Monitoring marine mammals is of broad interest to governments and individuals around the globe. Very high-resolution (VHR) satellites hold the promise of reaching remote and challenging locations to fill gaps in our knowledge of marine mammal distribution. The time has come to create an operational platform that leverages the increased resolution of satellite imagery, proof-of-concept research, advances in cloud computing, and machine learning to monitor the world’s oceans. The Geospatial Artificial Intelligence for Animals (GAIA) initiative was formed to address this challenge with collaborative innovation from government agencies, academia, and the private sector. In this paper, we share lessons learned, challenges faced, and our vision for how VHR satellite imagery can enhance our understanding of cetacean distribution in the future.

Weddell seals (Leptonychotes weddellii) are important predators in the Southern Ocean and are among the best-studied pinnipeds on Earth, yet much still needs to be learned about their year-round movements and foraging behaviour. Using biologgers, we tagged 62 post-moult Weddell seals in McMurdo Sound and vicinity between 2010 and 2012. Generalized additive mixed models were used to (i) explain and predict the probability of seal presence and foraging behaviour from eight environmental variables, and (ii) examine foraging behaviour in relation to dive metrics. Foraging probability was highest in winter and lowest in summer, and foraging occurred mostly in the water column or just above the bottom; across all seasons, seals preferentially exploited the shallow banks and deeper troughs of the Ross Sea, the latter providing a pathway for Circumpolar Deep Water to flow onto the shelf. In addition, the probability of Weddell seal occurrence and foraging increased with increasing bathymetric slope and where water depth was typically less than 600 m. Although the probability of occurrence was higher closer to the shelf break, foraging was higher in areas closer to shore and over banks. This study highlights the importance of overwinter foraging for recouping body mass lost during the previous summer.

Baleen whales reliably produce stereotyped vocalizations, enabling their spatio-temporal distributions to be inferred from acoustic detections. Soundscape analysis provides an integrated approach whereby vocal species, such as baleen whales, are sampled holistically with other acoustic contributors to their environment. Acoustic elements that occur concurrently in space, time and/or frequency can indicate overlaps between free-ranging species and potential stressors. Such information can inform risk assessment framework models. Here, we demonstrate the utility of soundscape monitoring in central New Zealand, an area of high cetacean diversity where potential threats are poorly understood. Pygmy blue whale calls were abundant in the South Taranaki Bight (STB) throughout recording periods and were also detected near Kaikōura during autumn. Humpback, Antarctic blue and Antarctic minke whales were detected in winter and spring, during migration. Wind, rain, tidal and wave activity increased ambient sound levels in both deep- and shallow-water environments across a broad range of frequencies, including those used by baleen whales, and sound from shipping, seismic surveys and earthquakes overlapped in time, space and frequency with whale calls. The results highlight the feasibility of soundscape analysis to quantify and understand potential stressors to free-ranging species, which is essential for conservation and management decisions.

Changes in habitat availability and prey abundance are predicted to adversely influence survival and reproduction of wildlife in the Southern Ocean. Some populations of southern right whale (SRW;Eubalaena australis) are showing dramatic changes in habitat use. Surveys were undertaken in the austral winters of 2020 and 2021 at the key nursery and socialising ground for New Zealand SRWs: Port Ross, Auckland Islands, with 548 encounters and 599 skin biopsy samples collected. Data from these two surveys spanned peak periods of use and were used to test the hypothesis there have been shifts in the phenology, demographic composition and behaviour of SRWs using the Auckland Islands over the past three decades. The behavioural phenology and demographic composition of SRW resembles that observed in the 1990s. In contrast, the proportion of groups containing cow-calf pairs increased from 20% in the 1998 survey to 50% in 2020/21. These changes are consistent with a growing population undergoing strong recruitment, not limited by food resources. Continued use of Port Ross by all SRW demographic classes confirms this as key habitat for SRW in New Zealand waters, and we support increased enforcement of existing management measures to reduce whale-vessel interactions in this remote subantarctic archipelago.

Few studies have assessed the influence of data quality on the predicted probability of occurrence and preferred habitat of marine predators. We compared results from four species distribution models (SDMs) for four southern-hemisphere albatross species, Buller’s (Thalassarche bulleri), Campbell (T. impavida), grey-headed (T. chrysostoma), and white-capped (T. steadi), based on datasets of differing quality, ranging from no location data to twice-daily locations of individual birds collected by geolocation devices. Two relative environmental suitability (RES) models were fit using minimum and maximum preferred and absolute values for each environmental variable based on (1) monthly 50% kernel density contours and background environmental data, and (2) primary literature or expert opinion. Additionally, two boosted regression tree (BRT) models were fit using (1) opportunistic sightings data, and (2) geolocation data from bird-borne electronic tags. Using model-specific threshold values, habitat was quantified for each species and model. Model variables included distance from land, bathymetry, sea surface temperature, and chlorophyll-a concentration. Results from both RES models and the BRT model fit with opportunistic sightings were compared to those from the BRT model fit using geolocation data to assess the influence of data quality on predicted occupancy and habitat. For all species, BRT models outperformed RES models. BRT models offer a predictive advantage over RES models by being able to identify relevant variables, incorporate environmental interactions, and provide spatially explicit estimates of model uncertainty. RES models resulted in larger, less refined areas of predicted habitat for all species. Our study highlights the importance of data quality in predicting the distribution and habitat of albatrosses and emphasises the need to consider the pros and cons associated with different levels of data quality when using SDMs to inform management decisions. Furthermore, we examine the overlap in preferred habitat predicted by each SDM with fishing effort. We discuss the influence of data quality on predicting the wide-scale distributions of pelagic seabirds and how these impacts could result in different protection measures.