Hard-to-find Papers
July-Aug 2009
Main Index
Jan-Feb 2007
Mar-Apr 2007
May-June 2007
Jul-Aug 2007
Sept-Oct 2007
Nov-Dec 2007
Jan-Feb 2008
Mar-Apr 2008
May-Aug 2008
Sept-Oct 2008
Mar-Apr 2009
May-June 2009
July-Aug 2009
Sept-Oct 2009

These aquaculture- and conservation-oriented commentaries are not abstracts written by the original authors.  They reflect the opinions of someone else -- usually Roger Doyle.  Direct quotations from the papers or abstracts are marked with inverted commas.

Note: E-mail addresses of authors have been modified in an obvious way to reduce the possibility of abuse.

 707. Up-to-date review of shrimp viral immunity
         Antiviral immunity in crustaceans. 2009. Liu, H., K. Söderhäll and P. Jiravanichpaisal. Fish & Shellfish Immunology 27:79-88. 
         This is a comprehensive but not severely technical review of the subject. "The aim of this review is to update recent knowledge of innate immunity against viral infections in crustaceans. Several antiviral molecules have been isolated and characterized recently from decapods. Characterization and identification of these molecules might provide a promising strategy for protection and treatment of these viral diseases. In addition dsRNA-induced antiviral immunity is also included." [Developments in the latter aspect of the invertebrate response to infection are especially interesting; See Apr 2008 #650, Oct 2007 #631, Oct 2006 #549 for other papers on dsRNA and siRNA.]. pikul.jir<%>biotec.or.th 

706.  Specialized tilapia selection lines not warranted 
         Genetic analysis of Nile tilapia (Oreochromis niloticus) selection line reared in two input environments. 2009. Khaw, H. L., H. Bovenhuis, R. W. Ponzoni, M. A. Rezk, H. Charo-Karisa and H. Komen. Aquaculture 294:37-42. 
         Lines originating from the same genetic base (compounded of several African populations) were selected in unfed ponds and ponds supplied with a 25% protein formulated diet. Response to selection was excellent in both environments. The genetic correlation between growth in the two environments ranged from 0.74 - 0.84, which is around the point where development of two specialized selection lines might be contemplated.
         The authors argue persuasively, however, that this is not the case in Egypt, where farming systems can be expected to improve and resources for genetics programs are limited. They write, "... for instance, if the selection program were conducted in the low input environment, the genetic correlation indicates that at least 74% of the gain would be captured in the high input environment." They also point out that more divergent environments might require specialized lines, e.g. ponds vs. cages (Feb 2008 #640), fresh vs. brackish water. Also, to my knowledge there still is no published study of GxE interaction between extensive pond and super-intensive tank culture.
         The mention of relative economic importance is a useful contribution to the GxE debate (#698, below). Also useful is the thorough, clear discussion of statistical procedures for distinguishing between environmental and genetic trends, especially when generations do not overlap and/or are synchronized with seasonal variation in the environment. See Feb 2007 #587 for an earlier report from this group on selection in low-input ponds. hans.komen<%>wur.nl

705.  Visualizing the complex evolution of quantitative traits 
         QST meets the G matrix: the dimensionality of adaptive divergence in multiple correlated quantitative traits. 2008. Chenoweth, S. F. and M. W. Blows. Evolution 62:1437-1449. 
         Aquaculturists and conservationists alike are concerned with adaptive changes in quantitative traits like size, growth and fecundity that you can see and measure. To what extent are such traits responding to natural or domestication selection?
         This question is not easy to answer because there are many mechanisms that cause quantitative traits to be correlated genetically, phenotypically, or both. Selection on one trait may or may not cause evolutionary change in another. Apparent evolution of any one trait may or may not be an indication that it is under selection. Multivariate approaches are called for.
         Most so far published consist of multiple, independent comparisons of FST divergence of neutral (marker) traits and its analogue, QST divergence of quantitative traits (Jan 2002 #281, Jul 2006 #515). Correlated the traits need to be analysed together, however, and to do this the G matrix is coming into play (Oct 2002 #356, Apr 2009 #673). The G matrix represents the genetic variances and covariances of a set of traits. This paper describes an approach to measuring quantitative divergence by a multivariate analogue of QST, called FSTq, which separates the G matrix into within- and between-population components.
         Implementation is well described and seems relatively straightforward using the popular, and free, WOMBAT software. The results of this study (on wild Drosophila populations) are interesting and relevant because it is clear that the multivariate approach reveals a lot about selection that analysing quantitative traits one at a time cannot. For this, the authors use factor analysis to combine groups of correlated trait into a smaller number of uncorrelated composite traits.
         Amon the conclusions: "... stronger multivariate than univariate genetic differentiation suggests that multiple rather than single traits may have been the targets of selection." And ,"... suggesting that sex-specific selection varies among geographic areas." And, "... large differences between the within and among-population genetic variance–covariance matrices [so that] there are several axes of adaptive divergence no longer available to selection within populations." s.chenoweth<%>uq.edu.au

