Changes in Individual Variability - an Important Component of Environmental Rsponses
Perjantai 31.5.2019 klo 16:35 - Mikko Nikinmaa
Sexual reproduction maximizes the variability of offspring to parents. Variability is the basis for natural selection and consecutively evolution. These statements are generally accepted. In view of this, it is amazing that individual variability is relegated to noise or error in much of experimental biology: we talk about “standard error of the mean” and “confidence interval”. However, the error component, which relates to inaccuracies in measurements represents only a small percentage of variation in biological measurements.
Variability is actually tested in most articles, but not because it is thought to be important in itself. The homogeneity of variance is tested in order to evaluate, if parametric statistical testing of the mean can be done, if data transformation (e.g. logarithmic transformation) must be done before parametric testing is possible or if one must resort to non-parametric statistical testing. When reading toxicological and physiological articles, one started to get the feeling that many if not most of them showed that data were heterogeneous. Therefore, we (Nikinmaa and Anttila, Aquatic Toxicology 207, 29-33;2019) went through a whole host of recent articles and found that more than 80 % of the ones which reported homogeneity/heterogeneity of data found heterogeneity. Thus, the original gut feeling that the different experimental groups had different variances, appeared correct.
The change in variability is important in identifying an occurrence of an environmental response, if it can occur without a change in the mean value measured after the treatment. Again, looking at the published literature, this seemed to be the case: in many if not most studies, a change in variability occurred even when no change in the mean value occurred. To ascertain that this was not only appearance, we tested if variance could change without a change in the mean using the water-soluble fraction (WSF) of crude oil and measured the oxygen consumption rate of Daphnia. The results unequivocally showed that WSF decreased the variability of oxygen consumption rate without affecting its mean value (Nikinmaa et al., Aquatic Toxicology 211, 137-140; 2019). Thus, a change in individual variance can signify an environmental response. Presently, it is not known, to what extent this occurs, since experimental designs have not explored this possibility.
There can be many different reasons for individual variability. One is that the studied animals usually present various genotypes, which may respond differently. This is, undoubtedly, an important component in generating variability. However, differences between individual Daphnia occur even though the animals are genetically identical. Actually, large non-genetic variation can have significant beneficial survival value for an organism, if the environment has cyclical changes with each portion of the cycle lasting for some generations: for example, if large variability with regard to temperature tolerance is achieved via the population consisting of several genotypes, each having small variation in tolerated temperature, an increase in temperature will wipe out all the other genotypes except the ones tolerating high temperatures. If the temperature then decreases, the high temperature-tolerant genotype cannot survive. If, on the other hand, the wide temperature tolerance is non-genetic, acclimation to high temperature will not affect the low-temperature tolerance, and some individual will survive the low temperatures. In natural populations it must also always be remembered that the exposures and environmental conditions by different individuals are not truly identical, whereby, e.g., life-time toxicant exposure can vary.
As a conclusion, it is quite clear that changes in individual variability can constitute an important component of environmental responses. It could be very valuable for environmental biology, if scientists re-explored their datasets concentrating on changes in variability instead of the mean. This would give us information about the role of changes in variability in environmental responses very rapidly.