Sci Rep. 2026 Jun 16. doi: 10.1038/s41598-026-58296-2. Online ahead of print.
ABSTRACT
Global biodiversity monitoring increasingly relies on open-access community science data, but these opportunistic records harbor complex biases that can severely distort macroecological inference. Here, we disentangle how human behavior, infrastructure, and taxonomy interact to shape perceived biodiversity patterns across the two distinct biogeographic regions of Malaysia. Analyzing 336,042 research-grade iNaturalist records, we quantified observer inequality and taxonomic disproportionality. We estimated true species richness (Chao2) to map spatial inventory completeness (median = 33.3%) and employed Zero-Inflated Negative Binomial GLMMs and effort-corrected Generalized Additive Models (GAMs) to test the effects of topography, accessibility, and observer classification. We demonstrate extreme observer inequality (Gini = 0.854), with data collection heavily anchored to urban centers. Crucially, a significant three-way interaction revealed that dedicated “Power Users” successfully penetrate roadless interiors in Peninsular Malaysia, whereas casual observers remain strictly road-bound. Taxonomically, the data exhibits a severe charismatic skew, massively over-representing Aves and Reptilia while under-sampling foundational hyper-diverse clades like Insecta and Fungi. Furthermore, explicitly modeling non-linear sampling effort (user-days) rendered the effects of elevation and terrain ruggedness statistically non-significant. This demonstrates that perceived biodiversity deficits in rugged, high-elevation terrains are anthropogenic artifacts of human inaccessibility rather than true ecological absences. To meet global conservation targets, state funding and structured monitoring should complement opportunistic data by actively targeting these remote, under-sampled geographic and taxonomic shortfalls.
PMID:42298119 | DOI:10.1038/s41598-026-58296-2
