bioRxiv Science⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/034819

Insect Wing Classification of Mosquitoes and Bees Using CO1 Image Recognition.

Nayna Vyas-Patel·Sai Ravela·Agenor Mafra-Neto·John D Mumford
View original source ↗Export metadata ↓

Abstract

The certainty that a species is accurately identified is the cornerstone of appearance based classification; however the methods used in classical taxonomy have yet to fully catch up with the digital age. Recognising this, the CO1 algorithm presented on the StripeSpotter platform was used to identify different species and sexes of mosquito wings (Diptera: Culicidae) and honey bee and bumblebee wings (Hymenoptera: Apidae). Images of different species of mosquito and bee wings were uploaded onto the CO1 database and test wing images were analysed to determine if this resulted in the correct species being identified. Out of a database containing 925 mosquito and bee wing images, the CO1 algorithm correctly identified species and sexes of test wing image presented, with a high degree of accuracy (80% to 100% depending on the species and database used, excluding sibling species) highlighting the usefulness of CO1 in identifying medically important as well as beneficial insect species. Using a larger database of wing images resulted in significantly higher numbers of test images being correctly identified than using a smaller database. The hind wings of Hymenoptera provided higher levels of correctly identified results than using the fore wings. The software should be used in conjunction with other identifying criteria (salient morphological features) in addition to the wings. CO1 is a powerful algorithm to use in identifying insect wings in its current form and has great potential if it is adapted and tailored for insect species identification. It is suggested that a primary aim in the digital age should be the production of a World Wide Database of insect images, where all known insect images can be made available to everyone, with image recognition and species knowledge at its core.

Source connections

Explore related subjects

EcologyPreprint

Keep this discovery

BibTeXRIS

Nayna Vyas-Patel, Sai Ravela, Agenor Mafra-Neto, John D Mumford. 2015-12-18. Insect Wing Classification of Mosquitoes and Bees Using CO1 Image Recognition.. https://doi.org/10.1101/034819

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2016

Temperature drives plant and soil microbial diversity patterns across an elevation gradient from the Andes to the Amazon

More than 200 years ago, von Humboldt reported decreases in tropical plant species richness with increasing elevation and decreasing temperature. Surprisingly, co-ordinated patterns in plant, bacterial and fungal diversity on tropical mountains are yet to be observed, despite the central role of soil microorganisms in terrestrial biogeochemistry. We studied an Andean transect traversing 3.5 km in elevation to test whether the species diversity and composition of tropical forest plants, soil bacteria and fungi can follow similar biogeographical patterns with shared environmental drivers. We found co-ordinated changes with elevation in all three groups: species richness declined as elevation increased, and the compositional-dissimilarity of communities increased with increased separation in elevation, although changes in plant diversity were larger than in bacteria and fungi. Temperature was the dominant driver of these diversity gradients, with weak influences of edaphic properties, including soil pH. The gradients in microbial diversity were strongly correlated with the activities of enzymes involved in organic matter cycling, and were accompanied by a transition in microbial traits towards slower-growing, oligotrophic taxa at higher elevations. We provide the first evidence of co-ordinated temperature-driven patterns in the diversity and distribution of three major biotic groups in tropical ecosystems: soil bacteria, fungi and plants. These findings suggest that, across landscape scales of relatively constant soil pH, inter-related patterns of plant and microbial communities with shared environmental drivers can occur, with large implications for tropical forest communities under future climate change.

Ecology↗
bioRxiv2016

A roadmap for a quantitative ecosystem-based environmental impact assessment

A new roadmap for quantitative methodologies of Environmental Impact Assessment (EIA) is proposed, using an ecosystem-based approach. EIA recommendations are currently based on case-by-case rankings, distant from statistical methodologies, and based on ecological ideas that lack proof of generality or predictive capacities. These qualitative approaches ignore process dynamics, scales of variations and interdependencies and are unable to address societal demands to link socio-economic and ecological processes (e.g. population dynamics). We propose to re-focus EIA around the systemic formulation of interactions between organisms (organized in populations and communities) and their environments but inserted within a strict statistical framework. A systemic formulation allows scenarios to be built that simulate impacts on chosen receptors. To illustrate the approach, we design a minimum ecosystem model that demonstrates non-trivial effects and complex responses to environmental changes. We suggest further that an Ecosystem-Based EIA - in which the socio-economic system is an evolving driver of the ecological one - is more promising than a socio-economic-ecological system where all variables are treated as equal. This refocuses the debate on cause-and-effect, processes, identification of essential portable variables, and a potential for quantitative comparisons between projects, which is important in cumulative effects determinations.

Ecology↗
bioRxiv2016

MaxEnt′s parameter configuration and small samples: Are we paying attention to recommendations?

Environmental niche modeling (ENM) is commonly used to develop probabilistic maps of species distribution. Among available ENM techniques, MaxEnt has become one of the most popular tools for modeling species distribution, with hundreds of peer-reviewed articles published each year. MaxEnts popularity is mainly due to the use of a graphical interface and automatic parameter configuration capabilities. However, recent studies have shown that using the default automatic configuration may not be always appropriate because it can produce non-optimal models; particularly when dealing with a small number of species presence points. Thus, the recommendation is to evaluate the best potential combination of parameters (feature classes and regularization multiplier) to select the most appropriate model. In this work we reviewed 244 articles from 142 journals between 2013 and 2015 to assess whether researchers are following recommendations to avoid using the default parameter configuration when dealing with small sample sizes, or if they are using MaxEnt as a \"black box tool\". Our results show that in only 16% of analyzed articles authors evaluated best feature classes, in 6.9% evaluated best regularization multipliers, and in a meager 3.7% evaluated simultaneously both parameters before producing the definitive distribution model. These results are worrying, because publications are potentially reporting over-complex or over-simplistic models that can undermine the applicability of their results. Of particular importance are studies used to inform policy making. Therefore, researchers, practitioners, reviewers and editors need to be very judicious when dealing with MaxEnt, particularly when the modelling process is based on small sample sizes.

Ecology↗