Non-random defaunation across China’s protected area network revealed by multidimensional dark diversity

Technical Summary: Non-random Defaunation Across China's Protected Area Network Revealed by Multidimensional Dark Diversity

Problem Statement
Current assessments of protected area (PA) effectiveness predominantly rely on observed species richness. This approach conflates true conservation success with baseline habitat quality, obscuring "biodiversity debt"—the gap between realized communities and their ecological potential. By focusing solely on what remains, traditional metrics fail to capture the extent of community incompleteness, particularly the loss of evolutionary history and functional roles that may persist even when species counts appear stable. Furthermore, existing dark diversity studies have largely been limited to taxonomic dimensions, lacking integration of phylogenetic and functional perspectives to determine if biodiversity losses are random or structured by specific traits and evolutionary lineages.

Methodology
The study synthesized a massive dataset comprising 7.4 million camera-trap days across 199 terrestrial nature reserves in China, covering medium- and large-sized mammals (>1 kg). The analysis focused on 105 species from 8 orders and 22 families.

1. Dark Diversity Estimation: The authors constructed reserve-specific species pools by filtering IUCN range maps against habitat affinities, elevational limits, and historical distribution records (CESIS) to account for range contractions. They calculated three dimensions of dark diversity:

   • Taxonomic (DDadj): Using a hypergeometric co-occurrence approach to estimate the probability of absent species belonging to the dark diversity pool.
   • Phylogenetic (EPD): Adapting Faith's Phylogenetic Diversity framework to calculate the expected evolutionary history (in Myr) missing from each reserve, utilizing a 5,911-species mammal phylogeny and accounting for tree uncertainty via posterior sampling.
   • Functional (FDI): Constructing a Gower-distance trait space based on body mass, diet, activity patterns, and IUCN threat status. The Functional Deficit Index (FDI) quantified the proportion of potential trait space currently unfilled.

2. Driver Analysis:

   • Species-Site Unified Model: A Bayesian hierarchical model partitioned dark diversity affinity into species-level traits (e.g., threat status, diet, activity) and site-level covariates (e.g., reserve area, sampling effort, governance level, environmental variables).
   • Spatial GAMs: Generalized Additive Models with spatial smoothers were used to identify non-linear environmental drivers for phylogenetic and functional dark diversity.

3. Non-randomness Testing: Regional null-model analyses (9,999 permutations) were conducted to test whether observed dark diversity in specific zoogeographic regions (e.g., South China, Qinghai-Tibet) significantly deviated from random expectations, distinguishing structured losses from stochastic absences.

Key Results

• Pervasive Incompleteness: On average, protected areas were missing 56% of the mammal species expected under their environmental conditions (mean DDadj = 0.56). This indicates that most reserves are depauperate remnants of historically richer communities.
• Multidimensional Deficits:
  • Phylogenetic: Reserves were missing an average of 468 million years of evolutionary history.
  • Functional: The average functional deficit index was 0.22, indicating significant gaps in ecological trait space.
• Decoupling from Richness: Observed species richness was strongly negatively correlated with taxonomic and functional dark diversity but uncorrelated with phylogenetic dark diversity. This suggests that species counts alone cannot reveal the loss of evolutionary history.
• Non-random, Trait-Selective Defaunation:
  • Species Traits: Missing species were disproportionately threatened, dietary specialists (herbivores and carnivores), and cathemeral/crepuscular (active during twilight/day). Omnivores and strictly nocturnal/diurnal species were more likely to be retained.
  • Drivers: Taxonomic dark diversity was driven by reserve area, sampling effort, and governance level. Phylogenetic dark diversity was driven by elevation, topographic heterogeneity, precipitation, and productivity. Functional dark diversity was linked to historical defaunation interacting with regional trait-space constraints.
• Biogeographic Patterns: A null-model analysis revealed that South China exhibited significantly elevated taxonomic, phylogenetic, and functional dark diversity relative to null expectations, indicating severe, structured biodiversity loss in this region. Conversely, regions like Northeast China and the Qinghai-Tibet Plateau showed lower-than-expected deficits.

Key Contributions

• Methodological Integration: The study provides the first simultaneous assessment of taxonomic, phylogenetic, and functional dark diversity at a national scale, demonstrating that these dimensions capture complementary, non-redundant conservation signals.
• Revealing Hidden Debt: It quantifies the "conservation debt" in China's PA network, showing that many reserves with moderate observed richness are functionally and evolutionarily incomplete.
• Mechanistic Insight: By linking specific traits (threat status, diet, activity) to absence, the study identifies a coherent syndrome of vulnerability, distinguishing between stochastic absences and deterministic, trait-selective defaunation driven by anthropogenic pressures.
• Biogeographic Specificity: The research identifies South China as a critical zone of non-random biodiversity loss, where high potential diversity contrasts sharply with realized community incompleteness.

Significance and Claims
The paper argues that relying on observed species richness substantially underestimates biodiversity erosion. By integrating multidimensional dark diversity, conservation assessments can shift from measuring "what remains" to identifying "what is missing but potentially recoverable." The authors claim this framework is essential for:

1. Prioritizing Restoration: Identifying reserves with high hidden deficits (particularly in South China) for active habitat recovery and connectivity enhancement.
2. Targeting Species Recovery: Focusing efforts on trait groups most likely to be absent (threatened specialists, large herbivores/carnivores).
3. Informing Policy: Providing a data-driven baseline for the Kunming–Montreal Global Biodiversity Framework to move beyond simple area protection toward restoring ecological integrity and evolutionary potential.

The study concludes that China's protected areas often represent incomplete remnants of historically richer communities and that addressing these multidimensional deficits is critical for advancing biodiversity recovery.