Arthropod ßiversity io a IropicaI Forest

¥ves 8asset, ¯ tukas Citek, Philippe Cuenoud, kaphael k. 0idham, françois 0uilhaumon,
0livier Hissa, vojtech Novotny, frode Ødegaard, Iomas koslin, )urgen Schmidl,
Alexey k. Iishechkin, Neville N. winchester, 0avid w. koubik, henri-Pierre Aberlenc,
)ohannes 8ail, hector 8arrios, )on k. 8ridle, 0abriela Castano-Heneses, 8runo Corbara,
0ianfranco Curletti, wesley 0uarte da kocha, 0omir 0e 8akker, )acques h. C. 0elabie,
Alain 0ejean, taura t. fagan, Andreas floren, koger t. kitching, fnrique Hedianero,
Scott f. Hiller, fvandro 0ama de 0liveira, )eróme 0rivel, Harc Pollet, Hathieu kapp,
Servio P. kibeiro, ¥ves koisin, )esper 8. Schmidt, tine Sørensen, Haurice teponce
Vost eukaryotìc organìsns are arthropods. Yet, theìr dìversìty ìn rìch terrestrìal ecosystens ìs
stìll unknoWn. here We produce tangìble estìnates of the total specìes rìchness of arthropods ìn
a tropìcal raìnforest. Usìng a conprehensìve range of structured protocols, We sanpled the
phylogenetìc breadth of arthropod taxa fron the soìl to the forest canopy ìn the San Lorento forest,
lanana. we collected 6144 arthropod specìes fron 0.48 hectare and extrapolated total specìes
rìchness to larger areas on the basìs of conpetìng nodels. Ihe Whole 6000-hectare forest reserve
nost lìkely sustaìns 25,000 arthropod specìes. Notably, |ust 1 hectare of raìnforest yìelds >60% of
the arthropod bìodìversìty held ìn the Wìder landscape. Vodels based on plant dìversìty fìtted the
accunulated specìes rìchness of both herbìvore and nonherbìvore taxa exceptìonally Well. Ihìs
lends credence to global estìnates of arthropod bìodìversìty developed fron plant nodels.
M
ost eukaryote species are terrestrial
arthropods (1), and most terrestrial
arthropods occur in tropical rainIorests
(2). However, considerably greater sampling
eIIort is required in tropical arthropod surveys
to yield realistic estimates oI global species
richness (37). A basic hindrance to estimat-
ing global biodiversity lies in a lack oI empir-
ical data that establish local biodiversity, which
can be scaled up to achieve a global estimate.
Although many studies reported species richness
Ior selected groups oI well-studied insect taxa,
no satisIactory estimate oI total arthropod species
richness exists Ior a single tropical rainIorest lo-
cation to date.
The unstructured collection and small-scale
survey oI tropical arthropods cannot yield con-
vincing estimates oI total species richness at a
speciIic Iorest (79). Most studies either target
Iew arthropod orders or trophic guilds, or use a
limited array oI sampling methods, or ignore the
diverse upper canopy regions oI tropical Iorests
(1015). Moreover, sampling protocols have
rarely been structured in such a way that, with
increased sampling, incomplete data on local
diversity (7) can be extrapolated to estimate total
species richness across multiple spatial scales
(16). Where such structured estimates are made,
it is invariably Ior insect herbivores on their host
plants (5). However, species accumulation rates
may diIIer markedly Ior nonherbivore guilds,
which include more than halI oI all described
arthropod species (1, 17). As the degree oI host
speciIicity (eIIective specialization) oI other guilds
can be much lower than that oI insect herbivores,
or may be driven by diIIerent Iactors (18, 19),
global estimates based on herbivores alone are
questionable. Consequently, extensive cross-taxon
surveys with structured protocols at reIerence
sites may be the only eIIective approach toward
estimating total arthropod species richness in
tropical Iorests (3).
To provide a comprehensive estimate oI total
arthropod species richness in a tropical rainIorest,
we established a collaboration involving 102 re-
searchers with expertise encompassing the Iull
breadth oI phylogenies and Ieeding modes present
among arthropods (20). This consortiuminvested
a total oI 24,354 trap- (or person-) days sam-
pling the San Iorenzo Iorest (SIPA) in Panama
using structured protocols (Iig. S1). We identiIied
129,494 arthropods representing 6144 Iocal spe-
cies (Fig. 1 and table S1) Irom 0.48 ha oI inten-
sively sampled mature Iorest. This allowed us to
extrapolate Iocal arthropod species richness to
a larger Iorest area with unprecedented power,
through a series oI best-inIormed species richness
estimates derived Irom six competing models Ior
each oI 18 Iocal data sets. Using taxon ratios to
estimate the species richness oI nonIocal taxa
|see Extrapolating results to nonIocal taxa in
materials and methods (20)|, we then predicted
the total species richness oI the study area. We
also evaluated diIIerences in relative species ac-
cumulation rates among arthropod guilds, across
spatial scales.
