Ecological Informatics最新文献_第4页

PantherAI: An autonomous behavioural monitoring tool for assessing activity budget and space use in a zoo-housed tiger PantherAI：用于评估动物园老虎活动预算和空间使用的自主行为监测工具

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-25 DOI: 10.1016/j.ecoinf.2025.103584

Li-Dunn Chen , Stephen Dodds , Molly McGuire , Maria Franke , Gabriela Mastromonaco

Machine learning (ML)-aided technologies can be applied to many of the existing wildlife science tools (e.g., camera traps) used to support conservation initiatives both in situ and ex situ. The automated nature of ML methods reduces manual labour, extends monitoring efforts past regular daylight/working hours, and improves the overall diagnostic capacity of tools routinely applied by wildlife biologists and animal care staff at zoological institutions. Though the conservation aims and expectations may differ among zoos and aquariums, simple monitoring tools that impose less demand on animal care staff should serve as an important aid for advancing management strategies for threatened species. We applied computer vision-based predictive models built on CCTV footage from a zoo-housed Panthera tigris individual to develop an automated behavioural monitoring tool (“PantherAI”) capable of rapidly assessing activity budget and space use across variable lighting and weather conditions. We applied YOLOv8 as the model backbone to detect and classify several tiger behaviours (e.g., stereotypical pacing, resting, enrichment interaction, feeding); the trained models were then applied with scripts to autonomously generate customized activity budgets and space use heatmaps from 24-h video samples. PantherAI yielded a mean average precision >75% on test data, where it detected and classified tiger behaviours with varying levels of accuracy (stereotypical pacing: 92.2%, resting: 72.2%, locomotion: 65.4%, feeding: 34.4%, object manipulation: 43.8%). Activity budgets varied (p < 0.05) across habitats and by time of day for several behaviours. PantherAI provided reliable estimates of behaviour and space usage, two important ecological metrics commonly used to establish baseline activity budgets and assess indicators of animal welfare. Overall, ML-coupled technologies can facilitate daily data collection and monitoring procedures, both of which are integral for objectively measuring behavioural outcomes as newly implemented husbandry practices (e.g., alterations to diet, environment, social group, enrichment) are enacted in zoological and other ex situ conservation settings.

机器学习（ML）辅助技术可以应用于许多现有的野生动物科学工具（例如，相机陷阱），用于支持原位和非原位保护计划。机器学习方法的自动化特性减少了体力劳动，延长了正常白天/工作时间的监测工作，并提高了野生生物学家和动物机构动物护理人员常规使用的工具的整体诊断能力。尽管动物园和水族馆的保护目标和期望可能有所不同，但简单的监测工具对动物护理人员的要求较低，应该成为推进濒危物种管理策略的重要辅助手段。我们应用基于闭路电视录像的计算机视觉预测模型，开发了一种自动行为监测工具（“PantherAI”），能够在不同的照明和天气条件下快速评估活动预算和空间使用情况。我们使用YOLOv8作为模型主干来检测和分类老虎的几种行为（如刻板踱步、休息、富集相互作用、摄食）；然后将训练好的模型与脚本一起应用于从24小时视频样本中自动生成定制的活动预算和空间使用热图。PantherAI在测试数据上的平均精确度为75%，它以不同的准确度检测和分类老虎的行为（常规踱步：92.2%，休息：72.2%，运动：65.4%，进食：34.4%，物体操纵：43.8%）。活动预算在不同的栖息地和不同的时间有不同的（p < 0.05）。PantherAI提供了行为和空间使用的可靠估计，这两个重要的生态指标通常用于建立基线活动预算和评估动物福利指标。总体而言，机器学习耦合技术可以促进日常数据收集和监测程序，这两者对于客观衡量在动物和其他非原位保护环境中实施的新实施的畜牧业实践（例如，改变饮食、环境、社会群体、富集）的行为结果是不可或缺的。

{"title":"PantherAI: An autonomous behavioural monitoring tool for assessing activity budget and space use in a zoo-housed tiger","authors":"Li-Dunn Chen , Stephen Dodds , Molly McGuire , Maria Franke , Gabriela Mastromonaco","doi":"10.1016/j.ecoinf.2025.103584","DOIUrl":"10.1016/j.ecoinf.2025.103584","url":null,"abstract":"<div><div>Machine learning (ML)-aided technologies can be applied to many of the existing wildlife science tools (e.g., camera traps) used to support conservation initiatives both in situ and ex situ. The automated nature of ML methods reduces manual labour, extends monitoring efforts past regular daylight/working hours, and improves the overall diagnostic capacity of tools routinely applied by wildlife biologists and animal care staff at zoological institutions. Though the conservation aims and expectations may differ among zoos and aquariums, simple monitoring tools that impose less demand on animal care staff should serve as an important aid for advancing management strategies for threatened species. We applied computer vision-based predictive models built on CCTV footage from a zoo-housed <em>Panthera tigris</em> individual to develop an automated behavioural monitoring tool (“PantherAI”) capable of rapidly assessing activity budget and space use across variable lighting and weather conditions. We applied YOLOv8 as the model backbone to detect and classify several tiger behaviours (e.g., stereotypical pacing, resting, enrichment interaction, feeding); the trained models were then applied with scripts to autonomously generate customized activity budgets and space use heatmaps from 24-h video samples. PantherAI yielded a mean average precision >75% on test data, where it detected and classified tiger behaviours with varying levels of accuracy (stereotypical pacing: 92.2%, resting: 72.2%, locomotion: 65.4%, feeding: 34.4%, object manipulation: 43.8%). Activity budgets varied (<em>p</em> < 0.05) across habitats and by time of day for several behaviours. PantherAI provided reliable estimates of behaviour and space usage, two important ecological metrics commonly used to establish baseline activity budgets and assess indicators of animal welfare. Overall, ML-coupled technologies can facilitate daily data collection and monitoring procedures, both of which are integral for objectively measuring behavioural outcomes as newly implemented husbandry practices (e.g., alterations to diet, environment, social group, enrichment) are enacted in zoological and other ex situ conservation settings.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103584"},"PeriodicalIF":7.3,"publicationDate":"2025-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925404","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Influence of seasonal canopy conditions on ICESat-2-based aboveground biomass estimation in deciduous forests 季节冠层条件对基于icesat -2的落叶森林地上生物量估算的影响

