REPORTS

129,494 arthropods representing 6144 focal spe-

Arthropod Diversity in a Tropical Forest cies (Fig. 1 and table S1) from 0.48 ha of inten-
sively sampled mature forest. This allowed us to
Yves Basset,1,2,3* Lukas Cizek,2,4 Philippe Cuénoud,5 Raphael K. Didham,6 François Guilhaumon,7 extrapolate focal arthropod species richness to
Olivier Missa,8 Vojtech Novotny,2,4 Frode Ødegaard,9 Tomas Roslin,10 Jürgen Schmidl,11 a larger forest area with unprecedented power,
Alexey K. Tishechkin,12 Neville N. Winchester,13 David W. Roubik,1 Henri-Pierre Aberlenc,14 through a series of best-informed species richness
Johannes Bail,11 Héctor Barrios,3 Jon R. Bridle,15 Gabriela Castaño-Meneses,16 Bruno Corbara,17 estimates derived from six competing models for
Gianfranco Curletti,18 Wesley Duarte da Rocha,19 Domir De Bakker,20 Jacques H. C. Delabie,19 each of 18 focal data sets. Using taxon ratios to
Alain Dejean,21 Laura L. Fagan,6 Andreas Floren,22 Roger L. Kitching,23 Enrique Medianero,3 estimate the species richness of nonfocal taxa
Scott E. Miller,24 Evandro Gama de Oliveira,25 Jérôme Orivel,26 Marc Pollet,27 Mathieu Rapp,28 [see “Extrapolating results to nonfocal taxa” in
Sérvio P. Ribeiro,29 Yves Roisin,30 Jesper B. Schmidt,31 Line Sørensen,31 Maurice Leponce20 materials and methods (20)], we then predicted
the total species richness of the study area. We
Most eukaryotic organisms are arthropods. Yet, their diversity in rich terrestrial ecosystems is also evaluated differences in relative species ac-
still unknown. Here we produce tangible estimates of the total species richness of arthropods in cumulation rates among arthropod guilds, across
a tropical rainforest. Using a comprehensive range of structured protocols, we sampled the spatial scales.
phylogenetic breadth of arthropod taxa from the soil to the forest canopy in the San Lorenzo forest, Although individual estimators adjusting for
Panama. We collected 6144 arthropod species from 0.48 hectare and extrapolated total species different aspects of sampling design offered slight-
ly different estimates (Fig. 1B), the total spe-

Downloaded from www.sciencemag.org on December 13, 2012

richness to larger areas on the basis of competing models. The whole 6000-hectare forest reserve
most likely sustains 25,000 arthropod species. Notably, just 1 hectare of rainforest yields >60% of cies richness for the entire San Lorenzo forest
the arthropod biodiversity held in the wider landscape. Models based on plant diversity fitted the (~6000 ha) was consistently quantified at be-
accumulated species richness of both herbivore and nonherbivore taxa exceptionally well. This tween 18,000 and 44,000 species (including focal
lends credence to global estimates of arthropod biodiversity developed from plant models. and nonfocal species). In particular, the most like-
ly lower bound of species richness was estimated
to be at least 21,833 species [95% confidence
ost eukaryote species are terrestrial Although many studies reported species richness level (CL) = 18,665, 29,420; model a1 in Fig.

