lifeTable_bootstrap: Bootstrap Standard Errors of Life Table Parameters

In insectecol: Insect Ecology Data Analysis Toolkit

Bootstrap Standard Errors of Life Table Parameters

Description

Estimates the standard errors and percentile confidence intervals of all scalar life table parameters with the bootstrap technique used by TWOSEX-MSChart: complete individual records (stage durations, adult days, sex and daily oviposition) are resampled with replacement B times, and every parameter is recomputed from each resampled cohort. Individuals that died before the adult stage are resampled as well - they carry their mortality and zero fecundity into the replicates.

Usage

lifeTable_bootstrap(lt, B = 1e+05, seed = NULL, conf.level = 0.95)

Arguments

lt	A life_table object returned by lifeTable_read or lifeTable_build.
B	Integer; number of bootstrap replicates. The published TWOSEX-MSChart standard is 100000 (the default); smaller values run faster but give rougher standard errors.
seed	Integer; seed for the random number generator. Set it to make the results exactly reproducible; NULL (default) uses the current R session state. The session state is restored when the function exits.
conf.level	Numeric; confidence level of the percentile intervals (default 0.95).

Details

The parameters covered are the mean developmental time of every immature stage (averaged over the individuals that entered the stage, including those that died during it), the female and male adult longevity, and - when oviposition data are present - the mean fecundity F, the net reproductive rate R0, the intrinsic rate of increase r, the finite rate of increase lambda and the mean generation time T. The intrinsic rate is solved per replicate from the Euler-Lotka equation with the same age convention as calc_r, but with a widened search interval so that declining cohorts (R0 < 1, i.e. r < 0) are also handled. Replicates in which no female is drawn have NA fecundity, and replicates without any offspring have NA r, lambda and T; such replicates are excluded from the means, standard errors and intervals of the affected parameters (column Valid_B shows how many remained). Curve-type outputs (s_xj, l_x, m_x, e_x) are not bootstrapped.

Attach the result to an analysis to have it exported by lifeTable_export: results$boot <- lifeTable_bootstrap(lt).

Value

An object of class life_table_boot: a list with

summary	data frame; one row per parameter with the original point estimate, the bootstrap mean, the bootstrap standard error and the percentile confidence interval
boot	numeric matrix; the raw B x p replicate values
B, conf.level, N, parameters	the settings used

References

Meyer, J. S., Ingersoll, C. G., McDonald, L. L. and Boyce, M. S. (1986) Estimating uncertainty in population growth rates: jackknife vs. bootstrap. Ecological Modelling 29, 251-271.

Efron, B. and Tibshirani, R. J. (1993) An Introduction to the Bootstrap. New York: Chapman and Hall.

Chi, H., You, M. S., Atlihan, R., Smith, C. L., Kavousi, A., Ozgokce, M. S., Guncan, A. and Tuan, S. J. (2020) Age-stage, two-sex life table: an introduction to theory, data analysis, and application. Entomologia Generalis 40(2), 103-124.

See Also

lifeTable_boot_test for the paired bootstrap test between two cohorts, lifeTable_calculate_all for the point estimates.

Examples

f <- system.file("extdata", "lifetable_example.csv", package = "insectecol")
lt <- lifeTable_read(f)

## B = 100000 is the recommended setting for publications; a smaller
## B is used here so that the example runs fast
bt <- lifeTable_bootstrap(lt, B = 2000, seed = 1)
bt$summary

insectecol documentation built on Oct. 5, 2026, 5:08 p.m.