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-25 DOI: 10.1016/j.ecoinf.2025.103585

Jialu Zhou , Pu Wang , Sheng Nie , Jinliang Wang , Cheng Wang , Zhou Yang , Feng Cheng , Xuebo Yang

Forest aboveground biomass (AGB) is a key indicator of terrestrial carbon storage, and its accurate estimation is essential for effective carbon stock monitoring. However, how seasonal canopy conditions (leaf-on vs. leaf-off) affect the accuracy of ICESat-2-based AGB estimation in deciduous forests remains underexplored. This study investigates the effects of seasonal canopy variability on ICESat-2-based AGB estimation and mapping. Predictive models were developed for leaf-on and leaf-off conditions by integrating ICESat-2 photon returns with airborne LiDAR-derived reference AGB, using multiple linear stepwise regression (MLSR) and random forest (RF). Subsequently, 30 m resolution AGB maps were generated by fusing ICESat-2 data with Sentinel-2 and ancillary remote sensing variables using RF and convolutional neural network (CNN) approaches. We further assessed the sensitivity of both estimation and mapping performance to seasonal canopy conditions. Results show that MLSR models produced low estimation accuracy and poor cross-seasonal transferability. RF substantially improved estimation performance compared to MLSR, with R² values of 0.53 (leaf-on) and 0.57 (leaf-off), and RMSEs of 37.06 Mg/ha and 34.03 Mg/ha, respectively. The RF model demonstrated moderate cross-seasonal transferability, achieving an R² of 0.51 when trained on leaf-off data. CNN further outperformed RF in AGB mapping, increasing R² from 0.28 to 0.42 (leaf-on) and from 0.46 to 0.56 (leaf-off), while reducing RMSE to 34.86 and 30.04 Mg/ha, respectively. The leaf-off CNN model achieved the highest accuracy (R² = 0.56, RMSE = 30.04 Mg/ha, rRMSE = 21.46 %). These results demonstrate that leaf-off ICESat-2 observations, combined with deep learning approaches, provide clear advantages for improving large-scale forest AGB estimation.

森林地上生物量（AGB）是陆地碳储量的重要指标，其准确估算是有效监测碳储量的关键。然而，季节性冠层条件（有无叶片）如何影响基于icesat -2的落叶林中AGB估算的准确性仍未得到充分研究。本文研究了季节冠层变化对基于icesat -2的AGB估算和制图的影响。利用多元线性逐步回归（MLSR）和随机森林（RF），将ICESat-2光子返回值与机载lidar导出的参考AGB相结合，建立了叶片生长和叶片脱落条件的预测模型。随后，使用RF和卷积神经网络（CNN）方法将ICESat-2数据与Sentinel-2和辅助遥感变量融合，生成30 m分辨率的AGB地图。我们进一步评估了估算和制图性能对季节冠层条件的敏感性。结果表明，MLSR模型的估计精度较低，跨季节可转移性较差。与MLSR相比，RF显著提高了估算性能，R2值分别为0.53（叶片上）和0.57（叶片下），rmse分别为37.06 Mg/ha和34.03 Mg/ha。RF模型显示出适度的跨季节可转移性，当对树叶数据进行训练时，其R2为0.51。CNN在AGB定位上进一步优于RF，将R2从0.28提高到0.42（叶片上），将R2从0.46提高到0.56（叶片下），将RMSE分别降低到34.86和30.04 Mg/ha。叶片CNN模型精度最高（R2 = 0.56, RMSE = 30.04 Mg/ha, rRMSE = 21.46%）。这些结果表明，ICESat-2叶片观测与深度学习方法相结合，为改善大尺度森林AGB估计提供了明显的优势。

{"title":"Influence of seasonal canopy conditions on ICESat-2-based aboveground biomass estimation in deciduous forests","authors":"Jialu Zhou , Pu Wang , Sheng Nie , Jinliang Wang , Cheng Wang , Zhou Yang , Feng Cheng , Xuebo Yang","doi":"10.1016/j.ecoinf.2025.103585","DOIUrl":"10.1016/j.ecoinf.2025.103585","url":null,"abstract":"<div><div>Forest aboveground biomass (AGB) is a key indicator of terrestrial carbon storage, and its accurate estimation is essential for effective carbon stock monitoring. However, how seasonal canopy conditions (leaf-on vs. leaf-off) affect the accuracy of ICESat-2-based AGB estimation in deciduous forests remains underexplored. This study investigates the effects of seasonal canopy variability on ICESat-2-based AGB estimation and mapping. Predictive models were developed for leaf-on and leaf-off conditions by integrating ICESat-2 photon returns with airborne LiDAR-derived reference AGB, using multiple linear stepwise regression (MLSR) and random forest (RF). Subsequently, 30 m resolution AGB maps were generated by fusing ICESat-2 data with Sentinel-2 and ancillary remote sensing variables using RF and convolutional neural network (CNN) approaches. We further assessed the sensitivity of both estimation and mapping performance to seasonal canopy conditions. Results show that MLSR models produced low estimation accuracy and poor cross-seasonal transferability. RF substantially improved estimation performance compared to MLSR, with R<sup>2</sup> values of 0.53 (leaf-on) and 0.57 (leaf-off), and RMSEs of 37.06 Mg/ha and 34.03 Mg/ha, respectively. The RF model demonstrated moderate cross-seasonal transferability, achieving an R<sup>2</sup> of 0.51 when trained on leaf-off data. CNN further outperformed RF in AGB mapping, increasing R<sup>2</sup> from 0.28 to 0.42 (leaf-on) and from 0.46 to 0.56 (leaf-off), while reducing RMSE to 34.86 and 30.04 Mg/ha, respectively. The leaf-off CNN model achieved the highest accuracy (R<sup>2</sup> = 0.56, RMSE = 30.04 Mg/ha, rRMSE = 21.46 %). These results demonstrate that leaf-off ICESat-2 observations, combined with deep learning approaches, provide clear advantages for improving large-scale forest AGB estimation.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103585"},"PeriodicalIF":7.3,"publicationDate":"2025-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925485","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Elevated extinction risk of sea moths under climate change 气候变化导致海蛾灭绝风险上升

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-24 DOI: 10.1016/j.ecoinf.2025.103578

Shuaishuai Liu , Ying Liu , Stefano Mammola , Songxi Yuan , Junmei Qu , Xin Wang , Qiang Lin , Zhixin Zhang

Climate change is increasingly associated with global biodiversity loss; therefore, it is essential to account for the threat of climate change when assessing species conservation status. Neglecting the effects of climate change may lead to a biased assessment of the extinction risk of focal species and misguided conservation strategies. In this study, we evaluated the extinction risk of five marine sea moth species under climate change by integrating the IUCN Red List Assessment criterion A3c and redistribution projection via species distribution models. These models showed relatively good predictive abilities and accurately described the spatial distributions of sea moths. Model projections indicated that future climate change would lead to the redistribution of sea moths and considerable range contractions especially in the Indo-Pacific. Because of climate-driven range shifts, sea moths were expected to face an increased extinction risk in the future. Our findings indicate that neglecting the threat of climate change might lead to underestimating the extinction risk faced by marine species. This has important implications for assessing or updating the extinction risk status of marine species and designing conservation measures.