704.  More realistic prediction of purging of deleterious recessives 
         A simple method to account for natural selection when predicting inbreeding depression. 2008. García-Dorado, A. Genetics 180:1559-1566. 
         Inbreeding must inevitably accumulate in small, closed populations. But natural or domestication selection will select against deleterious recessive homozygotes , thus reducing the rate of inbreeding below that predicted by simple theory. Selection may even eliminate recessive deleterious alleles entirely (purging controversy: Oct 2007 #629, April 2008 #653). In the absence of a proper model we are puzzled about how quickly purging should happen.
         This paper introduces a parameter called "purged inbreeding coefficient" and develops models that accurately predict inbreeding and purging rates (in simulated populations anyway). The coefficient can be measured in real populations as well so it is not an unobservable theoretical construct. The population needs to have experienced two different situations under which components of fitness or other traits of interest have been measured.
         "Ideally, one [estimation] situation could be a severe bottleneck not allowing for relevant purging, from which [the coefficient] could be estimated. The other situation should provide an estimate of the actual depression for an effective size large enough to allow for purge, but small enough to cause substantial inbreeding within a reasonable timescale." Together, these two situations encompass the typical trajectory of a captive endangered or aquacultural population. This approach may be useful in predicting what is going to happen to a captive population over the long haul. augardo<%>bio.ucm.es 

703.  Does TSV emerge quickly or does it hang around for a while? 
         A quick fuse and the emergence of Taura syndrome virus. 2009. Wertheim, J. O., K. F. J. Tang, S. A. Navarro and D. V. Lightner. Virology 390:324-329. 
         Does TSV begin to cause significant mortality as soon as it enters a region ("quick fuse") or does it linger in the background for years before some environmental factor triggers an epidemic, or slowly increasing prevalence in cultured and wild crustaceans passes a threshold for an epidemic? This question is addressed here by a type of phylogenetic analysis of one of the capsid protein (RNA) genes.
         A phylogenetic tree of TSV strains is generated and then calibrated to our ordinary chronological time scale (that is, years) with isolates taken between 1993 and 2008 and an assumed molecular clock. The inferred timing, geographic origin and phylogenetic roots of TSV infections in Asia and the Americas are presented in some detail. ".... the phylogeny was able to independently corroborate many of the [previously] suspected routes of TSV transmission around the world."
         The paper is especially interesting for its use of a Markov chain procedure for inferring phylogenic connections and timing from sequence data. The BEAST software for doing this, recently developed by Drummond et al, is freely available (http://beast.bio.ed.ac.uk/Main_Page  and http://beast.bio.ed.ac.uk/Tracer).
         Does TSV have a short fuse? Yes. "Using a relaxed molecular clock, we determined that TSV is almost always discovered within a year of entering a new region. This suggests that current monitoring programs are effective at detecting novel TSV outbreaks." wertheim<%>email.arizona.edu 

702.  Using compensatory weight gain to select for higher feed efficiency 
         Genetic variability in residual feed intake in rainbow trout clones and testing of indirect selection criteria. 2008. Grima, L., E. Quillet, T. Boujard, C. Robert-Granié, B. Chatain and M. Mambrini. Genetics Selection Evolution 40:607-624. 
         Aquaculture geneticists would like to develop strains that grow more quickly at a given level of feeding (or, equivalently, use less feed per unit of growth). "Residual feed intake", or RFI, is a more meaningful measure of feed efficiency than FCR (Sept 2006 #530). The problem is, RFI is hard to measure directly.
         The authors of this paper use loss of body weight during starvation and compensatory weight gain after re-feeding as indirect indicators of RFI. An index that incorporates both of these easy-to-measure variables does correlate well with RFI. Ten heterozygous clones were used in the experiment and the correlation is effectively inter-clonal. The ratio of genetic to phenotypic variance was reasonably high for growth, feed intake and RFI. 
         The authors conclude that "indirect criteria [for RFI] are good candidates for future selective breeding programs". This is important, because selection for rapid growth does not necessarily improve feed efficiency. Both traits deserve attention in a breeding program. Open access http://www.ifremer.fr/docelec/doc/2008/publication-4777.pdf ; laure.grima<%>jouy.inra.fr 

