Life-history trade-offs have been of primary interest in community ecology for decades. Recently, diverse forest communities have provided many compelling examples of trade-offs among traits, and correlations among traits and different aspects of performance. In principle, these correlations should help with modelling such communities by identifying which traits and trade-offs matter. But ironically, it is for these same communities that modelling progressing has been slowest, in part because the large number of species has prevented accurate parameter estimation of species parameters.
I present a general ‘dimension reduction’ statistical method that uses the reduction in freedom implied by the existence of life-history trade-offs, both to improve parameter estimates for each species, and to aid in the understanding of how trade-offs govern community dynamics. The method simultaneously solves for a life-history manifold, and the position of each species within the manifold, given any combination of data on measureable traits (e.g. wood density) and individual performance (e.g. growth rates). Varying the dimensionality of the manifold tunes between species equivalence (where all species share all parameters) and idiosyncrasy (where each species is allowed to have a unique combination of parameters).
Results/Conclusions
Applied to individual performance data from the US lake states, the method shows that: (1) a 3-d manifold provides the optimal fit to the data; this compares to the 11-d that would be needed for idiosyncrasy; (2) the trade-off axes defining the 3-d manifold are relatively complex, with each axis affecting most parameters; (3) for example, one axis is dominated by lifespan, and height, vs growth; whereas another axis is dominated by lifespan, the sensitivity of lifespan to shade, and growth, vs height; (4) the 3-d manifold captures most of the structural dynamics of the community (e.g. basal area vs time, size distributions); (5) the long-term behaviour of a simulated forest following the 3-d manifold is more reasonable than one based on species equivalence, or idiosyncrasy. The challenges and opportunities in applying similar methods to hyperdiverse tropical forests will be discussed.