气候变化与全球生物多样性丧失的关系日益密切；因此，在评估物种保护状况时，必须考虑气候变化的威胁。忽视气候变化的影响可能导致对焦点物种灭绝风险的有偏见的评估和错误的保护策略。本研究结合IUCN红色名录评估标准A3c和物种分布模型的再分布预测，对气候变化下5种海洋海蛾物种的灭绝风险进行了评估。这些模型具有较好的预测能力，能较准确地描述海蛾的空间分布。模式预测表明，未来的气候变化将导致海蛾重新分布，范围大幅缩小，尤其是在印度-太平洋地区。由于气候驱动的范围变化，预计海蛾在未来将面临更大的灭绝风险。我们的研究结果表明，忽视气候变化的威胁可能导致低估海洋物种面临的灭绝风险。这对评估或更新海洋物种灭绝风险状况和制定保护措施具有重要意义。

{"title":"Elevated extinction risk of sea moths under climate change","authors":"Shuaishuai Liu , Ying Liu , Stefano Mammola , Songxi Yuan , Junmei Qu , Xin Wang , Qiang Lin , Zhixin Zhang","doi":"10.1016/j.ecoinf.2025.103578","DOIUrl":"10.1016/j.ecoinf.2025.103578","url":null,"abstract":"<div><div>Climate change is increasingly associated with global biodiversity loss; therefore, it is essential to account for the threat of climate change when assessing species conservation status. Neglecting the effects of climate change may lead to a biased assessment of the extinction risk of focal species and misguided conservation strategies. In this study, we evaluated the extinction risk of five marine sea moth species under climate change by integrating the IUCN Red List Assessment criterion A3c and redistribution projection via species distribution models. These models showed relatively good predictive abilities and accurately described the spatial distributions of sea moths. Model projections indicated that future climate change would lead to the redistribution of sea moths and considerable range contractions especially in the Indo-Pacific. Because of climate-driven range shifts, sea moths were expected to face an increased extinction risk in the future. Our findings indicate that neglecting the threat of climate change might lead to underestimating the extinction risk faced by marine species. This has important implications for assessing or updating the extinction risk status of marine species and designing conservation measures.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103578"},"PeriodicalIF":7.3,"publicationDate":"2025-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925399","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

The conservation outcome of Indo-Pacific humpback Dolphin in China: A habitat suitability-based assessment 印度-太平洋座头海豚在中国的保护效果：基于栖息地适宜性的评估

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-24 DOI: 10.1016/j.ecoinf.2025.103580

Xue Zhang , Qinhua Fang , Ruiqing Qin , Zhiyuan Xiang , Ruiqiang Zheng , Boding Lin , Dian Zhang , Weiwei Yu

Effective planning of protected areas (PAs) for Indo-Pacific humpback dolphins (IPHDs) requires habitat identification and assessment of existing conservation outcomes. However, a systematic framework to evaluate the conservation outcomes of protected areas for this species is still absent. Therefore, we applied the MaxEnt model to estimate habitat suitability for eight IPHD populations in China. We further employed the Geographic Detector model to quantify the relationships between human activities and habitat suitability. To evaluate conservation outcomes, we developed a spatial framework integrating the Reasonable Site Selection Index (RSI) and the Sufficient Cover Index (SCI) to derive the Relative Protection Level (RPL). Results indicate: (1) Chlorophyll-a concentration, salinity, and sea surface temperature are the top-ranking influential factors shaping IPHD habitat preferences. Among anthropogenic variables, vessel density and offshore fixed infrastructure density showed strong associations with suitability, and their interactions with each other further amplified these effects. (2) All populations have less than 40 % coverage of high-suitability areas, and both the Longtou Bay and Raoping Chinese White Dolphin Nature Reserves encompass only low-suitability habitats. (3) The 14 PAs fall into three protection types: “High RSI - High SCI” (n = 1), “Low RSI - Low SCI” (n = 3), and “High RSI - Low SCI” (n = 10), indicating a critical need of expanding PAs to larger areas. (4) RPL ranges from 30 % to 55 % except for the alarming level of Shantou waters (≤4.76 %). The work can inform more strategic PA design and provide evidence-based recommendations for enhancing IPHD conservation.

印度太平洋座头海豚保护区的有效规划需要栖息地识别和现有保护成果的评估。然而，目前还没有一个系统的框架来评估保护区对该物种的保护效果。因此，我们应用MaxEnt模型对中国8个ipd种群的生境适宜性进行了估算。我们进一步利用地理探测器模型来量化人类活动与生境适宜性之间的关系。为了评估保护效果，我们建立了一个综合合理选址指数（RSI）和充分覆盖指数（SCI）的空间框架，以得出相对保护水平（RPL）。结果表明：(1)叶绿素a浓度、盐度和海温是影响ipd生境偏好的主要因素。在人为变量中，船舶密度和海上固定基础设施密度与适宜性表现出强烈的相关性，并且它们之间的相互作用进一步放大了这些影响。(2)高适宜区覆盖率均低于40%，龙头湾和饶平中华白海豚自然保护区均为低适宜区。(3) 14个保护区分为“高RSI -高SCI”（n = 1）、“低RSI -低SCI”（n = 3）和“高RSI -低SCI”（n = 10）三种保护类型，表明迫切需要将保护区扩展到更大的区域。(4)除汕头海域警戒水平（≤4.76%）外，RPL范围为30% ~ 55%。这项工作可以为更具战略性的PA设计提供信息，并为加强ipd保护提供基于证据的建议。

