Pub Date : 2024-11-08DOI: 10.1016/j.ecoinf.2024.102881
Yseult Héjja-Brichard , Kara Million , Julien P. Renoult , Tamra C. Mendelson
The sensory drive hypothesis of animal signal evolution describes how animal communication signals and preferences evolve as adaptations to local environments. While classical approaches to testing this hypothesis often focus on preference for one aspect of a signal, deep learning techniques like generative models can create and manipulate stimuli without targeting a specific feature. Here, we used an artificial intelligence technique called neural style transfer to experimentally test preferences for color patterns in a fish. Findings in empirical aesthetics show that humans tend to prefer images with the visual statistics of the environment because the visual system is adapted to process them efficiently, making those images easier to process. Whether this is the case in other species remains to be tested. We therefore manipulated how similar or dissimilar male body patterns were to their habitats using the Neural Style Transfer (NST) algorithm. We predicted that males whose body patterns are more similar to their native habitats will be preferred by conspecifics. Our findings suggest that both males and females are sensitive to habitat congruence in their preferences, but to different extents, requiring additional investigation. Nonetheless, this study demonstrates the potential of artificial intelligence for testing hypotheses about animal communication signals.
{"title":"Using neural style transfer to study the evolution of animal signal design: A case study in an ornamented fish","authors":"Yseult Héjja-Brichard , Kara Million , Julien P. Renoult , Tamra C. Mendelson","doi":"10.1016/j.ecoinf.2024.102881","DOIUrl":"10.1016/j.ecoinf.2024.102881","url":null,"abstract":"<div><div>The sensory drive hypothesis of animal signal evolution describes how animal communication signals and preferences evolve as adaptations to local environments. While classical approaches to testing this hypothesis often focus on preference for one aspect of a signal, deep learning techniques like generative models can create and manipulate stimuli without targeting a specific feature. Here, we used an artificial intelligence technique called neural style transfer to experimentally test preferences for color patterns in a fish. Findings in empirical aesthetics show that humans tend to prefer images with the visual statistics of the environment because the visual system is adapted to process them efficiently, making those images easier to process. Whether this is the case in other species remains to be tested. We therefore manipulated how similar or dissimilar male body patterns were to their habitats using the Neural Style Transfer (NST) algorithm. We predicted that males whose body patterns are more similar to their native habitats will be preferred by conspecifics. Our findings suggest that both males and females are sensitive to habitat congruence in their preferences, but to different extents, requiring additional investigation. Nonetheless, this study demonstrates the potential of artificial intelligence for testing hypotheses about animal communication signals.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102881"},"PeriodicalIF":5.8,"publicationDate":"2024-11-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142657844","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-08DOI: 10.1016/j.ecoinf.2024.102883
Salem Ibrahim Salem , Sakae Shirayama , Sho Shimazaki , Kazuo Oki
<div><div>Audio recordings have emerged as a pivotal tool in field observations, enriching environmental monitoring in both the spatial and temporal dimensions. However, the richness and complexity of these recordings pose significant challenges, primarily when extracting specific sound clips from long recordings owing to the presence of ambient noise and other irrelevant sounds. Traditional methods, such as manual extraction or a sliding window over audio segments, hinder practical bioacoustic applications. Therefore, we propose a framework that begins with a robust segmentation method for extracting sound clips that potentially contain deer vocalizations. This segmentation method relies on acoustic anomaly detection and can markedly improve computational efficiency, facilitating deployment in environments with limited resources. Subsequently, the isolated clips were classified into deer and non-deer categories using machine learning models. Our investigation assessed three state-of-the-art deep learning models, ResNet50, MobileNetV2, and EfficientNet-B2, considering various hyperparameter configurations to optimize the performance. We utilized 3842 clips from two sites, Oze National Park and Taki, for training and testing. The outcomes demonstrated that all models exhibited comparable performances, with median accuracies of 98.3 % and 92.9 % during the validation and testing stages, respectively. However, no single model outperformed the others across all the evaluation metrics. For instance, ResNet50 in different configurations led to the best accuracy, F1 score, precision, and specificity, whereas MobileNetV2 had the best recall. Therefore, we adopted a consensus-based ensemble scoring system in which an audio clip was classified as a deer call when at least two of three models concurred in their classification to enhance the reliability of our classifications. Our findings demonstrated that the Ensemble approach significantly enhanced the classification performance, achieving an accuracy of 99.2 % in the test stage. The proposed approach was successfully deployed during the deer rutting seasons in Oze and Taki in 2019 and 2021, respectively. We gained invaluable insights into deer behavior by analyzing deer calls' frequency, timing, and duration. Additionally, the spatial distribution of deer calls in Taki enabled us to detect a breach in the city's protective fencing and an association between the spatial patterns of deer calls and crop damage in the two fields. We aimed to draw a comprehensive picture of deer activity, which has significant implications for both conservation efforts and understanding animal behavior in various habitats. The insights gathered from this research contribute to the scientific understanding of deer behavior and serve as a foundation for future studies and conservation initiatives. By incorporating advanced machine learning models into environmental monitoring, we have paved the way for more data-driven approach
