Where information from government agencies, media, and others is limited and/or non-existent, local environmental knowledge (LEK) supports household and community-level decision-making about resource use and management during stable periods, and adaptation during uncertain times. The complexity of socio-ecological systems (SES) suggests that LEK users focus their attention on specific socioeconomic and/or biophysical indicators to forecast future environmental change, including climate uncertainty. LEK of the connections and interactions between various SES elements and processes is then drawn on a second time to analyze response risks and make decisions. This local SES model offers opportunities for exploring variation that can be lost in the large, aggregated datasets used for global and regional SES modeling. Comparisons of LEK indicators with regional and global indicators may also suggest variables for further analysis in studies of critical thresholds and regime shifts. In this paper, I draw on Complexity Theory to develop a framework for exploring SES complexity from a local perspective using LEK. Throughout, I highlight examples from previous fieldwork in Mozambique and Tanzania to show the value of combining visual methods, like photovoice, with other ethnographic and ecological approaches.