EGU General Assembly 2023
© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Hydrological driver for leptospiroses abundance in a small tropical catchment ? Example from the New Caledonian leptospirosis hot-spot

Pierre Genthon1, Roman Thibeaux2, Nazha Selmaoui-Folcher3, Caroline Tramier4, Malia Kainiu2, Marie-Estelle Soupé-Gilbert2, Kavya Wijesuriya1, and Cyrille Goarant1
Pierre Genthon et al.
  • 1IRD, University of Montpellier, Noumea, New Caledonia
  • 2Institut Pasteur of New Caledonia, Noumea, New Caledonia
  • 3ISEA University of New Caledonia, Noumea, New Caledonia
  • 4Province Nord, Koné, New Caledonia

Leptospirosis is a zoonosis caused by pathogenic Leptospira shed in the urine of mammals, able to survive in water and soils and remobilized during rainy events. Pathogenic Leptospira (PL) concentrations were measured together with hydrological variables in the upper Thiem river, near the Touho village, a hot spot for leptospirosis in the main island of New Caledonia (a small tropical island itself a hot spot for leptospirosis). Two hundred twenty-six water samples were collected at the outlet of as 3 km2 sub-watershed, which is frequented by invasive mammals (rodents, deer and wild pigs) known to be animal reservoirs for leptospirosis. The main features of our results highlight that (i) samples collected at the beginning of a rain event occurring after a dry period may contain high PL concentrations (ii) PL concentrations at the heart of a wet period exhibit significant correlation with rainfall, water level and suspended matter concentration (SMC) (iii) elevated PL concentration may be observed a few days after the main flood event and within weakly turbid waters, (iV) the largest PL concentrations were observed in the middle and at the end of a wet rain season. Comparison of PL concentrations with hydrological data (rainfall, water level, SMC, soil moisture) reveals that they cannot be explained by a linear combination of hydrological variables. Indeed, nonlinear machine learning models provided a fair fit to observed data (99% of explained variance on their decimal logarithm and a mean ratio of 2.5 between raw observed data and modeled values). Comparison of identical machine learning models of water levels, SMC and PL concentration shows that the remaining error in PL concentration data does not only result from the limited dataset but rather from the intrinsic characteristics of the Leptospira signal. Our results may help to refine recommendations for leptospirosis control towards local populations. Further studies in larger watersheds draining in more populated areas will be conducted to confirm and extend these findings

How to cite: Genthon, P., Thibeaux, R., Selmaoui-Folcher, N., Tramier, C., Kainiu, M., Soupé-Gilbert, M.-E., Wijesuriya, K., and Goarant, C.: Hydrological driver for leptospiroses abundance in a small tropical catchment ? Example from the New Caledonian leptospirosis hot-spot, EGU General Assembly 2023, Vienna, Austria, 24–28 Apr 2023, EGU23-1646,, 2023.