M arthropods (1), and most terrestrial

arthropods occur in tropical rainforests
(2). However, considerably greater sampling
for selected groups of well-studied insect taxa,
no satisfactory estimate of total arthropod species
richness exists for a single tropical rainforest lo-
1B], and the biologically and statistically best-
supported estimate of richness (criteria outlined
in table S2) was 25,246 species (95% CL =
effort is required in tropical arthropod surveys cation to date. 19,721, 33,181, model B+S in Fig. 1B). Accord-
to yield realistic estimates of global species The unstructured collection and small-scale ing to our estimates, a single hectare of rain-
richness (3–7). A basic hindrance to estimat- survey of tropical arthropods cannot yield con- forest will be inhabited by an average of 18,439
ing global biodiversity lies in a lack of empir- vincing estimates of total species richness at a species (95% CL = 17,234, 18,575; Fig. 2B).
ical data that establish local biodiversity, which specific forest (7–9). Most studies either target A relatively large proportion of the expected
can be scaled up to achieve a global estimate. few arthropod orders or trophic guilds, or use a species richness of the forest was recovered for
limited array of sampling methods, or ignore the most of our focal taxonomic groups (Fig. 2). For
1
Smithsonian Tropical Research Institute, Panama City, Re-
diverse upper canopy regions of tropical forests example, high proportions of all ant species and
public of Panama. 2University of South Bohemia, 370 05 (10–15). Moreover, sampling protocols have of the parasitoid species targeted in our study
Ceske Budejovice, Czech Republic. 3Universidad de Panamá, rarely been structured in such a way that, with were collected from our 12 sites, whereas fungal
Panama City, Republic of Panama. 4Czech Academy of Sciences, increased sampling, incomplete data on local feeders would require more intensive sampling
370 05 Ceske Budejovice, Czech Republic. 5Muséum d’histoire
diversity (7) can be extrapolated to estimate total to achieve adequate coverage (Fig. 2). Beta di-
naturelle de la Ville de Genève, 1208 Genève, Switzerland.
6
The University of Western Australia and CSIRO Ecosystem Sci- species richness across multiple spatial scales versity of all arthropods (in the broad sense of
ences, 6009 Perth, Australia. 7Catedra Rui Nabeiro, Universidade (16). Where such structured estimates are made, species turnover among sites) increased roughly
de Évora, 7004-516 Évora, Portugal. 8University of York,York it is invariably for insect herbivores on their host linearly with cumulative area surveyed (F1,3 =
YO10 5DD, UK. 9Norwegian Institute for Nature Research, 7485 plants (5). However, species accumulation rates 2422.5, P < 0.001). With increasing sampling
Trondheim, Norway. 10University of Helsinki, 00014 Helsinki,
Finland. 11University of Erlangen-Nuremberg, 91058 Erlangen, may differ markedly for nonherbivore guilds, effort, sample coverage [an unbiased measure of
Germany. 12Santa Barbara Museum of Natural History, Santa which include more than half of all described sample completeness, see (20)] was high and ac-
Barbara, CA 93105, USA. 13University of Victoria, Victoria, BC arthropod species (1, 17). As the degree of host cumulated at significantly different rates across
V8W 2Y2, Canada. 14Cirad, 34988 Montferrier-sur-Lez, France. specificity (effective specialization) of other guilds different arthropod orders and guilds, and across
15
University of Bristol, Bristol BS8 1UD, UK. 16Universidad
Nacional Autónoma de México, México 0510 DF, México. 17Uni- can be much lower than that of insect herbivores, the various guilds comprised by beetles (Fig. 3).
versité Blaise Pascal, 63000 Clermont-Ferrand, France. or may be driven by different factors (18, 19), However, despite the high sample coverage val-
18
Museo Civico di Storia Naturale, 10022 Carmagnola, Italy. global estimates based on herbivores alone are ues, we cannot discount the possibility that there
19
Centro de Pesquisas do Cacau, 45600-000, Itabuna, and questionable. Consequently, extensive cross-taxon were some vanishingly rare species that may not
Universidade Estadual de Santa Cruz, 45662-900 Ilhéus-Bahia,
Brazil. 20Institut Royal des Sciences Naturelles de Belgique, 1000
surveys with structured protocols at reference have been discovered with the sampling proto-
Brussels, Belgium. 21University of Toulouse III, 31062 Toulouse, sites may be the only effective approach toward cols used in this study.
France. 22Universität Würzburg, 97070 Würzburg, Germany. estimating total arthropod species richness in Despite idiosyncrasies in the rate of increase
23
Griffith University, Nathan QLD 4111, Australia. 24National tropical forests (3). in sample completeness across insects groups,
Museum of Natural History, Washington, DC 20008, USA.
25
Centro Universitário UNA, 30350-540 Belo Horizonte-MG,
To provide a comprehensive estimate of total the high proportion of overall species richness
Brazil. 26CNRS, 97379 Kourou, France. 27Research Institute for arthropod species richness in a tropical rainforest, detected at small spatial scales (Figs. 2 and 3)
Nature and Forest, 1070 Brussels, Belgium. 28Muséum d’histoire we established a collaboration involving 102 re- has a remarkable consequence. Based on a gen-
naturelle, 2000 Neuchâtel, Switzerland. 29Universidade Fed- searchers with expertise encompassing the full eral relationship between species numbers and
eral de Ouro Preto, 35400-000 Ouro Preto-MG, Brazil and breadth of phylogenies and feeding modes present area, we estimate that almost two-thirds (64%)
Universidade dos Açores, 9700-851 Terceira, Portugal. 30Uni-
versité Libre de Bruxelles, 1050 Brussels, Belgium. 31Natural among arthropods (20). This consortium invested of all species in SLPA occur in a single hectare
History Museum of Denmark, 2100 Copenhagen, Denmark. a total of 24,354 trap- (or person-) days sam- of rainforest (Fig. 2). Our plant models predicted
*To whom correspondence should be addressed. E-mail: pling the San Lorenzo forest (SLPA) in Panama total arthropod richness in the San Lorenzo forest
[email protected] using structured protocols (fig. S1). We identified to a precision of 1% (correlation between rich-

