Relaxed selection and trait loss in evolution


I was surprised to find that when African weaverbirds were introduced into habitats where they no longer needed a highly refined ability to recognize their own eggs, their abilities nevertheless remained highly refined for over 150 generations, whereas their egg colors evolved rapidly. This seemed to go directly contrary to the common observation, stated most famously by Ernst Mayr in 1960, that behavior evolves quickly and first, and morphological traits (body parts) evolve more slowly and have to play catch-up. I found that a couple of other studies had shown the reverse trend as well, and noticed that the three studies shared one thing in common: the traits in question were decaying. This made sense. Mayr, and just about everybody else who thinks about trait evolution, concentrates mostly on traits that natural selection favors, traits that are beneficial to their bearers. Fewer people think about traits that are useless, or becoming less useful over time, so a rule that assumes that we're talking about beneficial traits was able to slide by for fifty years. In reality, if a behavior is useful and can change quickly (like much behavior can), we can expect Mayr's trend-- the behavior will tend to adapt first, followed by morphological traits. But if the behavior in question is losing its utility, like the egg recognition abilities of the weaverbirds I was studying, an animal might just stop performing the behavior, but the ability to do it might remain latent in its brain. The morphological traits, that are more likely to stick out like sore thumbs and be a burden to an individual when useless, would be selected against to a greater extent, and so would tend to decay more quickly.

Other people I knew were studying similar things at the time. Susan Foster noted what she called "relaxed selection" in her stickleback fish, and Dick Coss reviewed this situation in a book edited by Susan, after working for several years on the loss vs. retention of antipredator defenses in ground squirrels. Dan Blumstein was also looking at antipredator behavior in large herbivores in areas safe from predators. At an Animal Behavior Society meeting these three biologists promptly got me excited about developing the conceptual advantages of this research strategy. I was leery about calling it "relaxed selection" because what we were really talking about is the relaxation or reversal of a single agent of selection, but we couldn't come up with a good alternative. Eventually, with their prompting, I assembled a group of twelve biologists into a working group on “Relaxed Selection and Trait Loss in Evolution” at the NSF-funded National Evolutionary Synthesis Center based at Duke University.  In August 2007 we began a series of meetings, to continue through 2009, to discuss and develop theory that may help explain what happens to traits upon relaxation of the source of selection that gave them their raison d'etre. Our group has researchers of all sorts of organisms, and expertise that spans genetic, developmental, and phenotypic/ecological levels of inquiry.  I look forward to using the theory and predictions we develop as the basis for future empirical work in my lab, although I haven't yet decided on an empirical study system. 

Our group has already published one major review/perspective article, and we are working on additional papers. Please let me know if you would like more information about our group, or if you have references or a study system you would like to share with us.


Consequences of relaxing a source of selection on a trait

See the flowchart below (from Lahti et al., invited and in review in TREE.) When an environmental change weakens an important source of selection, a trait may follow eight likely pathways with regard to its function, its level of expression, and its evolution.  Although these can follow any environmental change, the weakening of an important source of selection is distinctive in that it increases the likelihood that the trait loses its function, or adaptive value.  A central question in the study of relaxed selection is which path a trait will take and why. Here is a likely example of each pathway from nature (please see the manuscript for references).

  1. Persistence of function (due to alternative sources of selection):
    In the absence of mammalian predators, kangaroos lost an element of their predator avoidance behavior, but wallabies, subject also to eagle predation, retained it.

  2. Change of function (either immediately or following subsequent evolutionary modification):
    The pectoral muscles of flightless steamer ducks (Tachyeres) no longer function in flight, but are enhanced due to their new function in oaring on the water surface.

In the following, the trait loses its function and fitness advantage in the new environment:

  1. Full expression of a relict (persistent but nonfunctional) trait
    Bats are absent from the Arctic but some moths there (Gynaephora spp.) retain bat-defense reactions.

  2. Reduced expression of a relict trait without genetic change
    Vigilance is lessened in moose in the absence of predators but rapidly intensifies in mothers who lose calves to recolonizing wolves.

  3. Latency (plastic loss of expression) of a relict trait without genetic change:
    Captive meerkcats normally do not give alarm calls, but upon presentation of predator feces calls return with intact contextual and structural integrity.

  4. Vestigialization (reduced expression due to genetic change):
    Oil rewards and their secretory glands decay in Ceratandra orchids following a pollinator switch from bee to beetle.

  5. Latency of a vestigial trait (loss of expression, partially due to genetic change)
    Egg rejection behavior becomes unexpressed in Ploceus village weaverbirds separated from cuckoos, due to evolution of egg color and to lack of requisite environmental stimuli.

  6. Trait loss
    Several notothenioid icefishes near Antarctica have lost myoglobin as a result of evolution in a cold, oxygen-rich habitat.


How lost traits develop: dolphin hindlimbs and mouse hand webbing

The embryo of the spotted dolphin Stenella attenuata at 24 days (left) has a well-developed early hindlimb bud (h), which has regressed by 48 days (right), while the forelimb bud progresses, showing digital primordia (f). Photo and research by Brian K. Hall.
Development of the mouse hand illustrates formation and then programmed death (apoptosis) of cells between the digits; these are engulfed and cleared by macrophages. Photo and research by William Wood.


Selection knock-outs as a method for the study of evolution and adaptation

Cases where an important source of selection has been removed from a trait are like “knock-out” experiments, since removing a source of selection can provide insights into the role of that source of selection as well as the nature of trait interactions and how those interactions affect fitness and the evolvability of traits.  Just as pathologies have made contributions to physiology, and mutations and knockouts have aided genetic research, we are likely to gain a better understanding of trait evolution if we attend not only to cases where a trait confers a particular functional advantage, but also where that advantage has been removed.


The teleost Astyanax mexicanus has diverged into a surface form (left) and a cave form (right); the cavefish has lost pigmentation and eyes. Research and photos by Bill Jeffery. The orchid Ceratandra atrata (left) is pollinated by oil-collecting bees, and possesses a brown oil-producing callus in the center of the flower.  A more derived species, C. grandiflora, is pollinated exclusively by beetles that do not collect oil, resulting in relaxed selection on oil production.  In this species, flowers have either a vestigial oil gland (center) or none at all (right). Research and photos by Kim Steiner.


(These do not include publications; these links will open in a new browser window)



In addition to gaining perspective on the factors influencing the evolution of complex traits in general, many of my results would relate to public and environmental health issues.  For instance, an understanding of relaxed selection will help us understand and address the consequences of sheltering humans from natural selection due to medical and social advances.  Such study may also inform our attempts to relax selection that we have imposed on other organisms with undesirable results.  Our actions have unintentionally increased antibiotic resistance in pathogens and pesticide resistance in agricultural pests, and our exploitation of fisheries has caused life history evolution.  How can we most effectively relax anthropogenic sources of selection? To what extent can we expect these situations to be reversible? What are the best ways of promoting such reversals?  A major goal of my future work on relaxed selection and trait loss will be to place us in a better position to address these issues.