Abstract
1. Efforts to monitor and conserve populations and ecosystems in human-dominated landscapes can benefit from an empirical social–ecological systems approach.
2. Here we illustrate how latent variable structural equation modelling of regional time series data can effectively describe interconnected drivers of population fluctuations in dynamic landscapes and can help to reveal previously unknown system drivers.
3. Using a declining farmland-dependent bird species (Ammodramus savannarum) in the eastern United States (1994–2015) as a case study, our analysis reveals how farm management decisions drive population fluctuations (R2 = 20%), while management is in turn highly influenced by climate (R2 = 23%–51%), but not by regional conservation spending.
4. Synthesis and applications. Structural equation modelling revealed potential social–ecological pathways for halting regional population declines in a grassland bird, the Grasshopper Sparrow. Lower population growth rates followed years of higher hay yields (~4 percentage points per metric ton increase in hay yield) and later harvests (~2 percentage points per 10-day delay in harvest). Thus, one pathway for stabilising regional populations could involve compensating farmers for reducing hay harvests, potentially requiring a six-fold increase in current annual agri-environmental conservation spending.