Although individual estimators adjusting Ior
diIIerent aspects oI sampling design oIIered slight-
ly diIIerent estimates (Fig. 1B), the total spe-
cies richness Ior the entire San Iorenzo Iorest
(~6000 ha) was consistently quantiIied at be-
tween 18,000 and 44,000 species (including Iocal
and nonIocal species). In particular, the most like-
ly lower bound oI species richness was estimated
to be at least 21,833 species |95° conIidence
level (CI) ÷ 18,665, 29,420; model a1 in Fig.
1B|, and the biologically and statistically best-
supported estimate oI richness (criteria outlined
in table S2) was 25,246 species (95° CI ÷
19,721, 33,181, model B¹S in Fig. 1B). Accord-
ing to our estimates, a single hectare oI rain-
Iorest will be inhabited by an average oI 18,439
species (95° CI ÷ 17,234, 18,575; Fig. 2B).
A relatively large proportion oI the expected
species richness oI the Iorest was recovered Ior
most oI our Iocal taxonomic groups (Fig. 2). For
example, high proportions oI all ant species and
oI the parasitoid species targeted in our study
were collected Iromour 12 sites, whereas Iungal
Ieeders would require more intensive sampling
to achieve adequate coverage (Fig. 2). Beta di-
versity oI all arthropods (in the broad sense oI
species turnover among sites) increased roughly
linearly with cumulative area surveyed (F ÷
2422.5, P · 0.001). With increasing sampling
eIIort, sample coverage |an unbiased measure oI
sample completeness, see (20)| was high and ac-
cumulated at signiIicantly diIIerent rates across
diIIerent arthropod orders and guilds, and across
the various guilds comprised by beetles (Fig. 3).
However, despite the high sample coverage val-
ues, we cannot discount the possibility that there
were some vanishingly rare species that may not
have been discovered with the sampling proto-
cols used in this study.
Despite idiosyncrasies in the rate oI increase
in sample completeness across insects groups,
the high proportion oI overall species richness
detected at small spatial scales (Figs. 2 and 3)
has a remarkable consequence. Based on a gen-
eral relationship between species numbers and
area, we estimate that almost two-thirds (64°)
oI all species in SIPA occur in a single hectare
oI rainIorest (Fig. 2). Our plant models predicted
total arthropod richness in the San Iorenzo Iorest
to a precision oI 1° (correlation between rich-
Sulthsuulau !ruplcal 8esearch lustltute, lauaua Clt,, 8e-
publlc uí lauaua. uulverslt, uí Suuth ßuheula, 370 05
Cesle ßude|uvlce, Cìech 8epubllc. uulversldad de lauaua,
lauaua Clt,, 8epubllc uí lauaua. Cìech Acadeu, uí Scleuces,
370 05 Cesle ßude|uvlce, Cìech 8epubllc. \uséuu dhlstulre
uaturelle de la vllle de 6eueve, l208 6eueve, SWltìerlaud.
!he uulverslt, uí westeru Australla aud CSl80 Lcus,steu Scl-
euces, 6009 lerth, Australla. Catedra 8ul Nabelru, uulversldade
de Lvura, 7004-5l6 Lvura, lurtuqal. uulverslt, uí Yurl,Yurl
Y0l0 500, uK. NurWeqlau lustltute íur Nature 8esearch, 7485
!ruudhelu, NurWa,. uulverslt, uí helslull, 000l4 helslull,
llulaud. uulverslt, uí Lrlauqeu-Nureuberq, 9l058 Lrlauqeu,
6eruau,. Sauta ßarbara \useuu uí Natural hlstur,, Sauta
ßarbara, CA 93l05, uSA. uulverslt, uí vlcturla, vlcturla, ßC
v8w 2Y2, Cauada. Clrad, 34988 \uutíerrler-sur-Leì, lrauce.
uulverslt, uí ßrlstul, ßrlstul ßS8 lu0, uK. uulversldad
Nacluual Autuuuua de \étlcu, \étlcu 05l0 0l, \étlcu. uul-
verslté ßlalse lascal, 63000 Cleruuut-lerraud, lrauce.
