Stochastic simulations reveal few green wave surfing populations among spring migrating herbivorous waterfowl

Xin Wang,Lei Cao,Andrea Kölzsch,Richard Fuller,Oun-Kyong Moon,Zhenggang Xu,Carl Mitchell,Henk P Van Der Jeugd,David Cabot,Nyambayar Batbayar,Larry Griffin,Liding Chen,Jesper Madsen,Yunlin Zhao,Anthony D Fox,Ran Nathan

doi:10.1038/s41467-019-09971-8

Abstract

Tracking seasonally changing resources is regarded as a widespread proximate mechanism underpinning animal migration. Migrating herbivores, for example, are hypothesized to track seasonal foliage dynamics over large spatial scales. Previous investigations of this green wave hypothesis involved few species and limited geographical extent, and used conventional correlation that cannot disentangle alternative correlated effects. Here, we introduce stochastic simulations to test this hypothesis using 222 individual spring migration episodes of 14 populations of ten species of geese, swans and dabbling ducks throughout Europe, East Asia, and North America. We find that the green wave cannot be considered a ubiquitous driver of herbivorous waterfowl spring migration, as it explains observed migration patterns of only a few grazing populations in specific regions. We suggest that ecological barriers and particularly human disturbance likely constrain the capacity of herbivorous waterfowl to track the green wave in some regions, highlighting key challenges in conserving migratory birds.

Highlights

These studies used various vegetation greening metrics, such as the normalised difference vegetation index (NDVI)[16,17], green wave index (GWI, the scaled NDVI) and instantaneous rate of green-up (IRG, the acceleration of time-NDVI curve) calculated by fitting annual time-NDVI curves[10,11,18]
To assess the level of support for these predictions, one can first evaluate the robustness of migration–green wave associations using three methods: the Simple Conventional Correlation[12,20], the Correlation method evaluated by Stochastic Migrations[18] and the Metric Selection approach based on Stochastic Migrations (MSSMs) introduced in this study
Because the MSSM approach fulfilled both evaluation criteria, whereas both Simple Conventional Correlation and Correlation method evaluated by Stochastic Migrations did not, we focus on MSSM results in testing the green wave hypothesis and our three predictions

Summary

Introduction

These studies used various vegetation greening metrics, such as the normalised difference vegetation index (NDVI)[16,17], green wave index (GWI, the scaled NDVI) and instantaneous rate of green-up (IRG, the acceleration of time-NDVI curve) calculated by fitting annual time-NDVI curves[10,11,18]. The northward spring migration of northern hemisphere herbivores coincides with food availability (i.e., the green wave), avian spring migration could be associated with other environmental factors, such as day length and air temperature[20], which correlate with latitude. Such multiple associations cannot be disentangled using correlations alone. Testing whether the green wave determines spring migration requires going beyond correlations, to estimate the probability of detecting a match against the corresponding random (null) expectations of directional northward movement irrespective of the progress of the green wave. The hypothesis might not apply robustly even to grazers, as they can occasionally exploit more diverse food items such as non-leaf and even non-plant material[23,24], and migration timing can be affected by various other extrinsic (e.g., weather[25] and competition26) and internal factors (e.g., fat deposits[27])

Methods

Results

Conclusion