Hierarchical Bayesian models in ecology: Reconstructing species interaction networks from non-homogeneous species abundance data

Deep learning-based efficient drone-borne sensing of cyanobacterial blooms using a clique-based feature extraction approach

Recent advances in remote sensing techniques provide a new horizon for monitoring the spatiotemporal variations of harmful algal blooms (HABs) using hyperspectral data in inland water. In this study, a hierarchical concatenated variational autoencoder (HCVAE) is proposed as an efficient and accurate deep learning (DL) based bio-optical model. To demonstrate its usefulness in retrieving algal pigments, the HCVAE is applied to bloom-prone regions in Daecheong Lake, South Korea. By abstracting the similarity between highly related features using layer-wise clique-based latent-feature extraction, HCVAE reduces the computational loads in deriving outputs while preventing performance degradation. Graph-based clique-detection uses information theory-based criteria to group the related reflectance spectra. Consequently, six latent features were extracted from 79 spectral bands to consist of a multilevel hierarchy of HCVAE that can simultaneously estimate concentrations of chlorophyll-a (Chl-a) and phycocyanin (PC). Despite the parsimonious model architecture, the Chl-a and PC concentrations estimated by HCVAE closely agree with the measured concentrations, with test R2 values of 0.76 and 0.82, respectively. In addition, spatial distribution maps of algal pigments obtained from HCVAE using drone-borne reflectance successfully capture the blooming spots. Based on its multilevel hierarchical architecture, HCVAE can provide the importance of latent features along with their individual wavelengths using Shapley additive explanations. The most important latent features covered the spectral regions associated with both Chl-a and PC. The lightweight neural network DNNsel, which uses only the spectral bands of highest importance in latent-feature extraction, performed comparably to HCVAE. The study results demonstrate the utility of the multilevel hierarchical architecture as a comprehensive assessment model for near-real-time drone-borne sensing of HABs. Moreover, HCVAE is applicable to a wide range of environmental big data, as it can handle numerous sets of features.

An integrative approach to understanding diversity patterns and assemblage rules in Neotropical bats

Understanding the mechanisms shaping species composition of assemblages is critical for incorporating ecological and evolutionary perspectives into biodiversity conservation. Thus, we quantified the relative support of community assembly mechanisms by assessing how species richness relates to the functional and phylogenetic biodiversity of Neotropical bat assemblages. We assessed the association of functional diversity for functional categories and phylogenetic diversity with species richness for 20 assemblages of Neotropical bats. In addition, we contrasted functional and phylogenetic diversity against null models to determine the mechanisms that structure the assemblages. We hypothesize functional/phylogenetic overdispersion for high species sites and a positive relationship between those dimensions of diversity and richness. Functional divergence increased with species richness, indicating that the variability in ecological attributes among abundant bats increases as the assemblages contain more species. Taxa were more distantly related as richness increases, but distances among closely related species remained constant. We found a consistent tendency of clustering of functional traits in site assemblages, particularly in abundant species. We proposed competition between clades as a possible mechanism modulating the community structure in Neotropical bat assemblages. Our results suggest that decreasing overlap in functional traits between abundant species could promote coexistence with rare species that can buffer ecosystem function due to species loss.

Rapid, recurring, structured survey versus bioblitz for generating biodiversity data and analysis with a multispecies abundance model

A bioblitz inexpensively and quickly generates biodiversity data, but bioblitzes are often conducted with haphazard, unreplicated sampling. Results tend to be taxonomically, geographically, or temporally biased, lack metadata, and consist of lists of observed taxa that do not enable further analyses or correction for imperfect detection. A rapid, recurring, structured survey (RRSS) uses a structured sampling design and temporal and spatial replication to survey randomly selected sites on a conservation property. We participated in a loosely structured bioblitz and a subsequent RRSS at Big Canoe Creek Nature Preserve in Springville (St. Clair County), Alabama (USA) to compare observed richness derived from the 2 survey approaches. The RRSS data structure enabled us to fit models that accounted for imperfect detection to estimate abundances, occupancy probabilities, and habitat associations. The loosely structured bioblitz data could not be used in such models. We present a new integrated multispecies abundance model that we applied to avian RRSS data. Our model extension enables estimation for the community, employs data augmentation to estimate the number of undetected species, and incorporates covariates. The RRSS generated a more comprehensive and less biased list of observed taxonomic richness than the loosely structured bioblitz (e.g., 73 vs. 45 bird species and 104 vs. 63 insect families from the RRSS vs. loosely structured bioblitz, respectively). Models fit to the RRSS data identified seasonal patterns in avian community composition and allowed for estimation of habitat-occupancy relationships for insect taxa. The RRSS protocol has potential for broad transferability as a standardized, quick, and inexpensive way to inventory biodiversity and estimate ecological parameters while providing an outreach opportunity.