{"title":"The conservation outcome of Indo-Pacific humpback Dolphin in China: A habitat suitability-based assessment","authors":"Xue Zhang , Qinhua Fang , Ruiqing Qin , Zhiyuan Xiang , Ruiqiang Zheng , Boding Lin , Dian Zhang , Weiwei Yu","doi":"10.1016/j.ecoinf.2025.103580","DOIUrl":"10.1016/j.ecoinf.2025.103580","url":null,"abstract":"<div><div>Effective planning of protected areas (PAs) for Indo-Pacific humpback dolphins (IPHDs) requires habitat identification and assessment of existing conservation outcomes. However, a systematic framework to evaluate the conservation outcomes of protected areas for this species is still absent. Therefore, we applied the MaxEnt model to estimate habitat suitability for eight IPHD populations in China. We further employed the Geographic Detector model to quantify the relationships between human activities and habitat suitability. To evaluate conservation outcomes, we developed a spatial framework integrating the Reasonable Site Selection Index (RSI) and the Sufficient Cover Index (SCI) to derive the Relative Protection Level (RPL). Results indicate: (1) Chlorophyll-a concentration, salinity, and sea surface temperature are the top-ranking influential factors shaping IPHD habitat preferences. Among anthropogenic variables, vessel density and offshore fixed infrastructure density showed strong associations with suitability, and their interactions with each other further amplified these effects. (2) All populations have less than 40 % coverage of high-suitability areas, and both the Longtou Bay and Raoping Chinese White Dolphin Nature Reserves encompass only low-suitability habitats. (3) The 14 PAs fall into three protection types: “High RSI - High SCI” (<em>n</em> = 1), “Low RSI - Low SCI” (<em>n</em> = 3), and “High RSI - Low SCI” (<em>n</em> = 10), indicating a critical need of expanding PAs to larger areas. (4) RPL ranges from 30 % to 55 % except for the alarming level of Shantou waters (≤4.76 %). The work can inform more strategic PA design and provide evidence-based recommendations for enhancing IPHD conservation.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103580"},"PeriodicalIF":7.3,"publicationDate":"2025-12-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976638","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Detection algorithm for pine wilt disease in complex environments 复杂环境下松树萎蔫病的检测算法

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-23 DOI: 10.1016/j.ecoinf.2025.103572

Siyuan Chen, Qing Yang, Jiawen Zhang, Feng Xu

Pine wilt disease (PWD), caused by the pine wood nematode, is a highly destructive forest disease with severe ecological and economic consequences worldwide. Early and accurate detection of PWD is therefore crucial for effective prevention and control. To address the challenge of detecting small, sparsely distributed infected pine trees in complex environments, this study utilized RGB orthophotos captured by DJI drones flying at altitudes of 50–100 m across two representative pine forest regions in China - Jiangning District, Jiangsu Province (southern region) and Xinbin County, Liaoning Province (northern region).

We propose an improved detection model, termed YOLO-ESCF, designed to enhance detection performance under challenging conditions. The model integrates a C3ECA fused attention module and a SimCSPSPPF module into the backbone network, which effectively reduces the interference caused by overlapping tree crowns and environmental complexity while improving sensitivity to early-stage symptoms. In the neck structure, an enhanced coordinate convolution is introduced to enable the network to exploit spatial positional information, thereby improving its ability to learn target distribution patterns. Experimental results demonstrate that the proposed YOLO-ESCF model achieves an average detection accuracy of 82.9 % and an F1-score of 0.919 across both early and late PWD stages, outperforming conventional detection models. With a model size of only 18.8 MB and an FPS (frames per second, f/s) of 121.9, YOLO-ESCF offers a strong balance between accuracy and efficiency. These results highlight its potential for real-time monitoring and automated early warning systems, providing valuable support for timely intervention to minimize ecological and economic losses.

松材线虫引起的松材萎蔫病是一种具有高度破坏性的森林疾病，在世界范围内造成严重的生态和经济后果。因此，早期和准确发现残疾是有效预防和控制的关键。为了解决在复杂环境中检测小而稀疏的感染松树的挑战，本研究利用大疆无人机在50-100 m高度拍摄的RGB正射影像，跨越了中国两个具有代表性的松林区域——江苏省江宁区（南部地区）和辽宁省新宾县（北部地区）。我们提出了一种改进的检测模型，称为YOLO-ESCF，旨在提高在具有挑战性条件下的检测性能。该模型在骨干网中集成了C3ECA融合关注模块和SimCSPSPPF模块，有效降低了树冠重叠和环境复杂性带来的干扰，同时提高了对早期症状的敏感性。在颈部结构中，引入了增强的坐标卷积，使网络能够利用空间位置信息，从而提高其学习目标分布模式的能力。实验结果表明，YOLO-ESCF模型在PWD早期和晚期的平均检测准确率为82.9%，f1得分为0.919，优于传统的检测模型。模型大小仅为18.8 MB，每秒帧数为121.9，YOLO-ESCF在精度和效率之间提供了强有力的平衡。这些结果突出了实时监测和自动预警系统的潜力，为及时干预提供了宝贵的支持，以尽量减少生态和经济损失。

{"title":"Detection algorithm for pine wilt disease in complex environments","authors":"Siyuan Chen, Qing Yang, Jiawen Zhang, Feng Xu","doi":"10.1016/j.ecoinf.2025.103572","DOIUrl":"10.1016/j.ecoinf.2025.103572","url":null,"abstract":"<div><div>Pine wilt disease (PWD), caused by the pine wood nematode, is a highly destructive forest disease with severe ecological and economic consequences worldwide. Early and accurate detection of PWD is therefore crucial for effective prevention and control. To address the challenge of detecting small, sparsely distributed infected pine trees in complex environments, this study utilized RGB orthophotos captured by DJI drones flying at altitudes of 50–100 m across two representative pine forest regions in China - Jiangning District, Jiangsu Province (southern region) and Xinbin County, Liaoning Province (northern region).</div><div>We propose an improved detection model, termed YOLO-ESCF, designed to enhance detection performance under challenging conditions. The model integrates a C3ECA fused attention module and a SimCSPSPPF module into the backbone network, which effectively reduces the interference caused by overlapping tree crowns and environmental complexity while improving sensitivity to early-stage symptoms. In the neck structure, an enhanced coordinate convolution is introduced to enable the network to exploit spatial positional information, thereby improving its ability to learn target distribution patterns. Experimental results demonstrate that the proposed YOLO-ESCF model achieves an average detection accuracy of 82.9 % and an F1-score of 0.919 across both early and late PWD stages, outperforming conventional detection models. With a model size of only 18.8 MB and an FPS (frames per second, f/s) of 121.9, YOLO-ESCF offers a strong balance between accuracy and efficiency. These results highlight its potential for real-time monitoring and automated early warning systems, providing valuable support for timely intervention to minimize ecological and economic losses.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103572"},"PeriodicalIF":7.3,"publicationDate":"2025-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Bridging the Lab-to-Field gap in plant disease diagnosis through unsupervised domain adaptation enhanced by background recomposition 通过背景重组增强的无监督域适应，弥合植物病害诊断的实验室与田间差距

