Long-term pre-release evaluations of how invasive plants respond to herbivory in introduced ranges can help identify the most effective biological control agents. However, most evaluations have been conducted within only one generation of introduced invasive species. This study tested effects across seven generations of simulated herbivory (i.e., defoliation) and habitat types on the invasive weed Alternanthera philoxeroides. We found total biomass of A. philoxeroides was decreased by defoliation during the first three generations when grown in a simulated aquatic habitat, but was decreased by defoliation only in the first generation when grown in a simulated terrestrial habitat. Defoliation significantly decreased stem diameter and collenchyma thickness and increased cortex thickness and total phenol production in A. philoxeroides grown in a simulated terrestrial or aquatic habitat during the first generation, but showed little effect during the following six successive generations. The associations between stem anatomical structural parameters and biomass significantly differed between non-defoliation and defoliation treatments in the simulated aquatic habitat, but not in the simulated terrestrial habitat. Our results suggest simulated herbivory exerted successful biological control on A. philoxeroides during the first generation in a simulated terrestrial habitat and the first three generations in a simulated aquatic habitat, but failed to restrain the vegetative offspring of A. philoxeroides. This failure of long-term biological control on A. philoxeroides might be caused by changes in the stem anatomical structure and compensatory growth. Our study highlights the need for long-term pre-release evaluation when testing the efficiency of biological control agents.

对入侵植物如何对引入范围内的食草植物做出反应的长期释放前评估可以帮助确定最有效的生物防治剂。但是，大多数评估仅在引入的入侵物种的一代内进行。这项研究测试了七代模拟草食动物（即落叶）和栖息地类型对入侵杂草天南星的影响。我们发现了A. philoxeroides的总生物量在模拟水生生境中生长时，在前三代中通过脱叶而降低，但在模拟水生生境中生长时，仅在第一代中通过脱叶而降低。落叶显著降低茎粗和厚角组织厚度和增加的皮质厚度和总酚生产空心莲子草在第一代中生长在模拟的陆地或水生栖息地中，但在随后的六个连续世代中几乎没有显示出任何作用。在模拟水生生境中，非脱叶和脱叶处理之间茎解剖结构参数与生物量之间的关联显着不同，但在模拟陆地生境中则没有。我们的结果表明，模拟草食动物在模拟陆生生境中的第一代和模拟水生栖息地的前三代中均成功地对A. philoxeroides进行了生物防治，但未能抑制philoxeroides的营养后代。长期对菲律宾天竺葵进行生物防治的失败可能是由于茎解剖结构的变化和代偿性生长引起的。我们的研究强调，在测试生物防治剂的效率时，需要进行长期的释放前评估。