Assessing species distribution using Google Street View: a pilot study with the Pine Processionary Moth

Emerging technologies in citizen science and potential for insect monitoring

Emerging technologies are increasingly employed in environmental citizen science projects. This integration offers benefits and opportunities for scientists and participants alike. Citizen science can support large-scale, long-term monitoring of species occurrences, behaviour and interactions. At the same time, technologies can foster participant engagement, regardless of pre-existing taxonomic expertise or experience, and permit new types of data to be collected. Yet, technologies may also create challenges by potentially increasing financial costs, necessitating technological expertise or demanding training of participants. Technology could also reduce people's direct involvement and engagement with nature. In this perspective, we discuss how current technologies have spurred an increase in citizen science projects and how the implementation of emerging technologies in citizen science may enhance scientific impact and public engagement. We show how technology can act as (i) a facilitator of current citizen science and monitoring efforts, (ii) an enabler of new research opportunities, and (iii) a transformer of science, policy and public participation, but could also become (iv) an inhibitor of participation, equity and scientific rigour. Technology is developing fast and promises to provide many exciting opportunities for citizen science and insect monitoring, but while we seize these opportunities, we must remain vigilant against potential risks. This article is part of the theme issue 'Towards a toolkit for global insect biodiversity monitoring'.

Advances in understanding and predicting the spread of invading insect populations

Understanding and predicting the spread of invading insects is a critical challenge in management programs that aim to minimize ecological and economic harm to native ecosystems. Although efforts to quantify spread rates have been well studied over the past several decades, opportunities to improve our ability to estimate rates of spread, and identify the factors, such as habitat suitability and climate, that influence spread, remain. We review emerging sources of data that can be used to delineate distributional boundaries through time and thus serve as a basis for quantifying spread rates. We then address advances in modeling methods that facilitate our understanding of factors that drive invasive insect spread. We conclude by highlighting some remaining challenges in understanding and predicting invasive insect spread, such as the role of climate change and biotic similarity between the native and introduced ranges, particularly as it applies to decision-making in management programs.

Citizen Science and Monitoring Forest Pests: a Beneficial Alliance?

Purpose of the Review

One of the major threats to tree health, and hence the resilience of forests and their provision of ecosystem services, is new and emerging pests. Therefore, forest health monitoring is of major importance to detect invasive, emerging and native pest outbreaks. This is usually done by foresters and forest health experts, but can also be complemented by citizen scientists. Here, we review the use of citizen science for detection and monitoring, as well as for hypothesis-driven research and evaluation of control measures as part of forest pest surveillance and research. We then examine its limitations and opportunities and make recommendations on the use of citizen science for forest pest monitoring.

Recent Findings

The main opportunities of citizen scientists for forest health are early warning, early detection of new pests, monitoring of impact of outbreaks and scientific research. Each domain has its own limitations, opportunities and recommendations to follow, as well as their own public engagement strategies. The development of new technologies provides many opportunities to involve citizen scientists in forest pest monitoring. To enhance the benefits of citizen scientists' inclusion in monitoring, it is important that they are involved in the cocreation of activities.

Summary

Future monitoring and research may benefit from tailor-made citizen science projects to facilitate successful monitoring by citizen scientists and expand their practice to countries where the forest health sector is less developed. In this sense, citizen scientists can help understand and detect outbreaks of new pests and avoid problems in the future.

Mapping Tree Canopy in Urban Environments Using Point Clouds from Airborne Laser Scanning and Street Level Imagery

Resilient cities incorporate a social, ecological, and technological systems perspective through their trees, both in urban and peri-urban forests and linear street trees, and help promote and understand the concept of ecosystem resilience. Urban tree inventories usually involve the collection of field data on the location, genus, species, crown shape and volume, diameter, height, and health status of these trees. In this work, we have developed a multi-stage methodology to update urban tree inventories in a fully automatic way, and we have applied it in the city of Pamplona (Spain). We have compared and combined two of the most common data sources for updating urban tree inventories: Airborne Laser Scanning (ALS) point clouds combined with aerial orthophotographs, and street-level imagery from Google Street View (GSV). Depending on the data source, different methodologies were used to identify the trees. In the first stage, the use of individual tree detection techniques in ALS point clouds was compared with the detection of objects (trees) on street level images using computer vision (CV) techniques. In both cases, a high success rate or recall (number of true positive with respect to all detectable trees) was obtained, where between 85.07% and 86.42% of the trees were well-identified, although many false positives (FPs) or trees that did not exist or that had been confused with other objects were always identified. In order to reduce these errors or FPs, a second stage was designed, where FP debugging was performed through two methodologies: (a) based on the automatic checking of all possible trees with street level images, and (b) through a machine learning binary classification model trained with spectral data from orthophotographs. After this second stage, the recall decreased to about 75% (between 71.43 and 78.18 depending on the procedure used) but most of the false positives were eliminated. The results obtained with both data sources were robust and accurate. We can conclude that the results obtained with the different methodologies are very similar, where the main difference resides in the access to the starting information. While the use of street-level images only allows for the detection of trees growing in trafficable streets and is a source of information that is usually paid for, the use of ALS and aerial orthophotographs allows for the location of trees anywhere in the city, including public and private parks and gardens, and in many countries, these data are freely available.