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-23 DOI: 10.1016/j.ecoinf.2025.103579

Woosang Jeon , Taehyeong Kim , Sanghyeok Choi , Kyuseok Yang , Seong-Yeop Kim , Mungyeong Song

Early detection of plant diseases caused by pests, pathogens such as viruses is essential for preventing their spread and minimizing crop damage, particularly in indoor farming and orchards. Machine learning-based computer vision approaches have emerged as promising tools for automated disease diagnosis, but their implementation faces critical challenges due to the scarcity of labeled data from real agricultural environments. Consequently, models are primarily trained using laboratory-collected images with monotonous backgrounds, resulting in significantly reduced performance when deployed in real cultivation environments. To address these challenges, this study presents a two-step adaptation method to bridge the gap between laboratory and real-field environments in plant disease diagnosis. Our approach first applies field-adaptive background recomposition for image augmentation, followed by unsupervised domain adaptation, enabling effective disease diagnosis in real agricultural environments. We quantitatively and qualitatively validated the proposed method on disease identification tasks for tomato, chili pepper, grape, and apple leaves, achieving robust performance in field environments without requiring labeled real-field data.

早期发现由病虫害、病毒等病原体引起的植物疾病，对于防止其传播和尽量减少作物损害至关重要，特别是在室内农业和果园中。基于机器学习的计算机视觉方法已经成为自动化疾病诊断的有前途的工具，但由于缺乏来自真实农业环境的标记数据，它们的实施面临着严峻的挑战。因此，模型主要使用实验室收集的具有单调背景的图像进行训练，导致在实际种植环境中部署时性能显着降低。为了解决这些挑战，本研究提出了一种两步适应方法，以弥合实验室和实际环境在植物病害诊断中的差距。我们的方法首先应用自适应背景重组进行图像增强，然后进行无监督域自适应，从而在真实农业环境中实现有效的疾病诊断。我们在番茄、辣椒、葡萄和苹果叶片的疾病识别任务中对所提出的方法进行了定量和定性验证，在不需要标记的实际数据的情况下，在田间环境中获得了稳健的性能。

{"title":"Bridging the Lab-to-Field gap in plant disease diagnosis through unsupervised domain adaptation enhanced by background recomposition","authors":"Woosang Jeon , Taehyeong Kim , Sanghyeok Choi , Kyuseok Yang , Seong-Yeop Kim , Mungyeong Song","doi":"10.1016/j.ecoinf.2025.103579","DOIUrl":"10.1016/j.ecoinf.2025.103579","url":null,"abstract":"<div><div>Early detection of plant diseases caused by pests, pathogens such as viruses is essential for preventing their spread and minimizing crop damage, particularly in indoor farming and orchards. Machine learning-based computer vision approaches have emerged as promising tools for automated disease diagnosis, but their implementation faces critical challenges due to the scarcity of labeled data from real agricultural environments. Consequently, models are primarily trained using laboratory-collected images with monotonous backgrounds, resulting in significantly reduced performance when deployed in real cultivation environments. To address these challenges, this study presents a two-step adaptation method to bridge the gap between laboratory and real-field environments in plant disease diagnosis. Our approach first applies field-adaptive background recomposition for image augmentation, followed by unsupervised domain adaptation, enabling effective disease diagnosis in real agricultural environments. We quantitatively and qualitatively validated the proposed method on disease identification tasks for tomato, chili pepper, grape, and apple leaves, achieving robust performance in field environments without requiring labeled real-field data.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103579"},"PeriodicalIF":7.3,"publicationDate":"2025-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925551","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Improving salmon marine survival models with covariance map indices of sea surface temperature (CMISST) and sea surface height (CMISSH) 利用海表温度（CMISST）和海表高度（CMISSH）协方差图指数改进鲑鱼海洋生存模型

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-22 DOI: 10.1016/j.ecoinf.2025.103575

Brian J. Burke , Aimee H. Fullerton , Brian K. Wells , Jan Ohlberger

Pacific-basin scale atmospheric and oceanographic conditions cascade to regional environmental conditions. In turn, these regional conditions in the nearshore ocean have direct and indirect impacts on coastal ecology and influence everything from primary productivity to top predator abundance and distribution. Quantitatively representing these large-scale conditions with consolidated metrics allows the ocean environment to be represented in a small number of variables, or indices, and can aid in fisheries management and policy decisions. As an example, Pacific salmon face numerous challenges during their migration from freshwater rivers to the ocean and back, and managers must make accurate assessments of their marine survival to inform effective conservation and management strategies. From satellite data and earth systems models, we now have rich data sets with complete spatial and temporal coverage to use for these applications. Here, we create a spatial map of covariance values between salmon and sea surface temperature (or height) and describe a method to calculate a temporal index of similarity to that map. The resulting index, referred to as a Covariance Map Index (CMI), can be used in subsequent predictive models of salmon adult return abundance or survival. Importantly, and unlike standard indices such as PDO, this new metric is tailored for a specific stock of salmon, and can be easily applied to other marine species. By employing this innovative tool, researchers, policymakers, and conservationists can gain deeper insights into marine survival patterns, enabling more informed and effective strategies for the sustainable management of these iconic species.

