Species distribution modeling based on the automated identification of citizen observations.
Ontology highlight
ABSTRACT: Premise of the study
A species distribution model computed with automatically identified plant observations was developed and evaluated to contribute to future ecological studies.Methods
We used deep learning techniques to automatically identify opportunistic plant observations made by citizens through a popular mobile application. We compared species distribution modeling of invasive alien plants based on these data to inventories made by experts.Results
The trained models have a reasonable predictive effectiveness for some species, but they are biased by the massive presence of cultivated specimens.Discussion
The method proposed here allows for fine-grained and regular monitoring of some species of interest based on opportunistic observations. More in-depth investigation of the typology of the observations and the sampling bias should help improve the approach in the future.
SUBMITTER: Botella C
PROVIDER: S-EPMC5851560 | biostudies-literature | 2018 Feb
REPOSITORIES: biostudies-literature
Publications
Species distribution modeling based on the automated identification of citizen observations.
Botella Christophe C Joly Alexis A Bonnet Pierre P Monestiez Pascal P Munoz François F
Applications in plant sciences 20180201 2
<h4>Premise of the study</h4>A species distribution model computed with automatically identified plant observations was developed and evaluated to contribute to future ecological studies.<h4>Methods</h4>We used deep learning techniques to automatically identify opportunistic plant observations made by citizens through a popular mobile application. We compared species distribution modeling of invasive alien plants based on these data to inventories made by experts.<h4>Results</h4>The trained mode ...[more]
Similar Datasets
Project description:Citizen scientists play an increasingly important role in biodiversity monitoring. Most of the data, however, are unstructured-collected by diverse methods that are not documented with the data. Insufficient understanding of the data collection processes presents a major barrier to the use of citizen science data in biodiversity research. We developed a questionnaire to ask citizen scientists about their decision-making before, during and after collecting and reporting species observations, using Germany as a case study. We quantified the greatest sources of variability among respondents and assessed whether motivations and experience related to any aspect of data collection. Our questionnaire was answered by almost 900 people, with varying taxonomic foci and expertise. Respondents were most often motivated by improving species knowledge and supporting conservation, but there were no linkages between motivations and data collection methods. By contrast, variables related to experience and knowledge, such as membership of a natural history society, were linked with a greater propensity to conduct planned searches, during which typically all species were reported. Our findings have implications for how citizen science data are analysed in statistical models; highlight the importance of natural history societies and provide pointers to where citizen science projects might be further developed.
| S-EPMC9245884 | biostudies-literature
Project description:Technological advances and data availability have enabled artificial intelligence-driven tools that can increasingly successfully assist in identifying species from images. Especially within citizen science, an emerging source of information filling the knowledge gaps needed to solve the biodiversity crisis, such tools can allow participants to recognize and report more poorly known species. This can be an important tool in addressing the substantial taxonomic bias in biodiversity data, where broadly recognized, charismatic species are highly over-represented. Meanwhile, the recognition models are trained using the same biased data, so it is important to consider what additional images are needed to improve recognition models. In this study, we investigated how the amount of training data influenced the performance of species recognition models for various taxa. We utilized a large citizen science dataset collected in Norway, where images are added independently from identification. We demonstrate that while adding images of currently under-represented taxa will generally improve recognition models more, there are important deviations from this general pattern. Thus, a more focused prioritization of data collection beyond the basic paradigm that “more is better” is likely to significantly improve species recognition models and advance the representativeness of biodiversity data.
| S-EPMC9090737 | biostudies-literature
Project description:Species distribution modeling, which allows users to predict the spatial distribution of species with the use of environmental covariates, has become increasingly popular, with many software platforms providing tools to fit such models. However, the species observations used can have varying levels of quality and can have incomplete information, such as uncertain or unknown species identity.In this paper, we develop two algorithms to classify observations with unknown species identities which simultaneously predict several species distributions using spatial point processes. Through simulations, we compare the performance of these algorithms using 7 different initializations to the performance of models fitted using only the observations with known species identity.We show that performance varies with differences in correlation among species distributions, species abundance, and the proportion of observations with unknown species identities. Additionally, some of the methods developed here outperformed the models that did not use the misspecified data. We applied the best-performing methods to a dataset of three frog species (Mixophyes).These models represent a helpful and promising tool for opportunistic surveys where misidentification is possible or for the distribution of species newly separated in their taxonomy.