{"title":"Ensemble deep learning and anomaly detection framework for automatic audio classification: Insights into deer vocalizations","authors":"Salem Ibrahim Salem , Sakae Shirayama , Sho Shimazaki , Kazuo Oki","doi":"10.1016/j.ecoinf.2024.102883","DOIUrl":"10.1016/j.ecoinf.2024.102883","url":null,"abstract":"<div><div>Audio recordings have emerged as a pivotal tool in field observations, enriching environmental monitoring in both the spatial and temporal dimensions. However, the richness and complexity of these recordings pose significant challenges, primarily when extracting specific sound clips from long recordings owing to the presence of ambient noise and other irrelevant sounds. Traditional methods, such as manual extraction or a sliding window over audio segments, hinder practical bioacoustic applications. Therefore, we propose a framework that begins with a robust segmentation method for extracting sound clips that potentially contain deer vocalizations. This segmentation method relies on acoustic anomaly detection and can markedly improve computational efficiency, facilitating deployment in environments with limited resources. Subsequently, the isolated clips were classified into deer and non-deer categories using machine learning models. Our investigation assessed three state-of-the-art deep learning models, ResNet50, MobileNetV2, and EfficientNet-B2, considering various hyperparameter configurations to optimize the performance. We utilized 3842 clips from two sites, Oze National Park and Taki, for training and testing. The outcomes demonstrated that all models exhibited comparable performances, with median accuracies of 98.3 % and 92.9 % during the validation and testing stages, respectively. However, no single model outperformed the others across all the evaluation metrics. For instance, ResNet50 in different configurations led to the best accuracy, F1 score, precision, and specificity, whereas MobileNetV2 had the best recall. Therefore, we adopted a consensus-based ensemble scoring system in which an audio clip was classified as a deer call when at least two of three models concurred in their classification to enhance the reliability of our classifications. Our findings demonstrated that the Ensemble approach significantly enhanced the classification performance, achieving an accuracy of 99.2 % in the test stage. The proposed approach was successfully deployed during the deer rutting seasons in Oze and Taki in 2019 and 2021, respectively. We gained invaluable insights into deer behavior by analyzing deer calls' frequency, timing, and duration. Additionally, the spatial distribution of deer calls in Taki enabled us to detect a breach in the city's protective fencing and an association between the spatial patterns of deer calls and crop damage in the two fields. We aimed to draw a comprehensive picture of deer activity, which has significant implications for both conservation efforts and understanding animal behavior in various habitats. The insights gathered from this research contribute to the scientific understanding of deer behavior and serve as a foundation for future studies and conservation initiatives. By incorporating advanced machine learning models into environmental monitoring, we have paved the way for more data-driven approach","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102883"},"PeriodicalIF":5.8,"publicationDate":"2024-11-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142657791","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-08DOI: 10.1016/j.ecoinf.2024.102876
Chima J. Iheaturu , Samuel Hepner , Jonathan L. Batchelor , Georges A. Agonvonon , Felicia O. Akinyemi , Vladimir R. Wingate , Chinwe Ifejika Speranza
Unmanned aerial vehicle (UAV) technologies have emerged as promising tools to improve forest ecosystem assessments. These technologies offer high-resolution data that can significantly enhance evaluations of forest structure, condition, and disturbance severity. UAV sensors such as LiDAR and multispectral provide complementary information about forest attributes, capturing structural and spectral details, yet their integration for comprehensive forest assessment remains understudied. In this paper, we explored the potential of combining UAV LiDAR and multispectral data to assess the disturbance severity of a West African forest patch (Benin). We developed an integrated disturbance index (IDI) that fuses structural properties from LiDAR data and spectral characteristics from multispectral vegetation indices through principal component analysis (PCA). This allowed us to delineate low (> 0.65), medium (0.35–0.65), and high (< 0.35) forest disturbance levels. We applied the IDI to the 560-ha Ewe-Adakplame relict forest in Benin, West Africa, and achieved 95 % overall accuracy in disturbance detection, outperforming both LiDAR-only (80 %) and multispectral-only (75 %) approaches. The IDI revealed that 23 % of the forest area has experienced low disturbance, while 28 % and 49 % face medium and high disturbance levels, respectively. These findings indicate that more than three-quarters of this relict forest exhibits medium to high levels of disturbance, underscoring the urgent need for tailored conservation strategies to strengthen forest resilience. This method's ability to differentiate disturbance levels can inform resource allocation, prioritize conservation efforts, and guide the development of site-specific management plans. The integration of UAV LiDAR and multispectral data demonstrated here has potential for application across diverse tropical forest patches, providing an effective means to monitor forest health, assess disturbance severity, and support data-driven decision-making in forest conservation and sustainable management.