www.sciencemag.org SCIENCE VOL 338 14 DECEMBER 2012 1481

 REPORTS
ness estimates provided by the plant model and the limited heterogeneity of the study area com- (5, 21). Although the scope for direct compar-
best estimates: r = 0.992, P < 0.0001, N = 18; pared to larger geographical scales). ison is limited because of regional differences in
Fig. 1C). Notably, small discrepancies between Because this study targeted the full spectrum sampling effort, lowland tropical forest in Panama
observed arthropod species richness and esti- of arthropods, it offers a comprehensive test of seems to support 2.1 to 8.4 times as many ar-
mates derived from floristic diversity appeared previous estimates of species richness based only thropod species as observed in temperate forests
not to be scale-dependent (Fig. 1D). Hence, even on selected guilds or taxa. Reassuringly, our well- (table S3). While this supports the obvious truism
for arthropod guilds other than herbivores, plant resolved estimates of tropical arthropod species that tropical arthropods are indeed more diverse
diversity seems a powerful predictor of species richness are of the same order of magnitude as than their temperate counterparts (22), the mag-
richness across areas varying in size (at least prior estimates (table S3), adding credence to nitude of that difference is far lower than many
within the limits of our study design and given recent estimates of tropical arthropod diversity previous estimates would suggest (2).

Downloaded from www.sciencemag.org on December 13, 2012

Fig. 1. Number of arthropod species estimated at SLPA (20). (A) Number of calculated with local instead of global ratios (fig. S3); S: best estimates, in-
species (closed bars, log scale) and individuals (open bars, log scale) collected cluding only statistical arguments; and models a1 to a5. ?: optimization al-
in 0.48 ha for each data set (three-letter guild code as in table S1) and number gorithms did not converge to allow calculations of CL. Our estimates are robust
of species estimated in SLPA (best estimate, shaded boxes). Numbers above to even moderate to large shifts in taxon ratios (table S5). (C) Plot of the
bars identify the best model used for calculation (a1 to a6, fig. S2) and the number of species estimated in SLPA with the plant model against that es-
percentage of singletons. (B) Number of arthropod species estimated for SLPA timated with the best model, for each data set. Line denotes unity. (D) Plot of
(dots: all focal taxa; shaded boxes: focal and nonfocal taxa; table S4), as the percentage error between all arthropod species observed and estimated by
estimated by different methods: B+S: best estimates, including both biological the plant model against cumulative number of sites. Shaded boxes indicate
and statistical arguments (table S2); B+Sloc: same as B+S but estimates means and 95% CL.

