Pineapples grow best in hot environments, and wheat grows better in cold steppes; saguaro cacti live in deserts, and Venus fly-traps live in nutrient-poor wetlands. We all have very reasonable intuitive ideas of what organisms can be found in what environments, that are often borne out by careful observations. Alexander von Humboldt was one of the first westerners to systematically describe these patterns in his Essai sur la géographie des plantes, after completing an expedition to the Chimborazo volcano in what is now Ecuador.
Different species are clearly associated with different environments. This fact is not in itself surprising – two thousand years earlier, even Herodotus notes that some crops grow better in certain regions than others. The mystery is in why. What is it about wheat that means it grows better in Assyria than in Greece? Or about cacti that they live in deserts and not in swamps?
Imagine a challenge. Suppose someone shows you an organism, outside of its normal context – perhaps a tree in a pot, or an animal in a cage. How much would you be able to say about where it would be found in nature? You might propose that certain aspects of the organism’s phenotype can help answer this question. These functional traits should be linked to the performance and then distribution of the organism. For example, perhaps animals with more larger bodies do better in colder environments, or plants with smaller leaves do better in drier environments.
Being able to succeed at this challenge is important for two central problems. First, it allows reconstruction of the Earth’s past climate from paleoecological proxies (e.g. a fossil cactus might indicate a past warm climate). Second, it allows us to predict the planet’s response to climate change (e.g. warming environments might yield more cacti).
However, establishing robust trait-environment relationships for plants has been difficult. First, many linkages that one might imagine would be strong, like that between plant height and temperature, turn out to actually be very weak at global scales. Second, many linkages that appear to be strong (like between leaf teeth and temperature) are inconsistent across different evolutionary lineages and also have limited mechanisms underlying them. This makes it difficult to extrapolate to new scenarios – key issues for the two central problems. Put another way, it is often possible to statistically fit weak relationships between traits and climate – but we generally cannot make a priori predictions for the particular form of these relationships. If the temperature increases by 2°C, should a tree get taller by 1 cm or by 5 cm?
We recently had a paper come out in Ecology (PDF here) that tries to partially address this challenge. The approach is based on testing our previously-published model for linking the ecophysiology of leaf venation networks to temperature. Veins transport water within the leaf, and thus are central in constraining rates of water loss, carbon gain, and potentially overall growth. We proposed a ‘use it or lose it’ hypothesis, in which the maximum evaporative demand of an environment (related to its temperature) should be matched to the maximum transpiration rate of a leaf (related to the density of veins in the leaf). We were able to invert this model to write an equation for the relationship between temperature and vein density.
We then tested this prediction using a large dataset for vein density for hundreds of species spanning a 3000 meter elevation gradient from the Amazon basin to treeline in the Andes. This work was part of the intensive CHAMBASA project in eastern Peru, led by Yadvinder Malhi (Oxford), Brian Enquist (Arizona), Greg Asner (Carnegie), and Sandra Díaz (Córdoba).
Over several months in 2013, a large team of Peruvian biologists and foreign collaborators (too many to name here) intensively sampled ten one-hectare forest plots along this gradient for a range of functional traits. Under Dr. Salinas’ leadership, we were able to assemble a large leaf venation dataset via chemical work in Eric Cosio‘s laboratory in Lima.
Below you can see two contrasting species from these transects – the top, Pourouma bicolor (Urticaceeae), and the bottom Clusia alata (Clusiaceae) – varying in vein density by a factor of almost six.
Using these hard-won data, we were able to examine these hypothetical trait-environment relationships. We found that the relationship between vein density and temperature was very strong relative to previously published trait-environment relationships, suggesting that it has strong predictive power for understanding the environments in which different communities of plants occur. And we also found that the sign and slope of this relationship was consistent with our a priori prediction, though the overall prediction was also systematically under-biased.
The take-away from this study is that trait-environment relationships are elusive but not impossible to find, and that even simple theory provides ways to improve predictability in community ecology. There is still an opportunity to improve this theory, and to develop other theory for other key trait-environment relationships. But for now, this paper provides guidance for at least one important trait to measure, and hopefully stimulates conceptual thought on the limits to predictability in community ecology. Please give it a read!