• Open Access close
• Embargo close
• CZ close
• English close
• Knowmad Institut close

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 datasets, 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. Specifically, we extract rich representations of intermediate to high-level image features from the CNN architecture VGG16 pretrained on the ImageNet dataset. This information is fed into 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 reaches > 92 % accuracy on one dataset (884 face images of 11 families of Diptera, all specimens representing unique species), and > 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 outperforms a leading taxonomic expert on one dataset (339 images of three species of the Coleoptera genus Oxythyrea; 97 % accuracy), and both humans and traditional automated identification systems on another dataset (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 datasets are small and computational budgets limited. valan2018_SUPPLEMENT.tarThis directory contains: * metadata to obtain images from the three novel datasets we designed for our study * notebooks with thorough evaluation of off-the-shelf approach for image classification based on a feature extraction with a single feed forward pass trough pretrained VGG16 * script to run on your own dataset with what we found to be optimal settings. You can also access it here https://github.com/valanm/off-the-shelf-insect-identificationSupplementary_Figure1Impact of concatenating globally max pooled (MAX) and globally average pooled (AVG) features on identification accuracy for datasets D1, D2, D3 and D4. We used input images of size 416x416 and features are extracted after 4th convolutional block (c4). Concatenation of MAX and AVG features resulted in accuracy somewhere between the global average pooling (performed the best in all cases) and global max pooling (performed the worst in all cases).valan2018_latex

Human-driven environmental changes can induce marked shifts in the functional structure of biological communities with possible repercussion on important ecosystem functions and services. At the same time it remains unclear to which extent these changes may differently affect various types of organisms. We investigated species richness and community functional structure of species assemblages at the landscape scale (1km2 plots) for two contrasting model taxa, i.e., plants (producers and sessile organisms) and birds (consumers and mobile organisms), along topography, climate, landscape heterogeneity, and land-use (agriculture and urbanization) gradients in a densely populated region of Switzerland. Our study revealed that agricultural and urban land uses drove marked shifts in the functional structure of biological communities compared to changes along climate and topography gradients, especially for plants, while for birds these changes were comparable. Agricultural and urban land uses enhanced divergence in traits related to resource use for birds (diet and nesting), growth forms, dispersal, and reproductive traits for plants, while it induced convergence in vegetative plant traits (plant height and leaf dry matter content). These results suggest that contrasting assembly patterns may arise within and across taxonomic groups along the same environmental gradients as result of distinct underlying processes and 'organism-specific' environmental perceptions. Our results further suggest a potential homogenization of biological communities, as well as low functional diversity and redundancy levels of bird assemblages in our human-dominated study region. This might potentially compromise the maintenance of key ecological processes under future environmental changes. SES for plant and bird functional traits based on 1000 randomizationsSES_Data.zip

Predators attack conspicuous prey phenotypes that are present in the environment. Male display behaviour of conspicuous nuptial colouration becomes risky in the presence of a predator, and adult males face higher predation risk. High predation risk in one sex will lead to low survival and sex ratio bias in adult cohorts, unless the increased predation risk is compensated by higher escape rate. Here, we tested the hypothesis that sand lizards (Lacerta agilis) have sex-specific predation risk and escape rate. We expected the differences to manifest in changes in sex ratio with age, differences in frequency of tail autotomy, and in sex-specific survival rate. We developed a statistical model to estimate predation risk and escape rate, combining the observed sex ratio and frequency of tail autotomy with likelihood-based survival rate. Using a Bayesian framework, we estimated the model parameters. We projected the date of the tail autotomy events from growth rates derived from capture-recapture data measurements. We found statistically stable sex ratio in age groups, equal frequency of tail regenerates between sexes, and similar survival rate. Predation risk is similar between sexes, and escape rate increases survival by about 5%. We found low survival rate and a low number of tail autotomy events in females during months when sand lizards mate and lay eggs, indicating high predator pressure throughout reproduction. Our data show that gravid females fail to escape predation. The risks of reproduction season in an ectotherm are a convolution of morphological changes (conspicuous colouration in males, body allometry changes in gravid females), behaviour (nuptial displays), and environmental conditions which challenge lizard thermal performance. Performance of endotherm predators in cold spring months endangers gravid females more than displaying males in bright nuptial colouration.  Capture-recapture data of sand lizards (Lacerta agilis, Lacertidae, Reptilia) from Hustopeče, Czechia (48.93 N, 16.72 E). animal - identification of the animal; prefix indicates the latest possible year of hatching day - day of capture month - month of capture year - year of capture season - phase I started from the arousal from hibernation, and lasted until the first adult male started to lose nuptial colouration; phase II constituted the season after males started to lose nuptial colouration, and lasted until the beginning of hibernation sex - sex of the animal ventralia - number of scales in the second rows from the ventral medial line age - young animals were juveniles and subadults, the remaining animals were considered adult aged.month - approximate age of the animal in months regenerate - presence of tail regenerate predation.attempt - first record of tail regenerate or reduced tail length at recapture Lreg - tail regenerate length in mm; 0 - animal with intact tail, 0.01 - recent tail autotomy, no tail regenerate growth measurable tailL - tail length from cloaca to tail tip corrected for body length from rostrum to cloaca48.93 N, 16.72 E48.93 N, 16.72 E)

