Local adaptation, in which populations of a species become better able to survive and thrive in their home environment than in conditions found elsewhere in the species’ range, is a widespread pattern that evolutionary biologists have long used to study the causes and consequences of natural selection. My newest paper, which is now online ahead of print in The American Naturalist, combines data across many studies of local adaptation to answer a persistent question about the history of life on Earth — has evolution been influenced more by selection arising from environmental conditions, or by interactions among living things?

Factors in the nonliving environment, like temperature variation, water availability, or sunlight, have certainly shaped the evolution of biodiversity, as have species’ interactions with predators, prey, and mutualists. However, which of these two sources of natural selection more often contributes to local adaptation had not been systematically evaluated. This question links to some of the biggest ongoing mysteries about the distribution of biological diversity on Earth, such as why there is greater diversity of plants and animals in tropical regions, compared to locations closer to the poles.

The relative roles of biotic and abiotic environments came up as a topic in a journal discussion group I joined way back during my first postdoc, at the University of Minnesota. Ryan Briscoe Runquist, another postdoc in the group, and I had the idea to compare the relative effects of biotic and abiotic environmental factors by compiling experimental studies of local adaptation in which organisms from different populations of the same species had been grown in differing environments, either in experimental settings or by transplantation to different field sites. Meta-analysis methods would let us compare the relative effects of different environmental conditions, even between an experiment measuring (say) plants’ seed production in different soil types and an experiment measuring tadpoles’ growth in ponds with or without predators present.

A hazard of meta-analysis, though, is publication bias — experiments that don’t find a significant effect of a tested environmental factor may be less likely to be published and available for meta-analysis. We decided to get around this by focusing on studies that simultaneously tested the effects of at least one element of the nonliving environment (like soil chemistry or rainfall) and at least one interaction with another species (like competitors or predators). We reasoned that an experiment testing two factors might still be published if one had a significant effect but the other didn’t. Another member of the reading group, Jake Grossman, suggested we pair the meta-analysis with a meta-synthesis, a structured, systematic reading of the text of the papers we compiled, to better understand how the experiments were conceived, designed, and executed. We talked a bunch of other group members into helping with the tasks of finding papers that reported the right kind of experiments, reading and annotating them for meta-synthesis, and extracting and analyzing their data in the meta-analysis.

This all took place over a bunch of complicating life events — I moved to a second postdoc, then to my faculty position at CSU Northridge, and many of the other collaborators had similar transitions. Drafting the paper for submission took long enough, in between other commitments, that we actually went back for a second round of searching for studies and expanded the analysis last spring. But the final result was well worth it, I’d say.

Our meta-analysis found that, overall, the fitness effects of biotic environmental factors were slightly but significantly greater than those of abiotic factors. However, the relative impacts of abiotic and biotic factors varied with the type of organism tested, and the broader context of the experiment. Overall, local adaptation to interacting species tended to be stronger than local adaptation to the nonliving environment — but this pattern was mostly seen in experiments with animals, with plants showing the opposite pattern. The overall impact on fitness — growth or reproductive success — by abiotic and biotic factors also varied with the latitude at which experiments were conducted. In the tropics, nonliving factors had somewhat weaker effects and biotic factors had stronger ones; at higher latitudes, nonliving factors were stronger and biotic factors were weaker.

These results shed light on some large-scale patterns of biodiversity, but the meta-synthesis of the text of scientific papers we’d compiled revealed how much work remains to be done. That analysis found substantial differences in the ways that biologists think about the different environmental factors tested in local adaptation experiments, and potential biases introduced by the logistics of such experiments. Notably, nonliving environmental factors were often tested in terms of natural variation that existed prior to an experiment’s design, such as temperature differences at different elevations on a mountainside; but the effects of competitors or predators were more often tested by simply removing the interacting species. Experiments were also much more likely to be conducted with short-lived, immobile organisms, like annual plants, which are more amenable to experiments that can be completed in a year or two. Expanding the range of species tested in local adaptation experiments, and working to test more realistic variation in interactions among species, could make the results of these experiments more broadly informative.

For more detail, see the full article in The American Naturalist.