Deep learning increases the availability of organism photographs taken by citizens in citizen science programs

Citizen science programs using organism photographs have become popular, but there are two problems related to photographs. One problem is the low quality of photographs. It is laborious to identify species in photographs taken outdoors because they are out of focus, partially invisible, or under different lighting conditions. The other is difficulty for non-experts to identify species. Organisms usually have interspecific similarity and intraspecific variation, which hinder species identification by non-experts. Deep learning solves these problems and increases the availability of organism photographs. We trained a deep convolutional neural network, Xception, to identify bee species using various quality of bee photographs that were taken by citizens. These bees belonged to two honey bee species and 10 bumble bee species with interspecific similarity and intraspecific variation. We investigated the accuracy of species identification by biologists and deep learning. The accuracy of species identification by Xception (83.4%) was much higher than that of biologists (53.7%). When we grouped bee photographs by different colors resulting from intraspecific variation in addition to species, the accuracy of species identification by Xception increased to 84.7%. The collaboration with deep learning and experts will increase the reliability of species identification and their use for scientific researches.

The role of passive surveillance and citizen science in plant health

The early detection of plant pests and diseases is vital to the success of any eradication or control programme, but the resources for surveillance are often limited. Plant health authorities can however make use of observations from individuals and stakeholder groups who are monitoring for signs of ill health. Volunteered data is most often discussed in relation to citizen science groups, however these groups are only part of a wider network of professional agents, land-users and owners who can all contribute to significantly increase surveillance efforts through "passive surveillance". These ad-hoc reports represent chance observations by individuals who may not necessarily be looking for signs of pests and diseases when they are discovered. Passive surveillance contributes vital observations in support of national and international surveillance programs, detecting potentially unknown issues in the wider landscape, beyond points of entry and the plant trade. This review sets out to describe various forms of passive surveillance, identify analytical methods that can be applied to these "messy" unstructured data, and indicate how new programs can be established and maintained. Case studies discuss two tree health projects from Great Britain (TreeAlert and Observatree) to illustrate the challenges and successes of existing passive surveillance programmes. When analysing passive surveillance reports it is important to understand the observers' probability to detect and report each plant health issue, which will vary depending on how distinctive the symptoms are and the experience of the observer. It is also vital to assess how representative the reports are and whether they occur more frequently in certain locations. Methods are increasingly available to predict species distributions from large datasets, but more work is needed to understand how these apply to rare events such as new introductions. One solution for general surveillance is to develop and maintain a network of tree health volunteers, but this requires a large investment in training, feedback and engagement to maintain motivation. There are already many working examples of passive surveillance programmes and the suite of options to interpret the resulting datasets is growing rapidly.

Phylogenetic inference enables reconstruction of a long-overlooked outbreak of almond leaf scorch disease (Xylella fastidiosa) in Europe

The recent introductions of the bacterium Xylella fastidiosa (Xf) into Europe are linked to the international plant trade. However, both how and when these entries occurred remains poorly understood. Here, we show how almond scorch leaf disease, which affects ~79% of almond trees in Majorca (Spain) and was previously attributed to fungal pathogens, was in fact triggered by the introduction of Xf around 1993 and subsequently spread to grapevines (Pierceʼs disease). We reconstructed the progression of almond leaf scorch disease by using broad phylogenetic evidence supported by epidemiological data. Bayesian phylogenetic inference predicted that both Xf subspecies found in Majorca, fastidiosa ST1 (95% highest posterior density, HPD: 1990–1997) and multiplex ST81 (95% HPD: 1991–1998), shared their most recent common ancestors with Californian Xf populations associated with almonds and grapevines. Consistent with this chronology, Xf-DNA infections were identified in tree rings dating to 1998. Our findings uncover a previously unknown scenario in Europe and reveal how Pierce’s disease reached the continent.

Eduardo Moralejo et al. report a phylogenetic reconstruction tracing the origin and progression of a European outbreak of the almond scorch disease pathogen Xylella fastidiosa (Xf). Their data suggest Xf was introduced into Europe via grafting from infected Californian buds and was subsequently spread by the meadow spittlebug to multiple plant hosts.

