长期定位施肥条件下旱地小麦“产量差”影响因子评估

发布时间：2018-08-26 07:59

【摘要】：旱作小麦“产量差”即雨养小麦生产潜力与农田实际产量间的差值,其受气候要素、肥料、土壤养分以及作物品种等各种因素影响。由于小麦“产量差”影响因子间存在多重共线性,很难对其影响因子进行恰当评估。为了维持黄土旱地小麦高产和持续增产,解决小麦“产量差”影响因子评估不当的问题,本研究以长武长期定位试验(1984年开始)为依托,选取其中小麦连作6个施肥处理和粮豆轮作4个施肥处理,利用DSSAT模型结合田间试验法,模拟雨养小麦生产潜力,评估田块小麦“产量差”;同时分析气象要素、土壤养分、土壤水分等时间变化及对小麦“产量差”的影响,最后利用偏最小二乘回归分析排除变量间的共线问题,定量识别影响黄土旱地长期连作与粮豆轮作下小麦“产量差”的主控因子,对指导黄土旱地地区小麦生产具有重要意义。本研究的主要结论如下:(1)DSSAT模型适合在黄土旱地应用,研究区雨养小麦生产潜力为8324 kg·hm-2。长期定位试验种植31年(1984-2014)对小麦“产量差”分析表明,小麦连作不施肥(CK)和单施化肥中小麦“产量差”显著高于单施有机肥(M)、氮磷(NP)和氮磷有机肥(NPM)配施中小麦“产量差”,其中单施P中小麦“产量差”最高,达到6996 kg·hm-2;肥料贡献率在单施磷肥中呈抑制作用,大小为-18.3 kg·kg-1,降水利用率在P处理中最低,NPM处理中最高,分别为2.34 kg·mm-1·hm-2和7.45 kg·mm-1·hm-2。粮豆轮作中小麦“产量差”在CK中最高,NPM中最低;肥料贡献率和降水利用率最高值出现在NPM中,最低值在CK中。同一降水年型下不施肥和单施化肥中小麦“产量差”高于有机肥和肥料配施中小麦“产量差”,肥料贡献率和降水利用率在肥料配施下高于单施肥和不施肥处理;各个施肥处理中小麦“产量差”干旱年平水年丰水年,其中肥料配施中效果最明显。(2)气象要素在长期定位31年间分析发现,小麦生育期积温、平均温度、太阳辐射、休闲期降水和生育年(休闲期+生育期)降水呈现增加趋势,生育期降水呈现降低趋势,月降水中3、5、6、10、12月间呈现降低趋势,其他月份呈现增加趋势。气象要素间相关分析表明,干旱年、丰水年和31年中,小麦生育年降水与休闲期降水之间呈现出极显著相关关系,相关系数分别为0.816、0.832和0.901,干旱年下生育年降水和9月、12月份降水间相关系数分别为0.619和-0.688,丰水年下,生育年降水与8月份降水间呈现出极显著正相关关系。不同施肥处理下,小麦“产量差”与休闲期降水、生育年降水和9月份降水之间呈现出显著负相关,12月份降水间呈现正相关关系。(3)长期连作与粮豆轮作施肥下土壤养分分析发现,CK和单施N处理下土壤有机质、土壤全氮、有效磷、速效钾基本维持一个水平,年际间波动范围较小。截止至2014年,连作中单施P中有效磷比试验初期提高了12.39倍,NP配施中速效钾比试验初期降低了9.58%,NPM中各个养分与试验初期比呈显著增加趋势;小麦“产量差”与CK中速效钾呈现显著正相关,与P处理中土壤养分呈现显著正相关,其他处理中小麦“产量差”均与养分间呈现负相关。截止至2014年,粮豆轮作中P处理下土壤速效磷比试验初期提高了12.90倍,NP处理中土壤速效钾比试验初期降低了14.69%,NPM配施下各个养分显著高于试验初期;小麦“产量差”与各个施肥水平下土壤养分均呈现负相关关系。(4)长期连作与粮豆轮作施肥下土壤水分分析发现,连作中单施P中播种期和收获期储水量多年均值最高,NPM配施中最低;土壤消耗水绝对值和小麦耗水量在NP处理中最高,土壤消耗水绝对值和小麦耗水量在单施P中最低,分别为70.55 mm和347.65 mm;水分利用效率肥料配施中显著高于不施肥和单施肥;相关分析发现,小麦“产量差”与单施N和NPM配施下播种期储水量和小麦耗水量间存在显著负相关,与土壤耗水量间存在显著正相关。粮豆轮作中土壤耗水量绝对值和小麦耗水量在CK中最低;播种期和收获期储水量、水分利用效率不同施肥间变化趋势与连作中类似;小麦“产量差”与单施P中播种期储水量和小麦耗水量间呈显著负相关,与土壤耗水量呈显著正相关,相关系数分别为-0.700,-0.817和0.826。(5)针对小麦“产量差”与降水因子间做多重线性检验可知,许多降水因子间存在多重共线性,特别是生育年降水与休闲期降水和生育期降水间存在严重多重共线问题。使用偏最小二乘回归分析可有效解决自变量间多重共线问题,连作、粮豆轮作、连作与粮豆轮作组合模型中均表明,第一成分和第二成分对小麦“产量差”的解释性相对较大,两个成分的累积解释性在三个模型中,分别达到66.7%、84.4%和74.9%,偏最小二乘回归分析中,第一和第二成分主要是由氮肥、磷肥、有机肥主导;通过自变量的变量投影重要性指标(VIP值)可以看出三个模型中,氮、磷、有机肥、休闲期降水和生育年降水对小麦“产量差”起着重要影响(VIP1),而生育期积温、平均温度、太阳总辐射、小麦品种、种植方式等对小麦“产量差”作用相对较弱(VIP1);从变量的回归系数来看,小麦“产量差”随着施肥量和休闲期和生育年降水量的增加而降低。
[Abstract]:The "yield difference" of dry-farmed wheat is the difference between the potential productivity of rain-fed wheat and the actual yield of farmland. It is affected by various factors such as climatic factors, fertilizers, soil nutrients and crop varieties. In order to solve the problem of inappropriate evaluation of influencing factors of "poor yield" in wheat production, a long-term positioning experiment in Changwu (since 1984) was carried out. Six fertilization treatments and four fertilization treatments of grain and soybean rotation were selected for wheat continuous cropping. The productivity potential of rain-fed wheat was evaluated by DSSAT model combined with field experiment. At the same time, the temporal variation of meteorological factors, soil nutrients, soil moisture and their effects on wheat yield difference were analyzed. Finally, the collinearity between variables was eliminated by partial least squares regression analysis, and the main controlling factors of wheat yield difference under long-term continuous cropping and crop-bean rotation were identified quantitatively. The main conclusions of this study are as follows: (1) DSSAT model is suitable for the application in the Loess dryland. The potential productivity of rain-fed wheat in the study area is 8324 kg 65507 "Yield difference" of wheat was significantly higher than that of single organic fertilizer (M), nitrogen and phosphorus (NP) and nitrogen and phosphorus organic fertilizer (NPM). The highest "yield difference" of wheat was 6996 kg The highest yield difference was found in CK and the lowest in NPM. The highest fertilizer contribution rate and the lowest precipitation utilization rate were found in NPM, and the lowest was found in CK. The contribution rate of fertilizer and the utilization rate of precipitation under combined fertilization were higher than those under single fertilization and non-fertilization, and the effect of fertilization was the most obvious in the drought year with low yield and