Tropical forest tree mortality, recruitment and turnover rates: calculation, interpretation and comparison when census intervals vary
Article first published online: 16 SEP 2004
Journal of Ecology
Volume 92, Issue 6, pages 929–944, December 2004
How to Cite
LEWIS, S. L., PHILLIPS, O. L., SHEIL, D., VINCETI, B., BAKER, T. R., BROWN, S., GRAHAM, A. W., HIGUCHI, N., HILBERT, D. W., LAURANCE, W. F., LEJOLY, J., MALHI, Y., MONTEAGUDO, A., NÚÑEZ VARGAS, P., SONKÉ, B., NUR SUPARDI , M.N., TERBORGH, J. W. and VÁSQUEZ MARTÍNEZ, R. (2004), Tropical forest tree mortality, recruitment and turnover rates: calculation, interpretation and comparison when census intervals vary. Journal of Ecology, 92: 929–944. doi: 10.1111/j.0022-0477.2004.00923.x
- Issue published online: 16 SEP 2004
- Article first published online: 16 SEP 2004
- Received 10 December 2003 revision accepted 5 May 2004 Handling Editor: Joseph Wright
- carbon dioxide;
- global environmental change;
- long-term monitoring;
- permanent sample plot;
- rain forest;
- tropical forest dynamics
- 1Mathematical proofs show that rate estimates, for example of mortality and recruitment, will decrease with increasing census interval when obtained from censuses of non-homogeneous populations. This census interval effect could be confounding or perhaps even driving conclusions from comparative studies involving such rate estimates.
- 2We quantify this artefact for tropical forest trees, develop correction methods and re-assess some previously published conclusions about forest dynamics.
- 3Mortality rates of > 50 species at each of seven sites in Africa, Latin America, Asia and Australia were used as subpopulations to simulate stand-level mortality rates in a heterogeneous population when census intervals varied: all sites showed decreasing stand mortality rates with increasing census interval length.
- 4Stand-level mortality rates from 14 multicensus long-term forest plots from Africa, Latin America, Asia and Australia also showed that, on average, mortality rates decreased with increasing census interval length.
- 5Mortality, recruitment or turnover rates with differing census interval lengths can be compared using the mean rate of decline from the 14 long-term plots to standardize estimates to a common census length using λcorr = λ × t0.08, where λ is the rate and t is time between censuses in years. This simple general correction should reduce the bias associated with census interval variation, where it is unavoidable.
- 6Re-analysis of published results shows that the pan-tropical increase in stem turnover rates over the late 20th century cannot be attributed to combining data with differing census intervals. In addition, after correction, Old World tropical forests do not have significantly lower turnover rates than New World sites, as previously reported. Our pan-tropical best estimate adjusted stem turnover rate is 1.81 ± 0.16% a−1 (mean ± 95% CI, n = 65).
- 7As differing census intervals affect comparisons of mortality, recruitment and turnover rates, and can lead to erroneous conclusions, standardized field methods, the calculation of local correction factors at sites where adequate data are available, or the use of our general standardizing formula to take account of sample intervals, are to be recommended.