Table 1. Comparison of five richness estimators based on 1,000 simulation data sets under a homogeneous model with S = 1,000 and CV = 0.

The five estimators are: the Chao1 estimator (Chao, 1984) denoted as ${\hat{S}}_{\mathrm{Chao1}}$, the first-order Jackknife estimator (Burnham & Overton, 1978) denoted as ${\hat{S}}_{\mathrm{Jack1}}$, the estimator proposed by Chao & Bunge (2002) denoted as ${\hat{S}}_{\mathrm{CB}}$, the estimator proposed by Lanumteang & Böhning (2011) denoted as ${\hat{S}}_{\mathrm{LB}}$, and the newly proposed estimator denoted as ${\hat{S}}_{\mathrm{GP}}$.

Size n (Observed richness)	Estimator	Average estimate	Bias	Sample s.e.	Average estimated s.e.	Sample RMSE	95% CI coverage rate
1,000 (633.0)	${\hat{S}}_{\mathrm{Chao1}}$	1,006.4	6.4	51.5	51.4	51.9	0.94a
	${\hat{S}}_{\mathrm{Jack1}}$	1,002.2	2.2a	24.2	27.2	24.3a	0.97
	${\hat{S}}_{\mathrm{CB}}$	1,009	9	72.1	106.5	72.7	0.99
	${\hat{S}}_{\mathrm{LB}}$	1,019.9	19.9	122.4	118	124.0	0.93
	${\hat{S}}_{\mathrm{GP}}$	1,025.3	25.3	84.6	78.2	88.3	0.92
2,000 (864.4)	${\hat{S}}_{\mathrm{Chao1}}$	1,000.5	0.5	22.8	21.9	22.8	0.94a
	${\hat{S}}_{\mathrm{Jack1}}$	1,134.9	134.9	20.5	19.7	136.4	0
	${\hat{S}}_{\mathrm{CB}}$	1,000.2	0.2a	22	24.1	22a	0.96
	${\hat{S}}_{\mathrm{LB}}$	1,007.8	7.8	41.7	39.5	42.4	0.93
	${\hat{S}}_{\mathrm{GP}}$	1,006.4	6.4	32	30.4	32.6	0.93
4,000 (981.8)	${\hat{S}}_{\mathrm{Chao1}}$	1,000.4	0.4a	6	6.3	6	0.94
	${\hat{S}}_{\mathrm{Jack1}}$	1,055	55	8.7	9.4	55.7	0
	${\hat{S}}_{\mathrm{CB}}$	999.4	−0.6	5	5.3	5a	0.95a
	${\hat{S}}_{\mathrm{LB}}$	1,001.6	1.6	9.5	9.7	9.6	0.95a
	${\hat{S}}_{\mathrm{GP}}$	1,000.8	0.8	7.7	8.3	7.7	0.94
8,000 (999.7)	${\hat{S}}_{\mathrm{Chao1}}$	1,000.2	0.2	0.8	0.8	0.8	0.94a
	${\hat{S}}_{\mathrm{Jack1}}$	1,002.4	2.4	1.7	1.6	2.9	0.86
	${\hat{S}}_{\mathrm{CB}}$	999.9	−0.1a	0.6	0.6	0.6a	0.84
	${\hat{S}}_{\mathrm{LB}}$	1,001.1	1.1	4.1	2.9	4.2	0.84
	${\hat{S}}_{\mathrm{GP}}$	1,000.3	0.3	1.2	1.9	1.2	0.94a

Notes.

^adenotes the smallest bias, smallest RMSE, and figure closest to 95% coverage.