Figure 1.
Frequency distribution of sample sizes for estimating population niche widths in published studies using stable isotope methods (TA or SEA).
The data were sourced through literature search in ISI Web of Knowledge and Scopus and publications are listed in Appendix S1.
Figure 2.
Frequency distributions for δ13C and δ15N values in empirical data set used in this study.
Isotope values of perch (upper panels) and roach (lower panels) populations from Lake Jyväsjärvi showing their peaked and skewed distributions (Skewness: perch δ13C = 1.896, δ15N = −1.443; roach δ13C = 1.162, δ15N = −0.883 Kurtosis: perch δ13C = 4.816, δ15N = 2.501; roach δ13C = 1.239, δ15N = 0.746).
Figure 3.
Perch and roach isotope niche width estimates from Lake Jyväsjärvi.
Standard ellipse areas (SEA, solid lines) and convex hull TA (dashed lines) are estimated for perch (grey symbols and lines) and roach (black symbols and lines) populations using SIBER [18].
Figure 4.
Population isotopic niche width modelling using increasing sample size.
Perch (left panel) and roach (right panel) population datasets were used to estimate the convex hull TA, SEA and SEAc calculated for 5000 random selections of individuals with increasing sample size (n+5). Lines represent the upper 97.5%, 75%, 50% and lower 25% and 2.5% percentiles for the niche area estimates after each 5000 resamplings with increasing sample size. The solid grey line indicates the observed “true” total niche area for each metric (n = 202 for perch and n = 173 for roach).
Figure 5.
Isotopic niche widths calculated using simulated data.
The data followed a multivariate normal distribution, but otherwise matched the empirical roach and perch isotope data with identical sample size, sample means and variance-covariance matrix. Lines represent the upper 97.5%, 75%, 50% and lower 25% and 2.5% percentiles for the niche area estimates after each 5000 resamplings with increasing sample size (n+5) from the simulated perch (left panel) and roach (right panel) populations.