701.  Effective population sizes in fluctuating, pedigreed populations  
         Individual increase in inbreeding allows estimating effective sizes from pedigrees. 2008. Gutiérrez, J. P., I. Cervantes, A. Molina, M. Valera and F. Goyache. Genetics Selection Evolution 40:359-378. 
         This is an interesting way to calculate the effective population size Ne in captive populations where genealogical records are maintained and mating is far from random. It is estimated from the rate of accumulation of inbreeding in every individual between two discrete generations, delta-Fi, calculated as the t-root of (1-(1-Fi), where Fi is the inbreeding coefficient and t is the generation of the ith individual. The individual delta-Fi are averaged to get delta-F.
         In simple, ideal populations Ne=1/2*delta-F. In real-world populations with peculiar mating patterns and overlapping generations this simple formula breaks down because generations are hard to define and the population often has some kind of substructure. Yet the concept Ne, being an idealized population with the same inbreeding rate as the one observed, remains useful if one is able to calculate it in a meaningful way. (For ways of doing this in wild and semi-wild populations see Oct 2003 #429, Nov 2006 #553). Ne is useful to know, for example, in judging the adequacy of an aquaculture breeding program.
         The paper employs uses an interesting definition of t in such populations, meaning the ‘equivalent complete generations’ calculated for an individual as the sum over all of its known ancestors of the term of (½)n.  "The approach directly accounts for differences in pedigree knowledge and completeness at the individual level but also, indirectly, for the effects of mating policy, drift, overlap of generations, selection, migration and different contributions from a different number of ancestors ... [it is also] possible to obtain confidence intervals for the estimates of Ne." It also accounts for non-Poisson family size variation. Useful indeed, and intuitively acceptable.
         All the calculations of t, Fi etc are performed with the marvelous program ENDOG which is available free at http://www.ucm.es/info/prodanim/html/JP_Web.htm.  gutgar<%>vet.ucm.es 

700.  Highly heritable IPNv resistance in Atlantic salmon 
         Genetic parameters for resistance to Infectious Pancreatic Necrosis in pedigreed Atlantic salmon (Salmo salar) post-smolts using a Reduced Animal Model. 2009. Guy, D. R., S. C. Bishop, J. A. Woolliams and S. Brotherstone. Aquaculture 290:229-233. 
         This is an exceptionally thorough study of the quantitative genetics of a disease in an aquacultural context. It should serve as a model for the application of modern statistical procedures to this problem.
         The experimental plan reflects a situation in which survival data obtained from one or more field sites, which may differ in their environments as well as (to some degree) their family compositions, are used in the selection of breeders from a biosecure broodstock held at some other location. Several phenotypic traits are considered along with survival in the selection criteria and ranking of candidate breeders. The basic measure is family mean mortality.
         A noteworthy feature of this study is the use of a reduced animal model (RAM) rather than the full animal model which is now the norm in aquaculture genetics (Feb 2007 #586). Only animals which are parents are represented in a RAM pedigree relationship matrix. Breeding values of non-parents (e.g. candidates) are expressed in terms of the breeding value of their parents. "Heritabilities between 0.07 and 0.56 ... were obtained, genetic correlations between sites sharing the same families were uniformly high, 0.70 to 0.85,... and low levels of fullsib family effect due to common environment... were observed." 
         A noteworthy practical conclusion is that "These results therefore lend support to the finding that QTL of large effect are segregating in these populations".  All of this is very promising for development of IPNv-resistant strains of salmonids. The paper concludes with the interesting caution that evolution of the pathogen "will determine the long term sustainability of selective breeding as a strategy for controlling disease". drguy<%>swim-back.com 

699.  Selection on a new type of polymorphism in the immune system 
         Spatially and temporally fluctuating selection at non-MHC immune genes: evidence from TAP polymorphism in populations of brown trout (Salmo trutta, L.). 2008. Jensen, L. F., M. M. Hansen, K.-L. D. Mensberg and V. Loeschcke. Heredity 100:79-81. 
         Major histocompatability complex genes (MHC), a key component of the vertebrate immune system, experience disruptive local selection in fish populations that are bothered by different pathogens. Pathogen diversity also leads to balancing selection and, apparently, sexual selection on the MHC within fish populations. All of this is important in fish conservation genetics and possibly aquaculture as well. See March 2003 #398, Feb 2004 #464, June 2006 #501, Sept 2006 #540, April 2007 #596.
         The part of the MHC gene family studied by population biologists are for the most part peptide-binding proteins belong to the Class I and II MHC. The authors of this paper point out that variation in binding proteins has in fact been shown to account for only a fraction of the total variation in the immune response. The MHC Class I gene family encodes other key parts of the immune system as well, including antigen-processing molecules whose job is to transport antigenic peptides into the endoplasmic reticulum and there bind them to the MHC proteins.
         This paper addresses genetic variation in genes, called TAP, which are involved in transport of antigens across membranes. The variation being detected is not actually in the coding regions but in two embedded microsatellite loci that hitch-hike on selection on the TAP. Eight neutral microsatellites are used as controls. The study was based on brown trout from three geographical regions in Denmark. (See Feb 2004 #464.)
         It looks like variation in the antigen-processing function of the immune system could be every bit as interesting as variation in the MHC binding proteins themselves (for the latter see Apr 2008 #648). Several procedures were used to test for TAP selection, and variation in selection, in time and space. 
         "Analyses of the two TAP markers indicated an effect of selection at both a regional and micro-geographical spatial scale. Moreover, signals of divergent selection among temporal samples within localities suggest that selection also might fluctuate at a temporal scale." Whether temporal fluctuation is a sufficient cause of the diversity within populations remains unanswered, as it does in previous MHC studies. "The genes encoding the TAP molecules could be of importance to the efficiency of the immune response and thus of importance to fitness." lasse.fast<%>gmail.com