太平洋盆地尺度的大气和海洋条件级联到区域环境条件。反过来，近岸海洋的这些区域条件对沿海生态产生直接和间接的影响，并影响从初级生产力到顶级捕食者的丰度和分布的一切。用综合指标定量地表示这些大规模条件，可以用少量变量或指数表示海洋环境，并有助于渔业管理和政策决策。例如，太平洋鲑鱼在从淡水河流到海洋的迁徙过程中面临着许多挑战，管理者必须对它们的海洋生存做出准确评估，为有效的保护和管理战略提供信息。从卫星数据和地球系统模型中，我们现在拥有丰富的数据集，具有完整的空间和时间覆盖范围，可用于这些应用。在这里，我们创建了鲑鱼和海面温度（或高度）之间协方差值的空间图，并描述了一种计算与该图相似度的时间指数的方法。由此产生的指数被称为协方差图指数（CMI），可用于后续的鲑鱼成虫回归丰度或存活率预测模型。重要的是，与PDO等标准指数不同，这个新指标是为特定的鲑鱼种群量身定制的，可以很容易地应用于其他海洋物种。通过使用这一创新工具，研究人员、政策制定者和保护主义者可以更深入地了解海洋生存模式，为这些标志性物种的可持续管理提供更明智和有效的策略。

{"title":"Improving salmon marine survival models with covariance map indices of sea surface temperature (CMISST) and sea surface height (CMISSH)","authors":"Brian J. Burke , Aimee H. Fullerton , Brian K. Wells , Jan Ohlberger","doi":"10.1016/j.ecoinf.2025.103575","DOIUrl":"10.1016/j.ecoinf.2025.103575","url":null,"abstract":"<div><div>Pacific-basin scale atmospheric and oceanographic conditions cascade to regional environmental conditions. In turn, these regional conditions in the nearshore ocean have direct and indirect impacts on coastal ecology and influence everything from primary productivity to top predator abundance and distribution. Quantitatively representing these large-scale conditions with consolidated metrics allows the ocean environment to be represented in a small number of variables, or indices, and can aid in fisheries management and policy decisions. As an example, Pacific salmon face numerous challenges during their migration from freshwater rivers to the ocean and back, and managers must make accurate assessments of their marine survival to inform effective conservation and management strategies. From satellite data and earth systems models, we now have rich data sets with complete spatial and temporal coverage to use for these applications. Here, we create a spatial map of covariance values between salmon and sea surface temperature (or height) and describe a method to calculate a temporal index of similarity to that map. The resulting index, referred to as a Covariance Map Index (CMI), can be used in subsequent predictive models of salmon adult return abundance or survival. Importantly, and unlike standard indices such as PDO, this new metric is tailored for a specific stock of salmon, and can be easily applied to other marine species. By employing this innovative tool, researchers, policymakers, and conservationists can gain deeper insights into marine survival patterns, enabling more informed and effective strategies for the sustainable management of these iconic species.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103575"},"PeriodicalIF":7.3,"publicationDate":"2025-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925557","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

From pictures to numbers: Multi-species seabird surveys using drone imagery and neural networks 从图片到数字：使用无人机图像和神经网络的多物种海鸟调查

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-22 DOI: 10.1016/j.ecoinf.2025.103583

Mie P. Arnberg , Are Charles Jensen , James Sample , Arnt-Børre Salberg , Kasper Hancke , Hege Gundersen , Sindre Molværsmyr

Seabirds are among the most threatened avian taxa globally, with over half of all species in decline. Accurate population estimates are essential for tracking trends and informing conservation, yet traditional survey methods are limited by logistical challenges, high costs, and the potential for wildlife disturbance, particularly in remote coastal areas. Unoccupied aerial vehicles (UAVs or drones) offer an efficient and low-disturbance alternative, but the vast volumes of imagery they produce are often labour-intensive to analyse.

In this study, we combined drone imagery with deep learning techniques to estimate colony size and abundance of surface-nesting seabirds based on counts of visible individuals. High-resolution aerial imagery was collected from 163 colonies along the southern and central Norwegian coastline over three breeding seasons (2022–2024), covering a total of 7.67 km². A convolutional neural network (Faster R-CNN with ResNet-101 backbone) was trained on 131 annotated orthomosaics and evaluated on 32 additional annotated orthomosaics from geographically distinct colonies.

Across all data, 23,062 individual seabirds were annotated. Colonies hosted an average of 141.5 ± 193.9 individuals and 4.1 ± 2.3 focal species per site. At a confidence threshold of 0.7, the model achieved a detection rate of 87.5 % and a macro F1-score of 0.88. It performed well across multiple focal species, including terns, gulls, and cormorants, and remained robust in mixed-species colonies. Most errors involved false negatives or confusion among visually similar species.

Our results demonstrate the potential for deep learning models to support efficient, scalable, and low-disturbance seabird monitoring across diverse habitats, reducing manual annotation effort and informing conservation practice.

海鸟是全球最受威胁的鸟类分类群之一，超过一半的物种正在减少。准确的种群估计对于跟踪趋势和为保护提供信息至关重要，但传统的调查方法受到后勤挑战、高成本和潜在的野生动物干扰的限制，特别是在偏远的沿海地区。无人驾驶飞行器（uav或无人机）提供了一种高效且低干扰的替代方案，但它们产生的大量图像通常需要劳动密集型的分析。在这项研究中，我们将无人机图像与深度学习技术相结合，根据可见个体的数量来估计海面筑巢海鸟的种群规模和丰度。在三个繁殖季节（2022-2024年），从挪威南部和中部海岸线的163个殖民地收集了高分辨率航空图像，总面积为7.67平方公里。采用基于ResNet-101主干的Faster R-CNN卷积神经网络对131个标注正形图进行了训练，并对另外32个地理位置不同的标注正形图进行了评价。在所有数据中，对23,062只海鸟进行了注释。每个站点平均寄主141.5±193.9个个体和4.1±2.3个焦点物种。在置信阈值为0.7时，模型的检出率为87.5%，宏观f1得分为0.88。它在包括燕鸥、海鸥和鸬鹚在内的多个焦点物种中表现良好，并且在混合物种群体中保持稳健。大多数错误涉及假阴性或视觉上相似的物种之间的混淆。我们的研究结果表明，深度学习模型有潜力支持跨不同栖息地的高效、可扩展和低干扰的海鸟监测，减少人工注释工作并为保护实践提供信息。

