Robust and simplified machine learning identification of pitfall trap‐collected ground beetles at the continental scale

1. Insect populations are changing rapidly, and monitoring these changes is essential for understanding the causes and consequences of such shifts. However, large‐scale insect identification projects are time‐consuming and expensive when done solely by human identifiers. Machine learning offers a possible solution to help collect insect data quickly and efficiently.
2. Here, we outline a methodology for training classification models to identify pitfall trap‐collected insects from image data and then apply the method to identify ground beetles (Carabidae). All beetles were collected by the National Ecological Observatory Network (NEON), a continental scale ecological monitoring project with sites across the United States. We describe the procedures for image collection, image data extraction, data preparation, and model training, and compare the performance of five machine learning algorithms and two classification methods (hierarchical vs. single‐level) identifying ground beetles from the species to subfamily level. All models were trained using pre‐extracted feature vectors, not raw image data. Our methodology allows for data to be extracted from multiple individuals within the same image thus enhancing time efficiency, utilizes relatively simple models that allow for direct assessment of model performance, and can be performed on relatively small datasets.
3. The best performing algorithm, linear discriminant analysis (LDA), reached an accuracy of 84.6% at the species level when naively identifying species, which was further increased to >95% when classifications were limited by known local species pools. Model performance was negatively correlated with taxonomic specificity, with the LDA model reaching an accuracy of ~99% at the subfamily level. When classifying carabid species not included in the training dataset at higher taxonomic levels species, the models performed significantly better than if classifications were made randomly. We also observed greater performance when classifications were made using the hierarchical classification method compared to the single‐level classification method at higher taxonomic levels.
4. The general methodology outlined here serves as a proof‐of‐concept for classifying pitfall trap‐collected organisms using machine learning algorithms, and the image data extraction methodology may be used for nonmachine learning uses. We propose that integration of machine learning in large‐scale identification pipelines will increase efficiency and lead to a greater flow of insect macroecological data, with the potential to be expanded for use with other noninsect taxa.

     

Phylogeny of Philippine slender skinks (Scincidae: Brachymeles) reveals underestimated species diversity, complex biogeographical relationships, and cryptic patterns of lineage diversification

The spectacular, virtually endemic radiation of Philippine semi-fossorial skinks of the genus Brachymeles represent one of the few radiations of scincid lizards to possess both fully limbed and limbless species. And yet, nothing is known of the phylogenetic relationships of this exceptional group. Morphologically similar body plans have made it difficult to assess species-level diversity, and the genus has long been recognized as one of the more modest radiations of southeast Asian lizards. However, recent large-scale survey efforts have resulted in the discovery of numerous new species, and taxonomic studies indicate that the diversity within the genus Brachymeles is grossly underestimated. In this study we provide the first robust estimate of phylogenetic relationships within the genus Brachymeles using a multi-locus dataset and nearly complete taxonomic sampling. We provide statistical tests of monophyly for all polytypic species and two widespread limb-reduced species and our results indicate wholesale deviations from past summaries and taxonomic evaluations of the genus. With few exceptions, we are able to reject the monophyly of all polytypic and widespread species, thereby validating the need for large-scale taxonomic revisions. Our results reveal that the limbless, monotypic, genus Davewakeum is nested within Brachymeles. Mapping of body form on our preferred phylogenetic tree suggests that limb-reduction and digit loss has occurred on multiple occasions in the history of the genus. A Bayesian reconstruction of ancestral areas indicates strong statistical support for a minimum of five major dispersal events that have given rise to a major component of the observed species diversity on separate Pleistocene aggregate island platforms of the archipelago.     

AN ADAPTIVE RADIATION OF FROGS IN A SOUTHEAST ASIAN ISLAND ARCHIPELAGO

Living amphibians exhibit a diversity of ecologies, life histories, and species‐rich lineages that offers opportunities for studies of adaptive radiation. We characterize a diverse clade of frogs (Kaloula, Microhylidae) in the Philippine island archipelago as an example of an adaptive radiation into three primary habitat specialists or ecotypes. We use a novel phylogenetic estimate for this clade to evaluate the tempo of lineage accumulation and morphological diversification. Because species‐level phylogenetic estimates for Philippine Kaloula are lacking, we employ dense population sampling to determine the appropriate evolutionary lineages for diversification analyses. We explicitly take phylogenetic uncertainty into account when calculating diversification and disparification statistics and fitting models of diversification. Following dispersal to the Philippines from Southeast Asia, Kaloula radiated rapidly into several well‐supported clades. Morphological variation within Kaloula is partly explained by ecotype and accumulated at high levels during this radiation, including within ecotypes. We pinpoint an axis of morphospace related directly to climbing and digging behaviors and find patterns of phenotypic evolution suggestive of ecological opportunity with partitioning into distinct habitat specialists. We conclude by discussing the components of phenotypic diversity that are likely important in amphibian adaptive radiations.     

