Traceable Claims in Ecology

In the world's longest-running tropical throughfall-exclusion experiment (Caxiuanã, >20 years), drought-primed trees lost transpiration roughly 2× slower during the 2023 El Niño, held higher resistance at peak drought (0.75 vs 0.46), and fully recovered within 143 days while controls reached only 77.5% of pre-drought transpiration — with smaller crowns (~26.5% less crown area) as the structural cost.

Limitations: single plot pair; resilience bootstrap robust in only 71.5% of resamples; baseline structural state unknown; crown-reduction causality partly inferential.

A 40-year Great Lakes community reassembly and 50-year species-distribution projections for the yellow-bellied glider agree directionally — warm-affiliated species expand, cold-affiliated species contract (glider habitat loss up to 100% by 2070) — with a single temperature variable (BIO8) as the strongest predictor.

Limitations: projections are model-based, not observed; observed shifts limited to one region.

Sub-adult bearded vultures range 90× wider than adults, which likely buffers inbreeding in fragmented populations — yet southern African birds still show F_IS = 0.24, suggesting natal dispersal alone cannot offset habitat-driven isolation.

Limitations: genetic structure findings are population-specific; the buffering mechanism is plausible but not directly demonstrated.

Dark diversity — the suitable-but-absent fraction of regional species pools — makes invisible extinction debt measurable, flagging ecosystems where greenness metrics (NDVI) still look healthy; dyked-marsh data show anthropogenic soil filters can restructure pools even when cover appears intact.

Limitations: site-pool estimation depends on co-occurrence modeling assumptions; flagship examples are regional.

Necropsy of 85 Cassin's auklets from the December 2014 British Columbia mass mortality event found a 40% microplastic ingestion rate (1.6 ± 6.8 pieces per bird), with the study arguing winter exposure exceeds summer exposure due to seasonal foraging and oceanographic shifts.

Limitations: carcasses conflate exposure with vulnerability — a 40% ingestion rate in dead birds cannot show plastic caused mortality; summer ingestion is modeled, not measured.

A stratified paired design — 20 Key Biodiversity Areas sampled with identical pan-trap and netting transects at boundary, interior, and >5 km off-road plots — directly tests whether the documented roadside concentration of records (77.1% within 2.5 km of roads; 68.4% of KBAs unsampled) reflects true ecological gradients or collector-access bias.

Limitations: proposed design, not completed; separating access artifact from road-effect ecology requires the paired effort standardization itself.

Across 598 arthropod species and 36,269 individuals from 72 German grasslands, landscape simplification — not local management intensity — filters traits and replaces specialists with generalized feeders, driving homogenization that high land-cover diversity can buffer.

Limitations: single sampling year; correlational RLQ inference; bounded to temperate managed landscapes.

Homogenization is identity-dependent: two landscapes with identical richness trajectories can diverge into opposite biodiversity outcomes, so richness-based metrics structurally miss the process eroding regional distinctiveness — a few cosmopolitan winners (carp, trout, gambusia) carry disproportionate homogenizing power.

Limitations: examples are US/North America-centric; relies on uncorrected third-party datasets.

Terrestrial invasive plants reduce native cover and diversity, homogenize communities, alter soil chemistry and microbes, and persist via decade-long seedbanks — yet the same invader can facilitate in one habitat and compete in another, and some high-productivity invaders may even increase soil carbon.

Limitations: impacts are not uniform; management cannot assume uniform harm without habitat-specific measurement.

A descriptive survey reports mangrove sediment carbon stocks of 13.87–132.65 Mg C ha⁻¹ across 11 North Minahasa sites (~645.19 Mg C ha⁻¹ total), using loss-on-ignition methods — locally useful baseline data for blue-carbon management.

Limitations: severe unit inconsistencies, no statistics or replication, and the density–carbon association is untested — the lowest-density site holds the third-highest carbon stock.

A survey of 1,248 macrozoobenthos individuals (18 families, 41 species) across 9 Indonesian mangrove stations found near-zero correlations (r = 0.01–0.33) between soil organic carbon (23.95–96.56 Mg/ha) and community indices — reported without significance testing.

Limitations: near-zero r-values more parsimoniously suggest sediment texture, salinity, and tidal exposure dominate; internal inconsistencies and single-month sampling limit generality.

Altitude proxies such as elevation and growing-degree-day sums are frequently conflated with the actual causal variables — bud-level microclimate and prior-season carbon reserves — so a single-year anthesis absence may reflect resource carryover rather than 2026 thermal sums.

Limitations: BGPT odds of true: 35% — no supplied data quantify GDD thresholds, canopy effects at 0.5 closure, or 2026 flowering status at any site.

An early-career researcher's 2026 preprint documents Betula nana across four Bucovina peat bogs with transparent limitation disclosure and explicit falsification criteria — rare in small descriptive surveys — though one 25 m² relevé per site cannot statistically test the soil-type hypothesis.

Limitations: N=4 unreplicated plots, single-season phenology window, no environmental covariates, and no peer review yet.