Dynamic Global Vegetation Models (DGVMs) are an important tool widely used for understanding ecosystem dynamics and predicting climate change impacts, yet recent work suggests that DGVMs often do a poor job of capturing seasonal patterns in plant production. This problem stems in part from the relatively simplistic representation of phenology in most models. Because the phenological niche of many plant species is linked to both their spatial distribution and competitive interactions with other plants, errors in how process-based models represent phenology may hinder our ability to predict climate change impacts.
We used big sagebrush as a model species to test whether improving the representation of phenology in the LPJ-GUESS DGVM could improve the accuracy of simulated vegetation and carbon dynamics. Sagebrush defines a habitat type that is of high conservation concern due to its importance to wildlife, and there is widespread interest in modeling potential climate change impacts. Sagebrush phenology is also poorly represented in LPJ-GUESS, which cannot account for its ability to grow two distinct cohorts of leaves. Data from flux towers located in sagebrush ecosystems were used to test the accuracy of seasonal patterns in modeled GPP and the efficacy of different methods for model improvement.
While our standard LPJ-GUESS simulation showed a reasonable level of agreement with seasonal patterns of measured GPP, there were still several systematic inconsistencies. For example, LPJ-GUESS tended to under-estimate GPP during the summer months and over-estimate GPP during the fall. In contrast to several previous studies, we found that LPJ-GUESS was able to replicate the timing and magnitude of GPP increases associated with spring green-up. We were able to reduce the seasonal bias by developing a new phenology type that allows for partial leaf shed in response to summer moisture stress. In addition to this structural change, we identified several key parameters that affect simulated patterns of GPP. Many of these parameters were poorly constrained by empirical data, but we were able to determine appropriate ranges based on estimates from a literature review. Optimizing these parameters using flux tower data yielded further improvements in our ability to replicate seasonal patterns of GPP in sagebrush ecosystems. Our results suggest that for species with unusual phenology, improving both the model structure and parameter values may be necessary to accurately capture seasonal patterns of plant production and associated impacts on vegetation dynamics.