Most forest ecosystem variables in Yellowstone NP are more stable than in adjacent logged forests

Background/Question/Methods

We showed recently that an exact mathematical model relates aggregate variability of a process (or quantity) to its spatio-temporal pattern such that Aggregate Variation = ∑Spatial Variances + ∑Temporal Covariances - ∑Spatial Covariances. The model implies that landscape stability can be reduced by spatial differences between sites, their synchrony, or both and increased by spatial covariance (persistence of the pattern in space).  As the measured variation is often scale-dependent, the perception of stability of a particular process may change with change of scale.  Our study aims to both test the model in a large natural system (forested landscape) and to determine the patterns of sensitivity of ecosystem variables to temporal scale of observation. To accomplish this, we used satellite (MODIS – Terra and Aqua spacecraft) data on production (NPP, GPP), surface reflectivity, temperature, albedo, emissivity, leaf area index, and other indices of the state of a forested landscape. 41x21 km area (half in natural forested area of Targhee National Forest (Yellowstone) and half in logged area adjacent to it), centered at latitude 44.32090 and longitude -111.09993 and resolution of 250-1000m depending on the variable, provided data for evaluation of seasonal and inter-annual variation.

Results/Conclusions

We found that natural forests were more stable for most of 8 variables included in the analysis than the logged forest west of Yellowstone (r-square values: 0.379-0.611).  Furthermore, at both short- and long-term scales, slopes of regression lines between variation in space and stability of individual variables (=data points) were steeper for logged (disturbed) landscapes, with the greater difference arising at longer time scale.  These patterns imply several important processes: (1) disturbed landscapes tend to undergo a personality split between variables dominated by regional (environmental) and local (disturbance) factors; (2) natural landscapes show greater spatial diversity associated with greater (thus stabilizing) persistence of local differences as opposed to the disturbed habitat; and (3) the negative effects of anthropogenic disturbance become more prominent at longer time scales in spite of some post-disturbance stabilization of the variables of interest in the logged landscape.  In conclusion: the quantitative model we propose effectively summarizes dynamics of a spectrum of landscape variables and identifies the nature of disruption (decoupling) among processes they characterize under anthropogenic pressure.