Mapping the status of the North American beaver invasion in the Tierra del Fuego archipelago

Quantifying the presence and environmental impact of invasive species is the starting point for research on management and nature conservation. North American beavers (Castor canadensis) were introduced to Argentina from Canada in 1946, and the species has been identified as a major agent of environmental change in the Tierra del Fuego archipelago in the Anthropocene. We studied the invasion status (distribution and density) of beavers through analyses of the dam densities in the Tierra del Fuego landscapes. We identified beaver dams with a GIS using visual interpretation of high-resolution aerial imagery from Microsoft Bing, Google Earth and HERE and related them to natural environmental gradients. These factors comprised geographic (vegetation zones and distance to streams), climatic (temperature, precipitation, evapotranspiration and net primary productivity) and topographic (elevation and slope) data. The datasets (dams and factors) were combined, and the data from the different zonation classes were subsequently compared using ANOVAs and Tukey's mean comparison tests. Deviations from the mean density (x mean density-x total mean density) were calculated to visualize the deviations for the studied factors. The datasets were also evaluated using principal component analyses (PCA). Our results showed a total of 206,203 beaver dams (100,951 in Argentina and 105,252 in Chile) in the study area (73,000 km2). The main island of Tierra del Fuego presented a greater degree of invasion (73.6% of the total study area) than the rest of the archipelago, especially in areas covered by mixed-evergreen and deciduous forests. The studied geographic, climatic and topographic factors showed positive trends (higher beaver preference) with beaver spread, which were all significant (p <0.05) when compared across the landscape. Although beavers are flexible in their habitat use, our empirical records showed that they had marked preferences and were positively influenced by the most productive forests. Here, we describe a scientific panorama that identified the drivers of species invasion based on satellite data and the available ecological datasets. The identification of such drivers could be useful for developing new tools for management and/or control strategies of the beavers in the Tierra del Fuego archipelago.

Large scale detailed mapping of dengue vector breeding sites using street view images

Targeted environmental and ecosystem management remain crucial in control of dengue. However, providing detailed environmental information on a large scale to effectively target dengue control efforts remains a challenge. An important piece of such information is the extent of the presence of potential dengue vector breeding sites, which consist primarily of open containers such as ceramic jars, buckets, old tires, and flowerpots. In this paper we present the design and implementation of a pipeline to detect outdoor open containers which constitute potential dengue vector breeding sites from geotagged images and to create highly detailed container density maps at unprecedented scale. We implement the approach using Google Street View images which have the advantage of broad coverage and of often being two to three years old which allows correlation analyses of container counts against historical data from manual surveys. Containers comprising eight of the most common breeding sites are detected in the images using convolutional neural network transfer learning. Over a test set of images the object recognition algorithm has an accuracy of 0.91 in terms of F-score. Container density counts are generated and displayed on a decision support dashboard. Analyses of the approach are carried out over three provinces in Thailand. The container counts obtained agree well with container counts from available manual surveys. Multi-variate linear regression relating densities of the eight container types to larval survey data shows good prediction of larval index values with an R-squared of 0.674. To delineate conditions under which the container density counts are indicative of larval counts, a number of factors affecting correlation with larval survey data are analyzed. We conclude that creation of container density maps from geotagged images is a promising approach to providing detailed risk maps at large scale.

Can Field Crews Telecommute? Varied Data Quality from Citizen Science Tree Inventories Conducted Using Street-Level Imagery

Street tree inventories are a critical component of urban forest management. However, inventories conducted in the field by trained professionals are expensive and time-consuming. Inventories relying on citizen scientists or virtual surveys conducted remotely using street-level photographs may greatly reduce the costs of street tree inventories, but there are fundamental uncertainties regarding the level of data quality that can be expected from these emerging approaches to data collection. We asked 16 volunteers to inventory street trees in suburban Chicago using Google Street ViewTM imagery, and we assessed data quality by comparing their virtual survey data to field data from the same locations. We also compared virtual survey data quality according to self-rated expertise by measuring agreement within expert, intermediate, and novice analyst groups. Analyst agreement was very good for the number of trees on each street segment, and agreement was markedly lower for tree diameter class and tree identification at the genus and species levels, respectively. Interrater agreement varied by expertise, such that experts agreed with one another more often than novices for all four variables assessed. Compared to the field data, we observed substantial variability in analyst performance for diameter class estimation and tree identification, and some intermediate analysts performed as well as experts. Our findings suggest that virtual surveys may be useful for documenting the locations of street trees within a city more efficiently than field crews and with a high level of accuracy. However, tree diameter and species identification data were less reliable across all expertise groups, and especially novice analysts. Based on this analysis, virtual street tree inventories are best suited to collecting very basic information such as tree locations, or updating existing inventories to determine where trees have been planted or removed. We conclude with evidence-based recommendations for effective implementation of this type of approach.

