Rapid advances in digital imaging technology offer efficient and cost‐effective methods for measuring seabird abundance, breeding success, phenology, survival and diet. These methods can facilitate understanding of long‐term population trends, and the design and implementation of successful conservation strategies. This paper reviews the suitability of satellites, manned aircraft, unmanned aerial vehicles (UAVs), and fixed‐position, handheld and animal‐borne cameras for recording digital photographs and videos used to measure seabird demographic and behavioural parameters. It considers the disturbance impacts, accuracy of results obtained, cost‐effectiveness and scale of monitoring possible compared with ‘traditional’ fieldworker methods. Given the ease of collecting large amounts of imagery, image processing is an important step in realizing the potential of this technology. The effectiveness of manual, semi‐automated and automated image processing is also reviewed. Satellites, manned aircraft and UAVs have most commonly been used for population counts. Spatial resolution is lowest in satellites, limiting monitoring to large species and those with obvious signs of presence, such as penguins. Conversely, UAVs have the highest spatial resolution, which has allowed fine‐scale measurements of foraging behaviour. Time‐lapse cameras are more cost‐effective for collecting time‐series data such as breeding success and phenology, as human visits are only required infrequently for maintenance. However, the colony of interest must be observable from a single vantage point. Handheld, animal‐borne and motion‐triggered cameras have fewer cost‐effective uses but have provided information on seabird diet, foraging behaviour and nest predation. The last of these has been important for understanding the impact of invasive mammals on seabird breeding success. Advances in automated image analysis are increasing the suitability of digital photography and videography to facilitate and/or replace traditional seabird monitoring methods. Machine‐learning algorithms, such as Pengbot, have allowed rapid identification of birds, although training requires thousands of pre‐annotated photographs. Digital imaging has considerable potential in seabird monitoring, provided that appropriate choices are available for both image capture technology and image processing. These technologies offer opportunities to collect data in remote locations and increase the number of sites monitored. The potential to include such solutions in seabird monitoring and research will develop as the technology evolves, which will be of benefit given funding challenges in monitoring and conservation.

The consequences of climate change for biogeographic range dynamics depend on the spatial scales at which climate influences focal species directly and indirectly via biotic interactions. An overlooked question concerns the extent to which microclimates modify specialist biotic interactions, with emergent properties for communities and range dynamics. Here, we use an in-field experiment to assess egg-laying behaviour of a range-expanding herbivore across a range of natural microclimatic conditions. We show that variation in microclimate, resource condition and individual fecundity can generate differences in egg-laying rates of almost two orders of magnitude in an exemplar species, the brown argus butterfly ( Aricia agestis ). This within-site variation in fecundity dwarfs variation resulting from differences in average ambient temperatures among populations. Although higher temperatures did not reduce female selection for host plants in good condition, the thermal sensitivities of egg-laying behaviours have the potential to accelerate climate-driven range expansion by increasing egg-laying encounters with novel hosts in increasingly suitable microclimates. Understanding the sensitivity of specialist biotic interactions to microclimatic variation is, therefore, critical to predict the outcomes of climate change across species' geographical ranges, and the resilience of ecological communities.

doi: bioRxiv preprint python package (https://deepforest.readthedocs.io/). The data and models have been uploaded to 23

Advances in artificial intelligence for image processing hold great promise for increasing the scales at which ecological systems can be studied. The distribution and behavior of individuals is central to ecology, and computer vision using deep neural networks can learn to detect individual objects in imagery. However, developing computer vision for ecological monitoring is challenging because it needs large amounts of human-labeled training data, requires advanced technical expertise and computational infrastructure, and is prone to overfitting. This limits application across space and time. One solution is developing generalized models that can be applied across species and ecosystems. Using over 250,000 annotations from 13 projects from around the world, we develop a general bird detection model that achieves over 65% recall and 50% precision on novel aerial data without any local training despite differences in species, habitat, and imaging methodology. Fine-tuning this model with only 1000 local annotations increases these values to an average of 84% recall and 69% precision by building on the general features learned from other data sources. Retraining from the general model improves local predictions even when moderately large annotation sets are available and makes model training faster and more stable. Our results demonstrate that general models for detecting broad classes of organisms using airborne imagery are achievable. These models can reduce the effort, expertise, and computational resources necessary for automating the detection of individual organisms across large scales, helping to transform the scale of data collection in ecology and the questions that can be addressed.

No abstract