high water content. Mean temperature, solar radiation, leisure precipitation and growth year (leisure + growth period) precipitation showed an increasing trend, growth period precipitation showed a decreasing trend, monthly precipitation showed a decreasing trend in March, May, June, October and December, and other months showed an increasing trend. The correlation coefficients were 0.816, 0.832 and 0.901, 0.619 and - 0.688, respectively, in drought and September, and 0.619 and - 0.688 in December, respectively. There was a very significant positive correlation between the precipitation of growth year and August precipitation in flood years. The results showed that soil organic matter, total nitrogen, available phosphorus and available potassium in CK and N treatments maintained one level and fluctuated between years. By the end of 2014, the available phosphorus in P increased by 12.39 times, the available potassium in NP decreased by 9.58% and the ratio of each nutrient in NPM to the initial stage of the experiment increased significantly. The yield difference of wheat was positively correlated with the available potassium in CK and positively correlated with the soil nutrient in P treatment. As of 2014, soil available phosphorus under P treatment was 12.90 times higher than that at the beginning of the experiment, soil available potassium under NP treatment was 14.69% lower than that at the beginning of the experiment, and nutrients under NPM combined application were significantly higher than that at the beginning of the experiment. Soil nutrients were negatively correlated at all fertilization levels. (4) Soil water content under long-term continuous cropping and rotation of grain and soybean showed that the average water storage at sowing and harvesting stages was the highest in continuous cropping, and the lowest in NPM application. The lowest water consumption of wheat was 70.55 mm and 347.65 mm under single P application, respectively. The water use efficiency of wheat was significantly higher in combined fertilization than that of non-fertilization and single fertilization. The absolute value of soil water consumption and water consumption of wheat in grain-bean rotation were the lowest in CK; the change trend of water storage and water use efficiency between different fertilization was similar to that in continuous cropping at sowing and harvesting stages; the "yield difference" of wheat was negatively correlated with the water storage and water consumption of wheat at sowing stage of single P application, and positively correlated with soil water consumption. The correlation coefficients were - 0.700, - 0.817 and 0.826. (5) Multiple linear tests were conducted to test the relationship between the "yield difference" and precipitation factors. It was found that there were multiple collinearities among many precipitation factors, especially between the precipitation in the growth year and the fallow period and the precipitation in the growth period. Multiple collinearity among independent variables, continuous cropping, crop-soybean rotation, continuous cropping and crop-soybean rotation combination models showed that the explanations of the first component and the second component for wheat "yield difference" were relatively large. The cumulative explanations of the two components in the three models were 66.7%, 84.4% and 74.9%, respectively. The second component is mainly dominated by nitrogen, phosphorus and organic fertilizers. Through the independent variable projection importance index (VIP value), we can see that nitrogen, phosphorus, organic fertilizer, leisure precipitation and growth year precipitation play an important role in wheat "yield difference" (VIP1), and the growth period accumulated temperature, average temperature, total solar radiation, wheat varieties, seeds and varieties. The effect of planting methods on wheat "yield difference" was relatively weak (VIP1), and the regression coefficient of variables showed that the "yield difference" of wheat decreased with the increase of fertilizer application and precipitation in fallow period and growing year.
【学位授予单位】：中国科学院教育部水土保持与生态环境研究中心
【学位级别】：博士
【学位授予年份】：2017
【分类号】：S512.1

【相似文献】