Mapping the margin: comparing marginal values of tropical forest remnants for pollination services.

Natural ecosystems benefit human communities by providing ecosystem services such as water purification and crop pollination. Mapping ecosystem service values has become popular, but most are static snapshots of average value. Estimating instead the economic impacts of specific ecosystem changes can better inform typical resource decisions. Here we develop an approach to mapping marginal values, those resulting from the next unit of ecosystem change, across landscapes. We demonstrate the approach with a recent model of crop pollination services in Costa Rica, simulating deforestation events to predict resulting marginal changes in pollination services to coffee farms. We find that marginal losses from deforestation vary from zero to US$700/ha across the landscape. Financial risks for farmers from these losses and marginal benefits of forest restoration show similar spatial variation. Marginal values are concentrated in relatively few forest parcels not identified using average value. These parcels lack substitutes: nearby forest parcels that can supply services in the event of loss. Indeed, the marginal value of forest parcels declines exponentially with the density of surrounding forest cover. The approach we develop is applicable to any ecosystem service. Combined with information on costs, it can help target conservation or restoration efforts to optimize benefits to people and biodiversity.

Modelling pollination services across agricultural landscapes.

BACKGROUND AND AIMS: Crop pollination by bees and other animals is an essential ecosystem service. Ensuring the maintenance of the service requires a full understanding of the contributions of landscape elements to pollinator populations and crop pollination. Here, the first quantitative model that predicts pollinator abundance on a landscape is described and tested. METHODS: Using information on pollinator nesting resources, floral resources and foraging distances, the model predicts the relative abundance of pollinators within nesting habitats. From these nesting areas, it then predicts relative abundances of pollinators on the farms requiring pollination services. Model outputs are compared with data from coffee in Costa Rica, watermelon and sunflower in California and watermelon in New Jersey-Pennsylvania (NJPA). KEY RESULTS: Results from Costa Rica and California, comparing field estimates of pollinator abundance, richness or services with model estimates, are encouraging, explaining up to 80 % of variance among farms. However, the model did not predict observed pollinator abundances on NJPA, so continued model improvement and testing are necessary. The inability of the model to predict pollinator abundances in the NJPA landscape may be due to not accounting for fine-scale floral and nesting resources within the landscapes surrounding farms, rather than the logic of our model. CONCLUSIONS: The importance of fine-scale resources for pollinator service delivery was supported by sensitivity analyses indicating that the model's predictions depend largely on estimates of nesting and floral resources within crops. Despite the need for more research at the finer-scale, the approach fills an important gap by providing quantitative and mechanistic model from which to evaluate policy decisions and develop land-use plans that promote pollination conservation and service delivery.