Multilocus phylogeny and Bayesian estimates of species boundaries reveal hidden evolutionary relationships and cryptic diversity in Southeast Asian monitor lizards

Recent conceptual, technological and methodological advances in phylogenetics have enabled increasingly robust statistical species delimitation in studies of biodiversity. As the variety of evidence purporting species diversity has increased, so too have the kinds of tools and inferential power of methods for delimiting species. Here, we showcase an organismal system for a data‐rich, comparative molecular approach to evaluating strategies of species delimitation among monitor lizards of the genus Varanus. The water monitors (Varanus salvator Complex), a widespread group distributed throughout Southeast Asia and southern India, have been the subject of numerous taxonomic treatments, which have drawn recent attention due to the possibility of undocumented species diversity. To date, studies of this group have relied on purportedly diagnostic morphological characters, with no attention given to the genetic underpinnings of species diversity. Using a 5‐gene data set, we estimated phylogeny and used multilocus genetic networks, analysis of population structure and a Bayesian coalescent approach to infer species boundaries. Our results contradict previous systematic hypotheses, reveal surprising relationships between island and mainland lineages and uncover novel, cryptic evolutionary lineages (i.e. new putative species). Our study contributes to a growing body of literature suggesting that, used in concert with other sources of data (e.g. morphology, ecology, biogeography), multilocus genetic data can be highly informative to systematists and biodiversity specialists when attempting to estimate species diversity and identify conservation priorities. We recommend holding in abeyance taxonomic decisions until multiple, converging lines of evidence are available to best inform taxonomists, evolutionary biologists and conservationists.     

The amphibians and reptiles of Luzon Island,Philippines, VIII: the herpetofauna of Cagayan and Isabela Provinces,northern Sierra Madre Mountain Range

We provide the first report on the herpetological biodiversity (amphibians and reptiles) of the northern Sierra Madre Mountain Range (Cagayan and Isabela provinces), northeast Luzon Island, Philippines. New data from extensive previously unpublished surveys in the Municipalities of Gonzaga, Gattaran, Lasam, Santa Ana, and Baggao (Cagayan Province), as well as fieldwork in the Municipalities of Cabagan, San Mariano, and Palanan (Isabela Province), combined with all available historical museum records, suggest this region is quite diverse. Our new data indicate that at least 101 species are present (29 amphibians, 30 lizards, 35 snakes, two freshwater turtles, three marine turtles, and two crocodilians) and now represented with well-documented records and/or voucher specimens, confirmed in institutional biodiversity repositories. A high percentage of Philippine endemic species constitute the local fauna (approximately 70%). The results of this and other recent studies signify that the herpetological diversity of the northern Philippines is far more diverse than previously imagined. Thirty-eight percent of our recorded species are associated with unresolved taxonomic issues (suspected new species or species complexes in need of taxonomic partitioning). This suggests that despite past and present efforts to comprehensively characterize the fauna, the herpetological biodiversity of the northern Philippines is still substantially underestimated and warranting of further study.     

Molecular population genetics of the alcohol dehydrogenase locus in the Hawaiian drosophilid D. mimica

Sequence variation among 10 alleles of the alcohol dehydrogenase (Adh) gene of the Hawaiian drosophilid D. mimica was analyzed with reference to the evolutionary history of the Hawaiian subgroup as well as to levels and patterns of polymorphism of the Adh gene in continental drosophilid species. The Adh gene of D. mimica is less polymorphic than that of other drosophilid species, and no replacement substitutions were found. Statistical analyses of the Adh alleles suggested the action of balancing selection and revealed significant linkage disequilibrium among three of the variable sites. The effective population size was estimated to be only slightly smaller than that of continental species and, surprisingly, on the same order of magnitude as the actual size.      

Colonization and production of macroinvertebrates on artificial substrata: upstream–downstream responses to a leaf litter exclusion manipulation

1. Macroinvertebrate colonization dynamics were examined on artificial substrata in a stream with terrestrial litter inputs excluded, downstream of the litter-exclusion treatment, and in a reference stream. 2. Short-term examination of the rates of organic matter accrual and invertebrate colonization demonstrated significantly lower accumulation of leaf detritus and invertebrates in the litter-excluded reach and a short distance downstream of that reach. 3. All major fractions of organic matter and invertebrates declined on artificial substrata during the 3-year litter exclusion. Further, secondary production on artificial substrata in the litter-excluded reach decreased from 6.2 to 1.5 g AFDM m⁻² year⁻¹ from pretreatment to the third year of litter exclusion, respectively. 4. Downstream, fine particulate organic matter on artificial substrata decreased during litter exclusion, and there was a significant reduction in colonization of collector-filterers. Total secondary production downstream of the litter exclusion declined >70%, demonstrating that downstream colonization dynamics are linked to upstream detritus inputs and processing by stream invertebrates.     

