The influence of breeding strategy, reproductive stage, and tissue type on transcript variability in fish

Characterizing factors that contribute to transcript variability is necessary before molecular endpoints are widely adopted as biomarkers for environmental monitoring programs and risk assessment. Here, we employed a meta-analysis approach to understand how reproductive stage, breeding strategy, and tissue type influence transcript variability in multiple fish species. Transcript abundance from the scientific literature was examined by method of quantification (qPCR or microarray), and the extracted data were used to calculate the coefficient of variation (CoV) for each transcript. Based on qPCR data, variability in the abundance of estrogen receptor 1 and hydroxysteroid dehydrogenase 3b was dependent upon reproductive stage and/or breeding strategy in the female ovaries. The variability of other transcripts in the steroid biosynthesis pathway as well as other steroid receptors did not depend upon sex, breeding strategy, or reproductive stage. Variability estimates were used to determine sample size requirements for detecting specific critical effects in molecular endpoints. It was estimated that only 37.8% of published studies used in the qPCR meta-analysis had sufficient experimental power (0.8) to detect a 2-fold expression difference in a transcript. To build upon these analyses, microarray data were used to measure overall variability of the transcriptome, and it was determined that the vitellogenic reproductive stage had the lowest transcriptomic variability compared to other reproductive stages. This variability was lower in a single-spawning species (largemouth bass) compared to a multiple-spawner (fathead minnow). Following this, a meta-analysis of 777 microarrays for multiple fish species was performed to determine the influence of breeding strategy and tissue type on transcriptomic variability. In this analysis, single-spawning fish showed lower gonadal and hepatic transcriptome variability compared to multiple-spawning species. Thus, these species may be more appropriate for sampling molecular endpoints in monitoring programs. Transcript variability was lowest in the brain, followed by the gonads and liver, which may reflect fewer morphological changes relative to these tissues. The results of this study should be used in conjunction with other experimental and sampling recommendations to optimize the use of molecular endpoints in regulatory ecotoxicology and environmental monitoring programs.