Research

New ENM Dimensions

Recently, my work was funded by a European Research Council Marie Skłodowska-Curie Postdoctoral Fellowship project composed of two research lines designed to advance understanding of the processes that led to present-day species richness patterns of Atlantic marine fish. My first research line is to develop a method of efficiently and accurately inferring marine species distributions via ecological niche models (ENMs). Modern ENM methods typically employ artificial intelligence and other statistical methods to model suitable conditions for a species based on environmental observations at species’ occurrences (e.g., temperature, precipitation, salinity). Models can then be projected into geographic space to infer species’ distributions. However, ENMs were first developed for use in terrestrial systems. The majority of ENM workflows are structured to accept a set of points expressed as latitudinal and longitudinal coordinates representing where a species of interest has been observed. Environmental data are then extracted from a stack of two-dimensional rasters, one for each environmental variable of interest. In a marine context, this may lead to mis-estimation of species distributions and subsequent diversity estimates, especially among pelagic and benthic species. Species capable of inhabiting a wide range of depths may experience a correspondingly wide range of environmental conditions at a single horizontal coordinate. To solve this problem and facilitate a large scale marine ENM workflow for my fellowship project, I developed voluModel, a package of R tools to generate three-dimensional species distribution models based on environmental data extracted at the coordinates and depths where individuals were observed and calibrated by sampling a three-dimensional environment (Owens and Rahbek, 2022).

For my second MSCA research line, I am using my new 3D ENM workflow to map the Atlantic biodiversity of three clades of fishes, including many economically important species (Gadiformes, cods; Scombriformes, tunas; and Beloniformes, flyingfishes).  These three clades generally have very different life history strategies; my results show that the biodiversity patterns for these groups also differ substantially and become more distinct when applying 3D methods compared to more traditional ENM. Be sure to check back for updates!

Integrative Biodiversity

Ecological niche modeling requires large, robust occurrence datasets. I often look first to vouchered natural history collections for these data, as records can be verified by referring to physical specimens. For my previous US NSF-funded postdoc fellowship, I expanded this work to include the compilation of multiple types of data from natural history specimens to understand broad-scale biodiversity patterns. Specifically, I was interested in how the macroevolution, macroecology, and biogeographic history of swallowtail butterflies led to an observed latitudinal diversity gradient (LDG) in the group. This project demonstrated that the LDG in New World swallowtails is likely neither the result of an elevated diversification rate in the tropics, nor increased availability of abiotic niche space in tropical locations. Instead, while one clade of swallowtails shows a classic tropical origin with subsequent dispersal into temperate zones, another likely originated in temperate North America and subsequently dispersed into the tropics along the American Cordillera via high-elevation temperate-analog corridors (Owens et al. 2017). This work highlights the importance of considering both climatic and geographic definitions of “tropicality” in biogeography, as well as the role of temperate regions as under-appreciated sources of biodiversity.

Through the course of this project, I generated a workflow to photograph, transcribe labels of, and georeference occurrence records of over 1500 butterfly specimens from four natural history museums in two years. By coordinating the efforts of several undergraduate and graduate student researchers and community volunteers, I collected morphological landmark measurements from images of these specimens to test the hypothesis that macroecology (specifically range size and abiotic ecological niche breadth) is correlated with intraspecific morphological variation. Instead, strong evidence was found that forewing size and shape are generally conserved, whereas hindwing size and shape may be driven by natural selection for Batesian mimicry (Owens et al. 2020 in Systematic Biology).


Advances in Ecological Modeling Theory

Picture



While correlative ENM is an increasingly popular tool for macroecological research, many aspects of niche modeling are poorly understood, both in theory and practice. To address these issues, I collaborate with a large, necessarily-interdisciplinary research group of biologists, geologists, and geographers. In one project, I led the design of a new approach to assess the degree and direction to which different ENM methods extrapolate when transferred outside the training region of the model (Owens et al. 2013). This is an important consideration for several applications of ENM, including estimating potential past and future species distributions based on climate dynamics.

Ecological niche evolution reconstructed via comparative phylogenetic methods can provide valuable insight into the roles of climate change and ecological niche conservatism in the diversification of evolving lineages. However, recent simulation studies demonstrate comparative phylogenetic methods may inaccurately infer the history of ecological niche evolution due to the reliance of these methods on summary statistics rather than estimates of the full range of a species' environmental tolerances. To fix this problem, I led a team to develop R tools to characterize and reconstruct the ecological niche tolerances (nichevol; Owens et al. 2020 in Ecology and Evolution).


Tools for Repeatable Biodiversity Science

Picture





R is a flexible, open-source statistical programming language that enhances the accessibility to and repeatability of quantitative analyses. I write R packages to solve analytical challenges in ecology and evolutionary biology, and to help others more easily adopt the methods I devise to solve these challenges.

Tools for ecological niche modeling are well-established, but methods reporting in peer-reviewed publications for truly repeatable science remain under-developed. To facilitate simple, standardized reporting that permits direct comparison between niche models from different studies, I am part of a collaborative project to develop metadata reporting standards for niche modeling studies. In service of this work, we have developed an R package to document model metadata in concordance with these data standards (Merow et al. 2019).

I am also working to improve metadata documentation during the collection of occurrence data for ENMs to provide better transparency and repeatability in large-scale biodiversity datasets, especially those derived from digitized and publicly available natural history collections (occCite). occCite enables users to query database aggregators (e.g., the Global Biodiversity Information Facility) and automate primary data provider citations (Owens et al., 2021). Future development will include the assignment of DOIs to occurrence datasets to further improve repeatability and transparency of studies implementing large species occurrence datasets.