1482 14 DECEMBER 2012 VOL 338 SCIENCE www.sciencemag.org

 REPORTS
The implications of the results observed on in the tropical fauna (table S3), we may then ar- labor-intensive to survey, such an “umbrella” ap-
a local scale are clear. For every species in the gue that conservation planning for biodiversity proach may be an efficient way forward.
well-known vascular flora (1294 species), avifauna should be largely determined by the spatial scaling Nonetheless, our findings also suggest that
(306 species), and mammalian fauna (81 species) of arthropod diversity. In this context, the asso- large-scale, region-wide understanding of trop-
of SLPA, we estimate that there will be a mini- ciation between the species richness of plants and ical arthropod richness may actually be more
mum of 17, 71, and 270 arthropod species, re- arthropods detected across spatial scales suggests achievable than previously assumed. Our data
spectively (based on lower bound of species that conservation efforts targeted at floristically indicate that a thorough sampling of 1 ha of rain-
richness) and most likely as high as 20, 83, and diverse sites may also serve to conserve arthro- forest may reveal nearly two-thirds of all ar-
312 arthropod species, respectively (table S3). pod diversity across both taxonomic lineages thropod species present in a much larger area
Based on the dominance of arthropod species and trophic guilds. As arthropods are notoriously (6000 ha in our case; Fig. 2B), consistent with

Downloaded from www.sciencemag.org on December 13, 2012

Fig. 2. Accumulation of species richness with area at SLPA (20). For all data sets (for all arthropods, including nonfocal species, values are indi-
groups, a high proportion of overall species richness was detected at small cated but the curve not plotted for the sake of clarity). Each curve is
spatial scales. (A) Partitioning of species richness within arthropod guilds characterized by its function (Ex, Exponential; Lo, Lomolino; Po, Power; We,
at different spatial scales (a: single site of 0.04 ha; b3: three sites spaced cumulative Weibull), its value for 1 ha (intersection with vertical line,
apart totaling 0.12 ha; b6: six sites totaling 0.24 ha; b12: 12 sites totaling shaded boxes with mean and 95% CL), and the percentage of the number
0.48 ha; bha: 1.0 ha; bSLPA: 6000 ha; means T SEM are shown for a, b3, of species present in 1 ha relative to the number of species estimated to
and b6). (B) Species-area models for the main arthropod groups and large occur in SLPA.

www.sciencemag.org SCIENCE VOL 338 14 DECEMBER 2012 1483

 REPORTS

Downloaded from www.sciencemag.org on December 13, 2012

Fig. 3. Average sample coverage [TSEM; error bars, see methods (20)] plotted against the cumulative number of sites surveyed, for the main (A) arthropod
guilds and orders and (B) beetle guilds. For the sake of clarity, SEMs are omitted in (A).

