Data for: Spatiotemporal reorganization of soil aggregates driven by film mulching in drylands: A CoDA-SBP framework reveals spatially decoupled evolutionary pathways

1. Data Generation Process, Processing Methods, Equipment, and ToolsGeneration Process: The dataset was obtained from a field-based stationary experiment. The experiment was conducted in an agricultural field with a history of over a decade of uniform drip-irrigated film-mulched cotton cultivation, utilizing a randomized complete block design (4 replicates). Two treatments were established: film mulching (FM; 0.01 mm transparent polyethylene film, configured with one film covering three drip tapes and six rows) and a bare-soil control (CK). Undisturbed soil samples were collected during key phenological stages of cotton growth and transported to the laboratory for processing.Processing Steps: Water-stable aggregates were determined using a standard wet-sieving method. Air-dried soil samples were slowly submerged in deionized water until fully saturated, and then transferred to a nested sieve assembly (apertures: 2, 1, 0.5, 0.25, and 0.106 mm). The sieving system was vertically oscillated at a frequency of 30 cycles/min with an amplitude of 38 mm for 30 minutes. Subsequently, the different size fractions were collected and oven-dried at 105 °C to a constant weight.Data Processing and Analytical Methods:Fraction merging: Because the >2 mm fraction had negligible mass in most samples, it was merged with the 1–2 mm fraction to ensure statistical stability. Ultimately, soil aggregates were classified into five size fractions: >1 mm, 0.5–1 mm, 0.25–0.5 mm, 0.106–0.25 mm, and <0.106 mm.Compositional Data Analysis (CoDA): To address the "constant-sum constraint" (closure problem) inherent in mass proportion data, the Aitchison geometry framework was introduced. By constructing a sequential binary partitioning (SBP) matrix, the five components were transformed into four orthogonal isometric log-ratio (ILR) coordinates with defined pedological interpretations for subsequent statistical inference using linear mixed-effects models (LMMs).Equipment and Tools Used:Hardware: Aggregate analyzer, standard nested sieves, 105 °C drying oven.Software platform: R 4.3.2 (core dependent packages include zCompositions for zero-value imputation, compositions for CoDA transformation, lme4 and lmerTest for LMM analysis, and emmeans for marginal mean comparisons).2. Spatiotemporal Information and ResolutionTemporal Information: Covers two consecutive full cotton growing seasons from 2024 to 2025.Temporal Resolution: Consists of 6 observation time points (Time 1 to Time 6), corresponding to the seedling, flowering-boll formation, and harvest stages over the two years. This temporal scale captured 15 cumulative hydrological (wetting-drying) events.Spatial Information: Fukang Desert Ecosystem Observation and Experiment Station, Chinese Academy of Sciences (FDEOS, 87°45′–88°05′ E, 43°45′–44°30′ N).Spatial Resolution:Vertical depth: Covers 4 specific depth intervals (0–10, 10–30, 30–60, and 60–100 cm).Horizontal micro-positions: Divided into 3 functional micro-zones based on the drip tape and film configuration. Pos A (under-mulch drip core zone, high water and nutrient flux); Pos B (under-mulch inter-drip convergence zone, overlapping lateral infiltration area); Pos C (inter-mulch bare zone, exposed to a non-covered environment, exclusive to the FM treatment).3. Data Table Structure (Record Count, Variable Labels, and Units)Record Count: The core dataset contains a total of 480 independent experimental records. (Calculation basis: The CK treatment includes two micro-positions, Pos A and Pos B, yielding 2 positions × 4 depths × 6 time points × 4 replicates = 192 samples; the FM treatment includes three micro-positions, Pos A, Pos B, and Pos C, yielding 3 positions × 4 depths × 6 time points × 4 replicates = 288 samples).Row Meaning: Each row in the data table represents the complete physical structural attributes of an independent undisturbed soil sample collected at a specific time, depth, and micro-position.Column Labels and Units:SampleID: Sample identifier.Block: Randomized block / experimental replicate number (1 to 4).Plot: Plot identifier.Time: Time point (1 to 6).Treatment: Treatment identifier (FM = film-mulched drip irrigation, CK = bare-soil control).Position: Horizontal functional micro-position (Pos A, Pos B, Pos C).Depth: Soil layer depth (0-10, 10-30, 30-60, 60-100), Unit: cm.Frac_>1mm to Frac_<0.106mm: Mass proportions of the 5 aggregate size fractions, Unit: %.4. Missing Data and Data ImputationUnbalanced Structure due to Experimental Design: The dataset does not contain missing values in the conventional sense (e.g., from sampling or recording errors). However, users should note that data for Pos C (the inter-mulch bare zone) exist exclusively within the FM treatment, as this micro-position division does not apply to the CK treatment (which is entirely bare soil). This inherently unbalanced design was scientifically addressed in our statistical analyses using two-stage linear mixed-effects models (Two-Stage LMMs).Zero-values handling: In extremely rare cases where the mass proportion of a specific size fraction was recorded as "0" within the compositional data, these true zeros were strictly imputed using a non-parametric replacement method (via the zCompositions R package) prior to calculating the ILR coordinates. This was done to satisfy the mathematical requirements of logarithmic transformations.5. Error Assessment and MitigationSpatial Heterogeneity Control: Point-source drip irrigation in arid regions induces steep micro-environmental gradients. To manage this, our dataset explicitly converts the heterogeneity-induced error driven by these micro-gradients into a fixed effect (Position). Furthermore, random background field noise was effectively isolated using a mixed-effects model with nested block structures (Block × Treatment).6. File Description The repository contains 1 core data file:soil_data.xlsx: Contains the basic environmental and spatial information for all 480 samples, along with the raw mass proportion data of the water-stable aggregates across the 5 size fractions.