%0 Journal Article
%9 ACL : Articles dans des revues avec comité de lecture répertoriées par l'AERES
%A Gorretta, N.
%A Chevallier, Tiphaine
%A Gobrecht, A.
%A Benoît, M.
%A Roger, J. M.
%A Barthès, Bernard
%T Spatial and temporal assessment of soil microbial respiration at microscale using proximal hyperspectral imaging
%D 2026
%L fdi:010096086
%G ENG
%J Journal of near Infrared Spectroscopy
%@ 0967-0335
%K soil biological activity ; visible and near infrared reflectance ; spectroscopy (VNIRS) ; partial least squares regression (PLSR) ; thermal stress ; soil heterogeneity
%M ISI:001667830700001
%P [15 ]
%R 10.1177/09670335251408843
%U https://www.documentation.ird.fr/hor/fdi:010096086
%> https://www.documentation.ird.fr/intranet/publi/2026-03/010096086.pdf
%V [Early access]
%W Horizon (IRD)
%X The study aimed to test the usefulness of the Hyperspectral imaging (HSI) in the visible and near infrared (VNIR) to address temporal and spatial variations in surface soil microbial respiration during an incubation. We hypothesized that VNIR HSI could show the heterogeneity and different evolutions of soil respiration between thermal stress and control conditions. Two mesocosms (approximate to 1500 cm2) were incubated 28 days under controlled conditions. Prior to incubation, one mesocosm was heat-stressed (stress), while the other was not (control). HS images of mesocosms (pixel resolution 0.18 x 0.36 mm) were acquired at several incubation dates. To calibrate respiration measurement of pixels, 56 standard samples were built and used as a calibration subterfuge to represent all conditions that would occur at the mesocosm surface in terms of straw cover, soil moisture, incubation date and consequences of initial thermal stress. Images of standard samples were acquired at the same dates than mesocosm images, just after CO2 emitted by each standard sample (i.e. respiration) was measured. Partial least squares (PLS) regression was used to establish a cumulative respiration model based on the average spectra of each standard samples at each incubation date. The calibration and validation yielded R2 = 0.88 and 0.85, respectively. The prediction model of respiration was applied on each pixel of the mesocosms and yielded prediction maps. The maps showed hotspots of respiration around plant residues surrounded by large areas of low respiration. The approach did not only contrast pieces of straw versus the soil matrix, but also hotspots with different evolutions. The effect of thermal stress on soil respiration patterns was not noticeable on mesocosms but was noted on standard samples with high straw cover. Further development of VNIR HSI at proximal scale should be encouraged to investigate the spatio-temporal processes of biological activity after a disturbance.
%$ 068 ; 074