{"title":"From pictures to numbers: Multi-species seabird surveys using drone imagery and neural networks","authors":"Mie P. Arnberg , Are Charles Jensen , James Sample , Arnt-Børre Salberg , Kasper Hancke , Hege Gundersen , Sindre Molværsmyr","doi":"10.1016/j.ecoinf.2025.103583","DOIUrl":"10.1016/j.ecoinf.2025.103583","url":null,"abstract":"<div><div>Seabirds are among the most threatened avian taxa globally, with over half of all species in decline. Accurate population estimates are essential for tracking trends and informing conservation, yet traditional survey methods are limited by logistical challenges, high costs, and the potential for wildlife disturbance, particularly in remote coastal areas. Unoccupied aerial vehicles (UAVs or drones) offer an efficient and low-disturbance alternative, but the vast volumes of imagery they produce are often labour-intensive to analyse.</div><div>In this study, we combined drone imagery with deep learning techniques to estimate colony size and abundance of surface-nesting seabirds based on counts of visible individuals. High-resolution aerial imagery was collected from 163 colonies along the southern and central Norwegian coastline over three breeding seasons (2022–2024), covering a total of 7.67 km<sup>2</sup>. A convolutional neural network (Faster R-CNN with ResNet-101 backbone) was trained on 131 annotated orthomosaics and evaluated on 32 additional annotated orthomosaics from geographically distinct colonies.</div><div>Across all data, 23,062 individual seabirds were annotated. Colonies hosted an average of 141.5 ± 193.9 individuals and 4.1 ± 2.3 focal species per site. At a confidence threshold of 0.7, the model achieved a detection rate of 87.5 % and a macro F1-score of 0.88. It performed well across multiple focal species, including terns, gulls, and cormorants, and remained robust in mixed-species colonies. Most errors involved false negatives or confusion among visually similar species.</div><div>Our results demonstrate the potential for deep learning models to support efficient, scalable, and low-disturbance seabird monitoring across diverse habitats, reducing manual annotation effort and informing conservation practice.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103583"},"PeriodicalIF":7.3,"publicationDate":"2025-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925400","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Arctic diel and circadian acoustic pattern of Orcas, Fin, and Humpback whales revealed by deep learning from two months of continuous recordings 从两个月的连续录音中深度学习揭示了逆戟鲸、长须鲸和座头鲸的北极昼夜节律和昼夜节律声学模式

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-22 DOI: 10.1016/j.ecoinf.2025.103564

Justine Girardet , Hervé Glotin , Stéphane Chavin , Marion Poupard , Julie Guiderdoni , Véronique Sarano

In the northern Norwegian fjords, orcas, fin and humpback whales gather each winter to feed on herring, overlapping with intense human activities such as fishing and whale watching. To assess how anthropophony and geophony influence their acoustic behavior, we conducted two months of continuous passive acoustic monitoring in Kvænangen fjord during winter 2022–2023. Whale vocalizations were automatically detected using a deep learning framework based on YOLOv5, enabling quantification of species-specific acoustic presence and activity. Ambient noise was estimated from power spectral density. Low- and high-noise conditions were identified for geophony and anthropophony using a three-step filtering procedure. Model performances were evaluated under various noise conditions to ensure robust and consistent detection accuracy. Analyzes were then performed to characterize diel, circadian and daily patterns of acoustic activity. All three species were detected nearly continuously, with orcas activity peaking in November. Acoustic patterns were strongly influenced by noise: orcas and fin whales were less vocally active with increasing anthropophony (

ρ

<

−0.31, p

<

0.05), while humpback whales showed a time-dependent response, increasing vocal activity on short timescales (p

<

0.01) but decreasing over longer periods (

ρ

= −0.33, p = 0.008). Geophony was associated with reduced acoustic presence for all three species on a daily basis (

ρ

<

−0.34, p

<

0.01), suggesting changes in spatial distribution or vocal behavior. Positive correlations between orcas and humpback whales vocal behavior indicated potential concurrent feeding. These findings revealed species- and timescale-specific acoustic responses to noise and illustrate how deep-learning can enhance ecoacoustic monitoring.

在挪威北部的峡湾，逆戟鲸、长须鲸和座头鲸每年冬天聚集在一起以鲱鱼为食，与捕鱼和观鲸等激烈的人类活动重叠。为了评估人声和地声对其声学行为的影响，我们于2022-2023年冬季在Kvænangen峡湾进行了为期两个月的连续被动声学监测。使用基于YOLOv5的深度学习框架自动检测鲸鱼的发声，从而量化物种特定的声音存在和活动。根据功率谱密度估计环境噪声。使用三步滤波程序确定了地声和人声的低噪声和高噪声条件。在各种噪声条件下对模型性能进行了评估，以确保检测精度的鲁棒性和一致性。然后进行分析，以表征昼夜节律和声学活动的日常模式。这三个物种几乎连续被检测到，逆戟鲸的活动在11月达到顶峰。声学模式受到噪音的强烈影响：逆戟鲸和长须鲸的声音活动随着人类噪音的增加而减少（p < - 0.31, p < 0.05），而座头鲸表现出时间依赖性的反应，在短时间尺度上增加声音活动（p < 0.01），但在较长时间内减少（p = - 0.33, p = 0.008）。地质现象与这三个物种每天的声音存在减少有关（ρ < - 0.34, p < 0.01），表明空间分布或发声行为发生了变化。逆戟鲸和座头鲸的声音行为呈正相关，表明可能同时进食。这些发现揭示了物种和时间尺度对噪声的特定声学响应，并说明了深度学习如何增强生态声学监测。