Inter-Kingdom Networks of Canola Microbiome Reveal Bradyrhizobium as Keystone Species and Underline the Importance of Bulk Soil in Microbial Studies to Enhance Canola Production

The subterranean microbiota of plants is of great importance for plant growth and health, as root-associated microbes can perform crucial ecological functions. As the microbial environment of roots is extremely diverse, identifying keystone microorganisms in plant roots, rhizosphere, and bulk soil is a necessary step towards understanding the network of influence within the microbial community associated with roots and enhancing its beneficial elements. To target these hot spots of microbial interaction, we used inter-kingdom network analysis on the canola growth phase of a long-term cropping system diversification experiment conducted at four locations in the Canadian Prairies. Our aims were to verify whether bacterial and fungal communities of canola roots, rhizosphere, and bulk soil are related and influenced by diversification of the crop rotation system; to determine whether there are common or specific core fungi and bacteria in the roots, rhizosphere, and bulk soil under canola grown in different environments and with different levels of cropping system diversification; and to identify hub taxa at the inter-kingdom level that could play an important ecological role in the microbiota of canola. Our results showed that fungi were influenced by crop diversification, which was not the case on bacteria. We found no core microbiota in canola roots but identified three core fungi in the rhizosphere, one core mycobiota in the bulk soil, and one core bacterium shared by the rhizosphere and bulk soil. We identified two bacterial and one fungal hub taxa in the inter-kingdom networks of the canola rhizosphere, and one bacterial and two fungal hub taxa in the bulk soil. Among these inter-kingdom hub taxa, Bradyrhizobium sp. and Mortierella sp. are particularly influential on the microbial community and the plant. To our knowledge, this is the first inter-kingdom network analysis utilized to identify hot spots of interaction in canola microbial communities.

Learning hierarchical Bayesian networks to assess the interaction effects of controlling factors on spatiotemporal patterns of fecal pollution in streams

The dynamics of fecal indicator bacteria, such as fecal coliforms (FC) in streams, are influenced by the interactions of a myriad of factors. To predict complex spatiotemporal patterns of FC in streams and assess the relative importance of numerous controlling factors, the adoption of a hierarchical Bayesian network (HBN) was proposed in this study. By introducing latent variables correlated to the observed variables into a Bayesian network, the HBN can represent causal relationships among a large set of variables with a multilevel hierarchy. The study area encompasses 215 sites across the watersheds of the four major rivers in South Korea. The monitoring data collected during the 2012-2019 period included 32 input variables pertaining to meteorology, geography, soil characteristics, land cover, urbanization index, livestock density, and point sources. As model endpoints, the exceedance probability of the FC standard concentration as well as two pollution characteristics (i.e., pollution degree and type), derived from FC load duration curves were used. The probability of exceeding an FC threshold value (200 CFU/100 mL) showed spatiotemporal variations, whereas pollution degree and type showed spatial variations that represent long-term severity and relative dominance of nonpoint and point source fecal pollution, respectively. The conceptual model was validated using structural equation modeling to develop the HBN. The results demonstrate that the HBN effectively simplified the model structure, while showing strong model performance (AUC = 0.81, accuracy = 0.74). The results of the sensitivity analysis indicate that land cover is the most important factor in predicting the probability of exceedance and pollution degree, whereas the urbanization index explains most of the variability in pollution type. Furthermore, the results of the scenario analysis suggest that the HBN provides an interpretable framework in which the interaction of controlling factors has causal relationships at different levels that can be identified and visualized.

