Multi-Input Convolutional Neural Networks for Automatic Pollen Classification

Pollen allergies are a cause of much suffering for an increasing number of individuals. Current pollen monitoring techniques are lacking due to their reliance on manual counting and classification of pollen by human technicians. In this study, we present a neural network architecture capable of distinguishing pollen species using data from an automated particle measurement device. This work presents an improvement over the current state of the art in the task of automated pollen classification, using fluorescence spectrum data of aerosol particles. We obtained a relative reduction in the error rate of over 48%, from 27% to 14%, for one of the datasets, with similar improvements for the other analyzed datasets. We also use a novel approach for doing hyperparameter tuning for multiple input networks.

Lower pollen nutritional quality delays nest building and egg laying in Bombus terrestris audax micro-colonies leading to reduced biomass gain

The performance of Bombus terrestris micro-colonies fed five diets differing in pollen species composition and level of nine essential amino acids (EAA; leucine, lysine, valine, arginine, isoleucine, phenylalanine, threonine, histidine, methionine) was assessed for 37 days by recording total biomass gain, nest building initiation, brood production (eggs, small and large larvae, pupae, drones), nectar, and pollen collection. Stronger colony performance was linked to higher amino acid levels but no consistent differences in biomass gain were recorded between mono- and poly-species diets. Poorest performance occurred in micro-colonies offered pure oilseed rape (OSR) pollen which contained the lowest EAA levels. Reduced micro-colony development (delayed nest initiation and lower brood production) was related to OSR proportion in the diet and lower EAA levels. Results are discussed in relation to selection of plant species in the design of habitats to promote bee populations.

Survey of bee friendly flowering plants and bee-plant interaction in an urban green space in Bengaluru, India

Aim: To study bee friendly plant species, nutritional sources, flowering season, and the dynamic relationship between urban flora and native bee species in a centrally located urban green space in Bengaluru, India. Methodology: Flowers of different plant species visited by bees were observed and recorded from September 2018 to August 2019. Based on the foraging pattern of visiting bees, the plants were classified into nectar or pollen or both nectar and pollen species. The monthly abundance of nutritional resources was estimated based on the floral phenology. Results: A total of 51 plant species, from 25 families, were visited by bees for foraging. Polylectic social bees namely Apis florea and Tetragonula iridipennis, visited 45 and 39 plant species, and two species of solitary bees, namely Amigella cingulate and Xylocopa violacea, visited 26 and 23 plant species, respectively. The urban green landscape was dominated by a variety of ornamental plants (49%) and also included vegetables (17.6%), fruit trees (13.7%), and weeds (19.6%). Plants that served as a source of both nectar and pollen (60.8%) were predominant over those that supplied either nectar alone (24.5%) or pollen alone (13.7%). Moreover, 72% of the species bloomed all the year round, which meant that floral resources were available to bees throughout the year. Interpretation: The study underscores the role of bee friendly floral diversity in the urban green spaces in protection and conservation of bee diversity. Efficient management of urban green spaces can provide dynamic habitat for bee conservation and can prevent the loss of biodiversity.

Metabarcoding Analysis of Pollen Species Foraged by Osmia excavata Alfken (Hymenoptera: Megachilidae) in China

To meet the pollination need of economic crops, Osmia excavata has been successfully used to improve the pollination efficiency of Rosaceae and Brassicaceae plants. As a widely used pollinator of economic crops, a systematic study of flower-visiting species and diversities of O. excavata stocked in China was not found. To investigate the foraging pollen species and diversities of O. excavata, beebread from 20 experimental plots in China was collected by the trap-nesting method and analyzed by DNA metabarcoding technology. A total of 26 pollen plants in 14 genera and nine families were identified. A further analysis showed that the richness and abundance of the wild flowering plants in orchards and farmlands were lower than those in the nearby semi-natural habitats. The favorite pollen comes from economic crops apple and rape and wild flowering plants Juncus interior, Rosa gymnocarpa, and Rosa laevigata. Through a diversity index analysis, it was found that the Anhui region has the highest pollen plant diversity, while the Liaoning region has the lowest. Our results can provide a basis for flower-visiting species and diversities of O. excavata.

