Parentage allocation in a complex situation: A large commercial Atlantic cod (Gadus morhua) mass spawning tank
Parentage assignment is becoming increasingly popular as a means of monitoring the effects of aquaculture husbandry practices. As the complexity of such studies increase, likelihood-based methods for assignment are coming to the fore since, theoretically at least, they offer the possibility of accur...
| Published in: | Aquaculture |
|---|---|
| Main Authors: | , , , |
| Other Authors: | , , , , |
| Format: | Article in Journal/Newspaper |
| Language: | English |
| Published: |
Elsevier
2007
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/1893/20628 https://doi.org/10.1016/j.aquaculture.2007.08.018 http://dspace.stir.ac.uk/bitstream/1893/20628/1/Aquaculture%202007.pdf |
| Summary: | Parentage assignment is becoming increasingly popular as a means of monitoring the effects of aquaculture husbandry practices. As the complexity of such studies increase, likelihood-based methods for assignment are coming to the fore since, theoretically at least, they offer the possibility of accurate assignment with less screening overheads (fewer loci, less cost) compared to exclusion-based methods. We were able to explore this assertion through the analysis of the parental contribution to 300 Atlantic cod (Gadus morhua) fiy produced from a commercial mass spawning tank (containing 99 broodstock) on a single day. Five polymorphic DNA microsatellite loci were screened and three different datasets generated: 'error prone' raw genotypic data from automated allele calling; manually corrected data; and corrected data from a reduced number of loci (four). Parentage analysis was performed by three software packages-PAPA, CERVUS and FAP-which all have been used in aquaculture contexts. PAPA and CERVUS perform likelihood-based analyses, while FAP is exclusion-based. Striking differences in the allocation performance were noted both among programs and among datasets. With no allowance for error all three programs flagged up potential problems within datasets, particularly the error prone 'raw' genotypes. When allowance for a presumed low level of error was invoked using PAPA (uniform error rate = 0.0001-0.02) parentage was assigned for virtually all offspring, irrespective of the dataset used. Despite simulations suggesting that these assignments were likely to be accurate (correctness values >95%) substantial numbers of assignments (up to 18%) differed among the datasets. FAP gave more conservative (maximum of c. 78% of assignments resolved) and consistent assignments. The power of assignment declined significantly when only four loci were considered. CERVUS performed relatively poorly in terms of numbers of assignments made. Where comparisons were possible most of the identified parental-pairs agreed with FAP ... |
|---|