| S-EPMC8131797 | biostudies-literature
Project description:In its invasive range in Australia, the European rabbit threatens the persistence of native flora and fauna and damages agricultural production. Understanding its distribution and ecological niche is critical for developing management plans to reduce populations and avoid further biodiversity and economic losses.We developed an ensemble of species distribution models (SDMs) to determine the geographic range limits and habitat suitability of the rabbit in Australia. We examined the advantage of incorporating data collected by citizens (separately and jointly with expert data) and explored issues of spatial biases in occurrence data by implementing different approaches to generate pseudo-absences. We evaluated the skill of our model using three approaches: cross-validation, out-of-region validation, and evaluation of the covariate response curves according to expert knowledge of rabbit ecology.Combining citizen and expert occurrence data improved model skill based on cross-validation, spatially reproduced important aspects of rabbit ecology, and reduced the need to extrapolate results beyond the studied areas.Our ensemble model projects that rabbits are distributed across approximately two thirds of Australia. Annual maximum temperatures >25°C and annual minimum temperatures >10°C define, respectively, the southern and northern most range limits of its distribution. In the arid and central regions, close access to permanent water (≤~ 0.4 km) and reduced clay soil composition (~20%-50%) were the major factors influencing the probability of occurrence of rabbits. Synthesis and applications. Our results show that citizen science data can play an important role in managing invasive species by providing missing information on occurrences in regions not surveyed by experts because of logistics or financial constraints. The additional sampling effort provided by citizens can improve the capacity of SDMs to capture important elements of a species ecological niche, improving the capacity of statistical models to accurately predict the geographic range of invasive species.
| S-EPMC6802020 | biostudies-literature
Project description:Citizen science platforms are increasingly growing, and, storing a huge amount of data on species locations, they provide researchers with essential information to develop sound strategies for species conservation. However, the lack of information on surveyed sites (i.e., where the observers did not record the target species) and sampling effort (e.g., the number of surveys at a given site, by how many observers, and for how much time) strongly limit the use of citizen science data. Thus, we examined the advantage of using an observer-oriented approach (i.e., considering occurrences of species other than the target species collected by the observers of the target species as pseudo-absences and additional predictors relative to the total number of observations, observers, and days in which locations were collected in a given sampling unit, as proxies of sampling effort) to develop species distribution models. Specifically, we considered 15 mammal species occurring in Italy and compared the predictive accuracy of the ensemble predictions of nine species distribution models carried out considering random pseudo-absences versus observer-oriented approach. Through cross-validations, we found that the observer-oriented approach improved species distribution models, providing a higher predictive accuracy than random pseudo-absences. Our results showed that species distribution modeling developed using pseudo-absences derived citizen science data outperform those carried out using random pseudo-absences and thus improve the capacity of species distribution models to accurately predict the geographic range of species when deriving robust surrogate of sampling effort.
| S-EPMC7663073 | biostudies-literature
Project description:Correlative species distribution models (SDMs) are increasingly being used to predict suitable insect habitats. There is also much criticism of prediction discrepancies among different SDMs for the same species and the lack of effective communication about SDM prediction uncertainty. In this paper, we undertook a factorial study to investigate the effects of various modeling components (species-training-datasets, predictor variables, dimension-reduction methods, and model types) on the accuracy of SDM predictions, with the aim of identifying sources of discrepancy and uncertainty. We found that model type was the major factor causing variation in species-distribution predictions among the various modeling components tested. We also found that different combinations of modeling components could significantly increase or decrease the performance of a model. This result indicated the importance of keeping modeling components constant for comparing a given SDM result. With all modeling components, constant, machine-learning models seem to outperform other model types. We also found that, on average, the Hierarchical Non-Linear Principal Components Analysis dimension-reduction method improved model performance more than other methods tested. We also found that the widely used confusion-matrix-based model-performance indices such as the area under the receiving operating characteristic curve (AUC), sensitivity, and Kappa do not necessarily help select the best model from a set of models if variation in performance is not large. To conclude, model result discrepancies do not necessarily suggest lack of robustness in correlative modeling as they can also occur due to inappropriate selection of modeling components. In addition, more research on model performance evaluation is required for developing robust and sensitive model evaluation methods. Undertaking multi-scenario species-distribution modeling, where possible, is likely to mitigate errors arising from inappropriate modeling components selection, and provide end users with better information on the resulting model prediction uncertainty.
