Plastic trait integration across a CO2 gradient in Arabidopsis thaliana

Shifts across environments in patterns of trait integration may govern or alter adaptive responses. Changes in resource supply rates may be an especially important cause of plasticity of trait integration because they can lead to shifts in colimitation and coregulation of traits. Traditional evolutionary genetic characterization of trait integration relies on covariance analyses. Structural equation modeling (SEM) can complement such analyses. The SEM provides insights into causal structure not possible with a covariance analysis, thereby providing mechanistic understanding of shifts in integration and suggesting likely foci of selection in changing environments. We tested for changes in trait integration by growing 35 genotypes of Arabidopsis thaliana (Brassicaceae; mouse-eared cress) from throughout the species' range in four atmospheric CO2 concentrations: 250 ( past), 355 (similar to recent CO2), 530, and 710 ( future) mu M M-1. SEM revealed significant shifts in the integration of N, C, and H2O use and their effects on reproductive dry mass across the CO2 gradient. The low CO2 stress of 250 mu M M-1 had the most divergent integration structures. Standardized total effects of C assimilation, water loss, and early N mass changed in sign across the C supply gradient, and the total effect of quantum yield decreased from significant to nonsignificant values across the gradient. Transpiration exhibited significant genetic variation and is thus a candidate target for selection and adaptation under novel growth CO2 concentrations. The strength of the correlation between C assimilation and transpiration declined by 19% from 250 to 710 mu M M-1, indicating a partial decoupling of their current mutual evolutionary constraint in the atmosphere of the future. Structural equation analysis of functional integration provides unique insights into the mechanisms through which changes in limiting resources can alter the nature of trait integration.