Aim: Identifying highly vulnerable regions to climate change is increasingly incorporated in marine management planning given the redistribution of species in the three-dimensional space following temperature changes. Here, we developed a spatially explicit vulnerability framework incorporating sensitivity, exposure, and adaptive capacity of species living within the Azores Marine Park (AMP).

Location: Azores Marine Park, North Atlantic.

Methods: We quantified species sensitivity to temperature changes based on thermal affinity and georeferenced their distribution with quality-controlled records from various data compilators. To assess their degree of exposure we extracted historical (1995 - 2020) environmental temperature across latitudes, longitudes, and depths and calculated mean interannual temperature change (i.e., increase or decrease) and temperature variability. We estimated the adaptive capacity of species based on two traits representing the ability to relocate at adult and early life-stage (i.e., “Motility” and “Developmental Mechanism”) using the FUN Azores trait database. To map the results, we pooled the species into 3D-regions representing cubes of 0.25º x 0.25º resolution and 50 and 500 m depth bands at shallow and deep areas, respectively. We then assigned a sensitivity, exposure, and adaptive capacity score to each spatial unit based on species scores and occurrences, combined them to a final vulnerability class (i.e., “Highly Vulnerable”, “Advisable Monitoring”, “Expected Relocation”, and “Least Concern”).

Results: “Highly Vulnerable” and “Advisable Monitoring” regions exist only in the benthic environment across various MPAs and depths. The increased mobility of benthopelagic and pelagic species explain the absence of the most vulnerable categories in these environments.

Main conclusions: We advise strong conservations measures in “Highly Vulnerable” areas and monitoring of environmental variables and populations in areas classified as “Advisable Monitoring” and “Expected Relocation”, respectively. Our results suggests that the Azores deep-sea benthos is highly vulnerable to both warming and wide temperature variations.

Spatio-temporal resolution

We created a three-dimensional grid with fixed coordinates (0.25 x 0.25 degrees) and depths where species data and environmental data were grouped together. We created specific depth bands for every 50 m. above 200 m. (i.e., 0 – 50 m., 50 – 100 m., 100 – 150 m., 150 – 200 m.) and for every 500 m. below 200 m. down to 3500 m. (i.e., 200 – 500 m., 500 – 1000 m., 1000 – 1500 m., 1500 – 2000 m., 2000 – 2500 m., 2500 – 3000 m., 3000 – 3500 m.). We obtained a community baseline by pooling together data from all recorded years (1905 – 2018), because data were scarce and unbalanced through time.

Temperature exposure based on historical data

We extracted historical temperature (i.e., 1995 - 2020) datapoints in the 4D domain (i.e., Latitude, Longitude, Depth, and Time) from the Copernicus data portal⁷ using Python. We imported the net.cdf files into R using the package “ncdf4”. We assigned each datapoint into a cube of the original three-dimensional mesh according to its four dimensions and calculated the mean interannual temperature change per cube to inform magnitude of change and its associated standard deviation to estimate temperature variation within each 3D-region across years. We classified each region into general increasing temperature if mean interannual change per region was positive and into high or low variability in relation to 50^th percentile of temperature variability across years.

Final vulnerability score

Vulnerability of each 3D-region is determined by the region’s exposure to temperature change and the sensitivity and adaptive capacity of its component species. A final vulnerability category for each 3D-region was based on the combination of the three elements. Therefore, exposed regions where most species are sensitive and lack adaptive capacity classify into the “Highly Vulnerable” category. Regions where most species are sensitive and have a low adaptive capacity but are not exposed classify as “Advisable Monitoring”. We classified regions as “Expected Relocation”, if they are exposed and most of its inhabiting species are sensitive but adaptive. Regions of “Least Concern” were reserved for cases where species were not sensitive at all. We applied the same procedure to assess vulnerability to temperature change (i.e., increase tendency) and temperature variation producing two scores for each 3D-region.

Biogeography

We mapped the final derived vulnerability 3D-region classifications in R using the package “terra” (Hijmans, 2023) and “ggplot2”. Geographic coordinate system used for all the mappings correspond to WGS 84 (World Geodetic System 1984 - EPSG 4326) and for map projections WGS 84/ UTM Zone 26N (EPSG:32626 code; https://epsg.io/). For each environment (i.e., benthic, benthopelagic, and pelagic), we mapped vulnerability scores to temperature increases and vulnerability scores to temperature variation, resulting in six vulnerability maps.

Correlations between temperature and species traits

To assess the relationship between species characteristics and species temperature tolerances (i.e., temperature maximum, minimum, and range) we ran spearman correlations with the function cor from the “stats” package in R programming software. We selected traits that define a species ecological niche form the “FUN Azores” trait database (i.e., “Maximum Body Size”, “Motility”, “Environmental Position”, “Trophic Position”) or new traits adapted from the “FUN Azores” trait database (i.e., “Depth Ranges”, “Substrate Type”).