Species Distribution 2.0: An Accurate Time- and Cost-Effective Method of Prospection Using Street View Imagery

Species occurrence data provide crucial information for biodiversity studies in the current context of global environmental changes. Such studies often rely on a limited number of occurrence data collected in the field and on pseudo-absences arbitrarily chosen within the study area, which reduces the value of these studies. To overcome this issue, we propose an alternative method of prospection using geo-located street view imagery (SVI). Following a standardised protocol of virtual prospection using both vertical (aerial photographs) and horizontal (SVI) perceptions, we have surveyed 1097 randomly selected cells across Spain (0.1x0.1 degree, i.e. 20% of Spain) for the presence of Arundo donax L. (Poaceae). In total we have detected A. donax in 345 cells, thus substantially expanding beyond the now two-centuries-old field-derived record, which described A. donax only 216 cells. Among the field occurrence cells, 81.1% were confirmed by SVI prospection to be consistent with species presence. In addition, we recorded, by SVI prospection, 752 absences, i.e. cells where A. donax was considered absent. We have also compared the outcomes of climatic niche modeling based on SVI data against those based on field data. Using generalized linear models fitted with bioclimatic predictors, we have found SVI data to provide far more compelling results in terms of niche modeling than does field data as classically used in SDM. This original, cost- and time-effective method provides the means to accurately locate highly visible taxa, reinforce absence data, and predict species distribution without long and expensive in situ prospection. At this time, the majority of available SVI data is restricted to human-disturbed environments that have road networks. However, SVI is becoming increasingly available in natural areas, which means the technique has considerable potential to become an important factor in future biodiversity studies.

Google Street View as an alternative method to car surveys in large-scale vegetation assessments

Car surveys (CS) are a common method for assessing the distribution of alien invasive plants. Google Street View (GSV), a free-access web technology where users may experience a virtual travel along roads, has been suggested as a cost-effective alternative to car surveys. We tested if we could replicate the results from a countrywide survey conducted by car in Portugal using GSV as a remote sensing tool, aiming at assessing the distribution of Eucalyptus globulus Labill. wildlings on roadsides adjacent to eucalypt stands. Georeferenced points gathered along CS were used to create road transects visible as lines overlapping the road in GSV environment, allowing surveying the same sampling areas using both methods. This paper presents the results of the comparison between the two methods. Both methods produced similar models of plant abundance, selecting the same explanatory variables, in the same hierarchical order of importance and depicting a similar influence on plant abundance. Even though the GSV model had a lower performance and the GSV survey detected fewer plants, additional variables collected exclusively with GSV improved model performance and provided a new insight into additional factors influencing plant abundance. The survey using GSV required ca. 9 % of the funds and 62 % of the time needed to accomplish the CS. We conclude that GSV may be a cost-effective alternative to CS. We discuss some advantages and limitations of GSV as a survey method. We forecast that GSV may become a widespread tool in road ecology, particularly in large-scale vegetation assessments.

An exhaustive inventory of coniferous trees in an agricultural landscape

BACKGROUND

Various species of forest trees are commonly used for ornamental purposes and are therefore frequently found in non-forest ecosystems. These trees constitute a significant component of the trees outside forests (TOF). Although increasingly recognized as prominent feature of agricultural lands and built-up areas, not much is known, however, about TOF since they are generally absent from forest inventories.

NEW INFORMATION

In the present study, we focus on the coniferous tree species that constitute potential hosts for a forest defoliator, the pine processionary moth Thaumetopoeapityocampa Den. & Schiff. (Lepidoptera, Notodontidae). We carried out an exhaustive inventory of all pines (Pinus spp.), cedars (Cedrus spp.) and Douglas-fir (Pseudotsugamenziesii) in a 22 × 22 km study window located in the open-field region of Beauce in the centre of France. We recorded a total of 3834 individuals or small groups host trees corresponding a density of 7.9 occurrences per 100 ha. We provide the spatial coordinates of the points without differentiation between tree species.

In search of pathogens: transcriptome-based identification of viral sequences from the pine processionary moth (Thaumetopoea pityocampa)

Thaumetopoea pityocampa (pine processionary moth) is one of the most important pine pests in the forests of Mediterranean countries, Central Europe, the Middle East and North Africa. Apart from causing significant damage to pinewoods, T. pityocampa occurrence is also an issue for public and animal health, as it is responsible for dermatological reactions in humans and animals by contact with its irritating hairs. High throughput sequencing technologies have allowed the fast and cost-effective generation of genetic information of interest to understand different biological aspects of non-model organisms as well as the identification of potential pathogens. Using these technologies, we have obtained and characterized the transcriptome of T. pityocampa larvae collected in 12 different geographical locations in Turkey. cDNA libraries for Illumina sequencing were prepared from four larval tissues, head, gut, fat body and integument. By pooling the sequences from Illumina platform with those previously published using the Roche 454-FLX and Sanger methods we generated the largest reference transcriptome of T. pityocampa. In addition, this study has also allowed identification of possible viral pathogens with potential application in future biocontrol strategies.