<p>Zoonotic malaria caused by <i>Plasmodium knowlesi</i> is an emerging threat in Southeast Asia, where deforestation and habitat fragmentation are reshaping human–wildlife–vector interfaces. While deforestation is known to increase ecological overlap among humans, mosquitoes, and macaques, how changes in host communities and landscape structure jointly influence spillover risk remains unclear. Here, deforestation is represented as edge accumulation using the edge ratio—a continuous measure of forest–non-forest boundary—within a process-based simulation model parameterized with previously reported estimates. The model integrates mosquito abundance, host densities, biting preferences, and reservoir competence to evaluate how spillover risk responds to landscape fragmentation. Scenarios explore changes in host populations, and global sensitivity analysis quantifies the relative influence of key parameters. Model simulations suggest that spillover risk may peak at intermediate edge ratios, where mosquito density and host overlap are jointly elevated. Changes in host composition can alter both the magnitude and location of peak risk, and abrupt biodiversity loss may produce threshold-like increases under certain conditions. Notably, spillover risk is not estimated to be highest where human density is greatest, but where both humans and macaques are sufficiently represented in the mosquito biting pool, reflecting a nonlinear trade-off in host availability. Sensitivity analysis indicates that variation in macaque competence is a dominant driver of model outcomes. These results suggest that monitoring macaque infection and mosquito activity in fragmented forest–agriculture edges may help identify areas of elevated zoonotic malaria risk, although empirical validation is needed to assess their applicability in real-world settings.</p>