Birth-death models are widely used in combination with species phylogenies to study past diversification dynamics. Current inference approaches typically rely on likelihood-based methods. These methods are not generalizable, as a new likelihood formula must be established each time a new model is proposed; for some models such formula is not even tractable. Deep learning can bring solutions in such situations, as deep neural networks can be trained to learn the relation between simulations and parameter values as a regression problem. In this paper, we adapt a recently developed deep learning method from pathogen phylodynamics to the case of diversification inference, and we extend its applicability to the case of the inference of state-dependent diversification models from phylogenies associated with trait data. We demonstrate the accuracy and time efficiency of the approach for the time constant homogeneous birth-death model and the Binary-State Speciation and Extinction model. Finally, we illustrate the use of the proposed inference machinery by reanalyzing a phylogeny of primates and their associated ecological role as seed dispersers. Deep learning inference provides at least the same accuracy as likelihood-based inference while being faster by several orders of magnitude, offering a promising new inference approach for deployment of future models in the field.

Understanding how organisms adapt to environmental changes is a major question in evolution and ecology. In particular, the role of ancestral variation in rapid adaptation remains unclear because its trace on genetic variation, known as soft selective sweep, is often hardly recognizable from genome-wide selection scans. Here, we investigate the evolution of chemosensory genes in Drosophila yakuba mayottensis, a specialist subspecies on toxic noni (Morinda citrifolia) fruits on the island of Mayotte. We combine population genomics analyses and behavioral assays to evaluate the level of divergence in chemosensory genes and perception of noni chemicals between specialist and generalist subspecies of D. yakuba. We identify a signal of soft selective sweep on a handful of genes, with the most diverging ones involving a cluster of gustatory receptors expressed in bitter-sensing neurons. Our results highlight the potential role of ancestral genetic variation in promoting host plant specialization in herbivorous insects and identify a number of candidate genes underlying behavioral adaptation.

Mangroves are tidal wetlands that are often under strong anthropogenic pressures, despite the numerous ecosystem services they provide. Pollution from urban runoffs is one such threats, yet some mangroves are used as a bioremediation tool for wastewater (WW) treatment. This practice can impact mangrove crabs, which are key engineer species of the ecosystem. Using an experimental area with controlled WW releases, this study aimed to determine from an ecological and ecotoxicological perspective, the effects of WW on the red mangrove crab Neosarmatium africanum. Burrow density and salinity levels (used as a proxy of WW dispersion) were recorded, and a 3-week caging experiment was performed. Hemolymph osmolality, gill Na+/K+-ATPase (NKA) activity and gill redox balance were assessed in anterior and posterior gills of N. africanum. Burrow density decreased according to salinity decreases around the discharged area. Crabs from the impacted area had a lower osmoregulatory capacity despite gill NKA activity remaining undisturbed. The decrease of the superoxide dismutase activity indicates changes in redox metabolism. However, both catalase activity and oxidative damage remained unchanged in both areas but were higher in posterior gills. These results indicate that WW release may induce osmoregulatory and redox imbalances, potentially explaining the decrease in crab density. Based on these results we conclude that WW release should be carefully monitored as crabs are key players involved in the bioremediation process. Highlights ► Mangroves are suggested as biofiltering systems of domestic effluent. ► Mangrove crabs are involved through their bioturbation activities. ► Wastewater release impacts crab burrow density. ► Mangrove crabs are physiologically impacted by wastewaters.

To infer the processes that gave rise to past speciation and extinction rates across taxa, space and time, we often formulate hypotheses in the form of stochastic diversification models and estimate their parameters from extant phylogenies using Maximum Likelihood or Bayesian inference. Unfortunately, however, likelihoods can easily become intractable, limiting our ability to consider more complicated diversification processes. Recently, it has been proposed that deep learning (DL) could be used in this case as a likelihood-free inference technique. Here, we explore this idea in more detail, with a particular focus on understanding the ideal network architecture and data representation for using DL in phylogenetic inference. We evaluate the performance of different neural network architectures (DNN, CNN, RNN, GNN) and phylogeny representations (summary statistics, Lineage Through Time or LTT, phylogeny encoding and phylogeny graph) for inferring rates of the Constant Rate Birth-Death (CRBD) and the Binary State Speciation and Extinction (BISSE) models. We find that deep learning methods can reach similar or even higher accuracy than Maximum Likelihood Estimation, provided that network architectures and phylogeny representations are appropriately tuned to the respective model. For example, for the CRBD model we find that CNNs and RNNs fed with LTTs outperform other combinations of network architecture and phylogeny representation, presumably because the LTT is a sufficient and therefore less redundant statistic for homogenous BD models. For the more complex BiSSE model, however, it was necessary to feed the network with both topology and tip states information to reach acceptable performance. Overall, our results suggest that deep learning provides a promising alternative for phylogenetic inference, but that data representation and architecture have strong effects on the inferential performance.