reports of relatively low beta diversity of insect other tropical locations around the world, using 20. Materials and methods are available as supplementary
herbivores in tropical rainforests (23). Hence, the current template as a foundation (25). materials on Science Online.
21. V. Novotny et al., Nature 416, 841 (2002).
to determine the species diversity of a tropical 22. T. C. Bonebrake, L. C. Ponisio, C. L. Boggs, P. R. Ehrlich,
rainforest, the total area sampled need not be References and Notes Biol. Conserv. 143, 1831 (2010).
overly large—provided that the sampling design 1. Z.-Q. Zhang, Zootaxa 3148, 99 (2011). 23. V. Novotny et al., Nature 448, 692 (2007).
2. T. L. Erwin, Coleopt. Bull 36, 74 (1982). 24. A. J. Hamilton et al., Am. Nat. 177, 544 (2011).
adequately covers both microhabitats and plant 3. R. M. May, Philos. Trans. R. Soc. Lond. B Biol. Sci. 359, 25. R. L. Kitching et al., Mem. Queensl. Mus. Nat. 55, 235 (2011).
species. However, this does not imply that most 733 (2004).
arthropod species have self-supporting popula- 4. R. M. May, Science 329, 41 (2010). Acknowledgments: IBISCA-Panama is an initiative of
tions in small forest areas or fragments. 5. A. J. Hamilton et al., Am. Nat. 176, 90 (2010). Pro-Natura International, Océan Vert, the universities Blaise
6. C. Mora, D. P. Tittensor, S. Adl, A. G. B. Simpson, Pascal and of Panama, and the Smithsonian Tropical Research
On a global scale, our results have implica- B. Worm, PLoS Biol. 9, e1001127 (2011). Institute (STRI), with core funding from SolVin-Solvay SA, STRI,
tions for current estimates of total species rich- 7. J. A. Coddington, I. Agnarsson, J. A. Miller, M. Kuntner, the United Nations Environment Programme, the Smithsonian
ness, which have been weakened by the lack of G. Hormiga, J. Anim. Ecol. 78, 573 (2009). Institution (Walcott Fund), the European Science Foundation,
knowledge regarding the strength of association 8. Y. Basset, Plant Ecol. 153, 87 (2001). and the Global Canopy Programme. J. Herrera, E. Andrade,
9. T. M. Lewinsohn, T. Roslin, Ecol. Lett. 11, 398 (2008). M. Samaniego, S. J. Wright, N. Baiben, S. Bechet, J. Belleguic,
between vascular plant species and nonherbivore 10. V. Novotny et al., Conserv. Biol. 18, 227 (2004). T. Aubert, K. Jordan, G. Ebersolt, D. Cleyet-Marrel, L. Pyot,
guilds (5). Based on the close association ob- 11. V. Novotny et al., J. Anim. Ecol. 79, 1193 (2010). O. Pascal, P. Basset, and E. Bauhaus helped with logistics
served here between floristic diversity and both 12. F. Ødegaard, Biodivers. Conserv. 15, 83 (2006). in the field. A. Barba, R. Cabrera, A. Cornejo, I. Díaz,
herbivore and nonherbivore species richness, we 13. J. T. Longino, J. Coddington, R. K. Colwell, Ecology 83, A. F. R. do Carmo, I. C. do Nascimento, E. A. dos Santos,
689 (2002). M. González, A. Hernandez, M. Manumbor, M. Mogia,
tentatively conclude that the most recent estimate
14. Y. Basset, V. Novotny, S. E. Miller, R. L. Kitching, Eds., S. Pinzón, B. Pérez, L. S. Ramos-Lacau, and O. Valdez helped
of global tropical arthropod species [6.1 million Arthropods of Tropical Forests. Spatio-temporal Dynamics with initial sorting of the arthropod and plant material. Data
arthropod species (24)] does not require drastic and Resource Use in the Canopy (Cambridge Univ. Press, (as of 10 May 2012) have been deposited in the Dryad
correction to account for differential scaling rela- Cambridge, 2003). repository: http://dx.doi.org/10.5061/dryad.f3p75.
tionships of nonherbivore taxa. The robust es- 15. J. T. Longino, R. K. Colwell, Ecol. Appl. 7, 1263 (1997).
16. R. K. Colwell, J. A. Coddington, Philos. Trans. R. Soc. Supplementary Materials
timates of local arthropod diversity derived in our Lond. B Biol. Sci. 345, 101 (1994).
www.sciencemag.org/cgi/content/full/338/6113/1481/DC1
study thus support previous estimates of global Materials and Methods
17. D. R. Strong, J. H. Lawton, T. R. E. Southwood, Insects on
Supplementary Text
species richness. They also show how stratified Plants. Community Patterns and Mechanisms (Blackwell,
Figs. S1 to S3
sampling designs and broad scientific cooper- Oxford, ed. 1, 1984).
18. E. Siemann, D. Tilman, J. Haarstad, M. Ritchie, Am. Nat. Tables S1 to S5
ation may be developed to formulate efficient 152, 738 (1998).
References (26–151)
estimates of tropical arthropod diversity. Similar 19. V. Novotny, Y. Basset, Proc. R. Soc. Lond. 272, 1083 29 June 2012; accepted 1 November 2012
initiatives have recently been implemented in (2005). 10.1126/science.1226727

1484 14 DECEMBER 2012 VOL 338 SCIENCE www.sciencemag.org

View publication stats