used the same association study strategy to identify genetic markers associated with RODresistance. Two oyster families with differences in susceptibility to ROD were used. In order to minimize the impact of field environmental conditions on survival, the oysters were challenged with cultured Roseovarius crassostreae, the causative agent of ROD, in laboratory conditions. The study tested 257 genetic markers (90 MS, 2 SNP, and 155 AFLP markers). Twenty-eight markers (11 MS, 1 SNP and 16 AFLP markers) were identified to experience significant post-mortality frequency shifts and 22 of them were mapped (Zhang,Guo,Figure 1b). Two microsatellite markers that had significant post-mortality shifts and are publicly available are RUCV 66 and Cvi2m10. We also decided to include it in this study RUCV 270, a marker in linkage group 10 located in relative proximity to another marker (Cvi12) for which a significant shift was detected after ROD mortality. Guo et al. (2008). In red are microsatellite or SNP markers, in black are AFLP markers. Markers in bold showed significant shifts in frequency after disease-related mortalities. Lines encompassing regions within selected linkage groups indicate disease resistance QTLs. In black boxes are the candidate markers selected for this study. (b) Genetic map with genetic markers showing shifts in frequency after an ROD experimental challenge (Zhang, Guo, and Gomez-Chiarri, unpublished). Markers in red were genetic markers showing association with resistance to ROD. Markers in black boxes were candidate markers investigated in this study.

Goal of this Project
Previous family-based association studies have identified several markers (including 8 publically available microsatellites) associated with disease resistance in oysters (reviewed in Guo et al. 2008). However, due to families' low genetic diversity, a family-based approach might not truthfully reflect wild population situations. Thus, results obtained from family-based studies should be further confirmed by subsequent studies using multiple oyster lines (mixed families) or cultures stocks and ultimately in wild oyster populations.
This project aimed to verify 8 several previously identified microsatellite markers that had evidence of association to disease resistance in oysters, and to possibly identify new associated markers by testing 12 additional microsatellite markers without previous evidence of association using samples from an oyster stock deployed in two separate field locations that experienced mortality due to Dermo and MSX collected by a previous field performance study (Proestou et al. in preparation).
Oysters before and after the disease-caused mortality event were genotyped at the 20 genetic marker loci. Significant post-selection allele distribution shifts, supported fully or in part by other lines of evidence (consistent shifts in allele frequency in the same direction at the 2 different sites and/or mapping of these markers to disease resistant QTLs in other studies), were considered as evidence for potential association with disease resistance.

Oyster Samples
Oysters samples were available from a previous study that evaluated the performance of 6 oyster lines or stocks in 5 locations in the East Coast of the US in August 2012 at two locations (Cape Shore, NJ and York River, VA) were used for this study, because of the high disease-related mortality (more than 30%) that these oysters experienced 1 year after deployment at both sites during the period spanning the summer and early fall (July -September 2013, Supplementary Figures 1 and 2

DNA Extraction and Evaluation
Oysters were shucked, and tissue samples from mantles and gills were stored in 70% ethanol at -20 degrees Celsius until extraction. Genomic DNA was extracted from mantle and gill tissues using either a Chelex method (Aranishi and Okimoto 2006) or Autogen's Quickgene Mini80 system (Autogen Inc., Holliston, MA; Pereira et al. 2011). The Chelex method has the advantage of being fast and inexpensive, but the DNA extracted was fragmented to ~500bp fragments, which reduces the success of PCR amplification for genetic markers of longer fragment. Therefore, the Autogen method was used later in the study, for it yielded higher quality DNA. The change of extraction method should not affect the integrity of the data since the same genotypes were obtained from successful PCR amplification of DNA from tissues of selected individual oysters that was extracted using both methods (data not shown). DNA quantity and quality were evaluated using a NanoDrop 8000 instrument (Thermo Fisher Scientific Inc., Waltham, MA). DNA was diluted to 7.5-10 ng/ul for PCR amplification. DNA was stored at -20˚C until use.

Microsatellite Marker Selection
Twenty previously published microsatellite markers were used to genotype the control and selected populations (Reece et al. 2004, Wang and Guo 2007, Wang et al. 2009). These markers were chosen based on allele size range and ability to multiplex, level of polymorphism, and reported frequency of null alleles and are therefore appropriate for detecting genetic differentiation among oyster populations. Moreover, 9 of the 20 loci included in this study were located within previously defined disease resistance QTL or near (RUCV 270) a marker showing a significant shift in allele frequency after ROD mortality (Table 2, Guo et al. 2008, Guo, Zhang, and Gomez-Chiarri, unpublished). The 5'end of the forward primer from each locus was labeled with one of four fluorescent dyes (PET, NED, 6-FAM and VIC) and markers were assigned to one of five plexes, based on allele size range and dye color (Table 2). Table 2. Microsatellite markers used in this project. Markers shaded grey represent those located within disease-resistance QTLs. *Marker located near a marker showing a significant shift in allele frequency after a mortality event (Guo et al. 2008, Guo, Zhang, and Gomez-Chiarri, unpublished for 3 minutes, 10 cycles of 94℃ for 30s, 5℃ above annealing temperature (Ta) of the marker for 30s (decrease 0.5℃ every cycle), 72℃ for 30s, followed by 25 cycles of 94℃ for 30s, annealing temperature (Ta) for 30s and 72℃ for 30s, and a final extension at 72℃ for 30 min (Jaris 2014). Cvi2i4, RUCV 58, and Cvi2m10 were amplified as previously described (Reece et al. 2004, Wang andGuo 2007). RUCV 58 and Cvi2m10 required 2.5mM and 2mM MgCl 2 respectively for successful amplification.
Gel electrophoresis was used to confirm successful amplification using 2% agarose gel.
For each individual oyster sample, PCR products from the four loci belonging to a particular plex were pooled in equimolar amounts and purified with the Agencourt Allele binning was conducted using Allelogram software (Morin et al. 2009), where the length of the repeat motif for each marker provided a guideline for bin size.

