Automating identification and morphometric measurement of soil fauna with machine learning

Abstract

Body size is a master functional trait of soil fauna, reflecting interactions among developmental, life-history, physiological, and ecological processes. Though recognized as a critical parameter, traditional manual measurement remains a major bottleneck due to its low efficiency and subjective error from different labors, hindering progress in large-scale, trait-based soil ecology. To address this gap, we developed FaunaAIM, a novel automated tool for high-throughput extraction of key morphological characteristics of soil fauna from images using machine learning. The workflow introduces an attention-based U-Net model integrated with Convolutional Block Attention Modules (CBAM) for individual segmentation, followed by morphological feature calculation using a series of morphological methods. The model achieved high segmentation accuracy (97.3% precision and a mean Intersection over Union (mIoU) of 0.951) on the test set (n=6000). Automated measurements of body length, body width, and area showed strong agreement with manual assessments, with concordance correlation coefficients (CCC) exceeding 0.97 and no significant systematic bias. Notably, the model was trained exclusively on Collembola data, while accurately segmenting mites, demonstrating strong cross-taxon transferability suitable for community-wide analyses. Overall, FaunaAIM substantially improves measurement efficiency and demonstrates excellent transferability, enabling high-throughput morphological characterization of soil fauna and potentially other biological organisms.

Graphical abstract

Keywords

soil fauna / image segmentation / automatic morphology measurement / machine learning

Highlight

	● Key morphological traits can be successfully extracted by FaunaAIM.
	● Automated measurements agree with manual assessments and are transferable across taxa.
	● This approach facilitates large-scale, high-throughput trait-based ecology of soil fauna.

References

[1] Bjerge, K., Nielsen, J.B., Sepstrup, M.V., Helsing-Nielsen, F., Høye, T. T., 2021. An automated light trap to monitor moths (Lepidoptera) using computer vision-based tracking and deep learning. Sensors21, 343.

[2] Brose, U., Jonsson, T., Berlow, E.L., Warren, P., Banasek-Richter, C., Bersier, L.F., Blanchard, J.L., Brey, T., Carpenter, S.R., Blandenier, M.F.C., Cushing, L., Dawah, H.A., Dell, T., Edwards, F., Harper-Smith, S., Jacob, U., Ledger, M.E., Martinez, N.D., Memmott, J., Mintenbeck, K., Pinnegar, J.K., Rall, B.C., Rayner, T.S., Reuman, D.C., Ruess, L., Ulrich, W., Williams, R.J., Woodward, G., Cohen, J.E., 2006. Consumer–resource body-size relationships in natural food webs. Ecology87, 2411–2417.

[3] Brown, J.H., Gillooly, J.F., Allen, A.P., Savage, V.M., West, G.B., 2004. Toward a metabolic theory of ecology. Ecology85, 1771–1789.

[4] Comor, V., Thakur, M.P., Berg, M.P., De Bie, S., Prins, H.H.T., Van Langevelde, F., 2014. Productivity affects the density-body mass relationship of soil fauna communities. Soil Biology and Biochemistry72, 203–211.

[5] Dayrell, R.L.C., Ott, T., Horrocks, T., Poschlod, P., 2023. Automated extraction of seed morphological traits from images. Methods in Ecology and Evolution14, 1708–1718.

[6] Deharveng, L., 2004. Recent advances in Collembola systematics. Pedobiologia48, 415–433.

[7] Doğan, N.Ö., 2018. Bland-Altman analysis: a paradigm to understand correlation and agreement. Turkish Journal of Emergency Medicine18, 139–141.

[8] Eerola, T., Batrakhanov, D., Barazandeh, N.V., Kraft, K., Haraguchi, L., Lensu, L., Suikkanen, S., Seppälä, J., Tamminen, T., Kälviäinen, H., 2024. Survey of automatic plankton image recognition: challenges, existing solutions and future perspectives. Artificial Intelligence Review57, 114.

[9] Giavarina, D., 2015. Understanding bland Altman analysis. Biochemia Medica25, 141–151.

[10] Gibbs, J.A., Mcausland, L., Robles-Zazueta, C.A., Murchie, E.H., Burgess, A.J., 2021. A deep learning method for fully automatic stomatal morphometry and maximal conductance estimation. Frontiers in Plant Science12, 780180.

[11] Jähne, B., 2005. Digital Image Processing. 6th ed. Berlin: Springer.

[12] Kudureti, A., Zhao, S., Zhakyp, D., Tian, C.Y., 2023. Responses of soil fauna community under changing environmental conditions. Journal of Arid Land15, 620–636.

