Quantifying success share and plant density of direct seeding of Douglas fir using a hurdle negative binomial regression model

In recent years, direct seeding as a means of stand establishment has experienced a revival. Among other things it allows for an undisturbed root development and is assumed to be cost-efficient. While success factors have been worked out through numerous experiments, sound overviews of success and failure in practice are scarce. With the goals of (i) quantifying the success proportion and (ii) extracting the associated influencing factors, we conducted an inventory of direct seedings of Douglas fir in Northern Germany and fitted a hurdle negative binomial regression model to the data. The results reveal a high variability of plant density within, as well as between stands. We could attribute these differences, and thus the success, to stocking degree of the shelter, seed amount and age. The model shows both, a high precision and accuracy, and respects previous physical and biological knowledge of the data-generating mechanisms.