Species co‐occurrence networks show reptile community reorganization under agricultural transformation

Abstract

Agricultural transformation represents one of the greatest threats to biodiversity, causing degradation and loss of habitat, leading to changes in the richness and composition of communities. These changes in richness and composition may, in turn, lead to altered species co-occurrence, but our knowledge of this remains limited. We used a novel co-occurrence network approach to examine the impact of agricultural transformation on reptile community structure within two large ( 172 000 km 2 ; 224 sites) agricultural regions in southeastern Australia. We contrasted assemblages from sites surrounded by intact and modified landscapes and tested four key hypotheses that agricultural transformation leads to (H1) declines in species richness, (H2) altered assemblages, (H3) declines in overall co-occurrence, and (H4) complex restructuring of pairwise associations. We found that modified landscapes differed in composition but not richness compared with intact sites. Modified landscapes were also characterized by differences in co-occurrence network structure; with species sharing fewer sites with each other (reduced co-occurrence connectance), fewer highly-connected species (truncation of the frequency distribution of co-occurrence degree) and increased modularity of co-occurrence networks. Critically, overall loss of co-occurrence was underpinned by complex changes to the number and distribution of pair-wise co-occurrence links, with 41-44% of species also gaining associations with other species. Change in co-occurrence was not correlated with changes in occupancy, nor by functional trait membership, allowing a novel classification of species susceptibility to agricultural transformation. Our study reveals the value of using co-occurrence analysis to uncover impacts of agricultural transformation that may be masked in conventional studies of species richness and community composition.

Using co‐occurrence network topology in assessing ecological stress in benthic macroinvertebrate communities

“…Given these prior observations, with regard to changes in patterns of diversity across watersheds in relation to stress, we expect the taxonomic “space” for co‐occurrences to shrink with a rise in stress due to land use (Figure ), and with it a trend toward the fracturing of assembled co‐occurrence networks into weakly connected subnetworks. Similar relationships, between the modularity of co‐occurrence networks and ecological stress, have also been observed in various ecological communities (Hu et al, ; Kay et al, ). To assess these trends, we use modularity, the degree to which networks are organized into clusters of weakly interconnected subnetworks (Barberán, Bates, Casamayor, & Fierer, ; Clauset et al, ).…”

“…For example, the clustering of microbial species into distinct modules within co‐occurrence networks has been used to infer physiochemical niches for various prokaryotic groups (Fuhrman & Steele, ; Larsen & Ormerod, ; Mandakovic et al, ; Ruan et al, ; Steele et al, ; Widder et al, ). In studies of larger organisms, topological measures of these networks have also been used to illustrate a loss in both diversity and the number of significant co‐occurrences between reptiles in response to habitat degradation (Kay et al, ). The previous diversity of scenarios where co‐occurrence network topology has been used in ecological analysis then implies that it could also be used to develop a framework for the assessment of the biotic integrity of streams across an entire catchment area (Ahn & Kim, ; Moyle & Randall, ; Smith & Lamp, ).…”

“…For example, the clustering of microbial species into distinct modules within co-occurrence networks has been used to infer physiochemical niches for various prokaryotic groups (Fuhrman & Steele, 2008;Larsen & Ormerod, 2014;Mandakovic et al, 2018;Ruan et al, 2006;Steele et al, 2011;Widder et al, 2014). In studies of larger organisms, topological measures of these networks have also been used to illustrate a loss in both diversity and the number of significant co-occurrences between reptiles in response to habitat degradation (Kay et al, 2018).…”

Fundamental contradictions among observational and experimental estimates of non‐trophic species interactions

“…Recent papers suggest that frequencies of positive or negative associations reflect assembly mechanism (Levy and Borenstein 2013, Zelezniak et al 2015, Lyons et al 2016, but no study to date has determined which processes lead to more positive or negative associations. In other studies, changes in associations are thought to signal shifts in system stability (Griffith et al 2018, Kay et al 2018.…”

Network parameters quantify loss of assemblage structure in human‐impacted lake ecosystems

“…Network theory provides a framework for hypothesizing the expected structural changes within a stressed system (Griffith, Strutton, & Semmens, 2018;Kay et al, 2018). In general, system interactions between components are represented by self-organizing links amongst nodes of similar type (Dakos & Bascompte, 2014;Gao, Barzel, & Barabási, 2016;Montoya, Pimm, & Solé, 2006;Newman, 2010), for example, documents on the World Wide Web and humans on social media (Amaral, Scala, Barthelemy, & Stanley, 2000;Watts & Strogatz, 1998).…”