[13] Lavelle, P., Bignell, D., Lepage, M., Wolters, V., Roger, P., Ineson, P., Heal, O.W., Dhillion S., 1997. Soil function in a changing world: the role of invertebrate ecosystem engineers. European Journal of Soil Biology33, 159–193.

[14] Lin, L.I.K., 1989. A concordance correlation coefficient to evaluate reproducibility. Biometrics45, 255–268.

[15] Mnih, V., Heess, N., Graves, A., Kavukcuoglu, K., 2014. Recurrent models of visual attention. In: Proceedings of the 28th International Conference on Neural Information Processing Systems - Volume 2. Montreal: MIT Press2204–2212.

[16] Oriol, T., Pasquet, J, Cortet, J., 2024. Automatic identification of Collembola with deep learning techniques. Ecological Informatics81, 102606.

[17] Peng, Y., Peñuelas, J., Vesterdal, L., Yue, K., Peguero, G., Fornara, D.A., Heděnec, P., Steffens, C., Wu, F.Z., 2022. Responses of soil fauna communities to the individual and combined effects of multiple global change factors. Ecology Letters25, 1961–1973.

[18] Poynton, C.A., 1998. Rehabilitation of gamma. In: Proceedings of SPIE 3299, Human Vision and Electronic Imaging III. San Jose: SPIE232–249.

[19] Ronneberger, O., Fischer, P., Brox, T., 2015. U-Net: convolutional networks for biomedical image segmentation. In: 18th International Conference on Medical Image Computing and Computer-Assisted Intervention. Munich: Springer234–241.

[20] Saikai, K.K., Bresilla, T., Kool, J., De Ruijter, N.C.A., Van Schaik, C., Teklu, M.G., 2024. Counting nematodes made easy: leveraging AI-powered automation for enhanced efficiency and precision. Frontiers in Plant Science15, 1349209.

[21] Shorten, C., Khoshgoftaar, T.M., 2019. A survey on image data augmentation for deep learning. Journal of Big Data6, 60.

[22] Silvela, J., Portillo, J., 2001. Breadth-first search and its application to image processing problems. IEEE Transactions on Image Processing10, 1194–1199.

[23] Sun, X., Xie, Z.J., Qiao, Z.H., Gao, M.X., Yin, R., Chang, L., Wu, D.H., Liu, M.Q., Zhu, Y.G., 2024. Research advances in trait-based approaches in soil animal community ecology. Chinese Journal of Applied Ecology35, 1150–1158.

[24] Sys, S., Weißbach, S., Jakob, L., Gerber, S., Schneider, C., 2022. CollembolAI, a macrophotography and computer vision workflow to digitize and characterize samples of soil invertebrate communities preserved in fluid. Methods in Ecology and Evolution13, 2729–2742.

[25] Vincent, L., 1994. Morphological area openings and closings for grey-scale images. In: O, Y.L., Toet, A., Foster, D., eds. Shape in Picture: Mathematical Description of Shape in Grey-Level Images. Berlin: Springer197–208.

[26] Wold, S., Esbensen, K., Geladi, P., 1987. Principal component analysis. Chemometrics and Intelligent Laboratory Systems2, 37–52.

[27] Woo, S., Park, J., Lee, J.Y., Kweon, I.S., 2018. CBAM: convolutional block attention module. In: Proceedings of the 15th European Conference on Computer Vision. Munich: Springer3–19.

[28] Yin, R., Kardol, P., Thakur, M.P., Gruss, I., Wu, G. L., Eisenhauer, N., Schädler, M., 2020. Soil functional biodiversity and biological quality under threat: intensive land use outweighs climate change. Soil Biology and Biochemistry147, 107847.

[29] Yu, B.X., Pan, X., Wu, H.T., Liu, D., 2025. Variations in body size and reproductive mode of oribatid mites along an altitudinal gradient in a temperate mountain region. Geoderma454, 117173.

[30] Zhang, T.Y., Suen, C.Y., 1984. A fast parallel algorithm for thinning digital patterns. Communications of the ACM27, 236–239.

[31] Zhao, Z.Y., Lu, Y.Y., Tong, Y.J., Chen, X., Bai, M., 2023. PENet: a phenotype encoding network for automatic extraction and representation of morphological discriminative features. Methods in Ecology and Evolution14, 3035–3046.

[32] Zieliński, B., Plichta, A., Misztal, K., Spurek, P., Brzychczy-Włoch, M., Ochońska, D., 2017. Deep learning approach to bacterial colony classification. PLoS One12, e0184554.

RIGHTS & PERMISSIONS

Higher Education Press