\useu Clvlcu dl Sturla Naturale, l0022 Caruaquula, ltal,.
Ceutru de lesqulsas du Cacau, 45600-000, ltabuua, aud
uulversldade Lstadual de Sauta Cruì, 45662-900 llhéus-ßahla,
ßraìll. lustltut 8u,al des Scleuces Naturelles de ßelqlque, l000
ßrussels, ßelqluu. uulverslt, uí !uuluuse lll, 3l062 !uuluuse,
lrauce. uulversltat wurìburq, 97070 wurìburq, 6eruau,.
6rlíílth uulverslt,, Nathau 0L0 4lll, Australla. Natluual
\useuu uí Natural hlstur,, washluqtuu, 0C 20008, uSA.
Ceutru uulversltarlu uNA, 30350-540 ßelu hurlìuute-\6,
ßraìll. CN8S, 97379 Kuuruu, lrauce. 8esearch lustltute íur
Nature aud lurest, l070 ßrussels, ßelqluu. \uséuudhlstulre
uaturelle, 2000 Neuchatel, SWltìerlaud. uulversldade led-
eral de 0uru lretu, 35400-000 0uru lretu-\6, ßraìll aud
uulversldade dus Açures, 9700-85l !ercelra, lurtuqal. uul-
verslté Llbre de ßrutelles, l050 ßrussels, ßelqluu. Natural
hlstur, \useuu uí 0euuarl, 2l00 Cupeuhaqeu, 0euuarl.
¯!u Whuu currespuudeuce shuuld be addressed. L-uall:
basset,Qsl.edu
www.sciencemug.org 5CIENCE vÒL 338 14 DECEMBER 2012 1481

ness estimates provided by the plant model and
best estimates: r ÷ 0.992, P · 0.0001, N ÷ 18;
Fig. 1C). Notably, small discrepancies between
observed arthropod species richness and esti-
mates derived Irom Iloristic diversity appeared
not to be scale-dependent (Fig. 1D). Hence, even
Ior arthropod guilds other than herbivores, plant
diversity seems a powerIul predictor oI species
richness across areas varying in size (at least
within the limits oI our study design and given
the limited heterogeneity oI the study area com-
pared to larger geographical scales).
Because this study targeted the Iull spectrum
oI arthropods, it oIIers a comprehensive test oI
previous estimates oI species richness based only
on selected guilds or taxa. Reassuringly, our well-
resolved estimates oI tropical arthropod species
richness are oI the same order oI magnitude as
prior estimates (table S3), adding credence to
recent estimates oI tropical arthropod diversity
(5, 21). Although the scope Ior direct compar-
ison is limited because oI regional diIIerences in
sampling eIIort, lowland tropical Iorest in Panama
seems to support 2.1 to 8.4 times as many ar-
thropod species as observed in temperate Iorests
(table S3). While this supports the obvious truism
that tropical arthropods are indeed more diverse
than their temperate counterparts (22), the mag-
nitude oI that diIIerence is Iar lower than many
previous estimates would suggest (2).
llg. 1. Nunber of arthropod specìes estìnated at SLlA (ì. (Aì Nunber of
specìes (closed bars, log scaleì and ìndìvìduals (open bars, log scaleì collected
ìn 0.48 ha for each data set (three-letter guìld code as ìn table S1ì and nunber
of specìes estìnated ìn SLlA (best estìnate, shaded boxesì. Nunbers above
bars ìdentìfy the best nodel used for calculatìon (a1 to a6, fìg. S2ì and the
percentage of sìngletons. (8ì Nunber of arthropod specìes estìnated for SLlA
(dots: all focal taxa, shaded boxes: focal and nonfocal taxa, table S4ì, as
estìnated by dìfferent nethods: 8+S: best estìnates, ìncludìng both bìologìcal
and statìstìcal argunents (table S2ì, 8+Sloc: sane as 8+S but estìnates
calculated Wìth local ìnstead of global ratìos (fìg. S3ì, S: best estìnates, ìn-
cludìng only statìstìcal argunents, and nodels a1 to a5. !: optìnìtatìon al-
gorìthns dìd not converge to alloW calculatìons of CL. 0ur estìnates are robust
to even noderate to large shìfts ìn taxon ratìos (table S5ì. (Cì llot of the
nunber of specìes estìnated ìn SLlA Wìth the plant nodel agaìnst that es-
tìnated Wìth the best nodel, for each data set. Lìne denotes unìty. (bì llot of
the percentage error betWeen all arthropod specìes observed and estìnated by
the plant nodel agaìnst cunulatìve nunber of sìtes. Shaded boxes ìndìcate
neans and 95% CL.
