Tree Seed Dispersal in Hemlock Removal Experiment at Harvard Forest 2005

Throughout the northeast, the hemlock woolly adelgid (Adelges tsugae) threatens eastern hemlock (Tsuga canadensis) through direct mortality resulting from infestation followed by defoliation and indirect mortality in the form of pre-emptive logging. The efficacy of regeneration of vegetation following hemlock decline depends upon advance regeneration of seedlings and saplings, seed dispersal, and recruitment. In this study, we investigated (1) whether the basic parameters of height of release and wind velocity affected seed dispersal distance and (2) tested the fit of a basic ballistic model of seed dispersal to empirical data in areas both with and without canopies. We collected empirical data from seed dropping and seed rain experiments at Harvard Forest. Height and wind velocity only affected seed dispersal distance in open areas. Predicted values of dispersal distance generated by the basic ballistic model did not provide a good fit to observed dispersal data. Poor fits of the ballistic model to the data were due to the model’s inability to account for rare, long distance dispersal events. More complex models with additional parameters are necessary to model non-localized seed dispersal.

在北美东北部区域，铁杉球蚜（Adelges tsugae）对东部铁杉（Tsuga canadensis）形成重大威胁：其通过侵染引发落叶导致直接死亡，同时以预防性采伐的形式造成间接死亡。铁杉衰退后的植被恢复成效，取决于先期萌生的实生苗与幼树、种子扩散以及种群补充环节。本研究展开两项探究：其一，检验种子释放高度与风速这两项基础参数是否会影响种子扩散距离；其二，验证基础弹道种子扩散模型对有林冠与无林冠区域实测数据的拟合度。研究数据采集自哈佛森林（Harvard Forest）开展的种子脱落与种子雨实验。仅在开阔区域中，释放高度与风速才会对种子扩散距离产生影响。基础弹道模型生成的扩散距离预测值，与实测扩散数据的拟合效果不佳。弹道模型拟合效果欠佳的原因在于，其无法覆盖罕见的长距离扩散事件。若要模拟非局域性种子扩散过程，需构建包含额外参数的更为复杂的模型。