Human-Environment Interactions Explain Local and Regional Variation in Aboveground Biomass in Tropical Montane Forests

The magnitude and spatial patterns of aboveground biomass (AGB) in tropical mountains are expected to correlate with ecosystem structure and biodiversity. However, the scalability of such relationships remains poorly understood partly due to forest disturbances, such as land use. Here, we quantified AGB magnitude and distribution in tropical montane cloud forest (TMCF) in southern Mexico at landscape and regional scales. Specifically, we investigated the relative roles of land use and environmental factors (i.e., topography and climate) on AGB patterns, as well as its relationship to tree species diversity using 160 plots from the Mexican National Forest Inventory (FI) database established along an elevation gradient, and Landsat time series. We assessed the relationships among environmental variables, forest structure, and remote sensing variables using multiple linear regression models. Our results show that tropical montane landscapes are very dynamic, and they show wide variation in forest structure. AGB ranges from 8-400 Mg ha-1, depending on land use, climate, and topography, where land use explains most of the variation. The amount of AGB increased with elevation and slope and decreased with very high levels of precipitation. The time series analysis shows that 76 FI plots were cleared at least once between 1993 and 2004, and that the variation of vegetation indices over time is related to forest structure (particularly to basal area). Furthermore, forest plots with recent clearings exhibit statistically significant lower basal area, tree height, and AGB. Additionally, we found that AGB is weakly but positively correlated with tree species richness (but not tree diversity) at small spatial scales (but not at larger extents). We conclude that AGB spatial patterns can be best predicted by the interactive effects of land use and environmental factors, with land use having a larger role. Including recent forest disturbance as input in AGB models can greatly improve biomass predictability and mapping in dynamic landscapes. Our results reveal new patterns that go against general assumptions about the structural and compositional properties of these ecosystems, emphasizing a need for explicitly including interactions between environmental and human drivers in AGB estimations and forest management plans.