Despite their interest and threat status, the number of whales in world’s oceans remains highly uncertain. Whales detection is normally carried out from costly sighting surveys, acoustic surveys or through high-resolution images. Since deep convolutional neural networks (CNNs) are achieving great performance in several computer vision tasks, here we propose a robust and generalizable CNN-based system for automatically detecting and counting whales in satellite and aerial images based on open data and tools. In particular, we designed a two-step whale counting approach, where the first CNN finds the input images with whale presence, and the second CNN locates and counts each whale in those images. A test of the system on Google Earth images in ten global whale-watching hotspots achieved a performance (F1-measure) of 81% in detecting and 94% in counting whales. Combining these two steps increased accuracy by 36% compared to a baseline detection model alone. Applying this cost-effective method worldwide could contribute to the assessment of whale populations to guide conservation actions. Free and global access to high-resolution imagery for conservation purposes would boost this process.
The most recent climate change projections show a global increase in temperatures along with precipitation changes throughout the 21st century. However, regional projections do not always match global projections and species with global distributions may exhibit varying regional susceptibility to climate change. Here we show the effect of local climatic conditions on the hatchling output of leatherback turtles (Dermochelys coriacea) at four nesting sites encompassing the Pacific, Atlantic and Indian Oceans. We found a heterogeneous effect of climate. Hatchling output increased with long-term precipitation in areas with dry climatic conditions (Playa Grande, Pacific Ocean and Sandy Point, Caribbean Sea), but the effect varied in areas where precipitation was high (Pacuare, Caribbean Sea) and was not detected at the temperate site (Maputaland, Indian Ocean). High air temperature reduced hatchling output only at the area experiencing seasonal droughts (Playa Grande). Climatic projections showed a drastic increase in air temperature and a mild decrease in precipitation at all sites by 2100. The most unfavorable conditions were projected for Sandy Point where hatching success has already declined over time along with precipitation levels. The heterogeneous effect of climate may lead to local extinctions of leatherback turtles in some areas but survival in others by 2100.
Sea turtles are vulnerable to climate change since their reproductive output is influenced by incubating temperatures, with warmer temperatures causing lower hatching success and increased feminization of embryos. Their ability to cope with projected increases in ambient temperatures will depend on their capacity to adapt to shifts in climatic regimes. Here, we assessed the extent to which phenological shifts could mitigate impacts from increases in ambient temperatures (from 1.5 to 3°C in air temperatures and from 1.4 to 2.3°C in sea surface temperatures by 2100 at our sites) on four species of sea turtles, under a “middle of the road” scenario (SSP2‐4.5). Sand temperatures at sea turtle nesting sites are projected to increase from 0.58 to 4.17°C by 2100 and expected shifts in nesting of 26–43 days earlier will not be sufficient to maintain current incubation temperatures at 7 (29%) of our sites, hatching success rates at 10 (42%) of our sites, with current trends in hatchling sex ratio being able to be maintained at half of the sites. We also calculated the phenological shifts that would be required (both backward for an earlier shift in nesting and forward for a later shift) to keep up with present‐day incubation temperatures, hatching success rates, and sex ratios. The required shifts backward in nesting for incubation temperatures ranged from −20 to −191 days, whereas the required shifts forward ranged from +54 to +180 days. However, for half of the sites, no matter the shift the median incubation temperature will always be warmer than the 75th percentile of current ranges. Given that phenological shifts will not be able to ameliorate predicted changes in temperature, hatching success and sex ratio at most sites, turtles may need to use other adaptive responses and/or there is the need to enhance sea turtle resilience to climate warming.