Nutrient Limitation and Ecological Chemicalometry Reveal the Impacts of Long-Term Continuous Cropping on Lavender Rhizosphere Soil

To elucidate the mechanisms of nutrient cycling in rhizosphere soil and microbial metabolism during the prolonged continuous cropping of lavender, this study examined the rhizosphere soil of lavender with different continuous cropping years (1, 4, 7, 10, 15, and 20 years) in the Ili River Valley of Xinjiang, China, measuring physicochemical properties, microbial biomass C/N/P, and eight extracellular enzyme activities. Microbial carbon use efficiency (CUE) and nutrient limitation were quantified using vector analysis, threshold elemental ratios (TERs), and two derived indices (TEREEA and TERL). Soil properties exhibited distinct nonlinear patterns: SOC peaked at 4 years (p < 0.05), TN was highest at 20 years, and TP was lowest at 4–7 years. MBC and MBN peaked at 20 years, whereas MBP was significantly lower than in 1-, 4-, and 10-year fields (p < 0.05). EEC and EEN were highest at 20 years, while EEP was lowest at 4 years (p < 0.05). The activity of carbon-related acquisition enzymes increases from 134.81 μmol/g·h in the first year to 393.86 μmol/g·h in the 20th year, an increase of 192%; the activity of nitrogen acquisition enzymes increases from 686.11 μmol/g·h in the first year to 1430.58 μmol/g·h in the 20th year, an increase of 108%. This indicates that the decomposition of organic matter and the nutrient cycling capacity continue to enhance. Vector analysis showed a mean VA of 46° and VL of 0.25, with VA > 45° (P limitation) at 1–4 years shifting to VA < 45° (N limitation) at 20 years. Critically, TEREEA and TERL produced opposite dominant limitations due to differing normalization frameworks—TEREEA scales by microbial biomass stoichiometry—while TERL normalizes against enzyme-derived thresholds. CUET and CUEE ranged from 0.42 to 0.56, with the minimum at 10 years and relatively high values at 15–20 years (p < 0.05). RDA identified CBH (26.2%) and NO3−–N (19.8%) as primary drivers, with extractable phosphorus exhibiting the strongest regulatory effect (pseudo-F = 26.0). These results demonstrate that multi-model stoichiometric assessment is essential, as single indices may yield contradictory diagnoses. These results demonstrate that multi-model stoichiometric assessment is essential, as single indices may yield contradictory diagnoses, and the observed nonlinear shifts in dominant limitation type provide a mechanistic basis for targeted nutrient management in sustainable lavender cultivation.