Multilocus phylogeny and a new classification for Southeast Asian and Melanesian forest frogs (family Ceratobatrachidae)

We present a near comprehensive, densely sampled, multilocus phylogenetic estimate of species relationships within the anuran family Ceratobatrachidae, a morphologically and ecologically diverse group of frogs from the island archipelagos of Southeast Asia and the South‐West Pacific. Ceratobatrachid frogs consist of three clades: a small clade of enigmatic, primarily high‐elevation, semi‐aquatic Sundaland species currently assigned to Ingerana (for which we erect a new genus), which is the sister taxon of two large, monophyletic radiations, each situated on islands on either side of Wallace's Line. One radiation is composed of Philippine species of Platymantis and the other contains all taxa from the eastern Indonesian, New Guinean, Solomon, Bismarck, and Fijian archipelagos. Several additional genera (Batrachylodes, Discodeles, Ceratobatrachus, and Palmatorappia) are nested within Platymantis, and of these Batrachylodes and Discodeles are nonmonophyletic. To address the widespread paraphyly of the genus Platymantis and several additional nomenclatural issues, we undertook a wholesale nomenclatural reorganization of the family. Given our partially unresolved phylogeny, and in order to impart a conservative, stable taxonomy, involving a minimal number of genus‐species couplet changes, we propose a conservative classification representing a few compromises. These changes are designed to preserve maximally the presumed original intent of taxonomy (widely used group names associated with morphological and ecological diversity of particular species or groups of species) while implementing a hierarchical system that is consistent with the estimate of phylogeny based on new molecular data. © 2015 The Linnean Society of London     

The comparative biogeography of Philippine geckos challenges predictions from a paradigm of climate‐driven vicariant diversification across an island archipelago

A primary goal of biogeography is to understand how large‐scale environmental processes, like climate change, affect diversification. One often‐invoked but seldom tested process is the “species‐pump” model, in which repeated bouts of cospeciation are driven by oscillating climate‐induced habitat connectivity cycles. For example, over the past three million years, the landscape of the Philippine Islands has repeatedly coalesced and fragmented due to sea‐level changes associated with glacial cycles. This repeated climate‐driven vicariance has been proposed as a model of speciation across evolutionary lineages codistributed throughout the islands. This model predicts speciation times that are temporally clustered around the times when interglacial rises in sea level fragmented the islands. To test this prediction, we collected comparative genomic data from 16 pairs of insular gecko populations. We analyze these data in a full‐likelihood, Bayesian model‐choice framework to test for shared divergence times among the pairs. Our results provide support against the species‐pump model prediction in favor of an alternative interpretation, namely that each pair of gecko populations diverged independently. These results suggest the repeated bouts of climate‐driven landscape fragmentation have not been an important mechanism of speciation for gekkonid lizards across the Philippine Archipelago.     

Rapid kinetic studies of electron transfer in the three isoforms of nitric oxide synthase.

The nitric oxide synthases (NOSs) consist of a flavin-containing reductase domain, linked to a heme-containing oxygenase domain, by a calmodulin (CaM) binding sequence. The flavin-containing reductase domains of the NOS isoforms possess close sequence homology to NADPH-cytochrome P450 reductase (CPR). Additionally, the oxygenase domains catalyze monooxygenation of L-arginine through a cytochrome P450-like cysteine thiolate-liganded heme bound in the active site. With these considerations in mind, we conducted studies in an attempt to gain insight into the intermediates involved in flavoprotein-to-heme electron transfer in the NOSs. Static, steady-state, and stopped-flow kinetic studies indicated that nNOS must be reduced to a more than one-electron-reduced intermediate before efficient electron transfer can occur. Therefore, the possibility exists that the oxygenase domains of the NOS isoforms may receive their electrons from the reductase domains by a mechanism resembling the CPR-P450 interaction. Furthermore, the rate-limiting step in electron transfer appears to be the transfer of electrons from the flavoprotein to the oxygenase domain facilitated by the binding of CaM at increased intracellular Ca(2+) concentrations. Thus, modulation of electron transfer rates appears to be regulated at the level of the flavoprotein domains of the NOS isoforms.