698.  Factors influencing the economic value of breeding programs 
         Accounting for genotype by environment interaction in economic appraisal of genetic improvement programs in common carp Cyprinus carpio. 2008. Ponzoni, R. W., N. N.H., H. L. Khaw and N. H. Ninh. Aquaculture 285:47-55. 
         This is a convincing analysis of the economic benefit of a national, nuclear genetic improvement program. It consists of simulations with an underpinning of genetic data from a carp breeding program in Vietnam.
         Whereas simple discounted net present value analyses of genetic programs usually break up into "infinities" (not really infinity, but absurdly high benefit to cost ratios), this does not happen here. GxE interactions arising from differences between selection and production environment, various types of cost including feed, sale price, cost of money, discount rate and adoption rate are included. Some parameters, e.g. heritability of feed intake, had to be guestimated from published data on other species. The sensitivity of the economic outcomes to variation in the biological, social and economic parameters was determined. 
         A nuclear selection program is reckoned to be highly beneficial in all realistic situations. However, a lower genetic correlation between performance in the selection and production environments does reduce the benefit to cost ratio (= cost efficiency). Also, "...greater heritability being associated with greater [cost efficiency] ...price of fish and feed costs had a substantial effect on [cost efficiency] .... the greatest contribution to variations in [cost efficiency] came from increases in adoption rates of the improved fish by the industry." The latter point is most interesting.
         Benefit to cost ratios of breeding programs are analysed from a national perspective in this paper. Perhaps the convincing analysis in this paper will help persuade governments to use more of their resources for aquaculture genetics. The majority of the large, internally-financed programs that I know of are, however, in the private sector. Why should this be when national benefits are so great? I can see three reasons.
         Firstly, slow adoption is not a problem on a private sector farm developing its own broodstock. Secondly, the private sector sees continuous investment in genetics as a permanent "competitive edge". And finally, the intellectual property incorporated in a broodstock is seen as a growing capital asset. These adoption-motivators do not need to be absent from national programs but I have yet to see a serious attempt to incorporate them into the technology dissemination component of such programs. See #706 above. r.ponzoni<%>cgiar.org 

697.  Incorrectly assuming that founders are unrelated does little damage 
         The impact of assumptions about founder relationships on the effectiveness of captive breeding strategies. 2008. Rudnick, J. A. and R. C. Lacy. Conservation Genetics 9:1439-1450. 
         Here we have a simulation study of diversity and inbreeding in captive populations which -- like most breeding programs -- ignore the fact that founders may be related to each other. The populations are managed for 10 generations using minimal kinship techniques (Jul 2006 #512, June 2007 #603), with monitoring of gene diversity and F.
         "Overall ... long-term benefits gained from knowing founder relationships were generally small. Therefore, MK strategies probably often produce near optimal results when standard founder assumptions are made." By "standard founder assumptions" the authors mean the convention that founder animals are unrelated. It is useful, although not unexpected, to see that the value of founder relatedness information is highest in the first few generations. jarudnic<%>brookfieldzoo.org 

696.  Software for using markers to estimate REML BLUPs
         Technical note: Use of marker-based relationships with multiple-trait derivative-free restricted maximal likelihood. 2007. Zhang, Z., R. J. Todhunter, E. S. Buckler and L. D. Vleck. Journal of Animal Science 85:881-886. 
         This describes a useful addition to the MTDFREML program for estimating multiple-trait breeding values and variance components. Marker data can, potentially, be used instead of pedigree data. Those who have used MTDFREML know that it consists of several linked programs that are run sequentially. The job of the first program in the sequence is to calculate the inverse of the relationship matrix, using pedigree data as input. This paper introduced a new first program, MTDFARM, in which you input the relationship matrix itself to produce the same output as the original first program. The coefficients of the matrix can be estimated from marker data (see e.g. Apr 2008 #651). The FORTRAN code for MTDFREML is available from lvanvleck<&>unlnotes.unl.edu and MTDFARM from zz19<&>cornell.edu. I haven't tried MTDFARM but it seems straightforward and the paper is clearly written. lvanvleck<%>unlnotes.unl.edu