A corridor is nothing
without its network
We are building a multi-species, national connectivity model, and are looking for collaboration with other organizations.
Mapping local corridors requires thinking at the landscape level. The long-range dispersal that keeps populations connected does not respect a reserve boundary or a district line.
Protecting one corridor means understanding the country it sits within, and working with those who know it best, from the Caribbean slope to the Pacific coast.
If that includes you or your organization, we'd love to hear from you.
Read more about our offer
SPECIES DISTRIBUTION MODELS
A species distribution model is only as honest as the records it learns from. Citizen-science occurrence records describe where people encounter a species, not where it lives. Institutionally derived camera trap records tend to be focused in optimal habitat and say little about how a species moves when conditions aren't ideal. By combining these data and comparing them against background samples of where non-target species have been observed via both frameworks, we can reduce sampling bias and paint a more honest picture of habitat suitability.
We used ensemble models to predict habitat suitability for our four focal species (puma, ocelot, collared peccary, lowland paca) across mainland Costa Rica at 30m resolution, scaled appropriately for each species. All four passed model diagnostics, with the final pruned models displaying ecologically coherent response curves: closed, stable canopy selected, roads and settlement avoided, and open, disturbed land scoring low. Bias from the two input datasets can still be seen, however, particularly in species responses to biophysical predictors such as climate and topographic gradients.
That data exists. Camera traps placed in forest interior, by people already working in these landscapes, hold exactly what the pipeline is missing. It is scattered across reserves, corridors, and private lands from north to south, held by the people doing the work on the ground.
NATIONAL SCALE CONNECTIVITY
At the national scale, this kind of model produces something a land cover map cannot: a picture of where the landscape is permeable to a species, not just where forest exists. These are the paths the landscape itself creates, where terrain, cover, and resistance combine to concentrate flow.
At this scale, the stakes are national and beyond. Isolated populations lose genetic diversity. Locally extinct areas cannot recolonise. Species that cannot disperse across a fragmented landscape are not conserved by the reserves that hold them - they are held captive, until they eventually disappear. The connections between populations are not a secondary concern. They are the mechanism by which species survive across generations.
Which parts of the country are maintaining those connections? Which corridors are doing the connective work that no single reserve can do alone? They are questions this model can start to answer. How well depends on the data behind it.
REGIONAL SCALE CONNECTIVITY
At the scale of an individual territory, the picture changes. The broad national routes resolve into something finer - a web of potential flow threading through fragments, forest edges, and gaps in the matrix. At this resolution, the landscape is no longer a set of patches connected by corridors. It is a continuous field, where some parts carry flow and others do not.
At this scale, the questions are local. Which fragments are connected to the wider network and which are isolated? Where does the web thin to a thread? Where is a single parcel, a single treeline, the difference between connectivity and its absence?
This is the output that makes conservation action legible on the ground. Which landowners hold the vital linkages. Where reforestation would reconnect what fragmentation has severed. Where intervention has leverage and where it does not. These are not questions that can be answered by looking at forest cover alone. They require knowing how the landscape functions for the species that depend on it - and that is exactly what this model is built to show.
What we are offering
We are looking for organisations willing to share camera trap detection records and accurate trap locations from their study areas.
In return, contributing organisations receive:
-
Full attribution in the form of data citation and acknowledgment in any publication arising from the analysis of their data
-
A habitat suitability map per contributed species, clipped to your project area
-
Two connectivity maps per contributed species, clipped to your project area, at territory and dispersal scale
-
Future updates as the model develops
Individual camera locations are never published. Only modelled surfaces are released. Your data remains yours.
If your organisation holds camera trap data from Costa Rica and would like to discuss contributing, get in touch below. We will come back to you directly.
LET´S WORK TOGTHER
We are building something that no single organisation can build alone. A national connectivity model is only as strong as the field data behind it, and that data exists, held by the people already doing the work across this country.
If your organisation holds camera trap records, works in biological corridors, or operates in landscapes where these species are present, we would like to hear from you. Every contribution strengthens the model and comes with outputs your organisation can use.
Contributing to:
