Temperature and water availability drive insect seasonality across a temperate and a tropical region.

The more insects there are, the more food there is for insectivores and the higher the likelihood for insect-associated ecosystem services. Yet, we lack insights into the drivers of insect biomass over space and seasons, for both tropical and temperate zones. We used 245 Malaise traps, managed by 191 volunteers and park guards, to characterize year-round flying insect biomass in a temperate (Sweden) and a tropical (Madagascar) country. Surprisingly, we found that local insect biomass was similar across zones. In Sweden, local insect biomass increased with accumulated heat and varied across habitats, while biomass in Madagascar was unrelated to the environmental predictors measured. Drivers behind seasonality partly converged: In both countries, the seasonality of insect biomass differed between warmer and colder sites, and wetter and drier sites. In Sweden, short-term deviations from expected season-specific biomass were explained by week-to-week fluctuations in accumulated heat, rainfall and soil moisture, whereas in Madagascar, weeks with higher soil moisture had higher insect biomass. Overall, our study identifies key drivers of the seasonal distribution of flying insect biomass in a temperate and a tropical climate. This knowledge is key to understanding the spatial and seasonal availability of insects-as well as predicting future scenarios of insect biomass change.

Optimizing insect metabarcoding using replicated mock communities.

1: Metabarcoding (high-throughput sequencing of marker gene amplicons) has emerged as a promising and cost-effective method for characterizing insect community samples. Yet, the methodology varies greatly among studies and its performance has not been systematically evaluated to date. In particular, it is unclear how accurately metabarcoding can resolve species communities in terms of presence-absence, abundances, and biomass. 2: Here we use mock community experiments and a simple probabilistic model to evaluate the effect of different DNA extraction protocols on metabarcoding performance. Specifically, we ask four questions: (Q1) How consistent are the recovered community profiles across replicate mock communities?; (Q2) How does the choice of lysis buffer affect the recovery of the original community?; (Q3) How are community estimates affected by differing lysis times and homogenization?; and (Q4) Is it possible to obtain adequate species abundance estimates through the use of biological spike-ins? 3: We show that estimates are quite variable across community replicates. In general, a mild lysis protocol is better at reconstructing species lists and approximate counts, while homogenization is better at retrieving biomass composition. Small insects are more likely to be detected in lysates, while some tough species require homogenization to be detected. Results are less consistent across biological replicates for lysates than for homogenates. Some species are associated with strong PCR amplification bias, which complicates the reconstruction of species counts. Yet, with adequate spike-in data, species abundance can be determined with roughly 40% standard error for homogenates, and with roughly 50% standard error for lysates, under ideal conditions. In the latter case, however, this often requires species-specific reference data, while spike-in data generalizes better across species for homogenates. 4: We conclude that a non-destructive, mild lysis approach shows the highest promise for presence/absence description of the community, while also allowing future morphological or molecular work on the material. However, homogenization protocols perform better for characterizing community composition, in particular in terms of biomass.

Skyline Fossilized Birth-Death Model is Robust to Violations of Sampling Assumptions in Total-Evidence Dating.

Several total-evidence dating studies under the fossilized birth-death (FBD) model have produced very old age estimates, which are not supported by the fossil record. This phenomenon has been termed "deep root attraction (DRA)." For two specific data sets, involving divergence time estimation for the early radiations of ants, bees, and wasps (Hymenoptera) and of placental mammals (Eutheria), it has been shown that the DRA effect can be greatly reduced by accommodating the fact that extant species in these trees have been sampled to maximize diversity, so-called diversified sampling. Unfortunately, current methods to accommodate diversified sampling only consider the extreme case where it is possible to identify a cut-off time such that all splits occurring before this time are represented in the sampled tree but none of the younger splits. In reality, the sampling bias is rarely this extreme and may be difficult to model properly. Similar modeling challenges apply to the sampling of the fossil record. This raises the question of whether it is possible to find dating methods that are more robust to sampling biases. Here, we show that the skyline FBD (SFBD) process, where the diversification and fossil-sampling rates can vary over time in a piecewise fashion, provides age estimates that are more robust to inadequacies in the modeling of the sampling process and less sensitive to DRA effects. In the SFBD model we consider, rates in different time intervals are either considered to be independent and identically distributed or assumed to be autocorrelated following an Ornstein-Uhlenbeck (OU) process. Through simulations and reanalyses of Hymenoptera and Eutheria data, we show that both variants of the SFBD model unify age estimates under random and diversified sampling assumptions. The SFBD model can resolve DRA by absorbing the deviations from the sampling assumptions into the inferred dynamics of the diversification process over time. Although this means that the inferred diversification dynamics must be interpreted with caution, taking sampling biases into account, we conclude that the SFBD model represents the most robust approach currently available for addressing DRA in total-evidence dating.

FAVIS: Fast and versatile protocol for non-destructive metabarcoding of bulk insect samples.

Insects are diverse and sustain essential ecosystem functions, yet remain understudied. Recent reports about declines in insect abundance and diversity have highlighted a pressing need for comprehensive large-scale monitoring. Metabarcoding (high-throughput bulk sequencing of marker gene amplicons) offers a cost-effective and relatively fast method for characterizing insect community samples. However, the methodology applied varies greatly among studies, thus complicating the design of large-scale and repeatable monitoring schemes. Here we describe a non-destructive metabarcoding protocol that is optimized for high-throughput processing of Malaise trap samples and other bulk insect samples. The protocol details the process from obtaining bulk samples up to submitting libraries for sequencing. It is divided into four sections: 1) Laboratory workspace preparation; 2) Sample processing-decanting ethanol, measuring the wet-weight biomass and the concentration of the preservative ethanol, performing non-destructive lysis and preserving the insect material for future work; 3) DNA extraction and purification; and 4) Library preparation and sequencing. The protocol relies on readily available reagents and materials. For steps that require expensive infrastructure, such as the DNA purification robots, we suggest alternative low-cost solutions. The use of this protocol yields a comprehensive assessment of the number of species present in a given sample, their relative read abundances and the overall insect biomass. To date, we have successfully applied the protocol to more than 7000 Malaise trap samples obtained from Sweden and Madagascar. We demonstrate the data yield from the protocol using a small subset of these samples.

Online tree expansion could help solve the problem of scalability in Bayesian phylogenetics.

Bayesian phylogenetics is now facing a critical point. Over the last 20 years, Bayesian methods have reshaped phylogenetic inference and gained widespread popularity due to their high accuracy, the ability to quantify the uncertainty of inferences and the possibility of accommodating multiple aspects of evolutionary processes in the models that are used. Unfortunately, Bayesian methods are computationally expensive, and typical applications involve at most a few hundred sequences. This is problematic in the age of rapidly expanding genomic data and increasing scope of evolutionary analyses, forcing researchers to resort to less accurate but faster methods, such as maximum parsimony and maximum likelihood. Does this spell doom for Bayesian methods? Not necessarily. Here, we discuss some recently proposed approaches that could help scale up Bayesian analyses of evolutionary problems considerably. We focus on two particular aspects: online phylogenetics, where new data sequences are added to existing analyses, and alternatives to Markov chain Monte Carlo (MCMC) for scalable Bayesian inference. We identify 5 specific challenges and discuss how they might be overcome. We believe that online phylogenetic approaches and Sequential Monte Carlo hold great promise and could potentially speed up tree inference by orders of magnitude. We call for collaborative efforts to speed up the development of methods for real-time tree expansion through online phylogenetics.

ASV portal: an interface to DNA-based biodiversity data in the Living Atlas.

BACKGROUND: The Living Atlas is an open source platform used to collect, visualise and analyse biodiversity data from multiple sources, and serves as the national biodiversity data hub in many countries. Although powerful, the Living Atlas has had limited functionality for species occurrence data derived from DNA sequences. As a step toward integrating this fast-growing data source into the platform, we developed the Amplicon Sequence Variant (ASV) portal: a web interface to sequence-based biodiversity observations in the Living Atlas. RESULTS: The ASV portal allows data providers to submit denoised metabarcoding output to the Living Atlas platform via an intermediary ASV database. It also enables users to search for existing ASVs and associated Living Atlas records using the Basic Local Alignment Search Tool, or via filters on taxonomy and sequencing details. The ASV portal is a Python-Flask/jQuery web interface, implemented as a multi-container docker service, and is an integral part of the Swedish Biodiversity Data Infrastructure. CONCLUSION: The ASV portal is a web interface that effectively integrates biodiversity data derived from DNA sequences into the Living Atlas platform.

Towards Large-Scale Integrative Taxonomy (LIT): Resolving the Data Conundrum for Dark Taxa.

New, rapid, accurate, scalable, and cost-effective species discovery and delimitation methods are needed for tackling "dark taxa," here defined as groups for which $<$10$\%$ of all species are described and the estimated diversity exceeds 1,000 species. Species delimitation for these taxa should be based on multiple data sources ("integrative taxonomy") but collecting multiple types of data risks impeding a discovery process that is already too slow. We here develop large-scale integrative taxonomy (LIT), an explicit method where preliminary species hypotheses are generated based on inexpensive data that can be obtained quickly and cost-effectively. These hypotheses are then evaluated based on a more expensive type of "validation data" that is only obtained for specimens selected based on objective criteria applied to the preliminary species hypotheses. We here use this approach to sort 18,000 scuttle flies (Diptera: Phoridae) into 315 preliminary species hypotheses based on next-generation sequencing barcode (313 bp) clusters (using objective clustering [OC] with a 3$\%$ threshold). These clusters are then evaluated with morphology as the validation data. We develop quantitative indicators for predicting which barcode clusters are likely to be incongruent with morphospecies by randomly selecting 100 clusters for in-depth validation with morphology. A linear model demonstrates that the best predictors for incongruence between barcode clusters and morphology are maximum p-distance within the cluster and a newly proposed index that measures cluster stability across different clustering thresholds. A test of these indicators using the 215 remaining clusters reveals that these predictors correctly identify all clusters that are incongruent with morphology. In our study, all morphospecies are true or disjoint subsets of the initial barcode clusters so that all incongruence can be eliminated by varying clustering thresholds. This leads to a discussion of when a third data source is needed to resolve incongruent grouping statements. The morphological validation step in our study involved 1,039 specimens (5.8$\%$ of the total). The formal LIT protocol we propose would only have required the study of 915 (5.1$\%$: 2.5 specimens per species), as we show that clusters without signatures of incongruence can be validated by only studying two specimens representing the most divergent haplotypes. To test the generality of our results across different barcode clustering techniques, we establish that the levels of incongruence are similar across OC, Automatic Barcode Gap Discovery (ABGD), Poisson Tree Processes (PTP), and Refined Single Linkage (RESL) (used by Barcode of Life Data System to assign Barcode Index Numbers [BINs]). OC and ABGD achieved a maximum congruence score with the morphology of 89$\%$ while PTP was slightly less effective (84$\%$). RESL could only be tested for a subset of the specimens because the algorithm is not public. BINs based on 277 of the original 1,714 haplotypes were 86$\%$ congruent with morphology while the values were 89$\%$ for OC, 74$\%$ for PTP, and 72$\%$ for ABGD. [Biodiversity discovery; dark taxa; DNA barcodes; integrative taxonomy.].

Phylogenetic reconstruction of ancestral ecological networks through time for pierid butterflies and their host plants.

The study of herbivorous insects underpins much of the theory that concerns the evolution of species interactions. In particular, Pieridae butterflies and their host plants have served as a model system for studying evolutionary arms races. To learn more about the coevolution of these two clades, we reconstructed ancestral ecological networks using stochastic mappings that were generated by a phylogenetic model of host-repertoire evolution. We then measured if, when, and how two ecologically important structural features of the ancestral networks (modularity and nestedness) evolved over time. Our study shows that as pierids gained new hosts and formed new modules, a subset of them retained or recolonised the ancestral host(s), preserving connectivity to the original modules. Together, host-range expansions and recolonisations promoted a phase transition in network structure. Our results demonstrate the power of combining network analysis with Bayesian inference of host-repertoire evolution to understand changes in complex species interactions over time.

The effect of ethanol concentration on the morphological and molecular preservation of insects for biodiversity studies.

Traditionally, insects collected for scientific purposes have been dried and pinned, or preserved in 70% ethanol. Both methods preserve taxonomically informative exoskeletal structures well but are suboptimal for preserving DNA for molecular biology. Highly concentrated ethanol (95-100%), preferred as a DNA preservative, has generally been assumed to make specimens brittle and prone to breaking. However, systematic studies on the correlation between ethanol concentration and specimen preservation are lacking. Here, we tested how preservative ethanol concentration in combination with different sample handling regimes affect the integrity of seven insect species representing four orders, and differing substantially in the level of sclerotization. After preservation and treatments (various levels of disturbance), we counted the number of appendages (legs, wings, antennae, or heads) that each specimen had lost. Additionally, we assessed the preservation of DNA after long-term storage by comparing the ratio of PCR amplicon copy numbers to an added artificial standard. We found that high ethanol concentrations indeed induce brittleness in insects. However, the magnitude and nature of the effect varied strikingly among species. In general, ethanol concentrations at or above 90% made the insects more brittle, but for species with robust, thicker exoskeletons, this did not translate to an increased loss of appendages. Neither freezing the samples nor drying the insects after immersion in ethanol had a negative effect on the retention of appendages. However, the morphology of the insects was severely damaged if they were allowed to dry. We also found that DNA preserves less well at lower ethanol concentrations when stored at room temperature for an extended period. However, the magnitude of the effect varies among species; the concentrations at which the number of COI amplicon copies relative to the standard was significantly decreased compared to 95% ethanol ranged from 90% to as low as 50%. While higher ethanol concentrations positively affect long-term DNA preservation, there is a clear trade-off between preserving insects for morphological examination and genetic analysis. The optimal ethanol concentration for the latter is detrimental for the former, and vice versa. These trade-offs need to be considered in large insect biodiversity surveys and other projects aiming to combine molecular work with traditional morphology-based characterization of collected specimens.

Universal probabilistic programming offers a powerful approach to statistical phylogenetics.

Statistical phylogenetic analysis currently relies on complex, dedicated software packages, making it difficult for evolutionary biologists to explore new models and inference strategies. Recent years have seen more generic solutions based on probabilistic graphical models, but this formalism can only partly express phylogenetic problems. Here, we show that universal probabilistic programming languages (PPLs) solve the expressivity problem, while still supporting automated generation of efficient inference algorithms. To prove the latter point, we develop automated generation of sequential Monte Carlo (SMC) algorithms for PPL descriptions of arbitrary biological diversification (birth-death) models. SMC is a new inference strategy for these problems, supporting both parameter inference and efficient estimation of Bayes factors that are used in model testing. We take advantage of this in automatically generating SMC algorithms for several recent diversification models that have been difficult or impossible to tackle previously. Finally, applying these algorithms to 40 bird phylogenies, we show that models with slowing diversification, constant turnover and many small shifts generally explain the data best. Our work opens up several related problem domains to PPL approaches, and shows that few hurdles remain before these techniques can be effectively applied to the full range of phylogenetic models.

Connecting high-throughput biodiversity inventories: Opportunities for a site-based genomic framework for global integration and synthesis.

High-throughput sequencing (HTS) is increasingly being used for the characterization and monitoring of biodiversity. If applied in a structured way, across broad geographical scales, it offers the potential for a much deeper understanding of global biodiversity through the integration of massive quantities of molecular inventory data generated independently at local, regional and global scales. The universality, reliability and efficiency of HTS data can potentially facilitate the seamless linking of data among species assemblages from different sites, at different hierarchical levels of diversity, for any taxonomic group and regardless of prior taxonomic knowledge. However, collective international efforts are required to optimally exploit the potential of site-based HTS data for global integration and synthesis, efforts that at present are limited to the microbial domain. To contribute to the development of an analogous strategy for the nonmicrobial terrestrial domain, an international symposium entitled "Next Generation Biodiversity Monitoring" was held in November 2019 in Nicosia (Cyprus). The symposium brought together evolutionary geneticists, ecologists and biodiversity scientists involved in diverse regional and global initiatives using HTS as a core tool for biodiversity assessment. In this review, we summarize the consensus that emerged from the 3-day symposium. We converged on the opinion that an effective terrestrial Genomic Observatories network for global biodiversity integration and synthesis should be spatially led and strategically united under the umbrella of the metabarcoding approach. Subsequently, we outline an HTS-based strategy to collectively build an integrative framework for site-based biodiversity data generation.

The data of the Swedish Malaise Trap Project, a countrywide inventory of Sweden's insect fauna.

BACKGROUND: Despite Sweden's strong entomological tradition, large portions of its insect fauna remain poorly known. As part of the Swedish Taxonomy Initiative, launched in 2002 to document all multi-cellular species occurring in the country, the first taxonomically-broad inventory of the country's insect fauna was initiated, the Swedish Malaise Trap Project (SMTP). In total, 73 Malaise traps were deployed at 55 localities representing a wide range of habitats across the country. Most traps were run continuously from 2003 to 2006 or for a substantial part of that time period. The total catch is estimated to contain 20 million insects, distributed over 1,919 samples (Karlsson et al. 2020). The samples have been sorted into more than 300 taxonomic units, which are made available for expert identification. Thus far, more than 100 taxonomists have been involved in identifying the sorted material, recording the presence of 4,000 species. One third of these had not been recorded from Sweden before and 700 have tentatively been identified as new to science. NEW INFORMATION: Here, we describe the SMTP dataset, published through the Global Biodiversity Information Facility (GBIF). Data on the sorted material are available in the "SMTP Collection Inventory" dataset. It currently includes more than 130,000 records of taxonomically-sorted samples. Data on the identified material are published using the Darwin Core standard for sample-based data. That information is divided up into group-specific datasets, as the sample set processed for each group is different and in most cases non-overlapping. The current data are divided into 79 taxonomic datasets, largely corresponding to taxonomic sorting fractions. The orders Diptera and Hymenoptera together comprise about 90% of the specimens in the material and these orders are mainly sorted to family or subfamily. The remaining insect taxa are mostly sorted to the order level. In total, the 79 datasets currently available comprise around 165,000 specimens, that is, about 1% of the total catch. However, the data are now accumulating rapidly and will be published continuously. The SMTP dataset is unique in that it contains a large proportion of data on previously poorly-known taxa in the Diptera and Hymenoptera.

From Inquilines to Gall Inducers: Genomic Signature of a Life-Style Transition in Synergus Gall Wasps.

Gall wasps (Hymenoptera: Cynipidae) induce complex galls on oaks, roses, and other plants, but the mechanism of gall induction is still unknown. Here, we take a comparative genomic approach to revealing the genetic basis of gall induction. We focus on Synergus itoensis, a species that induces galls inside oak acorns. Previous studies suggested that this species evolved the ability to initiate gall formation recently, as it is deeply nested within the genus Synergus, whose members are mostly inquilines that develop inside the galls of other species. We compared the genome of S. itoensis with that of three related Synergus inquilines to identify genomic changes associated with the origin of gall induction. We used a novel Bayesian selection analysis, which accounts for branch-specific and gene-specific selection effects, to search for signatures of selection in 7,600 single-copy orthologous genes shared by the four Synergus species. We found that the terminal branch leading to S. itoensis had more genes with a significantly elevated dN/dS ratio (positive signature genes) than the other terminal branches in the tree; the S. itoensis branch also had more genes with a significantly decreased dN/dS ratio. Gene set enrichment analysis showed that the positive signature gene set of S. itoensis, unlike those of the inquiline species, is enriched in several biological process Gene Ontology terms, the most prominent of which is "Ovarian Follicle Cell Development." Our results indicate that the origin of gall induction is associated with distinct genomic changes, and provide a good starting point for further characterization of the genes involved.

Bayesian Inference of Ancestral Host-Parasite Interactions under a Phylogenetic Model of Host Repertoire Evolution.

Intimate ecological interactions, such as those between parasites and their hosts, may persist over long time spans, coupling the evolutionary histories of the lineages involved. Most methods that reconstruct the coevolutionary history of such interactions make the simplifying assumption that parasites have a single host. Many methods also focus on congruence between host and parasite phylogenies, using cospeciation as the null model. However, there is an increasing body of evidence suggesting that the host ranges of parasites are more complex: that host ranges often include more than one host and evolve via gains and losses of hosts rather than through cospeciation alone. Here, we develop a Bayesian approach for inferring coevolutionary history based on a model accommodating these complexities. Specifically, a parasite is assumed to have a host repertoire, which includes both potential hosts and one or more actual hosts. Over time, potential hosts can be added or lost, and potential hosts can develop into actual hosts or vice versa. Thus, host colonization is modeled as a two-step process that may potentially be influenced by host relatedness. We first explore the statistical behavior of our model by simulating evolution of host-parasite interactions under a range of parameter values. We then use our approach, implemented in the program RevBayes, to infer the coevolutionary history between 34 Nymphalini butterfly species and 25 angiosperm families. Our analysis suggests that host relatedness among angiosperm families influences how easily Nymphalini lineages gain new hosts. [Ancestral hosts; coevolution; herbivorous insects; probabilistic modeling.].

Completing Linnaeus's inventory of the Swedish insect fauna: Only 5,000 species left?

Despite more than 250 years of taxonomic research, we still have only a vague idea about the true size and composition of the faunas and floras of the planet. Many biodiversity inventories provide limited insight because they focus on a small taxonomic subsample or a tiny geographic area. Here, we report on the size and composition of the Swedish insect fauna, thought to represent roughly half of the diversity of multicellular life in one of the largest European countries. Our results are based on more than a decade of data from the Swedish Taxonomy Initiative and its massive inventory of the country's insect fauna, the Swedish Malaise Trap Project The fauna is considered one of the best known in the world, but the initiative has nevertheless revealed a surprising amount of hidden diversity: more than 3,000 new species (301 new to science) have been documented so far. Here, we use three independent methods to analyze the true size and composition of the fauna at the family or subfamily level: (1) assessments by experts who have been working on the most poorly known groups in the fauna; (2) estimates based on the proportion of new species discovered in the Malaise trap inventory; and (3) extrapolations based on species abundance and incidence data from the inventory. For the last method, we develop a new estimator, the combined non-parametric estimator, which we show is less sensitive to poor coverage of the species pool than other popular estimators. The three methods converge on similar estimates of the size and composition of the fauna, suggesting that it comprises around 33,000 species. Of those, 8,600 (26%) were unknown at the start of the inventory and 5,000 (15%) still await discovery. We analyze the taxonomic and ecological composition of the estimated fauna, and show that most of the new species belong to Hymenoptera and Diptera groups that are decomposers or parasitoids. Thus, current knowledge of the Swedish insect fauna is strongly biased taxonomically and ecologically, and we show that similar but even stronger biases have distorted our understanding of the fauna in the past. We analyze latitudinal gradients in the size and composition of known European insect faunas and show that several of the patterns contradict the Swedish data, presumably due to similar knowledge biases. Addressing these biases is critical in understanding insect biomes and the ecosystem services they provide. Our results emphasize the need to broaden the taxonomic scope of current insect monitoring efforts, a task that is all the more urgent as recent studies indicate a possible worldwide decline in insect faunas.