A taxonomic-based joint species distribution model for presence-only data

Species distribution models (SDMs) are an important class of model for mapping taxa spatially and are a key tool for tackling biodiversity loss. However, most common SDMs depend on presence–absence data and, despite the accumulation and exponential growth of biological occurrence data across the globe, the available data are predominantly presence-only (i.e. they lack real absences). Although presence-only SDMs do exist, they inevitably require assumptions about absences of the considered taxa and they are specified mostly for single species and, thus, do not exploit fully the information in related taxa. This greatly limits the utility of global biodiversity databases such as GBIF. Here, we present a Bayesian-based SDM for multiple species that operates directly on presence-only data by exploiting the joint distribution between the multiple ecological processes and, crucially, identifies the sampling effort per taxa which allows inference on absences. The model was applied to two case studies. One, focusing on taxonomically diverse taxa over central Mexico and another focusing on the monophyletic family Cactacea over continental Mexico. In both cases, the model was able to identify the ecological and sampling effort processes for each taxon using only the presence observations, environmental and anthropological data.

Ecological interactions and the Netflix problem

Species interactions are a key component of ecosystems but we generally have an incomplete picture of who-eats-who in a given community. Different techniques have been devised to predict species interactions using theoretical models or abundances. Here, we explore the K nearest neighbour approach, with a special emphasis on recommendation, along with a supervised machine learning technique. Recommenders are algorithms developed for companies like Netflix to predict whether a customer will like a product given the preferences of similar customers. These machine learning techniques are well-suited to study binary ecological interactions since they focus on positive-only data. By removing a prey from a predator, we find that recommenders can guess the missing prey around 50% of the times on the first try, with up to 881 possibilities. Traits do not improve significantly the results for the K nearest neighbour, although a simple test with a supervised learning approach (random forests) show we can predict interactions with high accuracy using only three traits per species. This result shows that binary interactions can be predicted without regard to the ecological community given only three variables: body mass and two variables for the species' phylogeny. These techniques are complementary, as recommenders can predict interactions in the absence of traits, using only information about other species' interactions, while supervised learning algorithms such as random forests base their predictions on traits only but do not exploit other species' interactions. Further work should focus on developing custom similarity measures specialized for ecology to improve the KNN algorithms and using richer data to capture indirect relationships between species.

Quantifying Preferences and Responsiveness of Marine Zooplankton to Changing Environmental Conditions using Microfluidics

Global environmental change significantly affects marine species composition. However, analyzing the impact of these changes on marine zooplankton communities was so far mostly limited to assessing lethal doses through mortality assays and hence did not allow a direct assessment of the preferred conditions, or preferendum. Here, we use a microfluidic device to characterize individual behavior of actively swimming zooplankton, and to quantitatively determine their ecological preferendum. For the annelid zooplankton model Platynereis dumerilii we observe a broader pH preferendum than for the copepod Euterpina acutifrons, and reveal previously unrecognized sub-populations with different pH preferenda. For Platynereis, the minimum concentration difference required to elicit a response (responsiveness) is ~1 μM for H⁺ and ~13.7 mM for NaCl. Furthermore, using laser ablations we show that olfactomedin-expressing sensory cells mediate chemical responsiveness in the Platynereis foregut. Taken together, our microfluidic approach allows precise assessment and functional understanding of environmental perception on planktonic behaviour.

Cross-biome comparison of microbial association networks

Clinical and environmental meta-omics studies are accumulating an ever-growing amount of microbial abundance data over a wide range of ecosystems. With a sufficiently large sample number, these microbial communities can be explored by constructing and analyzing co-occurrence networks, which detect taxon associations from abundance data and can give insights into community structure. Here, we investigate how co-occurrence networks differ across biomes and which other factors influence their properties. For this, we inferred microbial association networks from 20 different 16S rDNA sequencing data sets and observed that soil microbial networks harbor proportionally fewer positive associations and are less densely interconnected than host-associated networks. After excluding sample number, sequencing depth and beta-diversity as possible drivers, we found a negative correlation between community evenness and positive edge percentage. This correlation likely results from a skewed distribution of negative interactions, which take place preferentially between less prevalent taxa. Overall, our results suggest an under-appreciated role of evenness in shaping microbial association networks.