Traceable Claims in Ecology

 Quick Analysis Plan Copied

 Long Analysis Plan

Evidence Base From Supplied Sources

Three of the supplied papers directly address the query. In the Northern Great Lakes region, 14,076 museum and field records show southern species expanding and northern species contracting since 1981: white-footed mice (Peromyscus leucopus) rose from 38.3% to 77.7% of captures while woodland deer mice (P. maniculatus gracilis) fell from 28.0% to 5.9%, and opossums (Didelphis virginiana) appeared from zero to 35.5% — an observed, warming-consistent community reassembly attributed to climate change, though historical collection records carry collector-bias risks Great Lakes small mammal community shift.

Forward-projected SDMs predict the yellow-bellied glider (Petaurus australis) loses 62–78% of distribution under low warming by 2050–2070 and 100% under high warming, based on 57 Wet Tropics and 428 SE Queensland occurrence records with MaxEnt (AUC 0.991 on BIO8, mean temperature threshold ~22 °C vs 26.5 °C at random sites) Yellow-bellied glider range contraction projections.

A complementary paleo-perspective uses sedimentary ancient DNA (116 samples, 4 marine cores) to reconstruct 12,000 years of marine mammal distribution shifts in Northern Greenland, showing species responded to Holocene deglaciation earlier than fossil records suggest sedimentary aDNA reveals Holocene marine mammal shifts.

Bioinformatics/Coding Analysis Plan

Step 1 — Data acquisition: harvest occurrence records (GBIF, MaNIS, museum databases) with coordinate cleaning in Python (pandas + geopy), matching the glider study's 485-record and Great Lakes study's 14,076-record scales; download WorldClim bioclimatic layers, prioritizing BIO8-type temperature variables identified as strong predictors.

Step 2 — SDM pipeline: replicate and extend the MaxEnt framework in Python (scikit-learn Maxent replacement or R via rleaflet/sdm): fit current and future projections (2050/2070, low/high emissions), report AUC on spatially blocked cross-validation (the reported AUC 0.991 may reflect overfitting to a small Wet Tropics sample of 57 records — a key limitation), and quantify core-habitat loss as % distribution lost per scenario.

Step 3 — Community shift analysis: bin record sets by pre/post warming thresholds (as the 1981 split), compute relative-abundance change per species, and test latitude/temperature associations with GAMs (statsmodels/pyGAM), mirroring the Great Lakes approach while correcting for collector bias via effort-modeling.

Step 4 — Validation against deep time: use sedimentary aDNA presence/absence time series as an independent empirical benchmark for SDM skill on past climate analogs.

Limitations/blindspots: SDMs omit biotic interactions, dispersal, and habitat quality (acknowledged in the glider study); historical record biases (misidentification, uneven sampling) affect the observed shifts; both projected and observed patterns are correlational with warming, so attribution requires experimental or multi-region replication.