Data Analysis
Population multi-locus pairwise F ST (an index between 0 -1 that estimates the degree of genetic differentiation between oyster populations at each of sampling time points, with 0 indicating no differentiation; Wright 1978) was calculated to determine population structure within the NEH-RI stock in each site between each of the sampling time points. Arlequin software ver 3.5 was used to conduct data analysis (Excoffier and Lischer 2010). Data from the allele binning needed to be formatted for Arlequin software to recognize and process. The format conversions, along with the generation of allele frequency data was done by the software CONVERT (Glaubitz).  (Oka et al. 1999). Markers that showed significant shifts in both sites and markers that had significant shift in one site but were previously known to be in disease resistance QTLs were considered to be markers of interest.
In order to find out whether affected markers were related to known functional genes related to host-defense, translated protein (tblastx) and nucleotide (blastn) blast (Stephen et al. 1997) were used to conduct homology searches in the NCBI server.

Amplification and Genotyping
All 20 microsatellite markers were successfully amplified in most samples, with some samples that failed to amplify for one or more marker loci due to low DNA quality or errors in carrying out PCR. In order to reduce the amount of missing data, samples with failed amplifications for five or more ( 25% of total markers) markers were discarded (Table 3).

Determination of Pre-and Post-selection Time Points for the Association Study
For this association study, the control population will be oysters before selection by disease and the case (selected) population will be oysters after selection by disease. At

Mortality Event
Fisher's exact test was conducted comparing the allelic distributions between preand post-selection time points. In the samples collected in Virginia, 3 out of 20 markers had significantly different allelic distributions between the pre and postdisease-related mortality event, while 9 out of 20 were significantly different in oysters collected in New Jersey (  --