The Swedish Malaise Trap Project: A 15 Year Retrospective on a Countrywide Insect Inventory.

The Swedish Malaise Trap Project (SMTP) is one of the most ambitious insect inventories ever attempted. The project was designed to target poorly known insect groups across a diverse range of habitats in Sweden. The field campaign involved the deployment of 73 Malaise traps at 55 localities across the country for three years (2003-2006). Over the past 15 years, the collected material has been hand sorted by trained technicians into over 300 taxonomic fractions suitable for expert attention. The resulting collection is a tremendous asset for entomologists around the world, especially as we now face a desperate need for baseline data to evaluate phenomena like insect decline and climate change. Here, we describe the history, organisation, methodology and logistics of the SMTP, focusing on the rationale for the decisions taken and the lessons learned along the way. The SMTP represents one of the early instances of community science applied to large-scale inventory work, with a heavy reliance on volunteers in both the field and the laboratory. We give estimates of both staff effort and volunteer effort involved. The project has been funded by the Swedish Taxonomy Initiative; in total, the inventory has cost less than 30 million SEK (approximately 3.1 million USD). Based on a subset of the samples, we characterise the size and taxonomic composition of the SMTP material. Several different extrapolation methods suggest that the material comprises around 20 million specimens in total. The material is dominated by Diptera (75% of the specimens) and Hymenoptera (15% of specimens). Amongst the Diptera, the dominant groups are Chironomidae (37% of specimens), Sciaridae (15%), Phoridae (13%), Cecidomyiidae (9.5%) and Mycetophilidae (9.4%). Within Hymenoptera, the major groups are Ichneumonidae (44% of specimens), Diaprioidea (19%), Braconidae (9.6%), Platygastroidea (8.5%) and Chalcidoidea (7.9%). The taxonomic composition varies with latitude and season. Several Diptera and Hymenoptera groups are more common in non-summer samples (collected from September to April) and in the North, while others show the opposite pattern. About 1% of the total material has been processed and identified by experts so far. This material represents over 4,000 species. One third of these had not been recorded from Sweden before and almost 700 of them are new to science. These results reveal the large amounts of taxonomic work still needed on Palaearctic insect faunas. Based on the SMTP experiences, we discuss aspects of planning and conducting future large-scale insect inventory projects using mainly traditional approaches in relation to more recent approaches that rely on molecular techniques.

Using Parsimony-Guided Tree Proposals to Accelerate Convergence in Bayesian Phylogenetic Inference.

Sampling across tree space is one of the major challenges in Bayesian phylogenetic inference using Markov chain Monte Carlo (MCMC) algorithms. Standard MCMC tree moves consider small random perturbations of the topology, and select from candidate trees at random or based on the distance between the old and new topologies. MCMC algorithms using such moves tend to get trapped in tree space, making them slow in finding the globally most probable trees (known as "convergence") and in estimating the correct proportions of the different types of them (known as "mixing"). Here, we introduce a new class of moves, which propose trees based on their parsimony scores. The proposal distribution derived from the parsimony scores is a quickly computable albeit rough approximation of the conditional posterior distribution over candidate trees. We demonstrate with simulations that parsimony-guided moves correctly sample the uniform distribution of topologies from the prior. We then evaluate their performance against standard moves using six challenging empirical data sets, for which we were able to obtain accurate reference estimates of the posterior using long MCMC runs, a mix of topology proposals, and Metropolis coupling. On these data sets, ranging in size from 357 to 934 taxa and from 1740 to 5681 sites, we find that single chains using parsimony-guided moves usually converge an order of magnitude faster than chains using standard moves. They also exhibit better mixing, that is, they cover the most probable trees more quickly. Our results show that tree moves based on quick and dirty estimates of the posterior probability can significantly outperform standard moves. Future research will have to show to what extent the performance of such moves can be improved further by finding better ways of approximating the posterior probability, taking the trade-off between accuracy and speed into account. [Bayesian phylogenetic inference; MCMC; parsimony; tree proposal.].

