Megafauna extinctions produce idiosyncratic Anthropocene assemblages

Abstract

The "trophic downgrading of planet Earth" refers to the systematic decline of the world's largest vertebrates. However, our understanding of why megafauna extinction risk varies through time and the importance of site- or species-specific factors remain unclear. Here, we unravel the unexpected variability in remaining terrestrial megafauna assemblages across 10 Southeast Asian tropical forests. Consistent with global trends, every landscape experienced Holocene and/or Anthropocene megafauna extirpations, and the four most disturbed landscapes experienced 2.5 times more extirpations than the six least disturbed landscapes. However, there were no consistent size- or guild-related trends, no two tropical forests had identical assemblages, and the abundance of four species showed positive relationships with forest degradation and humans. Our results suggest that the region's megafauna assemblages are the product of a convoluted geoclimatic legacy interacting with modern disturbances and that some megafauna may persist in degraded tropical forests near settlements with sufficient poaching controls.

Fig. 1.. Idiosyncratic megafauna assemblages are not systematically shaped by disturbances.

(A) Highly variable Anthropocene megafauna assemblages across our 10 study landscapes, and each circle represents the Forest Landscape Integrity Index (FLII) (26) in a 10-km radius around the center of each landscape. In (B), the base layer of the map depicts the contemporary habitat quality using the FLII, binned into high (FLII scores 9 to 10), medium (6 to 9), and poor (<6) integrity forest. The tan coloration shows land area that was exposed 50 m above current water levels at the onset of the Holocene [~12,000 years ago (68)]. See Table 1 for more information about each landscape and table S3 for all megafauna detected and abundance relative to each landscape. (C and D) Variation in FLII and HFP (38) for each study site. The study landscapes with new camera trapping on all panels are numbered as follows: 1, Khao Yai National Park; 2, Khao Ban That Wildlife Sanctuary; 3, Ulu Muda Forest Reserve; 4, Pasoh Forest Reserve; 5, Singapore; 6, Gunung Leuser National Park; 7, Kerinci Seblat National Park; 8, Bukit Barisan Selatan National Park; 9, Lambir Hills National Park; and 10, DVCA.

Fig. 2.. Highly variable megafauna extirpation dynamics among species and sites in Southeast Asia.

(A) The Holocene and Anthropocene extirpations of 14 megafauna across 10 tropical forest landscapes (table S3). The relative abundance of extant species is shown by the size of points as determined from hierarchical abundance modeling. High abundance corresponds to abundance >75% quartile of each species’ relative abundance estimates, medium corresponds to relative abundance estimates between 25 and 75% quartiles, and low is <25% quartile. Bar chart (B) shows the number of landscapes each species was extirpated from, and bar chart (C) shows the number of species extirpated from each landscape. The 14 study species are grouped as large carnivores first followed by megaherbivores. Within each guild, species are listed from largest to smallest on the basis of average adult body size (Table 2). Landscapes are organized from left to right by increasing FLII scores (Table 1). Species listed in descending order are as follows: tiger, sun bear, leopard, dhole, clouded leopard, Asian elephant, Sumatran rhino, gaur, banteng, tapir, sambar, bearded pig, mainland serow, and wild boar (scientific names, descriptions, and relative abundance estimates in Table 2 and table S3).

Fig. 3.. Species traits (guild and body size) do not explain extirpation patterns.

We conducted Welch two-sample t tests and found no significant differences between the relative number of extirpations between large carnivores and megaherbivores (A) or between different sized species within each guild (B). The maximum number of Holocene extirpations possible was 10 for species whose range covered the whole region and fewer for species with smaller distributions, including leopards, dhole, gaur, banteng, tapir, bearded pigs, mainland serow, and wild boar.

Fig. 4.. Species-specific responses to habitat degradation and humans.

The response variable (y axis) is the predicted abundance from hierarchical abundance modeling in relation to covariates (x axis), with each panel showing an important variable from each species’ top model based on AICc model selection. (A) Six negative relationships between megafauna abundance and forest degradation or human pressure that are consistent with globally emergent trends. (B) Four relationships that contradict global trends and suggest positive relationships with forest degradation or human pressure. (C) Five species that had insufficient independent detections (≤30) to implement hierarchical abundance models. Table S5 provides the full model descriptions.