@article{fdi:010090236, title = {{M}achine and deep learning approaches to understand and predict habitat suitability for seabird breeding}, author = {{G}arcia-{Q}uintas, {A}. and {R}oy, {A}m{\'e}d{\'e}e and {B}arbraud, {C}. and {D}emarcq, {H}erv{\'e} and {D}enis, {D}. and {B}ertrand, {S}. {L}.}, editor = {}, language = {{ENG}}, abstract = {{T}he way animals select their breeding habitat may have great impacts on individual fitness. {T}his complex process depends on the integration of information on various environmental factors, over a wide range of spatiotemporal scales. {F}or seabirds, breeding habitat selection integrates both land and sea features over several spatial scales. {S}eabirds explore these features prior to breeding, assessing habitats' quality. {H}owever, the information-gathering and decision-making process by seabirds when choosing a breeding habitat remains poorly understood. {W}e compiled 49 historical records of larids colonies in {C}uba from 1980 to 2020. {T}hen, we predicted potentially suitable breeding sites for larids and assessed their breeding macrohabitat selection, using deep and machine learning algorithms respectively. {U}sing a convolutional neural network and {L}andsat satellite images we predicted the suitability for nesting of non-monitored sites of this archipelago. {F}urthermore, we assessed the relative contribution of 18 land- and marine-based environmental covariates describing macrohabitats at three spatial scales (i.e. 10, 50 and 100 km) using random forests. {C}onvolutional neural network exhibited good performance at training, validation and test ({F}1-scores >85%). {S}ites with higher habitat suitability (p > .75) covered 20.3% of the predicting area. {L}arids breeding macrohabitats were sites relatively close to main islands, featuring sparse vegetation cover and high chlorophyll-a concentration at sea in 50 and 100 km around colonies. {L}ower sea surface temperature at larger spatial scales was determinant to distinguish the breeding from non-breeding sites. {A} more comprehensive understanding of the seabird breeding macrohabitats selection can be reached from the complementary use of convolutional neural networks and random forest models. {O}ur analysis provides crucial knowledge in tropical regions that lack complete and regular monitoring of seabirds' breeding sites.}, keywords = {animal habitat modeling ; convolutional neural networks ; gulls and terns ; breeding ; remote sensing ; seabirds' colonies ; selection pattern ; {CUBA} ; {CARAIBE} ; {ATLANTIQUE} {ILES}}, booktitle = {}, journal = {{E}cology and {E}volution}, volume = {13}, numero = {9}, pages = {e10549 [13 ]}, ISSN = {2045-7758}, year = {2023}, DOI = {10.1002/ece3.10549}, URL = {https://www.documentation.ird.fr/hor/fdi:010090236}, }