Establishing arthropod community composition using metabarcoding: Surprising inconsistencies between soil samples and preservative ethanol and homogenate from Malaise trap catches.

DNA metabarcoding allows the analysis of insect communities faster and more efficiently than ever before. However, metabarcoding can be conducted through several approaches, and the consistency of results across methods has rarely been studied. We compare the results obtained by DNA metabarcoding of the same communities using two different markers - COI and 16S - and three different sampling methods: (a) homogenized Malaise trap samples (homogenate), (b) preservative ethanol from the same samples, and (c) soil samples. Our results indicate that COI and 16S offer partly complementary information on Malaise trap samples, with each marker detecting a significant number of species not detected by the other. Different sampling methods offer highly divergent estimates of community composition. The community recovered from preservative ethanol of Malaise trap samples is significantly different from that recovered from homogenate. Small and weakly sclerotized insects tend to be overrepresented in ethanol while strong and large taxa are overrepresented in homogenate. For soil samples, highly degenerate COI primers pick up large amounts of nontarget DNA and only 16S provides adequate analyses of insect diversity. However, even with 16S, very little overlap in molecular operational taxonomic unit (MOTU) content was found between the trap and the soil samples. Our results demonstrate that none of the tested sampling approaches is satisfactory on its own. For instance, DNA extraction from preservative ethanol is not a valid replacement for destructive bulk extraction but a complement. In future metabarcoding studies, both should ideally be used together to achieve comprehensive representation of the target community.

Automated Taxonomic Identification of Insects with Expert-Level Accuracy Using Effective Feature Transfer from Convolutional Networks.

Rapid and reliable identification of insects is important in many contexts, from the detection of disease vectors and invasive species to the sorting of material from biodiversity inventories. Because of the shortage of adequate expertise, there has long been an interest in developing automated systems for this task. Previous attempts have been based on laborious and complex handcrafted extraction of image features, but in recent years it has been shown that sophisticated convolutional neural networks (CNNs) can learn to extract relevant features automatically, without human intervention. Unfortunately, reaching expert-level accuracy in CNN identifications requires substantial computational power and huge training data sets, which are often not available for taxonomic tasks. This can be addressed using feature transfer: a CNN that has been pretrained on a generic image classification task is exposed to the taxonomic images of interest, and information about its perception of those images is used in training a simpler, dedicated identification system. Here, we develop an effective method of CNN feature transfer, which achieves expert-level accuracy in taxonomic identification of insects with training sets of 100 images or less per category, depending on the nature of data set. Specifically, we extract rich representations of intermediate to high-level image features from the CNN architecture VGG16 pretrained on the ImageNet data set. This information is submitted to a linear support vector machine classifier, which is trained on the target problem. We tested the performance of our approach on two types of challenging taxonomic tasks: 1) identifying insects to higher groups when they are likely to belong to subgroups that have not been seen previously and 2) identifying visually similar species that are difficult to separate even for experts. For the first task, our approach reached $CDATA[$CDATA[$>$$92% accuracy on one data set (884 face images of 11 families of Diptera, all specimens representing unique species), and $CDATA[$CDATA[$>$$96% accuracy on another (2936 dorsal habitus images of 14 families of Coleoptera, over 90% of specimens belonging to unique species). For the second task, our approach outperformed a leading taxonomic expert on one data set (339 images of three species of the Coleoptera genus Oxythyrea; 97% accuracy), and both humans and traditional automated identification systems on another data set (3845 images of nine species of Plecoptera larvae; 98.6 % accuracy). Reanalyzing several biological image identification tasks studied in the recent literature, we show that our approach is broadly applicable and provides significant improvements over previous methods, whether based on dedicated CNNs, CNN feature transfer, or more traditional techniques. Thus, our method, which is easy to apply, can be highly successful in developing automated taxonomic identification systems even when training data sets are small and computational budgets limited. We conclude by briefly discussing some promising CNN-based research directions in morphological systematics opened up by the success of these techniques in providing accurate diagnostic tools.