Altitudinal Migration and Seasonal Redistribution in Rainforest Bird Communities

Problem

Seasonal shifts in abundance along mountain gradients—known as altitudinal migration—remain one of the least quantified forms of animal movement. In tropical mountains, these redistributions are subtle and partial, involving only portions of populations moving uphill or downhill seasonally.
Goal: Detect and quantify altitudinal migration at the community scale using long-term bird monitoring data, accounting for imperfect detection and uneven sampling, and identify how species and communities dynamically “breathe” across elevation and season.

Approach

• Integrated 16 years of imperfect bird count data from the Australian Wet Tropics (2000–2016), spanning >100 rainforest sites across full elevational gradients.
• Built a hierarchical Bayesian N-mixture model to jointly estimate abundance and detection while isolating the seasonal signal of redistribution.
• Defined altitudinal migration as the season × elevation interaction, using ecologically centred season encoding (−0.5 = winter, +0.5 = summer) to directly interpret uphill vs. downhill movements.
• Pooled across mountains with species-level random slopes to maximise statistical power and capture consistent migration signals.
• Generated posterior predictions of abundance across continuous elevation bands to reconstruct system-wide patterns of seasonal change.
• Computed derived metrics (centroid shift, range shift, turnover) to quantify redistribution at both species and community levels.

Stack

• Bayesian hierarchical modelling: implemented in JAGS with structured priors, shrinkage, and multi-level random effects.
• Data engineering & post-processing: extensive data reshaping, full grid expansion, NA-filling, and prediction-block generation in R (tidyverse).
• Model validation: posterior predictive checks, convergence diagnostics, and cross-season model comparisons.
• Downstream analytics: abundance centroids, elevational range width, beta-diversity turnover (vegan & betapart).
• Visualisation: ggplot2 and patchwork pipelines for system-level “breathing” plots across elevation and season.

Results

• Most species exhibited predictable seasonal redistribution—moving uphill in summer and downhill in winter.
• These individual shifts aggregate into a striking community-level pattern, with total bird abundance peaking in lowlands during winter and uplands in summer.
• Species-specific centroid and range shifts revealed diverse strategies—from narrow-range specialists showing limited movement to generalists tracking resources more flexibly.

Impact

• Provides the first quantitative system-wide evidence of partial altitudinal migration in tropical rainforest birds.
• Establishes a generalisable Bayesian workflow for detecting redistribution in long-term monitoring datasets.
• Demonstrates how seasonal migration acts as a systemic process, reshaping community structure and potentially buffering biodiversity against climate change.

Links & Resources

• 📄 Manuscript: In prep for Global Ecology and Biogeography (Altitudinal migration and community breathing in the Australian Wet Tropics).
• 💻 Code repository: GitHub – Altitudinal Migration Workflow

Role

• Designed the model architecture and analytical workflow.
• Engineered the data processing and prediction pipelines.
• Led the Bayesian modeling, post-processing, and community-level synthesis.
• Wrote the full manuscript.