Best environmental predictors of breeding phenology differ with elevation in a common woodland bird species.

Temperatures in mountain areas are increasing at a higher rate than the Northern Hemisphere land average, but how fauna may respond, in particular in terms of phenology, remains poorly understood. The aim of this study was to assess how elevation could modify the relationships between climate variability (air temperature and snow melt-out date), the timing of plant phenology and egg-laying date of the coal tit (Periparus ater). We collected 9 years (2011-2019) of data on egg-laying date, spring air temperature, snow melt-out date, and larch budburst date at two elevations (~1,300 m and ~1,900 m asl) on a slope located in the Mont-Blanc Massif in the French Alps. We found that at low elevation, larch budburst date had a direct influence on egg-laying date, while at high-altitude snow melt-out date was the limiting factor. At both elevations, air temperature had a similar effect on egg-laying date, but was a poorer predictor than larch budburst or snowmelt date. Our results shed light on proximate drivers of breeding phenology responses to interannual climate variability in mountain areas and suggest that factors directly influencing species phenology vary at different elevations. Predicting the future responses of species in a climate change context will require testing the transferability of models and accounting for nonstationary relationships between environmental predictors and the timing of phenological events.

Ecological interactions and the Netflix problem.

Species interactions are a key component of ecosystems but we generally have an incomplete picture of who-eats-who in a given community. Different techniques have been devised to predict species interactions using theoretical models or abundances. Here, we explore the K nearest neighbour approach, with a special emphasis on recommendation, along with a supervised machine learning technique. Recommenders are algorithms developed for companies like Netflix to predict whether a customer will like a product given the preferences of similar customers. These machine learning techniques are well-suited to study binary ecological interactions since they focus on positive-only data. By removing a prey from a predator, we find that recommenders can guess the missing prey around 50% of the times on the first try, with up to 881 possibilities. Traits do not improve significantly the results for the K nearest neighbour, although a simple test with a supervised learning approach (random forests) show we can predict interactions with high accuracy using only three traits per species. This result shows that binary interactions can be predicted without regard to the ecological community given only three variables: body mass and two variables for the species' phylogeny. These techniques are complementary, as recommenders can predict interactions in the absence of traits, using only information about other species' interactions, while supervised learning algorithms such as random forests base their predictions on traits only but do not exploit other species' interactions. Further work should focus on developing custom similarity measures specialized for ecology to improve the KNN algorithms and using richer data to capture indirect relationships between species.