Seasonal Spatial Dynamics of Butterfly Migration

Abstract

Understanding the seasonal movements of migratory species underpins ecological studies. Nearly 600 butterfly species show migratory behaviour, yet the spatial pattern of these migrations is poorly understood. We developed climatic niche models for 405 migratory butterfly species globally to estimate patterns of seasonal movement and the distribution of seasonal habitat suitability. We discover strong seasonal variation in habitat suitability for most of the migratory butterflies with >75% of pixels showing seasonal change in predicted occupancy for 85% of species. In contrast to bird migrations, the greatest rate of seasonal switching occurs in the tropics. Several species showed extreme range fluctuations, exceeding 10-fold for 53 species (13%) and more than 100-fold for nine species (2%); such species may be at elevated extinction risk. Our results can be used to search for the ecological processes that underpin migration in insects, as well as to design conservation interventions for declining migratory insects.

Migration in butterflies: a global overview

“…Future studies should obtain regional records, document movement patterns, and conduct experimental work on migration in butterflies to reveal hotspots and their vulnerability to extinction, thus providing invaluable data for butterfly conservation. A fruitful avenue would be to focus on mapping the distributions of migratory butterflies at different times of year (Chowdhury et al ., 2020).…”

Local adaptation and migratory habits balance spatial‐genetic structure between continental and insular chestnut tiger butterflies in East Asia

“…Some British butterflies also use a sun compass during migration (Baker, 1968). Temperature seasonality in temperate regions and seasonal precipitation in the tropics may play key roles in cueing butterfly migration (Chowdhury et al, 2021). Our understanding of the migratory behaviour of ssp.…”

Protected areas fail to cover the full annual cycle of migratory butterflies

“…Further, we removed records that are incompatible with the countries given, or otherwise doubtful using a range of error-checking methods using the CoordinateCleaner package 32 in R 33 . We split years into four seasons (3 months each; Dec-Feb, Mar-May, Jun-Aug, Sep-Nov), and divided the entire world into 11 zoogeographic regions: Afrotropical, Australian, Madagascan, Nearctic, Neotropical, Oceania, Oriental, Palaearctic, Panamanian, Saharo-Arabian and Sino-Japanese 25,34 . We developed the ecological niche model at 4.65*4.65 km 2 resolution using the ENMeval package 35 in R. The ENMeval package can spatially group available geospatial records in four bins, so we removed species with < 4 unique occurrence records in species-season-region combinations.…”

“…We validated each model with 10-fold cross-validation, tune it using three feature class combinations ('L', 'LQ' and 'LQP') and four candidate regularisation multipliers (0.5, 1, 1.5, 2) using the ENMeval package 35 in R. We chose the best tting model, as determined by the AUC (the Area Under the Curve). We used the 'checkerboard2' evaluation method to handle model overin ation resulting from biased sampling 25,35 , which partitions both geospatial records and background points into evaluation bins to reduce spatial autocorrelation between points in the testing and training bins 35 . We only considered species with four seasonal maps.…”

“…Of the 90 Least Concern migratory butter y species, the population trend is unknown for 81 species, stable for 5 species, decreasing for 3 species, and increasing for 1 species. A recent study estimated that over 13% of all migratory butter ies are at risk of extinction (seasonal range size uctuation exceeded 10-fold) 25 and there are six species with a seasonal range uctuation exceeding 100-fold. Further investigations of these species are needed to determine if any are at particularly acute conservation risk, including considerations of their year-round life stage cycles.…”