Evidence-based. Community-led. Openly shared.

Phase 1 - How the network has changed Complete We modelled connectivity change between 2020 and 2025 across the eleven protected areas of Cóbano, Lepanto, and Paquera - parameterized across four wildlife guilds and four spatial scales, producing sixteen models per protected area per time period. For each protected area, candidate models related connectivity change to forest cover change, human population change, anthropogenic pixel density change, and a null model. Model selection followed information-theoretic criteria with fixed effects tested using bootstrapped inference to account for the residual distributions characteristic of connectivity data. For the six protected areas of the Cóbano network, connectivity declined in the majority of model parameterisations across all six sites. Cabo Blanco showed negative trends in all sixteen models, thirteen of which were statistically significant. The dominant predictor is human population change in the surrounding landscape - not forest cover change. This finding is consistent and robust across guilds and spatial scales. Full results, diagnostic plots, and model outputs are in preparation for peer-reviewed publication. We are happy to discuss findings, methodology, and raw outputs directly with scientific peers and prospective collaborators. Phase 2 - Species-specific habitat suitability modelling In progress The second phase derives habitat suitability models for each of our eight focal species across Costa Rica, using occurrence data from global biodiversity databases combined with environmental predictors covering climate, canopy structure, land use and land cover, vegetation productivity, and road networks. Predictor selection follows a multi-scale approach - different species perceive and respond to environmental variables at different spatial grains, and the analysis reflects that. Individual model outputs are combined into ensemble predictions weighted by performance, producing a single habitat suitability surface per species. These surfaces are then inverted to derive species-specific resistance surfaces - the inputs to Phase 2 connectivity modelling at 10 metre resolution, using both cumulative current flow and dispersal kernel approaches to characterise corridor function. Phase 3 - Field validation Planned Phase 3 integrates empirical field survey data with the modelled predictions from Phases 1 and 2. The goal is to assess whether the corridors identified as functional are actually being used by the species they were parameterised for, and whether habitat suitability predictions correspond to observed occupancy and density in the field. Where discrepancies exist, resistance surface parameters will be updated and corridor mapping refined - an iterative process rather than a single output. Property analytics The connectivity modelling programme produces more than corridor maps. Its outputs are translated into property-level intelligence - identifying which cadastral parcels correspond to the highest-priority locations for corridor conservation, what land use they are currently under, who owns them, and what conservation proposition is most viable for each landowner given their specific situation. This translation layer serves all three intervention strands. For payments for ecosystem services and reforestation, it identifies which ranching properties to approach first and what kind of scheme is most important for each. For urban planning and regulatory change, it produces a spatially explicit develop and no-develop zone classification that can be used directly as an evidence base for the regional planning board. For linear infrastructure, it maps which road segments and fence lines correspond to which land tenures, and what the access and engagement pathway looks like for each.

Community social research The social research programme works in two stages. The first involves in-depth semi-structured interviews with land owners in priority engagement areas. Interviews are open-ended and non-directive, designed to produce genuine accounts of how land use decisions are made, what motivates long-term stewardship, and what conditions would make a conservation scheme viable from their perspective. Interviews are recorded, transcribed, and analysed thematically to identify the motivations, constraints, and priorities that recur consistently across participants. The analysis produces the input parameters for the second stage - the specific scheme attributes that matter to these communities, the ranges participants consider acceptable, and the combinations most likely to achieve meaningful uptake. The second stage translates those findings into structured community workshops where participants work through choice scenarios - hypothetical scheme designs that vary systematically across the attributes identified as most important - and indicate which combinations they would genuinely consider accepting. The workshop outputs produce a quantitative account of how community members trade off different scheme attributes against each other, and what combinations of payment level, contract length, land use conditions, and monitoring requirements are most likely to work. This directly informs any PES scheme design that emerges from the process. Reforestation Before any planting decision is made on a new property, we assess the degree to which natural regeneration is already occurring - measuring seed rain and surveying seedling and sapling presence. Some properties, depending on their proximity to remnant forest, may have sufficient seed source populations dispersed through wind and animals for natural regeneration to proceed without further intervention. Where regeneration needs support, assisted natural regeneration is the first response - weed suppression, mulching, tending of naturally regenerating seedlings, enhanced seed dispersal, and protection from burning, grazing, and logging. Where soil condition or seed source distance makes natural regeneration insufficient, active planting is required. To identify the most effective planting approach for this landscape specifically, experimental trials are being conducted across repeated plots - testing variables including tree density, planting patch size and shape, species composition across pioneer, secondary, and climax species, and mature tree height. The trials are designed to produce locally grounded evidence rather than applying methods developed for different landscapes. Once restoration sites are agreed with landowners, permanent monitoring plots are established. Monitoring is stratified across three conditions: the intervention area, a control area where no action is taken, and a reference state based on an inventory of the nearest mature healthy forest. This protocol tracks both the effectiveness of the restoration method and the trajectory of recovery toward a defined ecological end state. Monitoring continues at minimum annually until canopy closure is reached.