Aim Temporal dynamics of biodiversity along tropical elevational gradients are unknown. We studied seasonal changes of Lepidoptera biodiversity along the only complete forest elevational gradient in the Afrotropics. We focused on shifts of species richness patterns, seasonal turnover of communities, and seasonal shifts of species’ elevational ranges, the latter often serving as an indicator of the global change effects on mountain ecosystems. Location Mount Cameroon, Cameroon. Taxon Butterflies and moths (Lepidoptera) Methods We quantitatively sampled nine groups of Lepidoptera by bait-trapping (16,800 trap-days) and light-catching (126 nights) at seven elevations evenly distributed along the elevational gradient from sea level (30 m asl) to timberline (2,200 m asl). Sampling was repeated in three seasons. Result Altogether, 42,936 specimens of 1,099 species were recorded. A mid-elevation peak of species richness was detected for all groups but Eupterotidae. This peak shifted seasonally for five groups, most of them ascending during the dry season. Seasonal shifts of species’ elevational ranges were mostly responsible for these diversity pattern shifts along elevation: we found general upward shifts in fruit-feeding butterflies, fruit-feeding moths and Lymantriinae from beginning to end of the dry season. Contrarily, Arctiinae shifted upwards during the wet season. The average seasonal shifts of elevational ranges often exceeded 100 metres and were even several times higher for numerous species. Main conclusion We report seasonal uphill and downhill shifts of several lepidopteran groups. The reported shifts can be driven by both delay in weather seasonality and shifts in resource availability, causing phenological delay of adult hatching and/or adult migrations. Such shifts may lead to misinterpretations of diversity patterns along elevation if seasonality is ignored. More importantly, considering the surprising extent of seasonal elevational shifts of species, we encourage taking account of such natural temporal dynamics while investigating the global climate change impact on communities of Lepidoptera in tropical mountains. The dataset was collected by two methodologies: 1/ bait-trapping and 2/ manual catching of target group at light. See Maicher et al. (2019) for details.

Bioluminescent beetles of the superfamily Elateroidea (fireflies, fire beetles, glow-worms) are the most speciose group of terrestrial light-producing animals. The evolution of bioluminescence in elateroids is associated with unusual morphological modifications, such as soft-bodiedness and neoteny, but the fragmentary nature of the fossil record discloses little about the origin of these adaptations. We report the discovery of a new bioluminescent elateroid beetle family from the mid-Cretaceous of northern Myanmar (ca. 99 Ma), Cretophengodidae fam. nov. Cretophengodes azari gen. et sp. nov. belongs to the bioluminescent lampyroid clade, and represents a transitional fossil linking the soft-bodied Phengodidae + Rhagophthalmidae clade and hard-bodied elateroids. The fossil male possesses a light organ on the abdomen which presumably served a defensive function, documenting a Cretaceous radiation of bioluminescent beetles coinciding with the diversification of major insectivore groups such as frogs and stem-group birds. The discovery adds a key branch to the elateroid tree of life and sheds light on the timing of the evolution of soft-bodiedness and historical biogeography of elateroid beetles. The Burmese amber specimen studied here originates from amber mines near the Noije Bum Hill (26°20' N, 96°36' E), Hukawng Valley, Kachin State, northern Myanmar. The specimen is deposited in the Nanjing Institute of Geology and Palaeontology (NIGP), Chinese Academy of Sciences, Nanjing, China. The amber piece was trimmed with a small table saw, ground with emery papers of different grain sizes, and finally polished with polishing powder. Photographs under incident light were taken with a Zeiss Discovery V20 stereo microscope. Widefield fluorescence images were captured with a Zeiss Axio Imager 2 light microscope combined with a fluorescence imaging system. Confocal images were obtained with a Zeiss LSM710 confocal laser scanning microscope. Images under incident light and widefield fluorescence were stacked in Helicon Focus 7.0.2 or Zerene Stacker 1.04. Confocal images were manually stacked in Adobe Photoshop CC. Images were further processed in Adobe Photoshop CC to enhance contrast. 

