21.07.2015 Views

Gladstone Fish Health Investigation 2011 - 2012 - Western Basin ...

Gladstone Fish Health Investigation 2011 - 2012 - Western Basin ...

Gladstone Fish Health Investigation 2011 - 2012 - Western Basin ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Appendix C – Statistical report: Histology results1. METHODSThere were 45 barramundi, 10 bull sharks, 25 grinner and 40 mud crabs submitted for histologicalanalyses . As each species had a different sampling scheme over locations, these were analysed separately.The statistical methods have been previously detailed; in summary: General linear models (GLMs; McCullagh and Nelder 1989) were adopted for the analyses, usingGenStat (<strong>2011</strong>). Each fish or crab was taken as an independent experimental unit. Weighting factors were used to adjust for the stratified sampling of both diseased and unaffectedanimals. As the grinner were all rated as ‘normal’ (not diseased), no weightings were adopted fortheir analyses. Length was generally significant for barramundi, bull sharks and grinner, so was retained in allanalyses to adjust for size (length being approximately the same as age). The primary variable of importance is cumulative score. Analyses are also conducted for eachindividual component of this score. These were all considered as continuous variables, as thesignificance‐difference testing and interpretation are more understandable on this basis. Liver lipid (fish) and hepatopancreas lipid (mud crabs) are categorical variables, so wereanalysed as counts in these classes under a Poisson GLM with a log‐link. The means, and significant differences between these, are primarily reported at the location bytrip level. These values are also then compared at the regional level. The pooled standard error wascalculated according to the usual sum‐of‐variances method (Raj 1968), and these were used toconduct a t‐test for the degree of difference between these regional means. Bull sharks (with fewer observations) were only analysed and reported at the regional level. Secondary analyses were also conducted, including ‘disease status’ in the model (except forgrinner, where all fish were rated as undiseased).2. RESULTS – BARRAMUNDISkin lesions were missing for 14 of the 45 observations, unfortunately including all five from Bundabergon trip 1. Muscle lesions were missing for two observations. This also results in missing values for thecumulative score for these fish. As shown in Figure 2.1, the distribution of the residuals justifies theanalyses using the assumptions of untransformed continuous data and the Normal distribution.129

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!