Machine learning prediction of particle-size distribution from infrared spectra, methodologies and soil features

Parent, E.; Parent, S.-E.; Parent, L. E.

doi:10.1101/2020.05.04.076471

Home/Archive/bioRxiv

bioRxiv · 10.1101/2020.05.04.076471

Machine learning prediction of particle-size distribution from infrared spectra, methodologies and soil features

Parent, E.·Parent, S.-E.·Parent, L. E.

View original source ↗Export metadata ↓

Abstract

Accuracy of infrared (IR) models to measure soil particle-size distribution (PSD) depends on soil preparation, methodology (sedimentation, laser), settling times and relevant soil features. Compositional soil data may require log ratio (ilr) transformation to avoid numerical biases. Machine learning can relate numerous independent variables that may impact on NIR spectra to assess particle-size distribution. Our objective was to reach high IRS prediction accuracy across a large range of PSD methods and soil properties. A total of 1298 soil samples from eastern Canada were IR-scanned. Spectra were processed by Stochastic Gradient Boosting (SGB) to predict sand, silt, clay and carbon. Slope and intercept of the log-log relationships between settling time and suspension density function (SDF) (R2 = 0.84-0.92) performed similarly to NIR spectra using either ilr-transformed (R2 = 0.81-0.93) or raw percentages (R2 = 0.76-0.94). Settling times of 0.67-min and 2-h were the most accurate for NIR predictions (R2 = 0.49-0.79). The NIR prediction of sand sieving method (R2 = 0.66) was more accurate than Bouyoucos (R2 = 0.53). The NIR 2X gain was less accurate (R2 = 0.69-0.92) than 4X (R2 = 0.87-0.95). The MIR (R2 = 0.45-0.80) performed better than NIR (R2 = 0.40-0.71) spectra. Adding soil carbon, reconstituted bulk density, pH, red-green-blue color, oxalate and Mehlich3 extracts returned R2 value of 0.86-0.91 for texture prediction. In addition to slope and intercept of the SDF, 4X gain, method and pre-treatment classes, soil carbon and color appeared to be promising features for routine SGB-processed NIR particle-size analysis. Soil Classification (Soil Taxonomy): Inceptisols, Spodosols

Source connections

Published article ↗

Explore related subjects

ecology Preprint

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Parent, E., Parent, S.-E., Parent, L. E.. 2020-05-04. Machine learning prediction of particle-size distribution from infrared spectra, methodologies and soil features. https://doi.org/10.1101/2020.05.04.076471

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

A leaf-carbon economy model to explain the adaptive value of low vs. high leaf mass per area as a function of habitat productivity

Plant traits interact with environmental conditions to determine plant fitness. In particular, plant species with low leaf mass per area (LMA) are more common in high fertility soils and forest gaps, while plant species with high LMA are more common in low fertility soils and in shaded forest understories. It is not yet known why the adaptive value of LMA changes as a function of environmental productivity. We present a model for plant growth as a function of LMA and productivity. Our model relies on an empirically derived relationship between LMA and leaf lifespan. We are able to explain why species with lower LMA are successful in high productivity environments including environments with high availability of either light or soil nutrients, by showing that the LMA which maximizes growth rate decreases with increasing rate of carbon assimilation. We show that low LMA species are better at taking advantage of the 'high interest' environment of high productivity habitats by reinvesting assimilated carbon in new leaves and earning compound interest, while high LMA species grow more efficiently in the 'low interest' environment of low productivity habitats. This tradeoff sheds light on the adaptive value of LMA as it relates to environmental productivity.

ecology↗

bioRxiv2026

Species differentiate water uptake depths within a narrow vertical zone in temperate grasslands

Whether and under which conditions coexisting species in temperate grasslands differentiate their water uptake within the shallow soil layer they share is central to understanding hydrological niche differentiation in these ecosystems. In this field study, we quantified water uptake depth of 28 coexisting herbaceous species across 74 temperate grassland plots in three regions of Germany using oxygen stable isotopes and soil water content measured at multiple depths. Species consistently differed in water uptake depth across all regions, consistent with hydrological niche differentiation. This differentiation was largely confined to the upper 30 cm of soil and became stronger as overall soil moisture declined, with plants accessing water at greater depths in drier plots. In contrast, the vertical distribution of soil moisture was unrelated to either uptake depth or species differentiation. Despite pronounced species level variation, broad herbaceous functional classifications, including grasses versus forbs and fibrous versus taprooted species, showed substantial overlap in water uptake depth. Together, our results show that hydrological niche differentiation in temperate grasslands is expressed through fine scale differences among species within a shared shallow soil domain, becomes more pronounced under drier conditions, and is poorly captured by broad herbaceous functional classifications.

ecology↗

bioRxiv2026

Fungal-bacterial interactions enhance bacterial but constrain fungal resistance and recovery to extreme heat across soil moisture levels

Soil fungi and bacteria are key drivers of nutrient cycling and plant productivity in terrestrial ecosystems, yet their responses to climate extremes are rarely studied as interacting partners rather than in isolation. Moreover, it remains largely unknown how fungal-bacterial interactions regulate fungal and bacterial stability under climate extremes, particularly resistance (immediate responses to climate extremes) and recovery (responses after climate extremes cease). Here, we conducted a soil microcosm experiment with a bacterial species (Pseudomonas putida KT2440) paired with one of four Trichoderma species (T. hamatum, T. atroviride, T. koningiopsis, and T. harzianum). These fungal-bacterial pairs were then exposed to a 5-day heat event (40 {degrees}C/32 {degrees}C, day/night) followed by a 14-day recovery period (25 {degrees}C/17 {degrees}C, day/night), at two soil moisture levels (30% and 5% gravimetric water content). Our results show that fungal-bacterial interactions asymmetrically affect fungal and bacterial responses to extreme heat. Trichoderma species facilitated the resistance and recovery of P. putida density (CFUs) to extreme heat, whereas P. putida constrained the resistance and recovery of fungal biomass (ergosterol) to extreme heat, regardless of soil moisture conditions. Moreover, faster-growing Trichoderma species were associated with greater facilitation of bacterial resistance and recovery. Bacterial density also covaried positively with fungal biomass during heat stress under both soil moisture conditions, suggesting that net fungal facilitation enhances bacterial persistence under extreme heat. Together, our findings reveal that fungal-bacterial interactions can stabilize one microbial partner while increasing the vulnerability of the other, highlighting the contrasting consequences of biotic interactions under climate extremes.

ecology↗