{"title":"Integrating UAV LiDAR and multispectral data to assess forest status and map disturbance severity in a West African forest patch","authors":"Chima J. Iheaturu , Samuel Hepner , Jonathan L. Batchelor , Georges A. Agonvonon , Felicia O. Akinyemi , Vladimir R. Wingate , Chinwe Ifejika Speranza","doi":"10.1016/j.ecoinf.2024.102876","DOIUrl":"10.1016/j.ecoinf.2024.102876","url":null,"abstract":"<div><div>Unmanned aerial vehicle (UAV) technologies have emerged as promising tools to improve forest ecosystem assessments. These technologies offer high-resolution data that can significantly enhance evaluations of forest structure, condition, and disturbance severity. UAV sensors such as LiDAR and multispectral provide complementary information about forest attributes, capturing structural and spectral details, yet their integration for comprehensive forest assessment remains understudied. In this paper, we explored the potential of combining UAV LiDAR and multispectral data to assess the disturbance severity of a West African forest patch (Benin). We developed an integrated disturbance index (IDI) that fuses structural properties from LiDAR data and spectral characteristics from multispectral vegetation indices through principal component analysis (PCA). This allowed us to delineate low (> 0.65), medium (0.35–0.65), and high (< 0.35) forest disturbance levels. We applied the IDI to the 560-ha Ewe-Adakplame relict forest in Benin, West Africa, and achieved 95 % overall accuracy in disturbance detection, outperforming both LiDAR-only (80 %) and multispectral-only (75 %) approaches. The IDI revealed that 23 % of the forest area has experienced low disturbance, while 28 % and 49 % face medium and high disturbance levels, respectively. These findings indicate that more than three-quarters of this relict forest exhibits medium to high levels of disturbance, underscoring the urgent need for tailored conservation strategies to strengthen forest resilience. This method's ability to differentiate disturbance levels can inform resource allocation, prioritize conservation efforts, and guide the development of site-specific management plans. The integration of UAV LiDAR and multispectral data demonstrated here has potential for application across diverse tropical forest patches, providing an effective means to monitor forest health, assess disturbance severity, and support data-driven decision-making in forest conservation and sustainable management.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102876"},"PeriodicalIF":5.8,"publicationDate":"2024-11-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142657790","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-04DOI: 10.1016/j.ecoinf.2024.102880
Erica Karolina Barros de Oliveira , Alba Valéria Rezende , Leonidas Soares Murta Júnior , Lucas Mazzei , Renato Vinícius Oliveira Castro , Marcus Vinicio Neves D'Oliveira , Rafael Coll Delgado
The mortality of trees in humid tropical forests plays a fundamental role in understanding forest development, particularly after disturbances such as those caused by logging and extreme weather events. The aim of this study was to evaluate estimates of individual tree mortality following Reduced Impact Logging (RIL) in the Eastern Brazilian Amazon at biennial intervals from 2005 to 2012. RIL is based on operations planning, personnel training, and investments in forest management, and harvesting through RIL must: (a) minimize environmental damage, (b) diminish operation cost by increasing work efficiency, and (c) reduce operational waste. A mortality model was constructed based on the estimation of three distance-independent competition-indices (DII) and five models for predicting the probability of individual tree mortality. The Kolmogorov-Smirnov statistical test was used to determine the most representative model, from which a Neural Network Autoregressive (NNAR) model was constructed to forecast mortality after RIL. Mortality data was correlated with the El Niño–Southern Oscillation (ENSO) and climate (Rainfall, Maximum, Minimum, and Average air temperature). The tested models showed similar and accurate estimates with R2 exceeding 0.90, although underestimation and overestimation trends were observed. The NNAR satisfactorily represented species mortality over the simulated years. The period from 2012 to 2014 was characterized by a Neutral and Weak El Niño event, and exhibited the highest mortality value for a 25 cm DBH (diameter at breast height), the smallest DBH class measured in this study. In the correlation matrix analysis, maximum air temperature showed the highest positive correlation with trees mortality. Despite the challenges in estimating individual tree mortality in tropical forests after selective logging, accurate estimates were achieved using traditional regression techniques and NNAR. These results can support technical and silvicultural decisions regarding forest management in the Eastern Amazon region of Brazil.
