中文摘要：

目的揭示农田尺度下冬小麦晚霜冻害与产量关系及其空间分布规律，探讨其空间差异性原因，为冻害风险早期预判及影响因子调控提供先验知识和依据。方法选择矮抗58冬小麦品种作为研究对象，以2013年4月发生的3次自然霜冻为契机，构建死穗率、残穗率、残穗指数和减产率等晚霜冻害评价指标。基于商丘市一农户麦田内100个采样点（以5 m间隔定点）的小麦产量、冻害考察以及土壤肥力测定数据，运用多元线性逐步回归、地统计学、系统聚类、方差分析以及多重比较等方法对穗数、实际产量和晚霜冻害评价指标进行空间统计分析，探讨其与小麦发育进程、返青期土壤肥力因子的关系。结果逐步回归分析表明，死穗率是影响穗数的关键因子，呈负效应。影响实际产量的因子是残穗指数、死穗率和残穗率，3个因子均呈负效应，值越大，实际产量越低，其中残穗指数的影响最大（直接通径系数为-0.453）。影响减产率的因子为死穗率、残穗率和残穗指数，3个因子均为正效应，值越大，减产率越高，其中死穗率的影响最大（直接通径系数为0.626）。晚霜冻害具有显著的正空间自相关特性，冻害程度相近的样点在局域空间上呈集聚分布状态；在所有冻害评价指标中，减产率的空间集聚性最强（Moran’s I=0.5538）。冻害分区结果表明，随着冻害程度加深，穗数和实际产量显著降低（P＜0.05）；死穗率增幅最大（达271.3%），其次是残穗率和残穗指数（分别为36.4%和31.8%），它们共同成为导致减产率大幅攀升（增幅达132.1%）的因素；冻害程度最重的区域几乎连片分布，空间集聚性明显。小麦发育进程和返青期土壤养分的空间差异明显，与晚霜冻害具有一定空间关联性。土壤全氮、水解氮、速效磷、速效钾和有机质等与冻害指标之间达显著负相关（P?

英文摘要：

Objective]At the field scale, the study was conducted to investigate spatial differences of late frost injury to winter wheat and their influences on yields. The reasons for the spatial differences were discussed, which might provide a priori knowledge for predicting late frost damage and regulating influence factors.[Method]A frost-prone farmland in Shangqiu was selected as the＆amp;nbsp;study area where a winter wheat cultivar Aikang58 was sowed. Natural frost occurred in April 7, 10 and 21, 2013, when the winter wheat was in the periods from middle jointing to booting stage. At maturity, ear number and actual yield per square meter, and soil fertilities were respectively measured based on one hundred sampling points （the interval between samples is 5 m）. Four indexes, such as dead ear rate （DER）, injured ear rate （IER）, injured ear index （IEI） and yield loss rate （YLR）, were established to evaluate late frost damage. Their spatial differences and relations with ear number and actual yield were estimated using multiple linear stepwise regression method, geo-statistics method, cluster method, ANOVA and multiple comparison methods. In order to reveal the reasons for the spatial differences, late frost indexes related to developmental progressions of winter wheat and soil fertility factors were measured at early returning green stage.[Result] Through the stepwise regression, DER was found to be the only factor affected ear number and showed a negative effect. Factors, including IEI, DER and IER, all showed negative effects on actual yield. Of all the factors, IEI was the most powerful （its direct path coefficient reached -0.453） to actual yield. DER, IER and IEI were the factors impacted YLR and all showed positive effects, indicating that YLR increased with their increments. DER showed the most influence （its direct path coefficient was up to 0.626） on YLR. Late frost damage had a positive spatial autocorrelation （P〈0.05）. Sampling points with similar frost damage degrees tended