基于GA-BP人工神经网络的樟树林土壤呼吸对施氮响应的研究

Responses of Camphor forest soil respiration to nitrogen addition determined based on the GA-BP network

目前开展的施氮对土壤呼吸影响研究大多基于实验观测结果,受实验地自然条件的限制,不能研究在一定条件范围内土壤呼吸对施氮响应的连续变化。通过喷洒NH_4NO_3水溶液,设置对照(C,no N added),低氮(L,5 gNm^(-2)a^(-1)),中氮(M,15gNm^(-2)a^(-1)),高氮(H,30 gNm^(-2)a^1)4种处理水平,使用GA-BP人工神经网络建立樟树林土壤呼吸对施氮响应的模型,并将模拟结果使用响应曲面法展示,研究土壤呼吸对施氮响应的变化。研究结果表明,施氮对樟树林土壤呼吸既有抑制作用又有促进作用,其程度是由土壤温湿度条件决定的,总体上使得施氮对土壤呼吸在低土壤湿度的条件下主要表现为促进作用,在高土壤湿度条件下主要表现为抑制作用,在一部分土壤温湿度组合下表现为无明显作用。GA-BP人工神经网络模型以其特性,可以模拟土壤呼吸对施氮响应的连续变化,并在一定程度上解释了施氮量、土壤呼吸、土壤温度和土壤湿度之间复杂的数学关系。

Studies to date have examined the responses of soil respiration to nitrogen addition, mostly based on the experimental observation results. Owing to restrictions imposed by natural conditions, it is not possible to assess the response of soil respiration to variations in nitrogen addition across a continuum. A simulated nitrogen deposition experiment was conducted in a Camphor forest located in the Hnnan Forest Botanical Garden in subtropical China between May 2010 and June 2012. Soil respiration rate was measured twice a month under four levels of N treatments. Using a sprayed NH4 NO3 aqueous solution, this study employed the following conditions ： control （ C, no N added） , low nitrogen （ L, 5 g Nm-2 a-1 ） , medium nitrogen （ M, 15 g N m-2 a-1 ） , and high nitrogen （H, 30 g N m-2 a-1 ） ; 4 processing levels, using the GA-BP artificial neural network, were used to establish a model for the response of forest soil respiration to nitrogen addition, and the simulation results were indicated by the corresponding surface method. The results showed that, affected by control factors such as solar radiation, precipitation, vegetation types, and soil properties, the soil respiration showed a significant seasonal variation--the maximum value of soil respiration was recorded in June and August, and the minimum value in January and March; nitrogen addition had an effect on soil respiration rate, but the seasonal dynamics of soil respiration were not changed. Nitrogen addition could not only inhibit but also promote soil respiration, and the degree of the response was determined by soil temperature and humidity conditions--soil respiration showed a significant increase, significant decrease, and no significant change under 3 of the total nitrogen application conditions, under different conditions of soil temperature and humidity. Lower soil respiration rates mainly appeared in the lower soil temperature region, while higher soil respiration rates mainly appeared in regions with higher soil temperature and humidity. At the maximum value of soil respiration, soil temperature was significantly affected by nitrogen addition, and changes in soil moisture change were not significant. However, soil temperature was not significantly （P 〉 0.05 ） affected by nitrogen application when the soil respiration was minimum; however in this condition, a significant change was noted in soil moisture （P 〈 0.05 ）. Soil respiration at different humidity ranges decreased as the amount of nitrogen addition increased. Nitrogen generally inhibited soil respiration, but this was the result of the interaction between nitrogen application and soil respiration and the promotion of two kinds of interactions--nitrogen addition to soft respiration under low soil moisture mainly promoted this effect; under high soil moisture, the effect was mainly characterized by inhibition; under the combination of soil temperature and humidity, no obvious effect was observed. The GA-BP artificial neural network model with its characteristics can be used to simulate the response of soil respiration to continuous changes in nitrogen addition and, to a certain extent, explains the complex mathematical relationships between the amount of nitrogen addition, soil respiration, soil temperature, and soil moisture. At the same time, the input data of the model can be adjusted based on the experimental conditions, so as to provide considerable flexibility for the study of the relationship between soil respiration and multiple factors.