{"title":"Individual tree mortality: Risks of climate change in the eastern Brazilian Amazon region","authors":"Erica Karolina Barros de Oliveira , Alba Valéria Rezende , Leonidas Soares Murta Júnior , Lucas Mazzei , Renato Vinícius Oliveira Castro , Marcus Vinicio Neves D'Oliveira , Rafael Coll Delgado","doi":"10.1016/j.ecoinf.2024.102880","DOIUrl":"10.1016/j.ecoinf.2024.102880","url":null,"abstract":"<div><div>The mortality of trees in humid tropical forests plays a fundamental role in understanding forest development, particularly after disturbances such as those caused by logging and extreme weather events. The aim of this study was to evaluate estimates of individual tree mortality following Reduced Impact Logging (RIL) in the Eastern Brazilian Amazon at biennial intervals from 2005 to 2012. RIL is based on operations planning, personnel training, and investments in forest management, and harvesting through RIL must: (a) minimize environmental damage, (b) diminish operation cost by increasing work efficiency, and (c) reduce operational waste. A mortality model was constructed based on the estimation of three distance-independent competition-indices (<em>DII</em>) and five models for predicting the probability of individual tree mortality. The Kolmogorov-Smirnov statistical test was used to determine the most representative model, from which a Neural Network Autoregressive (NNAR) model was constructed to forecast mortality after RIL. Mortality data was correlated with the El Niño–Southern Oscillation (ENSO) and climate (Rainfall, Maximum, Minimum, and Average air temperature). The tested models showed similar and accurate estimates with R<sup>2</sup> exceeding 0.90, although underestimation and overestimation trends were observed. The NNAR satisfactorily represented species mortality over the simulated years. The period from 2012 to 2014 was characterized by a Neutral and Weak El Niño event, and exhibited the highest mortality value for a 25 cm DBH (diameter at breast height), the smallest DBH class measured in this study. In the correlation matrix analysis, maximum air temperature showed the highest positive correlation with trees mortality. Despite the challenges in estimating individual tree mortality in tropical forests after selective logging, accurate estimates were achieved using traditional regression techniques and NNAR. These results can support technical and silvicultural decisions regarding forest management in the Eastern Amazon region of Brazil.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102880"},"PeriodicalIF":5.8,"publicationDate":"2024-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142657843","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-02DOI: 10.1016/j.ecoinf.2024.102873
Simon Bull, Robert Sharrad, Michael G. Gardner
Research data are almost always assumed to be reliable, but there are many reasons why data can be unreliable. Manual data-entry error rates are typically observed in the 1 to 4 % range and can be statistically impactful. This has encouraged techniques to mitigate the risk of transcription error, among which the double-entry method remains the most effective. Unfortunately, these techniques are rarely applied retrospectively to datasets collected years or decades ago, including to highly valued long-term ecological datasets that continue to contribute to active research.
This study defines an approach for the retrospective double-entry of long-term ecological datasets and then applies it to one such dataset: the 34-year (and counting) Mt Mary Lizard Survey. Software was used to execute comparisons of c.760,000 individual data value pairs across c.56,000 records to corroborate matching values and identify unmatched values.
The key findings are: a) from 760,967 value pair comparisons between the originally keyed dataset and a retrospectively re-keyed version of the same dataset, 18,637 differences (2.5 %) were detected, b) almost half (48 %) of the differences detected were intentional alterations made to the original dataset during data curation efforts, c) data differences were not uniformly distributed across data fields but concentrated in the animal identity data field, and d) a three-way comparison of the identity field corroborated a recorded value in almost all cases.
Landmark, long-term ecological studies continue to be the evidentiary framework for ecological science. However, data quality metrics—including how faithfully digital transcriptions represent the originally recorded values—are rarely reported. Given that manual transcription errors are virtually assured and the realistic possibility of post hoc, intentional alterations made during data curation, one could legitimately ask whether a manually transcribed and curated dataset is a genuine representation of the originally recorded values. The retrospective double-entry approach is one way to find out.
