Disentangling the factors driving tree reproduction

Abstract Seed production is a strong indicator of plant fitness and plays a major role in population dynamics. However, the environmental and endogenous factors driving seed production are still poorly described and are often hard to disentangle. Consequently, we combined principal components analysis and mixed‐effects linear models that can consider the multicollinearity of the explanatory variables and quantify their respective influence on the spatio‐temporal variability in reproduction. We applied this method to analyze the relationships between cone production in Abies alba Mill. trees (6829 individual reports of cone production). We estimated the relationships between cone production and climate, elevation, tree size (diameter and height), age, crown defoliation rate, and past radial growth. We found that the distribution of annual cone production was highly skewed; 21% of the trees did not produce any cones, whereas 3.7% produced more than 100 cones in a single year. Among the endogenous factors, tree size explained 57% of the variation in cone production with large trees being the most productive. Low radial growth rates in previous years were mostly associated with higher cone production (14% of the variation), while elevation and crown defoliation had non‐monotonic effects on reproduction. Finally, years of high cone production were strongly correlated with the difference between the April temperatures of the two previous years and were also associated with a dry spring 2 yr prior to cone production followed by a humid spring the previous year. These results highlight the complexity of the abiotic and biotic factors involved in reproduction and their respective and interactive influence on the interannual and interindividual variability in cone production.