Overcoming challenges in measuring how seed dispersal and climatic niche evolution are connected in plants

In this blog post, I discuss some of the “behind the scenes” factors in writing “Linking mode of seed dispersal and climatic niche evolution in flowering plants”, including our main motivation and some challenges we had to overcome with data curation and trait evolution modeling in the project

Above: Miconia sp. (Melastomataceae), a diverse lineage of flowering plants with fleshy fruits and biotically dispersed seeds in tropical forests; Photograph by Vanessa Staggemeier.

My co-authors and I started talking about writing a paper linking mode of seed dispersal and climatic niche evolution during the Covid-19 pandemic lockdown in 2020. At the time, I was reading a lot of the classic literature on plant biogeography and, led by Jeremy Beaulieu, we were all working on developing new models to better understand plant macroevolution in a quantitative way. Inspired by the interdisciplinary synthesis of Stebbins (1974), I became particularly interested in understanding how certain traits are important modulators in the biogeography of plants which, as sessile organisms, move in space mainly during seed dispersal, and so seed dispersal should have a strong impact on their spatial distribution. For example, it had been suggested that species with biotically dispersed seeds tend to be more common in warmer and wetter environments and that the behavior of frugivores that disperse their seeds would lead these lineages to have slower rates of climate niche evolution over time.

Reasoning behind one of the hypotheses tested in our study. (a,b) Dry fruits are used as a proxy for abiotic seed dispersal, whereas fleshy fruits are used as a proxy for biotic seed dispersal. (c) Different modes of seed dispersal are expected to impact rates of climatic niche evolution in flowering plants differently, based on the assumption that abiotic factors promote erratic dispersal and increased opportunity for niche shifts through time.

Despite that, we noticed that much of the more recent literature on climatic niche evolution in plants focused mainly on the importance of vegetative traits, such as leaves and woodiness, while fewer investigate reproductive structures, such as flowers and fruits. The few studies that had analyzed the connection between climatic niche evolution and mode of seed dispersal in some capacity did so within a restricted sample of species or in a non-quantitative way. Consequently, it was difficult to tell if certain patterns were general enough to be replicated in many unrelated plant lineages. Conveniently, James Boyko was working on developing new trait evolution models for his PhD at the time and the hypotheses we wanted to test matched perfectly the way in which these models were parameterized. When we heard about the JBI Innovation Awards, we felt that that was the final push we needed to work together on this collaborative project.

Editors’ choice article: (Free to read online for two years.)
Vasconcelos, T., Boyko, J. D., & Beaulieu, J. M. (2023). Linking mode of seed dispersal and climatic niche evolution in flowering plants. Journal of Biogeography. 50:46-53. https://onlinelibrary.wiley.com/toc/13652699/2023/50/1 

In my opinion, one of our most interesting result was that no matter which clade we look at, we always found support for lineages with abiotic seed dispersal shifting to colder climatic optima and lineages with fleshy fruits shifting towards warmer climatic optima. This is true for more temperate clades like Ericaceae as well as more tropical ones like Melastomataceae. We also found some support for lineages with biotically dispersed seeds having slower rates of climatic niche evolution, though in this case the pattern is not as general and the story is perhaps a bit more complicated, as we discuss in the paper. But, to me, that is the main appeal to combine and compare results from multiple lineages under the same analytical framework – we can use these results to set apart rules (patterns observed in most lineages) and exceptions (lineages where the general pattern is not observed) in how biodiversity evolves in space over time.

Optima estimates for temperature (in Celcius) for lineages with different fruit types. Note that the climatic optima of Ericaceae lineages are still significantly colder than those of Melastomataceae, reflecting their global distribution. However, both families still show the same general pattern where lineages fleshy fruits shift to warmer climatic optima. Distribution maps taken from Stevens (2001-onwards).

Regarding challenges we had to overcome, I think two of the most time consuming parts of the project were data curation and improving computational speed for modeling continuous traits. Much of our data had been scored using fruit type as a proxy for mode of seed dispersal. Fleshy fruits, such as berries and pomes, are typically dispersed by frugivore animals while dry fruits, such as achenes and capsules, are typically dispersed by abiotic factors such as wind and gravity. However, this is an often simplistic view of how seed dispersal works, and during data curation we had to perform a more thorough literature review of some groups where fruit type did not necessarily correspond to the main mode of seed dispersal. One other problem was the computational time and effort needed to fit our models. Not only were the trait evolution models very complicated, but they needed to be fit many times in each dataset to make sure our parameter estimates were correct. This sort of computational workload was not feasible using standard algorithms available at the time, so my co-authors implemented an algorithm which allowed for the fast calculation of model parameters. The increase in speed is difficult to understate as models which once may have taken weeks to finish, could now be completed in hours!

Today, our team keeps working on similar projects linking trait-evolution and biogeography in plants, and we have several papers to be published soon in the same vein as this. For instance, in Vasconcelos (in press.), I give a brief historical overview of trait-environment correlation analyses and indicate some limitations in our current approaches to quantify rules in plant biogeography. In Boyko et al. (in press.), James and Jeremy developed a new software that improves our current analytical framework. Using this new software, our team is working on answering other trait-environment correlation questions in plants, such as, for instance, what are the evolutionary processes that lead to a prevalence of annual life history strategy in deserts and mediterranean areas (Boyko et al., in review).

Written by:
Thais Vasconcelos
Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, MI 48109, USA

Additional information:
Twitter: @TNCVasconcelos

References:
Boyko, J. D., O’Meara, B., & Beaulieu, J. M. (in press.). Jointly Modeling the Evolution of Discrete and Continuous Traits. Evolution. Preprint available at: 10.32942/osf.io/fb8k7
Boyko, J. D., Hagen, E. R., Beaulieu, J. M., & Vasconcelos, T. (in review). Long-term responses of life-history strategies to climatic variability in flowering plants. Preprint available at: 10.1101/2022.10.19.512857
Stebbins, G. L. 1974. Flowering plants: evolution above the species level. Harvard University Press, Cambridge, MA.
Stevens, P. F. (2001 onwards). Angiosperm Phylogeny Website. Version 14, July 2017 [and more or less continuously updated since] http://www.mobot.org/MOBOT/research/APweb/.
Vasconcelos, T., Boyko, J. D., & Beaulieu, J. M. (2023). Linking mode of seed dispersal and climatic niche evolution in flowering plants. Journal of Biogeography. 50: 46-53. 10.1111/jbi.14292 Vasconcelos, T. (in press.) A trait-based approach to the rules of plant biogeography. American Journal of Botany . Preprint available at: 10.32942/osf.io/azytc