Using network of species interactions to value biodiversity conservation in a megadiverse country: a comparison of latent class and mixed logit models

Abstract This study examines whether different biodiversity proxies – species, habitat and functionality – satisfy the scope sensitivity and plausibility criteria in willingness to pay (WTP) estimation using a choice experiment in Manu National Park, Peru. We introduce the network of species interactions as a proxy for functionality and apply latent class (LC) models, including attribute non-attendance (ANA), to account for heterogeneity in preferences. Our results indicate that functionality is the only proxy consistently meeting both validity criteria across all specifications. LC analysis reveals two segments: one (74.4 per cent) displaying coherent, scope-sensitive WTP across biodiversity attributes, and another (25.6 per cent) less engaged, disregarding standard proxies but still valuing networks. Even under ANA constraints, networks remain salient for less attentive respondents, underscoring their cognitive accessibility in complex ecological contexts. These findings highlight the methodological and policy relevance of functionality-based proxies for biodiversity valuation in megadiverse environments, where conventional measures may fail to elicit behaviourally consistent responses.