Phenotypic Engineering: Using Hormones to Explore the Mechanistic and Functional Bases of Phenotypic Variation in Nature

Document Type


Publication Date


Publication Title







Perhaps the best way to determine whether and how traits of organisms are currently adaptive is to alter them experimentally and compare the relative fitness of altered and unaltered individuals. We call this method phenotypic engineering. To the extent that natural selection moulds organisms on a trait‐by‐trait basis, we would expect fitness of unmanipulated (control) individuals to be higher than that of experimentally altered individuals. However, other outcomes are possible and of interest. If, for example, a single trait were altered and the fitness of manipulated and unmanipulated organisms were found to be similar, we might conclude that selection is not currently operating on the altered trait. Phenotypic engineering with hormones describes an experimental approach to the study of adaptive variation in suites of traits that are hormonally mediated and correlated in their expression. A likely outcome of such manipulations is that some traits would be altered so as to elevate fitness but that changes in other, correlated traits would lower fitness. If the net effect were to depress fitness, a process by which natural selection shapes and maintains organisms as integrated units would be demonstrated. We have employed this approach in studies of the Dark‐eyed Junco Junco hyemalis, a small passerine whose reproductive success varies with the abundance of nest predators. We treated males with testosterone, documented the phenotypic consequences and related these to various measures of fitness. Summarizing results to date: Behavioural comparison of males treated with testosterone (T‐males) and control males (C‐males) shows that T‐males sing more frequently, are less attentive to offspring, have larger home ranges and are more attractive to females. Physiologically, testosterone accelerates entry into breeding condition in spring (loss of winter lipid stores) and results in higher levels of corticosterone. If exposure to testosterone is prolonged beyond the breeding season, pre‐basic moult is delayed or prevented. We are currently comparing T‐ and C‐males with respect to corticosteroid binding proteins, sperm reserves, response to nestling vocalizations and neuroanatomy. The relationship between testosterone‐induced phenotypic variation and fitness is still under study. When treatment extends well beyond the breeding season, testosterone significantly reduces survivorship; otherwise it does not. With respect to apparent reproductive success (i.e. estimates of paternity that are not based on genetic analysis), more young leave the nests of C‐males than of T‐males, but treatment groups do not differ in the number of young that reach independence. Preliminary data on realized reproductive success (i.e. number of genetic offspring sired) suggest that production as the result of extra‐pair fertilizations is greater in T‐ than in C‐males but that T‐males lose paternity of more of the offspring of their social mates to other males. Continued investigation will, we hope, reveal the factors governing the trade‐offs between male mating effort and parental effort and between survival and current reproduction, as well as the frequency with which the typical phenotype outperforms one that has been experimentally altered.