Quantification of R–T curves and Model Comparison.

MH Mary A. Heskel
OO Odhran S. O’Sullivan
PR Peter B. Reich
MT Mark G. Tjoelker
LW Lasantha K. Weerasinghe
AP Aurore Penillard
JE John J. G. Egerton
DC Danielle Creek
KB Keith J. Bloomfield
JX Jen Xiang
FS Felipe Sinca
ZS Zsofia R. Stangl
AT Alberto Martinez-de la Torre
KG Kevin L. Griffin
CH Chris Huntingford
VH Vaughan Hurry
PM Patrick Meir
MT Matthew H. Turnbull
OA Owen K. Atkin
request Request a Protocol
ask Ask a question
Favorite

The 673 RT curves collected by the methods described above required thorough quantification for comparison across replicates, species, sites, biomes, and plant functional types. For each replicate R–T response curves, we assessed the fits commonly applied R–T models, including: (i) an exponential model with a fixed-Q10 across the entire T range (though not specifically a fixed Q10 of 2, as is applied in some biosphere models of R); (ii) an Arrhenius model; (iii) a model of R responding to the UTD as defined by Gillooly et al. (15), which contains an activation energy parameter and uses Boltzmann’s constant; (iv) a model presented by Lloyd and Taylor (17) to describe the response of soil R to T that includes a temperature-sensitive activation energy; (v) a model that incorporates a variable-Q10 response across the T range as described by two parameters; and (vi) a simple second-order polynomial model. Equations for these models are shown in Supporting Information. To compare how these models fit to data, we fitted each of the aforementioned models to all replicate R–T response curves in JMP (Version 11; SAS Institute), with parameters calculation controlled by the minimal residuals produced from each individual fit for each model. In cases where model convergence was not possible via the curve-fitting software, those replicate curves were not included to calculate mean residuals for the model fit over all replicates. Further, to evaluate the impact of different measurement temperature span (i.e., 10–45 °C vs. 20–45 °C) on model fits, we compared fit coefficients across all replicate curves at different segmented intervals of the response curve (Table S2, Fig. S3, and Supporting Information). Using these data, we also compared model fit coefficients from the approximate 20 °C T range that best represents the climate of that species (the “ecologically relevant” T range; Table S3 and Supporting Information) to the fit coefficients calculated from all available data from the entire measurement T range.

Do you have any questions about this protocol?

Post your question to gather feedback from the community. We will also invite the authors of this article to respond.

post Post a Question
0 Q&A