{"title":"The retrospective double-entry of a long-term ecological dataset","authors":"Simon Bull, Robert Sharrad, Michael G. Gardner","doi":"10.1016/j.ecoinf.2024.102873","DOIUrl":"10.1016/j.ecoinf.2024.102873","url":null,"abstract":"<div><div>Research data are almost always assumed to be reliable, but there are many reasons why data can be unreliable. Manual data-entry error rates are typically observed in the 1 to 4 % range and can be statistically impactful. This has encouraged techniques to mitigate the risk of transcription error, among which the double-entry method remains the most effective. Unfortunately, these techniques are rarely applied retrospectively to datasets collected years or decades ago, including to highly valued long-term ecological datasets that continue to contribute to active research.</div><div>This study defines an approach for the retrospective double-entry of long-term ecological datasets and then applies it to one such dataset: the 34-year (and counting) Mt Mary Lizard Survey. Software was used to execute comparisons of c.760,000 individual data value pairs across c.56,000 records to corroborate matching values and identify unmatched values.</div><div>The key findings are: a) from 760,967 value pair comparisons between the originally keyed dataset and a retrospectively re-keyed version of the same dataset, 18,637 differences (2.5 %) were detected, b) almost half (48 %) of the differences detected were intentional alterations made to the original dataset during data curation efforts, c) data differences were not uniformly distributed across data fields but concentrated in the animal identity data field, and d) a three-way comparison of the identity field corroborated a recorded value in almost all cases.</div><div>Landmark, long-term ecological studies continue to be the evidentiary framework for ecological science. However, data quality metrics—including how faithfully digital transcriptions represent the originally recorded values—are rarely reported. Given that manual transcription errors are virtually assured and the realistic possibility of post hoc, intentional alterations made during data curation, one could legitimately ask whether a manually transcribed and curated dataset is a genuine representation of the originally recorded values. The retrospective double-entry approach is one way to find out.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102873"},"PeriodicalIF":5.8,"publicationDate":"2024-11-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142578323","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-01DOI: 10.1016/j.ecoinf.2024.102875
Mo Zhang , Yong Ge , Jianghao Wang
Soil moisture is a key variable in the water, energy, and carbon cycles. Mapping sub-surface soil moisture with fine spatial resolution requires integrating downscaling approaches and process-based models. However, the effectiveness of hybrid methods, such as regression kriging (RK), in enhancing soil moisture estimates through process-based parameter predictions remains inconclusive. This study aims to integrate infiltration processes into downscaling models to predict 1-km multi-layer soil moisture, while comparing performance of nonlinear and linear models, and evaluating RK improvements. Random forests (RF) and generalized linear model (GLM) were used to downscale surface soil moisture (0–5 cm) from 36-km Soil Moisture Active Passive satellite products to 1 km across the Qinghai-Tibet Plateau. Next, the soil moisture analytical relationship (SMAR) model was applied to simulate infiltration processes and obtain site-scale parameters. RK variants (RFRK and GLMRK) were applied to jointly predict the spatial distribution of multiple infiltration parameters, which were used in SMAR at 1-km grids to estimate sub-surface soil moisture (5–40 cm). The results showed that parameter calibration significantly enhanced sub-surface soil moisture simulation, reducing root mean square error (RMSE) by 61.2 % to 69.8 %, from 0.09 to 0.03. RF outperformed GLM across all depth intervals, providing higher prediction accuracy (average RMSE, RF: 0.07; GLM: 0.09). Moreover, RK enhanced the Nash-Sutcliffe efficiency coefficient (RFRK: 0.34; GLMRK: 0.28) and coefficient of determination (RFRK: 0.5; GLMRK: 0.38) by 7.7 %–13.3 % and 2.2 %–2.4 %. This study provides a reference for mapping multi-layer soil moisture through the integration of data-driven and knowledge-driven approaches in regional-scale study areas.
{"title":"Integrating infiltration processes in hybrid downscaling methods to estimate sub-surface soil moisture","authors":"Mo Zhang , Yong Ge , Jianghao Wang","doi":"10.1016/j.ecoinf.2024.102875","DOIUrl":"10.1016/j.ecoinf.2024.102875","url":null,"abstract":"<div><div>Soil moisture is a key variable in the water, energy, and carbon cycles. Mapping sub-surface soil moisture with fine spatial resolution requires integrating downscaling approaches and process-based models. However, the effectiveness of hybrid methods, such as regression kriging (RK), in enhancing soil moisture estimates through process-based parameter predictions remains inconclusive. This study aims to integrate infiltration processes into downscaling models to predict 1-km multi-layer soil moisture, while comparing performance of nonlinear and linear models, and evaluating RK improvements. Random forests (RF) and generalized linear model (GLM) were used to downscale surface soil moisture (0–5 cm) from 36-km Soil Moisture Active Passive satellite products to 1 km across the Qinghai-Tibet Plateau. Next, the soil moisture analytical relationship (SMAR) model was applied to simulate infiltration processes and obtain site-scale parameters. RK variants (RFRK and GLMRK) were applied to jointly predict the spatial distribution of multiple infiltration parameters, which were used in SMAR at 1-km grids to estimate sub-surface soil moisture (5–40 cm). The results showed that parameter calibration significantly enhanced sub-surface soil moisture simulation, reducing root mean square error (RMSE) by 61.2 % to 69.8 %, from 0.09 to 0.03. RF outperformed GLM across all depth intervals, providing higher prediction accuracy (average RMSE, RF: 0.07; GLM: 0.09). Moreover, RK enhanced the Nash-Sutcliffe efficiency coefficient (RFRK: 0.34; GLMRK: 0.28) and coefficient of determination (RFRK: 0.5; GLMRK: 0.38) by 7.7 %–13.3 % and 2.2 %–2.4 %. This study provides a reference for mapping multi-layer soil moisture through the integration of data-driven and knowledge-driven approaches in regional-scale study areas.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102875"},"PeriodicalIF":5.8,"publicationDate":"2024-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142571647","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-30DOI: 10.1016/j.ecoinf.2024.102861
Kim Bjerge , Henrik Karstoft , Hjalte M.R. Mann , Toke T. Høye
Arthropods, including insects, represent the most diverse group and contribute significantly to animal biomass. Automatic monitoring of insects and other arthropods enables quick and efficient observation and management of ecologically and economically important targets such as pollinators, natural enemies, disease vectors, and agricultural pests. The integration of cameras and computer vision facilitates innovative monitoring approaches for agriculture, ecology, entomology, evolution, and biodiversity. However, studying insects and their interactions with flowers and vegetation in natural environments remains challenging, even with automated camera monitoring.