| S-EPMC6468778 | biostudies-literature
Project description:Information on species' distributions, abundances, and how they change over time is central to the study of the ecology and conservation of animal populations. This information is challenging to obtain at landscape scales across range-wide extents for two main reasons. First, landscape-scale processes that affect populations vary throughout the year and across species' ranges, requiring high-resolution, year-round data across broad, sometimes hemispheric, spatial extents. Second, while citizen science projects can collect data at these resolutions and extents, using these data requires appropriate analysis to address known sources of bias. Here, we present an analytical framework to address these challenges and generate year-round, range-wide distributional information using citizen science data. To illustrate this approach, we apply the framework to Wood Thrush (Hylocichla mustelina), a long-distance Neotropical migrant and species of conservation concern, using data from the citizen science project eBird. We estimate occurrence and abundance across a range of spatial scales throughout the annual cycle. Additionally, we generate intra-annual estimates of the range, intra-annual estimates of the associations between species and characteristics of the landscape, and interannual trends in abundance for breeding and non-breeding seasons. The range-wide population trajectories for Wood Thrush show a close correspondence between breeding and non-breeding seasons with steep declines between 2010 and 2013 followed by shallower rates of decline from 2013 to 2016. The breeding season range-wide population trajectory based on the independently collected and analyzed North American Breeding Bird Survey data also shows this pattern. The information provided here fills important knowledge gaps for Wood Thrush, especially during the less studied migration and non-breeding periods. More generally, the modeling framework presented here can be used to accurately capture landscape scale intra- and interannual distributional dynamics for broadly distributed, highly mobile species.
| S-EPMC7187145 | biostudies-literature
Project description:Species distribution models (SDM) have been increasingly developed in recent years, but their validity is questioned. Their assessment can be improved by the use of independent data, but this can be difficult to obtain and prohibitive to collect. Standardized data from citizen science may be used to establish external evaluation datasets and to improve SDM validation and applicability.We used opportunistic presence-only data along with presence-absence data from a standardized citizen science program to establish and assess habitat suitability maps for 9 species of amphibian in western France. We assessed Generalized Additive and Random Forest Models' performance by (1) cross-validation using 30% of the opportunistic dataset used to calibrate the model or (2) external validation using different independent datasets derived from citizen science monitoring. We tested the effects of applying different combinations of filters to the citizen data and of complementing it with additional standardized fieldwork.Cross-validation with an internal evaluation dataset resulted in higher AUC (Area Under the receiver operating Curve) than external evaluation causing overestimation of model accuracy and did not select the same models; models integrating sampling effort performed better with external validation. AUC, specificity, and sensitivity of models calculated with different filtered external datasets differed for some species. However, for most species, complementary fieldwork was not necessary to obtain coherent results, as long as the citizen science data were strongly filtered.Since external validation methods using independent data are considered more robust, filtering data from citizen sciences may make a valuable contribution to the assessment of SDM. Limited complementary fieldwork with volunteer's participation to complete ecological gradients may also possibly enhance citizen involvement and lead to better use of SDM in decision processes for nature conservation.
| S-EPMC8019030 | biostudies-literature
Project description:BackgroundTo determine species distributions and the factors underlying them, reliable occurrence data are crucial. Assembling such data can be challenging for species with cryptic life histories or that occur at low densities.MethodsWe developed species-specific eDNA protocols, from sampling through data interpretation, to detect the common musk turtle (Sternotherus odoratus) and tested whether eDNA occurrences change our understanding of the species distribution and the factors that shape its northern range limit. We used Species Distribution Models (SDMs) with full parameter optimization on citizen science observations of S. odoratus in Southern Ontario alone and together with eDNA occurrences.ResultsOur eDNA protocol was robust and sensitive. SDMs built from traditional observations and those supplemented with eDNA detections were comparable in prediction accuracy. However, models with eDNA detections suggested that the distribution of S. odoratus in Southern Ontario is underestimated, especially near its northern range limit, and that it is shaped by thermal conditions, hydrology, and elevation. Our study underscores the promise of eDNA for surveying cryptic aquatic organisms in undocumented areas, and how such insights can help us to improve our understanding of species distributions.
| S-EPMC10040185 | biostudies-literature
Project description:BackgroundWhen data are collected subject to a detection limit, observations below the detection limit may be considered censored. In addition, the domain of such observations may be restricted; for example, values may be required to be non-negative.MethodsWe propose a method for estimating population mean and variance from censored observations that accounts for known domain restriction. The method finds maximum likelihood estimates assuming an underlying truncated normal distribution.ResultsWe show that our method, tcensReg, has lower bias, Type I error rates, and mean squared error than other methods commonly used for data with detection limits such as Tobit regression and single imputation under a range of simulation settings from mild to heavy censoring and truncation. We further demonstrate the consistency of the maximum likelihood estimators. We apply our method to analyze vision quality data collected from ophthalmology clinical trials comparing different types of intraocular lenses implanted during cataract surgery. All of the methods yield similar conclusions regarding non-inferiority, but estimates from the tcensReg method suggest that there may be greater mean differences and overall variability.ConclusionsIn the presence of detection limits, our new method tcensReg provides a way to incorporate known domain restrictions in dependent variables that substantially improves inferences.
| S-EPMC7322891 | biostudies-literature