Table 1. Predicting the biodiversity in tropical forests.

Predicted total number of species, S_pred, at the whole-forest scale (corresponding to p = 1) for each of the 15 tropical forests in our database. Predictions are determined by using information on the sampled scale p* (fourth column), where we observe N* trees belonging to S* species (second and third columns). In the fifth column, we show the predictions obtained by using the NB framework with a single NB for fitting the sampled SAD. SEs were computed by propagating the errors in the fitting parameters of the SAD (obtained by the bootstrapping method) and of S*. The latter has been determined as follows: For each data set, we created the corresponding predicted forest at the scale p = 1 by generating S_pred numbers distributed according to an NB with parameters (r, ξ). We then sampled the p% of the list of individuals, as in the original data. The last two columns show the predictions of the LS and Chao methods.

Forest	S*	N*	p*%	Spred (NB)	Spred (LS)	Spred (Chao)
Amazonia	4962	553949	0.00016	13602 ± 711	14984	5561
Barro Colorado	301	222602	3.20513	366 ± 15	419	315
Bukit Barisan	340	14974	0.00169	471 ± 40	1020	346
Bwindi	128	18490	0.01813	163 ± 15	288	129
Caxiuana	386	32701	0.01818	437 ± 14	915	386
Cocha Cashu	489	16640	0.00035	731 ± 63	1674	501
Korup	226	17427	0.00473	282 ± 23	591	226
Manaus	946	38933	0.06000	1016 ± 14	2242	956
Nouabalé- Ndoki	110	7196	0.00143	125 ± 8	316	110
Pasoh Forest Reserve	927	310520	0.35714	1193 ± 36	1590	1049
Ranomafana	269	34580	0.01463	336 ± 22	620	269
Udzungwa	109	18447	0.00302	146 ± 20	269	114
Volcan Barva	392	44439	0.02025	448 ± 16	895	395
Yanachaga	209	2041	0.00372	802 ± 211	802	259
Yasuni	481	13817	0.61100	565 ± 20	974	484