This paper presents a comprehensive methodology to monitor abundance and diversity of arthropods in the wild and to quantify floral cover as a key resource. We apply the methods across more than 10 million images recorded over two years using 48 insect camera traps placed in three main habitat types. The cameras monitor arthropods, including insect visits, on a specific mix of Sedum plant species with white, yellow and red/pink colored of flowers. The proposed deep-learning pipeline estimates flower cover and detects and classifies arthropod taxa from time-lapse recordings. However, the flower cover serves only as an estimate to correlate insect activity with the flowering plants.Color and semantic segmentation with DeepLabv3 are combined to estimate the percent cover of flowers of different colors. Arthropod detection incorporates motion-informed enhanced images and object detection with You-Only-Look-Once (YOLO), followed by filtering stationary objects to minimize double counting of non-moving animals and erroneous background detections. This filtering approach has been demonstrated to significantly decrease the incidence of false positives, since arthropods, occur in less than 3% of the captured images.
The final step involves grouping arthropods into 19 taxonomic classes. Seven state-of-the-art models were trained and validated, achieving -scores ranging from 0.81 to 0.89 in classification of arthropods. Among these, the final selected model, EfficientNetB4, achieved an 80% average precision on randomly selected samples when applied to the complete pipeline, which includes detection, filtering, and classification of arthropod images collected in 2021. As expected during the beginning and end of the season, reduced flower cover correlates with a noticeable drop in arthropod detections. The proposed method offers a cost-effective approach to monitoring diverse arthropod taxa and flower cover in natural environments using time-lapse camera recordings.
{"title":"A deep learning pipeline for time-lapse camera monitoring of insects and their floral environments","authors":"Kim Bjerge , Henrik Karstoft , Hjalte M.R. Mann , Toke T. Høye","doi":"10.1016/j.ecoinf.2024.102861","DOIUrl":"10.1016/j.ecoinf.2024.102861","url":null,"abstract":"<div><div>Arthropods, including insects, represent the most diverse group and contribute significantly to animal biomass. Automatic monitoring of insects and other arthropods enables quick and efficient observation and management of ecologically and economically important targets such as pollinators, natural enemies, disease vectors, and agricultural pests. The integration of cameras and computer vision facilitates innovative monitoring approaches for agriculture, ecology, entomology, evolution, and biodiversity. However, studying insects and their interactions with flowers and vegetation in natural environments remains challenging, even with automated camera monitoring.</div><div>This paper presents a comprehensive methodology to monitor abundance and diversity of arthropods in the wild and to quantify floral cover as a key resource. We apply the methods across more than 10 million images recorded over two years using 48 insect camera traps placed in three main habitat types. The cameras monitor arthropods, including insect visits, on a specific mix of <em>Sedum</em> plant species with white, yellow and red/pink colored of flowers. The proposed deep-learning pipeline estimates flower cover and detects and classifies arthropod taxa from time-lapse recordings. However, the flower cover serves only as an estimate to correlate insect activity with the flowering plants.Color and semantic segmentation with DeepLabv3 are combined to estimate the percent cover of flowers of different colors. Arthropod detection incorporates motion-informed enhanced images and object detection with You-Only-Look-Once (YOLO), followed by filtering stationary objects to minimize double counting of non-moving animals and erroneous background detections. This filtering approach has been demonstrated to significantly decrease the incidence of false positives, since arthropods, occur in less than 3% of the captured images.</div><div>The final step involves grouping arthropods into 19 taxonomic classes. Seven state-of-the-art models were trained and validated, achieving <span><math><mrow><mi>F</mi><mn>1</mn></mrow></math></span>-scores ranging from 0.81 to 0.89 in classification of arthropods. Among these, the final selected model, EfficientNetB4, achieved an 80% average precision on randomly selected samples when applied to the complete pipeline, which includes detection, filtering, and classification of arthropod images collected in 2021. As expected during the beginning and end of the season, reduced flower cover correlates with a noticeable drop in arthropod detections. The proposed method offers a cost-effective approach to monitoring diverse arthropod taxa and flower cover in natural environments using time-lapse camera recordings.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102861"},"PeriodicalIF":5.8,"publicationDate":"2024-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142578321","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-29DOI: 10.1016/j.ecoinf.2024.102871
Long Duong, Rowdy White, Brad Dabbert, Hamed Sari-Sarraf
Passive monitoring of wildlife has proven to be a highly effective tool in management and conservation. This work describes an end-to-end system for acoustic localization within the context of a specific use case. The system is described in terms of its constituent modules and the functionality of each module, as it relates to the use case of Northern bobwhite (Colinus virginianus) localization, is detailed. First, we address the field deployment of acoustic recorders in terms of optimal configuration, spacing, and number in a manner that is at once utilitarian and mathematically rigorous. Then, we propose novel methods used to automatically detect the calls from recordings, match the detected calls across recordings, and calculate the time difference of arrivals (TDOAs). Finally, a new hyperbolic localization approach is presented that uses the TDOAs to estimate the position of the calls. Each module is formulated within a theoretical framework, implemented numerically in an efficient manner, and shown to compare favorably against existing methods. Moreover, the performance of the complete system is evaluated using field recorded data and the impact of environmental factors such as field relief, vegetation features, and wind speed are illustrated and discussed. We assert and demonstrate that the factor with the most immediate and profound impact on advancing the state of the art in acoustic monitoring of wildlife is open access to high-volume, diverse field data that is accompanied by high-quality ground truth.
