Home

Home

Biodiversity and global change

Research in the Clark lab focuses on biodiversity and global change, including how species coexist and how they are influenced by changing climate and natural and human disturbance.  Studies range in scale from field plots to continental data sets.  Large, long-term experiments are central to the approach.  There is emphasis on modeling innovation to synthesize evidence from many sources.

Subalpine at Niwot Ridge, CO one of many sites where tree fecundity is monitored.

Biodiversity change (NSF, NASA).

Generalized Joint Attribute Modeling (gjam) is being implemented to predict biodiversity responses to climate change.  One of our main data streams is the NEON, which is monitoring a number of species groups across the US.

US  Forecasts will be updated as data accumulate for NEON, land cover, climate, and remote sensing products. Forecasts will include species and community types at risk now based on habitat and biotic neighborhoods.  They will include community reorganization expected with climate change.

Hardwoods at Ordway-Swisher Biological Preserve, FL, another study site for tree fecundity and dynamic food webs.

A few examples:

Dynamic food webs respond to climate change

Clark, J. S., C. L. Scher, and M. Swift. 2020. The emergent interactions that govern biodiversity change. Proceedings of the National Academy of Sciences, 202003852, https://doi.org/10.1073/pnas.2003852117. clarkPNAS2003852117.full

Dynamic food web models can predict the non-linear effects that come through species interactions. Equilibrium abundances of species along a temperature gradient (cold green to warm brown), with a response curve for each of six land covers,including (A) chimney swift, (B) common grackle, and (C) American goldfinch. Maps show mean counts per observation effort by BBS route since 1996.Temperature responses across cover types demonstrate nonlinearities and interactions, neither of which are specified in the model. The land-cover type “dev” refers to developed lands.

Predicting how ecological communities respond to change requires an understanding of the direct effects of environment and the indirect effects that emerge when environment is propagated through food webs of interacting species. A probabilistic, dynamic framework for inference identifies these environment–species interactions. Results show how environmental effects translate into nonlinear responses, how they can be estimated from data, and the insight they provide on the relative importance of direct and indirect (through other species) responses to change. Because these effects include uncertainty from the model and data, they can guide management that has to weigh the utility of efforts to protect critical habitat (or not) against the risk for species that respond through the responses of others.

Seed dispersal near and far: patterns across temperate and tropical forests

Models allow us to predict fruit, seed, and nut supply across the forest floor.

The capacity for replacement through reproduction is one of the first requirements for persistence of any species.  The variables that control maturation, fecundity, and seed dispersal must all be inferred together from highly indirect data.  This paper is cited as the framework for accurate inference and prediction.  It has been implemented by labs all over the globe.  Software is here.

Continent-wide tree fecundity driven by indirect climate effects

Clark, J.S., R. Andrus, M. Aubry-Kientz, Y. Bergeron, M. Bogdziewicz, D.C. Bragg, D. Brockway, N.L. Cleavitt, S. Cohen, B. Courbaud, R. Daley, A.J. Das, M. Dietze, T.J. Fahey, I. Fer, J.F. Franklin, C.A. Gehring, G.S. Gilbert, C.H. Greenberg, Q. Guo, J. Hille Ris Lambers, I. Ibanez, J. Johnstone, C.L. Kilner, J. Knops, W.D. Koenig, G. Kunstler, J.M. Lamontagne, K.L. Legg, J. Luongo, J.A. Lutz, D. Macias, E.J. Mcintire, Y. Messaoud, C.M. Moore, E. Moran, J.A. Myers, O.B. Myers, C. Nunez, R. Parmenter, S. Pearson, R. Poulton-Kamakura, E. Ready, M.D. Redmond, C.D. Reid, K.C. Rodman, C.L. Scher, W.H. Schlesinger, A.M. Schwantes, E. Shanahan, S. Sharma, M. Steele, N.L. Stephenson, S. Sutton, J.J. Swenson, M. Swift, T.T. Veblen, A.V. Whipple, T.G. Whitham, A.P. Wion, K. Zhu, and R. Zlotin. 2020. Continent-wide tree fecundity driven by indirect climate effects. Nature Communications in press.

The effects of changing climate (left panels) on fecundity F (center) is mediated by climate effects on tree growth and stand structure (right). Clark et al. (2020) Nature Communications.

Trends in species abundances predicted from meta-analyses and species distribution models will be misleading if they depend on the conditions of individuals. Here we find from a synthesis of tree species in North America that climate-condition interactions dominate responses through two pathways, i) effects of growth that depend on climate, and ii) effects of climate that depend on tree size. Because tree fecundity first increases and then declines with size, climate change that stimulates growth promotes a shift of small trees to more fecund sizes, but the opposite can be true for large sizes. Change that depresses growth also affects fecundity. We find a biogeographic divide, with these interactions reducing fecundity in the West and increasing it in the East.  Continental-scale responses of these forests are thus driven largely by indirect effects, recommending management for climate change that considers multiple demographic rates.

The seasonal timing of warming that controls onset of the growing season.

Clark, J.S., J. Melillo, J. Mohan, and C. Salk. 2014.  Global Change Biology20:1136-1145.

Experimental warming at Duke Forest

Forecasting how global warming will affect onset of the growing season is essential for predicting terrestrial productivity. We show that accurate estimates require ways to connect discrete observations of changing tree status (e.g., pre- vs. post budbreak) with continuous plant responses to fluctuating temperatures. Results identify a critical window concentrated in late winter, weeks ahead of the main budbreak period. By late February/early March forecasts can predict early vs late onset of growth.

Ecological forecasts: an emerging imperative

JS Clark, SR Carpenter, M Barber, S Collins, A Dobson, JA Foley, …. Science 293, 657-660

This paper laid the foundation for a new initiative in ecology, focused on using information to anticipate, and potentially mitigate, responses to global change.  It is cited as the motivation for many new efforts to provide predictive understanding of change.

Individuals and the variation needed for high species diversity in forest trees

JS Clark. Science 327, 1129-1132

This paper can resolve a decades-old question in ecology, that of explaining why so many competing species could coexist on a small number of limiting resources.  Species-level differences are not apparent from species-level data, masked by averaging over individuals in species-level data.  Species differences masked by the phenomenon known as Simpson’s Paradox or the ‘ecological fallacy’ become apparent when analyzed at the individual scale: individuals respond more like others of the same species, thus concentrating competition within the species.

Failure to migrate: lack of tree range expansion in response to climate change

K Zhu, CW Woodall, JS Clark. Global Change Biology 18, 1042-1052

Changing geographic distributions have been the focus of climate change studies—evidence that populations can move to sites that become favorable.  More concerning is the possibility that they cannot. From a large analysis of inventory data across North America, results showed that tree species lack the capacity to track rapid contemporary climate change.

Generalized joint attribute modeling for biodiversity analysis: Median‐zero, multivariate, multifarious data

JS Clark, D Nemergut, B Seyednasrollah, PJ Turner, S Zhang. Ecological Monographs, 87, 34–56.

Synthesis of biodiversity data has not had a valid basis for inference and prediction.  Species are observed on different scales, and most are absent from most observations. They respond to one another at the same time that each is responding in its own way a changing environment.  Generalized joint attribute modeling (GJAM) provides a first advance towards integration, providing excellent parameter recovery and precise prediction of entire ecological communities.  More on GJAM software here.