The order of impact of each influencing factor on TOC, N, P, K, C/N, C/P, C/K, N/P, N/K and P/K is different, revealing different soil nutrients and their ecological stoichiometry, the heterogeneity of the mechanism of chemical change is caused by the joint action of various influencing factors. In general, the main factors affecting the spatial distribution of soil and its ecological stoichiometry are soil properties. Large-scale planting of fruits and vegetables and horticultural nurseries in the study area, agricultural activities such as the use of nutrient soil, fertilization, irrigation, and pesticide application directly make the soil C, N, P, and K change, which in turn leads to changes in its ecological stoichiometry31,32. Correlation analysis also showed that soil nutrients and ecological stoichiometry had extremely significant correlations with soil heavy metals33. On the one hand, nutrients have a good inhibitory effect on the toxicity of certain heavy metal elements. On the other hand, many chemical fertilizers or pesticides contain some heavy metal elements and also contain a large amount of C, N, P and K components34. As one of the important indicators of soil physical and chemical properties, pH has weak explanatory power to the spatial heterogeneity of soil nutrients and eco-stoichiometry in the study area, which may be due to the weak acidity and small variability of soil pH in the study area as a whole, while the spatial variability of soil nutrients and eco-stoichiometric characteristics in the study area is large, and the location environment of different sampling sites is different, resulting in differences in the change trend of the same soil index in the study area.
Topography is one of the important factors affecting soil development, which affects soil nutrients by regulating the spatial redistribution of soil moisture and solar radiation35. The elevation of the study area has an obvious influence on the spatial distribution of K, N, C/P, N/K and N/P, but the effects of slope and aspect on soil nutrients and their ecological chemometrics are not obvious, which may be due to the small size of the study area and the small variation range of slope and aspect. Land use types have no significant effect on the spatial distribution of soil C, N, P and K and their ecological stoichiometry, which is significantly different from other people's studies36,37, which may be due to the rapid change of land use types and the lack of obvious boundaries in the study area.
The distance from railway, industrial land, river, residential area, commercial service land and highway all have important influence on one or more spatial distributions of TOC, N, P, K, C/N, C/P, C/K, N/P, N/K and P/K in the soil of the study area, which is reflected in the comprehensive influence of human activities on the spatial distribution of soil nutrients and their ecological stoichiometry. Human activities have changed the distribution characteristics of soil nutrients and their eco-stoichiometric space in the natural state, forming new spatial characteristics. River system is an important source of agricultural irrigation. The enrichment of water source P, N, K is caused by industrial wastewater discharge and domestic sewage discharge, and then farmland soil enrichment is caused by agricultural irrigation38–40. Relevant studies have pointed out that residential areas are the areas with the most frequent human activities. Residents will produce a large amount of domestic waste containing C, N, P and K in their daily life, which will cause changes in the soil around residential areas, at the same time, frequent human activities in cities and towns will make some C, N, P and K are enriched into the soil through diffusion methods such as atmospheric deposition, which leads to changes in soil nutrients and their ecological stoichiometry41–43. In addition, factories and enterprises such as building materials, plastics and printing are distributed in the study area, the C, N, P and K carried by the waste generated by industrial activities are enriched into the soil through atmospheric deposition, rainwater erosion and infiltration. It is found that the distance from the railway has the strongest explanation for soil N, but there is a difference between the two results by comparing the Pearson correlation analysis, that is, there is no significant correlation between the distance from the railway and soil N, this is because the geographic detector analyzes the correlation between soil nutrients and influencing factors, including linear and nonlinear relationships, while the Pearson correlation coefficient is not significant, indicating that there is no significant linear relationship between soil nutrients and influencing factors, but that doesn't mean there's no nonlinear relationship22.
The factor detector can achieve good results in explaining the spatial variation of soil nutrients (Fig. 2), but from the specific explanatory power value, it is obvious that these factors cannot fully explain the variation of nutrients. Therefore, many factors of soil formation and change should be considered comprehensively, and representative and stronger influencing factors should be extracted, such as soil parent material36, soil type36, vegetation type37, soil bulk density44, irrigation method45, crop rotation method46 and climate and environmental shadows36, to a greater extent explain the spatial variability of soil nutrients and their ecological stoichiometry.
As a new method to study spatial differentiation and explain driving factors, geographic detectors have obvious advantages in dealing with soil parent material, soil type and other type variables and revealing the interaction between variables, and can be better applied to the analysis of influencing factors of soil nutrients and their ecological chemometrics. However, it can’t be ignored that since the input data of the model must be type variables, the classification of continuous variables largely determines the accuracy of the model detection results. On the one hand, it is necessary to ensure that there are obvious differences between various types of sample point data, so as to fully express the explanatory power of various factors causing spatial variation of soil nutrients; on the other hand, it is also necessary to ensure that each category includes enough samples to avoid errors caused by abnormal values.