Stagnating potato tuber yields in Prince Edward Island (PEI) are a major economic concern. Identification of factors influencing within-field yield variation may provide insight into strategies for overcoming yield limitations. A survey approach was used where soil samples were collected from 49 commercial fields from 2014 to 2017 from a total of 289 sampling points, with points identified as having either “high” or “low” yield, determined by yield monitor as being 5.6–11.2 t ha⁻¹ above or below average field yield, respectively. A suite of 33 measures of soil physical and chemical properties and soil pathogens were measured. Principal component analysis identified three principal components (PCs) which accounted for 85.6% of the total variation. The PC1 (reflecting 42.3% of the total variance) was associated primarily with soil texture (i.e., sand, clay) and parameters which were highly correlated with soil texture. Under the rainfed potato production on sandy-loam soils in PEI, finer soil texture is likely related to increased yield through its effect on improved soil water holding capacity. The PC2 (reflecting 29.0% of the total variance) was primarily associated with soil fertility and the PC3 (reflecting 14.4% of the total variance) was associated primarily with soil organic matter quality and soil structure. Although soil pathogens were measured at levels high enough to impact yield, they did not differ significantly between high and low yield locations. The findings of this study highlight the value in using multivariate approaches to overcome the challenges in identifying factors which control within-field yield variability.

爱德华王子岛（PEI）的马铃薯块茎产量停滞不前是一个主要的经济问题。识别影响田间产量变化的因素可能有助于洞悉克服产量限制的策略。使用调查方法，从2014年至2017年从总共289个采样点收集了49个商业田地的土壤样品，这些点被确定为“高”或“低”产量，由产量监控器确定为5.6–11.2 t ha ^-1高于或低于平均田间产量。对土壤的理化性质和土壤病原体进行了33项测量。主成分分析确定了三个主成分（PC），占总变异的85.6％。PC1（反映总方差的42.3％）主要与土壤质地（即沙子，粘土）和与土壤质地高度相关的参数相关。在PEI的沙壤土上雨养马铃薯生产中，较细的土壤质地可能通过提高土壤持水能力而与产量增加有关。PC2（占总变化的29.0％）主要与土壤肥力相关，PC3（占总变化的14.4％）主要与土壤有机质和土壤结构相关。尽管对土壤病原体的测定水平足以影响产量，但在高产和低产地区之间差异并不明显。这项研究的结果突出了使用多变量方法来克服在确定控制田间单产可变性因素方面的挑战的价值。