{"title":"A complete framework for hyperbolic acoustic localization with application to northern bobwhite covey calls","authors":"Long Duong, Rowdy White, Brad Dabbert, Hamed Sari-Sarraf","doi":"10.1016/j.ecoinf.2024.102871","DOIUrl":"10.1016/j.ecoinf.2024.102871","url":null,"abstract":"<div><div>Passive monitoring of wildlife has proven to be a highly effective tool in management and conservation. This work describes an end-to-end system for acoustic localization within the context of a specific use case. The system is described in terms of its constituent modules and the functionality of each module, as it relates to the use case of Northern bobwhite (<em>Colinus virginianus</em>) localization, is detailed. First, we address the field deployment of acoustic recorders in terms of optimal configuration, spacing, and number in a manner that is at once utilitarian and mathematically rigorous. Then, we propose novel methods used to automatically detect the calls from recordings, match the detected calls across recordings, and calculate the time difference of arrivals (TDOAs). Finally, a new hyperbolic localization approach is presented that uses the TDOAs to estimate the position of the calls. Each module is formulated within a theoretical framework, implemented numerically in an efficient manner, and shown to compare favorably against existing methods. Moreover, the performance of the complete system is evaluated using field recorded data and the impact of environmental factors such as field relief, vegetation features, and wind speed are illustrated and discussed. We assert and demonstrate that the factor with the most immediate and profound impact on advancing the state of the art in acoustic monitoring of wildlife is open access to high-volume, diverse field data that is accompanied by high-quality ground truth.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102871"},"PeriodicalIF":5.8,"publicationDate":"2024-10-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142571635","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-10-28DOI: 10.1016/j.ecoinf.2024.102872
Mingjun Jiang , Yuan Tian , Yulei Sun , Xinqing Guo , Xinfei Zhao , Le Yin , Baolei Zhang
The establishment of nature reserves (NRs) is widely acknowledged as one of the most effective measures to mitigate the threats on habitat quality (HB) posed by human activities (HAs). Precise and scientific assessment of the effectiveness of NRs holds crucial significance in improving management and promoting conservation. In this study, key indicators were creatively selected and applied to the propensity score matching (PSM) model to comprehensively assess the variations in HAs and HB within national NRs on the Qinghai-Tibet Plateau. The results indicated that between 2000 and 2020, 67.4 % of the NR area experienced a decline in HA-related impacts, while 53.8 % of the area saw improvements in HB. Additionally, with the exclusion of external environmental factors, in 2020, the difference in HAs and HB between NRs and non-protected areas was −0.131 and 0.179, respectively. Finally, based on an assessment of the overall conservation effectiveness, seven NRs were classified as “Class I", 18 as “Class II", and another seven as “Class III". These results not only confirmed the effectiveness of national NRs in alleviating anthropogenic pressure and enhancing HB but also served as an important basis for accurately assessing the conservation effectiveness of other NRs and formulating more scientifically sound and appropriate management policies.