Homology Search
As all RUCV markers were developed from coding sequences (ESTs), there is a chance that they are part of the functional genes. A homology search with tblastx in the non-redundant database (Stephen et al. 1997) was conducted for markers that showed significant shift in allelic distribution in either one site or two sites. Nucleotide blast was also conducted for markers that did not detect a hit initially with tblastx.
RUCV 68 was found to be homologous to the 40S ribosomal protein S28 (Table 7).
RUCV 27 and RUCV 28 showed identity with a predicted cell-number regulator 3like protein that contained a PLAC8 (Placenta-specific 8) domain in the closely related species C. gigas. Cv02i23 found was found to be homologous to another hypothetical protein in C. gigas, but tblastx did not detect any putative conserved domains (Table   7). The aim of this study was to gather new or strengthen existing evidence supporting the association of previously published markers with disease-resistant phenotypes as performed in family-based studies by determining if these markers also experience a significant shift in allele frequency in an oyster stock experiencing Dermo and MSX-related mortality in 2 different sites. Among our panel of 20 microsatellites, it was expected that the 8 markers located in regions previously associated with disease resistance (Guo et al. 2008, Zhang, Guo and Gomez-Chiarri, unpublished) would have a higher probability of discriminating among populations pre-and post-a disease-related mortality event. However, none of these markers showed significant differentiation in allelic distribution before and after mortality at both the Virginia and New Jersey deployment sites. Of these 8 markers, only 2 could differentiate between populations pre-and post-the disease-related mortality event at one of the two sites (Cv02i23 and RUCV 97-New Jersey). Interestingly, a signal of potential selection was detected at two markers (RUCV 270 and RUCV 68)  proteins. This is not unexpected since many microsatellite markers do not code for proteins, but they may experience shifts in allelic distribution in response to a diseaserelated mortality event because the functional genes under selection are nearby on the chromosome. RUCV 270 mapped to linkage group 10 (Guo et al. 2008) and there are 8 other genetic markers (2 MS and 7 AFLP markers) within 5cM of RUCV 270 (Zhang, Guo and Gomez-Chiarri, unpublished). However, none of these markers showed significant shifts in allele frequency after ROD mortality except Cvi12, the one closest to RUCV270 according to the genetic map ( Figure 1). The next logical step is to fine-map this region and test more genetic markers in the close vicinity of RUCV 270 and Cvi12 for association with disease resistance, in order to determine if this is a disease-resistant QTL and ultimately identify the functional gene under selection.
Results of the Fisher's exact test suggested a high probability that allelic distribution at marker RUCV 68 was different before and after selection in both sites.
However, closer examination of individual allele frequency shifts at this marker show no significant shifts in the same direction at the two deployment sites. The mostly differing response to selection of this alleles at the two deployment sites does not support an association of this marker with disease resistance.
Several of the markers used in this study were located in a region of linkage group 2 associated with disease-resistance that also contains the serine protease gene, a functional gene that has been associated with disease resistance (Guo et al. 2008).
Serine protease inhibitors have demonstrated ability to inhibit the proliferation of the Dermo pathogen, Perkinsus marinus (La Peyre et al. 2010). A polymorphism in the serine protease inhibitor promoter region has been shown to be associated with resistance to Dermo, and was verified comparing susceptible and resistant oyster lines (Yu et al. 2011, He et al. 2012. Of the 5 markers in linkage group 2 tested in this study, only Cv02i23 and RUCV 97, showed significant shifts in allele frequency after the mortality event in our study, and in only one of the sites (New Jersey). A lack of detection of a significant association with disease resistance may be due to the limitations of our study, including a small sample size, especially for the Virginia site (28 pre-selection samples and 36 post-selection samples), which would limit the power to detect noticeable post-mortality shifts. The Virginia site also experienced high over-wintering mortality (over 50%), and the fact that pre-winter samples have significantly differentiated genetic composition suggests that over-wintering already had a selection effect on the oysters. If the alleles associated with disease resistance were reduced to a low level by over-wintering selection, it would reduce our ability to detect them by analyzing post-mortality allele frequency shifts. More significant postmortality shifts in oyster samples from New Jersey site were detected, which could be explained by the larger sample size and low over-wintering mortality. Moreover, using highly polymorphic microsatellite markers proved to be a bit problematic, as sample sizes allocated for each allele can be extremely small. Their favorably high variability needs to be backed up with a large sample size. In future studies, SNPs might be better candidates if a large sample size is not available. It would be interesting to genotype the polymorphism in the serine protease inhibitor gene of our samples to determine whether it was associated with better survival and whether its change in allele frequency is in accordance with those detected for Cv02i23 or RUCV97.
Five other markers without previous evidence of association showed significant allelic distribution shift in one site in our study. We cannot confidently say that these markers are associated with disease resistance, and more association study needs to be done for these markers. However, RUCV 3 in this group had the largest allele frequency shift in the same direction of all the affected markers, although no statistically significant allelic distribution shift was detected in the Virginia site.
RUCV 27 and RUCV 28 matched to the same gene, and both showed allele frequency shift of more than 10% in the same direction in both sites, which indicated that they could be under selection. These 2 markers showed homology to a predicted cellnumber-regulator-like protein in C. gigas that contains a PLAC8 domain, a cysteinerich domain found in a variety of proteins in animals, plants, fungi and algae (Song et al. 2011). The study of PLAC8 containing proteins only started recently and variety of functions have been proposed or observed, including controlling fruit size, cell number, and heavy metal transport in plants (Song et al. 2011). Association studies with more populations need to be done to confirm whether RUCV 27 and RUCV 28 are associated with disease resistance, and investigate the potential functional role in disease resistance of PLAC8-containing proteins.
Marker-assisted selection in oysters is still in its very early stage. The main challenge is the lack of validated genetic markers associated with disease resistance.
This study provides extra support for several candidate markers of disease-resistant markers identified by previous studies. The scientific community should prioritize these markers in further investigations, with a particular focus on fine mapping the QTL in linkage group 2 and determining if a disease resistant QTL could be present in linkage group 10.

APPENDIX Field Evaluation of Disease Resistance in Oyster Lines.
Our study took advantage of samples collected from a previous unpublished collaborative study funded by the Agricultural Research Service (ARS) was done to evaluate the performance of six lines (Table 1)  This study provides a good opportunity to conduct association studies for disease resistance since we have samples collected before and after mortality events to serve as control and case groups. Samples collected before mortality can be considered as the control population, which largely consists of susceptible oysters, with some resistant oysters in the mix. Samples collected after mortality can be considered as the case population, as they had undergone selection from disease. When choosing which oyster line and site to use for this project, two criteria were used: 1) Oyster population with high mortality is preferred, since the selection on the genetic marker loci by disease would be easier to identify. However, populations with 100% mortality should be avoided as there would be no case populations. 2) High MSX and/or Dermo prevalence and intensity during mortality period is preferred, for disease resistance is the trait we are interested in. Oyster lines in Maine site suffered high mortality early in the study from a ROD outbreak and no oysters were left, so Maine site is excluded.
NEH-RI line had high mortality in both Virginia and New Jersey site and still had survivors at the end of the study (Figure 2). qPCR (Quantitative Polymerase Chain Reaction) data confirmed high prevalence and intensity of Dermo and MSX in the summer mortality period (Supplementary Table 1). Other lines/sites did not have as high mortality rates, so NEH-RI line at Virginia and New Jersey were chosen for this study.