At a global scale, mean air or soil temperatures appear to be drivers of treeline position. However, at finer scales, seed availability and microsite conditions may limit the germination, establishment, and growth of tree seedlings—and therefore the position of treeline. Tree islands are features of many treelines, and they can alter microsite conditions by producing seed, providing shelter, and redistributing snow. Near Churchill, Manitoba, Canada, tree islands have higher seedling establishment than nearby forest and treeline sites and could be hot spots for treeline expansion. However, seedling establishment and tree recruitment events may be stochastic, making predictions based on the relationship between seedling counts and microsite conditions difficult. We asked whether current tree island seedling establishment can be predicted by (1) historical recruitment and/or (2) tree and snowpack characteristics. To answer these questions, we measured tree island tree characteristics and monitored seedling establishment and snowpack in five tree islands at latitudinal treeline near Churchill, Manitoba. We fit hierarchical, generalized linear mixed effect models to assess the influence of our hypothesized predictors. We found that tree recruitment in the past was strongly positively associated with current seedling density. Mean basal area and age of trees were also positively associated with seedling density, whereas tree density was negatively associated with seedling density. We found weak positive effects of snowpack snow water equivalent. Our results provide evidence for positive feedbacks within tree islands and suggest useful factors to include in models for predicting future treeline change.

在全球范围内，平均气温或土壤温度似乎是树线位置的驱动因素。但是，在较小的规模上，种子的可利用性和微地点的条件可能会限制树木幼苗的发芽，定植和生长，从而限制树木的位置。树木岛是许多林木的特征，它们可以通过产生种子，提供庇护所和重新分配雪来改变微场所的条件。在加拿大马尼托巴省丘吉尔附近，树木岛的树苗生长率高于附近的森林和林木遗址，并且可能是林木扩张的热点。但是，苗木的建立和树木的募集活动可能是随机的，因此难以根据苗木数量和微地点条件之间的关系进行预测。我们询问是否可以通过（1）历史募集和/或（2）树和积雪的特征来预测当前树岛幼苗的建立。为了回答这些问题，我们测量了曼尼托巴省丘吉尔附近纬向树线的五个树岛上的树的树木特征，并监测了幼苗的生长和积雪。我们使用分层的广义线性混合效应模型来评估假设的预测变量的影响。我们发现，过去的树木募集与当前的幼苗密度密切相关。树木的平均基础面积和年龄也与幼苗密度呈正相关，而树木密度与幼苗密度呈负相关。我们发现积雪堆雪水当量的弱正面影响。