The volatility and periodicity of tree masting at the global scales: The role of species traits and phylogeny

James Clark, Statistical Science, Duke University, Durham, NC, Benoit Courbaud, Mountain Ecosystem Research Unit, Cemagref - Grenoble, Saint-Martin-d'Hères, France, Valentin Journe, LESSEM, University of Grenoble Alpes, INRAE, St-Martin-d'Heres, France, Christopher L. Kilner, Duke University, Durham, NC, Georges Kunstler, UR EMGR, Irstea, St-Martin-d'Heres, France, Tong Qiu and Jennifer J. Swenson, Nicholas School of the Environment, Duke University, Durham, NC, C. Lane Scher, University Program in Ecology, Duke University, Durham, NC

Background/Question/Methods
Understanding the masting phenomenon is a goal of forest and evolutionary ecology. Masting is the quasi-periodic, quasi-synchronous seed production at lagged intervals. Understanding the drivers and the consequences of masting require ways to quantify its distinguishing feature: the concentration of variance in low frequencies (long lags) and between individuals, all in the context of a variable environment. Current methods based on the coefficient of variation and its derivatives omit the basic attribute of frequency. Extracting mean intervals between events does not work, because there is no threshold value that distinguishes an event from background. Using the Masting Inference and Forecasting (MASTIF) network we estimated 18 million tree-years, incorporating uncertainty in parameters, process, and data, including dependence between trees and over time. We developed the mast score consisting of dual components, volatility and period, both of which emerge from the spectral density. Volatility quantifies the variance that is concentrated at low frequency (long lags). Periodicity is a variance-weighted (inverse) frequency (years), describing the tendency to lag reproductive effort. Application to 500 species across five continents revealed the relationships between volatility, periodicity, and seed mass and their dependence on phylogeny.
Results/Conclusions
Relationships between seed mass, volatility, and periodicity emerge from the year effects that remain after extracting effects of climate, tree size, and crowding. Across all species, seed mass is positively correlated with period (0.21 (CI = 0.12, 0.30)) and negatively correlated with volatility (-0.16 (-0.25, -.06)). For the majority of phylogenetic groups there is a positive correlation between period and volatility, although it is negative in one of the most volatile groups, Pinaceae (r = -0.27). The strongest positive correlations were observed within the Lauraceae (0.71), Ulmaceae (0.59), Euphorbiaceae (0.53), and Sapindaceae (0.49). Due to these phylogenetic differences, there is no overall correlation between period and volatility. Temperate and tropical species broadly overlap in terms of volatility and periodicity—there is no systematic difference between these major biomes in their tendency to generate volatile or periodic seed production. Thus, there is a general tendency for large-seeded species to mast at longer intervals, but it does not represent the dominant source of variation in reproductive effort. Results highlight the role of phylogenetic conservatism, while showing the connections between the components variation and seed mass at the level of families.