{"title":"Arctic diel and circadian acoustic pattern of Orcas, Fin, and Humpback whales revealed by deep learning from two months of continuous recordings","authors":"Justine Girardet , Hervé Glotin , Stéphane Chavin , Marion Poupard , Julie Guiderdoni , Véronique Sarano","doi":"10.1016/j.ecoinf.2025.103564","DOIUrl":"10.1016/j.ecoinf.2025.103564","url":null,"abstract":"<div><div>In the northern Norwegian fjords, orcas, fin and humpback whales gather each winter to feed on herring, overlapping with intense human activities such as fishing and whale watching. To assess how anthropophony and geophony influence their acoustic behavior, we conducted two months of continuous passive acoustic monitoring in Kvænangen fjord during winter 2022–2023. Whale vocalizations were automatically detected using a deep learning framework based on YOLOv5, enabling quantification of species-specific acoustic presence and activity. Ambient noise was estimated from power spectral density. Low- and high-noise conditions were identified for geophony and anthropophony using a three-step filtering procedure. Model performances were evaluated under various noise conditions to ensure robust and consistent detection accuracy. Analyzes were then performed to characterize diel, circadian and daily patterns of acoustic activity. All three species were detected nearly continuously, with orcas activity peaking in November. Acoustic patterns were strongly influenced by noise: orcas and fin whales were less vocally active with increasing anthropophony (<span><math><mi>ρ</mi></math></span> <span><math><mo><</mo></math></span> −0.31, p <span><math><mo><</mo></math></span> 0.05), while humpback whales showed a time-dependent response, increasing vocal activity on short timescales (p <span><math><mo><</mo></math></span> 0.01) but decreasing over longer periods (<span><math><mi>ρ</mi></math></span> = −0.33, p = 0.008). Geophony was associated with reduced acoustic presence for all three species on a daily basis (<span><math><mi>ρ</mi></math></span> <span><math><mo><</mo></math></span> −0.34, p <span><math><mo><</mo></math></span> 0.01), suggesting changes in spatial distribution or vocal behavior. Positive correlations between orcas and humpback whales vocal behavior indicated potential concurrent feeding. These findings revealed species- and timescale-specific acoustic responses to noise and illustrate how deep-learning can enhance ecoacoustic monitoring.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103564"},"PeriodicalIF":7.3,"publicationDate":"2025-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977422","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Benthic habitat influences sea scallop distributions and swimming behavior based on underwater imagery and machine learning 基于水下图像和机器学习的底栖生物栖息地影响海扇贝分布和游泳行为

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-22 DOI: 10.1016/j.ecoinf.2025.103577

Liese A. Siemann , Matthew Dawkins , Luisa M. Garcia , Jonathan Crall , Tasha E. O'Hara

Understanding the drivers of Atlantic sea scallop distribution and abundance is a priority for fisheries management. While factors influencing scallop habitat selection and recruitment remain poorly understood, the increasing use of optical surveys for scallop assessment has opened new opportunities for studying juvenile and adult scallop distributions. Yet manual image annotation is time-consuming, limiting the potential of these datasets. Automating this process could improve biomass estimates and advance research on scallop ecology by enabling the analysis of larger datasets in a fraction of the time required for manual annotation. This study aimed to (1) use automated scallop detectors and benthic habitat classifiers to generate accurate scallop plus habitat datasets for large numbers of images from Habitat Mapping Camera surveys of scallop grounds and (2) use these datasets to develop models to assess how benthic habitat influences sea scallop distributions and age-specific swimming behavior. Scallop detectors and habitat classifiers were developed for the computer vision platform Video and Image Analytics for Marine Environments. Density models using automated datasets indicated that benthic habitat components (gravel, shell hash, bryozoans, sea star beds, sand dollar beds, and sand waves) influenced scallop densities, with some trends for habitat preference shifting with age. Swimming models confirmed that juvenile scallops swim more frequently than adults, particularly in areas with lower bryozoan densities. This work highlights the value of using automated tools to process large-scale optical datasets and provides new insights into habitat-specific scallop behavior across life stages.

了解大西洋扇贝分布和丰度的驱动因素是渔业管理的首要任务。虽然影响扇贝栖息地选择和招募的因素仍然知之甚少，但越来越多地使用光学调查来评估扇贝，为研究幼年和成年扇贝的分布开辟了新的机会。然而，手动图像注释是耗时的，限制了这些数据集的潜力。自动化这一过程可以改善生物量估算，并推进扇贝生态学的研究，因为它可以在人工注释所需的一小部分时间内分析更大的数据集。本研究的目的是：(1)利用自动扇贝探测器和底栖生物栖息地分类器，对来自生境测绘相机（habitat Mapping Camera）对扇贝地调查的大量图像生成精确的扇贝和栖息地数据集；(2)利用这些数据集开发模型，评估底栖生物栖息地对海洋扇贝分布和年龄特异性游泳行为的影响。为海洋环境视频与图像分析计算机视觉平台开发了扇贝检测器和生境分类器。利用自动化数据集建立的密度模型表明，底栖生物栖息地组成（砾石、贝壳堆、苔藓虫、海星床、沙元床和沙波）会影响扇贝的密度，并且栖息地偏好随年龄的变化有一定的趋势。游泳模型证实，幼年扇贝比成年扇贝游得更频繁，特别是在苔藓虫密度较低的地区。这项工作强调了使用自动化工具处理大规模光学数据集的价值，并为扇贝在生命阶段的栖息地特定行为提供了新的见解。

{"title":"Benthic habitat influences sea scallop distributions and swimming behavior based on underwater imagery and machine learning","authors":"Liese A. Siemann , Matthew Dawkins , Luisa M. Garcia , Jonathan Crall , Tasha E. O'Hara","doi":"10.1016/j.ecoinf.2025.103577","DOIUrl":"10.1016/j.ecoinf.2025.103577","url":null,"abstract":"<div><div>Understanding the drivers of Atlantic sea scallop distribution and abundance is a priority for fisheries management. While factors influencing scallop habitat selection and recruitment remain poorly understood, the increasing use of optical surveys for scallop assessment has opened new opportunities for studying juvenile and adult scallop distributions. Yet manual image annotation is time-consuming, limiting the potential of these datasets. Automating this process could improve biomass estimates and advance research on scallop ecology by enabling the analysis of larger datasets in a fraction of the time required for manual annotation. This study aimed to (1) use automated scallop detectors and benthic habitat classifiers to generate accurate scallop plus habitat datasets for large numbers of images from Habitat Mapping Camera surveys of scallop grounds and (2) use these datasets to develop models to assess how benthic habitat influences sea scallop distributions and age-specific swimming behavior. Scallop detectors and habitat classifiers were developed for the computer vision platform Video and Image Analytics for Marine Environments. Density models using automated datasets indicated that benthic habitat components (gravel, shell hash, bryozoans, sea star beds, sand dollar beds, and sand waves) influenced scallop densities, with some trends for habitat preference shifting with age. Swimming models confirmed that juvenile scallops swim more frequently than adults, particularly in areas with lower bryozoan densities. This work highlights the value of using automated tools to process large-scale optical datasets and provides new insights into habitat-specific scallop behavior across life stages.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103577"},"PeriodicalIF":7.3,"publicationDate":"2025-12-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925559","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0