Use of 3D Laser Microscopy to Determine Plant Pollen

Pollen analysis as a part of palynology deals with the morphological determination of pollen and spores. Different technologies with different resolutions varying from simple light microscopy to highly elaborate electron microscopy are used for the examination, depending on the area of application (e.g. sedimentology, melissopalynology, forensic palynology, etc.). To answer the question of whether laser scanning microscopy (LSM) can replace scanning electron microscopy (SEM) for the determination of pollen species, 168 species were examined using LSM. It was concluded that LSM is both efficient and easy to handle. After preparing the fresh pollen, a 3D laser scan takes 5-10 minutes and unlike using SEM, the pollen does not have to be sputtered or processed. The 3D scans can be measured quickly and easily with the integrated software and there were no observable artifacts. At magnifications up to 8545x, the image quality is comparable to that of a sputtered SEM sample whereas at higher magnifications, the SEM method is superior. Overall, pollen display by LSM is much less time consuming and more cost effective than with the SEM method.

Operational aspects of machine learning in a met-service

<p>Machine Learning has a big potential for various tasks along the whole value chain of a national Met-Service. Indeed, many research groups, private and national weather services have started to explore the possibilities and first real-time operational implementations are in place already. However, the building up of the expertise is difficult, large amounts of data have to be made available in an efficient way and the necessary tools have special and demanding requirements concerning infrastructure and maintenance. Also, the transition from research results towards operational tools being operated in realtime is a particular challenge. Not least, trust from end-users must be built, while trying to avoid falling into the short-term hype trap.</p><p>In this presentation, we want to present some examples of machine-learning at MeteoSwiss that are in operational use or soon to be. This includes the use in a measurement system to identify pollen species, the quality control of meteorological observations, the postprocessing of numerical weather forecasts and the condensation of weather forecast information for the meteorologists. These examples have different characteristics and cover a wide range of applications, but also share some common properties. We want to juxtapose these properties with the incentives and conditions how machine learning methods are developed and employed in a more research oriented context like in academia. It turns out that an operational setup of machine learning has very different requirements than machine learning in a research context. The identification of these differences, but also the similarities, could help to understand the challenge of bringing research results into operation and how to alleviate this challenge in the future.</p>

50 Years of Pollen Monitoring in Basel (Switzerland) Demonstrate the Influence of Climate Change on Airborne Pollen

Climate change and human impact on vegetation modify the timing and the intensity of the pollen season. The 50 years of pollen monitoring in Basel, Switzerland provide a unique opportunity to study long-term changes in pollen data. Since 1969, pollen monitoring has been carried out in Basel with a Hirst-type pollen trap. Pollen season parameters for start dates, end dates and duration were calculated with different pollen season definitions, which are commonly used in aerobiology. Intensity was analyzed by the annual pollen integral (APIn), peak value and the number of days above specific thresholds. Linear trends were calculated with the non-parametric Mann Kendall method with a Theil-Sen linear trend slope. During the last 50 years, linear increase of the monthly mean temperatures in Basel was 0.95–1.95°C in the 3 winter months, 2–3.7°C in spring months and 2.75–3.85°C in summer months. Due to this temperature increase, the start dates of the pollen season for most of the spring pollen species have advanced, from 7 days for Poaceae to 29 days for Taxus/Cupressaceae. End dates of the pollen season depend on the chosen pollen season definition. Negative trends predominate, i.e., the pollen season mostly ends earlier. Trends in the length of the pollen season depend even more on the season definitions and results are contradictory and often not significant. The intensity of the pollen season of almost all tree pollen taxa increased significantly, while the Poaceae pollen season did not change and the pollen season of herbs decreased, except for Urticaceae pollen. Climate change has a particular impact on the pollen season, but the definitions used for the pollen season parameters are crucial for the calculation of the trends. The most stable results were achieved with threshold definitions that indicate regular occurrence above certain concentrations. Percentage definitions are not recommended for trend studies when the annual pollen integral changed significantly.

How to really help bees: nutrient demand and supply in a changing environment

Local and global changes affect which pollen varieties are available to bees in the environment. Therefore, wild bees cannot always access the optimally balanced diet required for their survival. Our feeding experiment showed that the nutritional quality of the pollen diet eaten by bee larvae is shaped not by pollen diversity but by a specific pollen species composition that results in specific nutrients being scarce or sufficient; this species composition influences bee survivability, development and mass. We proposed that the functioning of bee populations and communities may depend on the floral diversity of the local habitat, which determines whether a nutritionally balanced pollen diet obtained from specific species can be provided to bee larvae. Holistically considering wild bee fitness and health and the different characteristics of the food base at both the ecosystem and bee biology levels can provide new, important knowledge for conserving bees and their critical ecological roles.