The spatial evidence for this strand comes from the property analytics output of the connectivity modelling programme - described in full above. The planning-specific output is a ranked connectivity surface across the Cóbano district, identifying which land carries the highest biological value for corridor function. That surface can be interpreted as a spectrum - showing the relative connectivity contribution of every parcel in the landscape - or applied with a threshold to produce a binary develop and no-develop zone classification, depending on what the planning process requires. Translating that output into planning action requires more than a map. The connectivity surface is overlaid onto cadastral boundary data from the national registry, giving each high-priority zone a finca number and, through the national registry, an ownership record, converting a landscape-level finding into a specific, actionable evidence base: this parcel, this owner, this conservation proposition. The evidence is intended to be submitted to the regional planning board as both a technical report and a direct presentation - with an offer to assist in the collaborative design of the spatial planning framework itself rather than simply delivering findings and stepping back. The timeline of the planning process is not publicly known. What is known is that planning windows open and close on political cycles, and that connectivity science submitted after zoning decisions are made arrives too late to influence them.

Guild-specific road and fence analysis The infrastructure programme derives its target list from the connectivity modelling outputs - focusing specifically on road segments and fence lines identified as critical fragmentation points for one or more wildlife guilds. Prioritisation reflects both the severity of fragmentation at each location and the number of species affected - locations where a single modification would benefit the widest range of guilds simultaneously are ranked highest. A road that presents moderate resistance to a large terrestrial mammal may present near-absolute resistance to a ground-dwelling bird. The guild-specific analysis produces different ranked lists for different species groups, which are overlaid to identify convergence points where multiple guilds share the same critical fragmentation location. Fence permeability is assessed separately. Collared peccary movement is particularly sensitive to fencing - their social movement dynamics mean that a fence passable by a solitary animal may still function as a barrier to herd movement along an established route. Fence modification recommendations are designed around the movement ecology of the most affected species, not just physical passability. Where road and fence modifications are implemented, roadside reforestation forms part of the intervention - establishing native species cover that reduces effective barrier width, improves microclimate conditions for crossing species, and accelerates natural colonisation of the modified zone. Species selection draws from the experimental trials underway under the reforestation programme. Monitoring crossing structure use Crossing structures and fence modifications will be monitored for wildlife at structure entrances. Detection data will be analysed to assess which species are using each structure, at what frequency, and whether use patterns change over time as animals learn crossing locations. This data feeds into Phase 3 of the connectivity modelling programme - providing empirical evidence of whether modifications are delivering the permeability improvement the models predicted

Mammal surveys Mammal surveys use systematic camera trap arrays across the study area. Array design is species-informed - placement, spacing, and survey effort reflect the movement ecology and detection characteristics of the focal species, while maintaining sufficient spatial coverage to produce landscape-level estimates rather than point estimates at individual locations. Population density across all focal mammals will be estimated using a standardised distance-based camera trap approach that accounts for imperfect detection. To assess the reliability of that approach under Cóbano's specific field conditions, it will be validated against two independent methods applied to different species. For ocelot, individual identification from pelage markings allows the application of spatially explicit mark-recapture methods - the most precise population estimation technique available for a tropical felid without physical capture. For agouti, where animals may be present in sufficient numbers and group detectability is more tractable, conventional distance sampling with multiple covariate adjustment provides a second independent estimate. Comparing our primary approach against both gold standards - across two species with substantially different detectability characteristics - tests its reliability with a rigour that single-method validation cannot provide. Puma survey methodology will be determined on the basis of first-season detection rates. Where detections are sufficient, density estimation will be attempted. Where they are not, occupancy modelling will provide the most robust available population index. The decision will be made on the data. Bird surveys Bird surveys target our four focal species alongside the broader bird community across established transects within the study area. Survey protocols are designed to produce estimates of both occupancy and detection probability - accounting for the fact that not every bird present is detected on every survey occasion, and that detection probability varies with time of day, weather, observer, and season. This is particularly important for forest birds, where detection is primarily acoustic. Our survey design and analytical approach are being developed in consultation with leading researchers in avian population estimation to meet the highest available methodological standards for tropical forest bird monitoring. Survey data will be contributed to BirdLife DataZone and assessed against Key Biodiversity Area trigger thresholds for relevant species. Amphibians Amphibian monitoring is planned for a future phase. Methodology, species list, and survey protocols will be published here when the programme is established. Open data All occurrence data will be deposited to GBIF, the Wildlife Picture Index, and BirdLife DataZone as field seasons are completed and data quality checks are passed. We do not withhold data beyond the period necessary for quality assurance and, where applicable, peer review. For collaboration enquiries, methodology questions, data sharing requests, or access to pre-publication findings, contact us directly at: admin@caboblancoconectado.org