{"title":"Assessment of the conservation effectiveness of nature reserves on the Qinghai-Tibet plateau using human activity and habitat quality indicators","authors":"Mingjun Jiang , Yuan Tian , Yulei Sun , Xinqing Guo , Xinfei Zhao , Le Yin , Baolei Zhang","doi":"10.1016/j.ecoinf.2024.102872","DOIUrl":"10.1016/j.ecoinf.2024.102872","url":null,"abstract":"<div><div>The establishment of nature reserves (NRs) is widely acknowledged as one of the most effective measures to mitigate the threats on habitat quality (HB) posed by human activities (HAs). Precise and scientific assessment of the effectiveness of NRs holds crucial significance in improving management and promoting conservation. In this study, key indicators were creatively selected and applied to the propensity score matching (PSM) model to comprehensively assess the variations in HAs and HB within national NRs on the Qinghai-Tibet Plateau. The results indicated that between 2000 and 2020, 67.4 % of the NR area experienced a decline in HA-related impacts, while 53.8 % of the area saw improvements in HB. Additionally, with the exclusion of external environmental factors, in 2020, the difference in HAs and HB between NRs and non-protected areas was −0.131 and 0.179, respectively. Finally, based on an assessment of the overall conservation effectiveness, seven NRs were classified as “Class I\", 18 as “Class II\", and another seven as “Class III\". These results not only confirmed the effectiveness of national NRs in alleviating anthropogenic pressure and enhancing HB but also served as an important basis for accurately assessing the conservation effectiveness of other NRs and formulating more scientifically sound and appropriate management policies.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102872"},"PeriodicalIF":5.8,"publicationDate":"2024-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142552290","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Water quality analysis is a vital component of the water resources management and has to be undertaken promptly to make sure environmental regulations are being followed and to eliminate any pollution that could harm the ecosystem. The main objective of this study to retrieve and map the water quality parameters from Sentinel-2 and ResourceSat-2 [Linear Imaging Self-Scanning Sensor (LISS)–IV] multi-spectral satellite data, using Support Vector Machines (SVM), Random Forests (RF), and Multi-Linear regression (MLR) models. This study represents the first attempt to demonstrate the applicability and performance of high-spatial resolution ResourceSat-2 remote sensing satellite's LISS-4 sensor, which operates in three spectral bands in the Visible and Near Infrared Region (VNIR), to predict water quality. Spectral bands of each satellite were used as independent parameter to generate the algorithms for pH, Dissolved Oxygen (DO), Total Suspended Solids (TSS) and Total Dissolved Solids (TDS). The model performance was evaluated based on coefficient of determination (R2), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and the Root Mean Square Error (RMSE) statistical indices. The result of this study indicates that the SVM yielded the highest accuracy followed by the RF and MLR. The R2, MAE, MAPE and RMSE ranged between 0.78 and 0.99, 0.049–0.24, 0.01–10.9 % and 0.05–0.28 respectively for all the four SVM models across both the sensors. Based on the spatial trend Sentinel-2 was found to be slightly superior to the ResourceSat-2 (LISS-IV) for the estimation of water quality parameters owing to its superior spectral and radiometric resolution, nevertheless ResourceSat-2 (LISS-IV) has its own advantage in terms of high spatial resolution. The results of this study highlight the high potential of machine learning models in conjunction with multispectral satellite images to manage water quality.
{"title":"Machine learning approach for water quality predictions based on multispectral satellite imageries","authors":"Vicky Anand , Bakimchandra Oinam , Silke Wieprecht","doi":"10.1016/j.ecoinf.2024.102868","DOIUrl":"10.1016/j.ecoinf.2024.102868","url":null,"abstract":"<div><div>Water quality analysis is a vital component of the water resources management and has to be undertaken promptly to make sure environmental regulations are being followed and to eliminate any pollution that could harm the ecosystem. The main objective of this study to retrieve and map the water quality parameters from Sentinel-2 and ResourceSat-2 [Linear Imaging Self-Scanning Sensor (LISS)–IV] multi-spectral satellite data, using Support Vector Machines (SVM), Random Forests (RF), and Multi-Linear regression (MLR) models. This study represents the first attempt to demonstrate the applicability and performance of high-spatial resolution ResourceSat-2 remote sensing satellite's LISS-4 sensor, which operates in three spectral bands in the Visible and Near Infrared Region (VNIR), to predict water quality. Spectral bands of each satellite were used as independent parameter to generate the algorithms for pH, Dissolved Oxygen (DO), Total Suspended Solids (TSS) and Total Dissolved Solids (TDS). The model performance was evaluated based on coefficient of determination (R<sup>2</sup>), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), and the Root Mean Square Error (RMSE) statistical indices. The result of this study indicates that the SVM yielded the highest accuracy followed by the RF and MLR. The R<sup>2</sup>, MAE, MAPE and RMSE ranged between 0.78 and 0.99, 0.049–0.24, 0.01–10.9 % and 0.05–0.28 respectively for all the four SVM models across both the sensors. Based on the spatial trend Sentinel-2 was found to be slightly superior to the ResourceSat-2 (LISS-IV) for the estimation of water quality parameters owing to its superior spectral and radiometric resolution, nevertheless ResourceSat-2 (LISS-IV) has its own advantage in terms of high spatial resolution. The results of this study highlight the high potential of machine learning models in conjunction with multispectral satellite images to manage water quality.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"84 ","pages":"Article 102868"},"PeriodicalIF":5.8,"publicationDate":"2024-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142571646","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号