1. When searching for food, great tits (Parus major) can use herbivore-induced plant volatiles (HIPVs) as an indicator of arthropod presence. Their ability to detect HIPVs was shown to be learned, and not innate, yet the flexibility and generalization of learning remains unclear. 2. We studied if, and if so how, naïve and trained great tits (Parus major) discriminate between herbivore-induced and non-induced saplings of Scotch elm (Ulmus glabra) and cattley guava (Psidium cattleyanum). We chemically analysed the used plants and showed that their HIPVs differed significantly and overlapped only in a few compounds. 3. Birds trained to discriminate between herbivore-induced and non-induced saplings preferred the herbivore-induced saplings of the plant species they were trained to. Naïve birds did not show any preferences. Our results indicate that the attraction of great tits to herbivore-induced plants is not innate, rather it is a skill that can be acquired through learning, one tree species at a time. 4. We demonstrate that the ability to learn to associate HIPVs with food reward is flexible, expressed to both tested plant species, even if the plant species has not coevolved with the bird species (i.e. guava). Our results imply that the birds are not capable of generalising HIPVs among tree species but suggest that they either learn to detect individual compounds or associate whole bouquets with food rewards. Birds were split into three groups. Two groups were trained to respective smells or kept naive (Trained to: Psidium, Ulmus, Control). Bird = ID of the bird and each bird entered severa pair-wise tests in which it was suppose to search on trees familiar/unfamiliar to it and trees either chemicall induced or non-induced. The "induced psidium or induced ulmus " were considered as a focal tree, to which the data were related. The type of the tree on the other side of the cage was used as an explanatory factor and is noted in column "Other side". Measured variable were TimeClose = number of seconds spent in the proximity of the tree, TimeOn - time spent onthe tree by real searching for prey, these two variables were then related to the total length of the time spent (TotalTimeCloseTree, TotalTimeOnTree) at any of the offered tree. THe values were expressed as bonamial (Binom) or in percentages (perc). Experiments were run at two cages, their nick-names are "wooden" and "metal".

To compare long-term real-life results of dasatinib therapy to clinical trial data among chronic phase chronic myeloid leukemia patients resistant or intolerant to imatinib, we retrospectively analysed data of 118 patients treated in centres participating in the database INFINITY. Presented dataset was prepared as a data export from INFINITY database, comprehensive Czech database of patients with chronic myeloid leukemia treated in participated specialised hematological centers in the Czech Republic.

Despite the dramatic phenological responses of fungal fruiting to recent climate warming, it is unknown whether spatial distributions of fungi have changed and to what extent such changes are influenced by fungal traits, such as ectomycorrhizal (ECM) or saprotrophic lifestyles, spore characteristics, or fruit body size. Our overall aim was to understand how climate and fungal traits determine whether and how species-specific fungal fruit body abundances have shifted across latitudes over time, using the UK national database of fruiting records. The data employed were recorded over 45 years (1970 – 2014), and include 853,278 records of Agaricales, Boletales and Russulales, though we focus only on the most common species (with more than 3,000 records each). The georeferenced observations were analysed by a Bayesian inference as a Gaussian Additive Model with a specification following a joint species distribution model. We used an offset, random contributions and fixed effects to isolate different potential biases from the trait-specific interactions with latitude/climate and time. Our main aim was assessed by examination of the three-way-interaction of trait, predictor (latitude or climate) and time. The results show a strong trait-specific shift in latitudinal abundance through time, as ECM species have become more abundant relative to saprotrophic species in the north. Along precipitation gradients, phenology was important, in that species with shorter fruiting seasons have declined markedly in abundance in oceanic regions, whereas species with longer seasons have become relatively more common overall. These changes in fruit body distributions are correlated with temperature and rainfall, which act directly on both saprotrophic and ECM fungi, and also indirectly on ECM fungi, through altered photosynthate allocation from their hosts. If these distributional changes reflect fungal activity, there will be important consequences for the responses of forest ecosystems to changing climate, through effects on primary production and nutrient cycling. Gange et al EcographyInput raw data for all analyses. The file includes record information about taxonomy (species), location, traits and environmental data associated with each record.

A space-time extremely randomised trees model was used to estimate daily (between 10 a.m. and 2 p.m.) PM2.5 concentrations with 1 km spatial resolution for a three-year period 2018–2020 over Europe. Satellite remote sensing, meteorological data, and land variables were used as the independent variables, PM2.5 ground-observations were used as the dependent variable while building the model. This work is co-financed under Grant Agreement Connecting Europe Facility (CEF) Telecom project 2018-EU-IA-0095 by the European Union.