TY - JOUR

T1 - Mitigating bias in long‐term terrestrial ecoacoustic studies

AU - Jarrett, David

AU - Barnett, Ross

AU - Bradfer‐Lawrence, Tom

AU - Froidevaux, Jérémy S. P.

AU - Gibb, Kieran

AU - Guinet, Pauline

AU - Greenhalgh, Jack

AU - Heath, Becky

AU - Johnston, Alison

AU - Monfort, José Joaquín Lahoz

AU - Rogers, Alex

AU - Willis, Stephen G.

AU - Metcalf, Oliver

PY - 2025/4/30

Y1 - 2025/4/30

N2 - Long‐term biodiversity monitoring is needed to track progress towards ambitious global targets to reduce species loss and restore ecosystems. The recent development of cheap and robust acoustic recording devices offers a cost‐effective means of gathering standardised long‐term datasets. Accounting for sources of bias in ecological monitoring and research is a fundamental part of the study design process. To highlight this issue in the context of long‐term terrestrial ecoacoustic monitoring, here we collate and discuss sources of bias arising from (i) hardware devices, (ii) firmware, software and analysis tools and (iii) the deployment environment. One important source of bias is unavoidable changes in recording hardware—to demonstrate how this potentially introduces bias, we present two case studies comparing the output from simultaneous recordings from different recorders. To mitigate biases, we recommend effective documentation of environmental and hardware‐related variables, as well as a long‐term data storage strategy that facilitates reanalysis. Additionally, the use of regular calibration tests to measure variation in the acoustic detection space will facilitate analytical approaches or post‐hoc AI solutions that remove unwanted biases. Synthesis and applications: The sources of bias and suggested mitigations described here will be of relevance to hardware manufacturers, ecological researchers and conservation practitioners. Researchers and conservation practitioners must be fully aware of relevant biases when designing long‐term ecoacoustic studies and should incorporate appropriate mitigations into their study design.

AB - Long‐term biodiversity monitoring is needed to track progress towards ambitious global targets to reduce species loss and restore ecosystems. The recent development of cheap and robust acoustic recording devices offers a cost‐effective means of gathering standardised long‐term datasets. Accounting for sources of bias in ecological monitoring and research is a fundamental part of the study design process. To highlight this issue in the context of long‐term terrestrial ecoacoustic monitoring, here we collate and discuss sources of bias arising from (i) hardware devices, (ii) firmware, software and analysis tools and (iii) the deployment environment. One important source of bias is unavoidable changes in recording hardware—to demonstrate how this potentially introduces bias, we present two case studies comparing the output from simultaneous recordings from different recorders. To mitigate biases, we recommend effective documentation of environmental and hardware‐related variables, as well as a long‐term data storage strategy that facilitates reanalysis. Additionally, the use of regular calibration tests to measure variation in the acoustic detection space will facilitate analytical approaches or post‐hoc AI solutions that remove unwanted biases. Synthesis and applications: The sources of bias and suggested mitigations described here will be of relevance to hardware manufacturers, ecological researchers and conservation practitioners. Researchers and conservation practitioners must be fully aware of relevant biases when designing long‐term ecoacoustic studies and should incorporate appropriate mitigations into their study design.

KW - bias

KW - passive acoustics

KW - bioacoustics

KW - ecoacoustics

KW - acoustic indices

KW - biodiversity

KW - monitoring

U2 - 10.1111/1365-2664.70000

DO - 10.1111/1365-2664.70000

M3 - Journal article

VL - 62

SP - 761

EP - 772

JO - Journal of Applied Ecology

JF - Journal of Applied Ecology

SN - 0021-8901

IS - 4

ER -