To succeed locally
​
We had to think nationally
Accurate connectivity modelling requires understanding not just where our focal species live, but how they move - the long-range dispersal events that sustain genetic flow between populations and determine whether a corridor functions over time.
That understanding cannot be built from local occurrence data alone.
So that is the scale we worked at.
Using publicly available occurrence records across mainland Costa Rica, we built species distribution models (SDMs) for all eight of our focal species plus jaguar.
A persistent challenge in SDMs is that occurrence records do not tell you where a species is - they tell you where people have looked. We addressed this at two stages:
-
Pseudo-absence points were drawn from a target-group background pool of co-occurring mammal species, so that sampling effort was matched between presence and background.
​
-
Anthropogenic predictors - roads, urban areas, and related variables - were each tested for distributional bias against the background pool using Kolmogorov-Smirnov tests before being admitted to modelling; those showing significant bias dropped from the analysis pipeline for that species
​
Models were trained using a multi-algorithm ensemble framework with predictor variables covering climate, terrain, canopy structure, land use composition measured at species-appropriate spatial scales, and distance-decay surfaces for roads and anthropogenic features. Ensemble outputs were converted into resistance surfaces and used to produce predicted connectivity maps covering mainland Costa Rica at both the individual territory scale (30m resolution) and the dispersal event scale (100m resolution).
We are making these outputs available to other organisations working in conservation across Costa Rica.
We know that safeguarding the corridor at the tip of the Nicoya Peninsula depends on what happens well beyond our own jurisdiction.
Once our write-up is hosted on a relevant pre-print server, we will be offering offer two tiers of access to our data to qualifying organizations, free of charge :
​
-
Tier 1 - Habitat suitability and circuit theory-based connectivity predictions clipped to your project area, at 300m (individual territory scale) and 1km (dispersal event scale). No data contribution required.​
​
-
Tier 2 - Full resolution output data at our most precise resolution - 30m (individual territory scale) and 100m (dispersal event scale), in exchange for camera trap detection data and precise trap locations from your study area. Contributing organisations will be offered co-authorship on any publication arising from the analysis.
​
To discuss either tier, get in touch via the form below.
​​
Additionally, if your organisation funds or equips field research in Costa Rica, we are actively seeking support for camera trap deployment, and for what will be the first long-term population monitoring programme in the district of Costa Rica's first protected area - producing the empirical data needed to validate and refine our habitat suitability and connectivity models over time.
Contributing to:
