This depends on how "germination rate" is defined. If this is "numbers of seeds germinating after x weeks", then the data you have are binomial, leading to a chi² test (or a binomial GLM, to also estimate the probabilities of germination [within x weeks]).
Note that the treatment effect may depend on environmental conditions. In such a case you may include the respective factors in your experiment and in your statistical model to provide conclusions of a wider population structure that are more applicable in practice.