1
Nature, Nurture, and Nature-by-Nurture - Killing the Dichotomy

David J. Hosken1, John Hunt1,2 and Nina Wedell1

1 Centre for Ecology & Conservation, University of Exeter, Penryn Campus, Penryn, TR10 9EZ, UK

2 School of Science and Health & Hawkesbury Institute for the Environment, Western Sydney University, Hawkesbury, NSW 2793, Australia

The primary purpose of this book is to provide a broad snapshot of recent findings showing how the environment and genes influence behaviour. At face value, this should be uncontroversial but unfortunately, the history of genetics includes eugenic movements and Lysenkoism. As a result, discussions of how nature and nurture affect behaviour have been dogged by polemic disputes because ideological views about their contributions have tended to cloud what is really an empirical question. This is in some ways exemplified by the book Not in Our Genes (Lewontin et al. 1984), which begins with a political confession from the authors - we are committed socialists - and starts with a chapter on right-wing politics and determinism. For us, the evidence, and not political or any other beliefs, is what counts and any 'belief' approach puts the desire for the world to be a certain way ahead of the evidence that it is not so, ultimately committing a version of the naturalistic fallacy - if something is 'natural', it is morally correct, which is clearly rubbish (also see Chapter 10). Infanticide, cannibalism, forced copulation (rape), and killing other members of your species (murder) are rife in nature, but it would be difficult to convince anyone of intelligence that these acts are moral because they are natural. Furthermore, 'politically' motivated arguments against 'reductionism', reducing complex behaviours to single causes, are frequently concocted to protect against a biological determinism that must be fought at all costs. However, as we hope to explain, acknowledging that there are genes underlying behaviour, even genes of large effect, is imperative if that is what the data tell us. After all, it is no use playing music to cows if milk yield is totally determined by genes and unaffected by the environment, and as we outline below, in a polygenic world that includes inevitable environmental effects and all manner of interactions, prediction is tricky and determinism dubious because of the probabilistic and complex nature of the gene-behaviour link. But again, even if single genes were completely responsible for single behaviours, which they cannot be in the strictest sense (see below), let us not fall into a naturalistic fallacy.

Rather than engage in further fruitless arguments about world-views, this book explores exciting new findings about behaviour and where we go from here. Before moving on to these new advances and the interesting questions that arise from them, we wish to make another - a final? - attempt to kill the nature versus nurture polarity that has plagued the study of behaviour. This dichotomy is largely, but not totally, dead in academic circles but still haunts many debates outside academia, from views on teaching and punishment to politics and the media more generally. It potentially has grave consequences and is a serious distraction to the much more fruitful and interesting discussion about the determinants and influences of behaviour.

Most behaviours, like any aspect of the phenotype, are not influenced by either nature or nurture but by both and by the statistical interaction between nature and nurture (see reviews in Boake 1994; Sokolowski 2001; Bucan and Abel 2002; van Oers et al. 2005; Hunt and Hosken 2014; Anholt and Mackay 2015) (see also Chapters 6, and 7). To explain, starting with the genetic effects, behaviours (and other characters, for that matter) are typically polygenic (Anholt and Mackay 2004). That is, they have complicated genetic architecture that involves many segregating genes with pleiotropic effects and are characterized by complicated epistatic interactions (Anholt and Mackay 2004). In other words, there are lots of genes, each can affect many characters, and the effects of any one gene frequently depend on the other genes it is associated with. There are exceptions to some of this (see Chapter 05), with, for example, foraging movement in Drosophila melanogaster having two distinct behavioural phenotypes that are largely determined by a single gene (reviewed in Sokolowski 2001), and aggression being altered by transposon upregulation of a cytochrome P450 gene (Rostant et al. 2017). However, even these large single-gene effects can be complicated by epistasis (gene-gene interactions) (e.g. Smith et al. 2011; Rostant et al. 2015).

Nonetheless, most behaviours are influenced by many genes, often of small effect, and because of this, we may never uncover all the precise genes that influence a behavioural phenotype. As a result, a statistical approach is needed to describe the average effects of genes on a behaviour and, importantly, to show how genes affect the variation around the mean. The distinction between an average effect and the variation around it is crucial, because for the most part there is not a single gene for phenotype A or B; rather, there are many genes that alter the probability of expressing phenotype A or B. Thus, many interesting traits do not vary discretely but are continuous (Falconer 1981; Roff 1997; Lynch and Walsh 1998), and genes influence the likelihood that an individual will express more or less of the trait in question.

The simplest statistical approach to understanding these relationships involves partitioning the variation in the behaviour of interest into the sum of the genetic effects and the variance unexplained is then due to the environment (which includes maternal/paternal effects, indirect genetic effects, ecology and abiotic factors like temperature, food, and water), or alternatively, testing a range of genotypes across environments and then partitioning effects into genes, environment, and their interaction (how genes and environment affect each other to determine phenotypic variation) (see Chapter 04). This reveals exactly how genes, the environment, and their interaction can affect phenotypes, including behavioural phenotypes.

Figure 1.1 A pictorial explanation of genotype-by-environment interactions (GxE). In (a) we show a plant GxE - for simplicity's sake (see explanation below) - and in (b) cricket calling behaviour as a hypothetical behavioural example. (a) Three plant genotypes (clones) grown in two environments that only differ in how much water each plant receives, but everything else about the environments is identical. This means that each plant experiences exactly the same conditions within each environment and differences in water between environments. Therefore, plant size differences within each environment are due to just the genetic differences between plants. However, because each plant genotype is found in each environment, any difference in the average plant phenotype across environment is due to the environmental (water) differences alone. The changes in relative size across environments (i.e. Clone 1 is biggest in Environment 1, but smallest in Environment 2) represents a genotype-by-environment interaction. So plant size variation is due to genetic differences, environmental differences and an interaction between the genetic and environmental differences. The same principles apply to any phenotype, including behaviour. (b) This figure shows the same interaction-type across cricket calls where the sonograms above and below the cricket images show the hypothetical songs females of each hypothetical genotype are most attracted to across two imaginary environments. In Environment 1, call rates are slower than in Environment 2 (there is an environmental effect on preferred calls), and each genotype prefers different calls (a genetic effect), but the type of call preferred depends on the environment sampled (gene-by-environment effect).

To use a simple morphological example to make this point very clearly while noting the principles are exactly the same for behaviour: if we could take three plant-clones (three distinct plant genotypes (Figure 1.1) and grow each of them in two highly controlled environments that only differed from each other by how much water was available and all else was exactly the same, then the differences in plant heights within each environment would be due to just the genes, and the average difference in heights between environments would be due to environmental differences alone. And if the effects of the genes on the phenotype varied across the two environments (i.e. the biggest genotype in environment 1 is the smallest in environment 2), then we have a genotype-by-environment interaction (we additionally include a hypothetical behavioural example as well; see Figure 1.1). To put that into the simplest terms:

where P = the phenotype, G = the genotype, E = the environment, and GxE = the interaction between genotype and environment, and this is as true of behaviour as it is of morphology. And if we are talking about variation around average behaviours, then we have:

where VP = phenotypic variation, VG = genetic variation...