14 DECEMBER 2012 vÒL 338 5CIENCE www.sciencemug.org 1482

The implications oI the results observed on
a local scale are clear. For every species in the
well-known vascular Ilora (1294 species), aviIauna
(306 species), and mammalian Iauna (81 species)
oI SIPA, we estimate that there will be a mini-
mum oI 17, 71, and 270 arthropod species, re-
spectively (based on lower bound oI species
richness) and most likely as high as 20, 83, and
312 arthropod species, respectively (table S3).
Based on the dominance oI arthropod species
in the tropical Iauna (table S3), we may then ar-
gue that conservation planning Ior biodiversity
should be largely determined by the spatial scaling
oI arthropod diversity. In this context, the asso-
ciation between the species richness oI plants and
arthropods detected across spatial scales suggests
that conservation eIIorts targeted at Iloristically
diverse sites may also serve to conserve arthro-
pod diversity across both taxonomic lineages
and trophic guilds. As arthropods are notoriously
labor-intensive to survey, such an umbrella ap-
proach may be an eIIicient way Iorward.
Nonetheless, our Iindings also suggest that
large-scale, region-wide understanding oI trop-
ical arthropod richness may actually be more
achievable than previously assumed. Our data
indicate that a thorough sampling oI 1 ha oI rain-
Iorest may reveal nearly two-thirds oI all ar-
thropod species present in a much larger area
(6000 ha in our case; Fig. 2B), consistent with
llg. 2. Accunulatìon of specìes rìchness Wìth area at SLlA (ì. lor all
groups, a hìgh proportìon of overall specìes rìchness Was detected at snall
spatìal scales. (Aì lartìtìonìng of specìes rìchness Wìthìn arthropod guìlds
at dìfferent spatìal scales (: sìngle sìte of 0.04 ha, 3: three sìtes spaced
apart totalìng 0.12 ha, 6: sìx sìtes totalìng 0.24 ha, 12: 12 sìtes totalìng
0.48 ha, ha: 1.0 ha, SLlA: 6000 ha, neans  SLV are shoWn for , 3,
and 6ì. (8ì Specìes-area nodels for the naìn arthropod groups and large
data sets (for all arthropods, ìncludìng nonfocal specìes, values are ìndì-
cated but the curve not plotted for the sake of clarìtyì. Lach curve ìs
characterìted by ìts functìon (Lx, Lxponentìal, Lo, Lonolìno, lo, loWer, we,
cunulatìve weìbullì, ìts value for 1 ha (ìntersectìon Wìth vertìcal lìne,
shaded boxes Wìth nean and 95% CLì, and the percentage of the nunber
of specìes present ìn 1 ha relatìve to the nunber of specìes estìnated to
occur ìn SLlA.
www.sciencemug.org 5CIENCE vÒL 338 14 DECEMBER 2012 1483

reports oI relatively low beta diversity oI insect
herbivores in tropical rainIorests (23). Hence,
to determine the species diversity oI a tropical
rainIorest, the total area sampled need not be
overly largeprovided that the sampling design
adequately covers both microhabitats and plant
species. However, this does not imply that most
arthropod species have selI-supporting popula-
tions in small Iorest areas or Iragments.
On a global scale, our results have implica-
tions Ior current estimates oI total species rich-
ness, which have been weakened by the lack oI
knowledge regarding the strength oI association
between vascular plant species and nonherbivore
guilds (5). Based on the close association ob-
served here between Iloristic diversity and both
herbivore and nonherbivore species richness, we
tentatively conclude that the most recent estimate
oI global tropical arthropod species |6.1 million
arthropod species (24)| does not require drastic
correction to account Ior diIIerential scaling rela-
tionships oI nonherbivore taxa. The robust es-
timates oI local arthropod diversity derived in our
study thus support previous estimates oI global
species richness. They also show how stratiIied
sampling designs and broad scientiIic cooper-
ation may be developed to Iormulate eIIicient
estimates oI tropical arthropod diversity. Similar
initiatives have recently been implemented in
other tropical locations around the world, using
the current template as a Ioundation (25).
keferences and Notes
l. Z.-0. Zhauq, ¿oo|a\a 3148, 99 (20llì.
2. !. L. LrWlu, co|eoµ|. 8a|| 36, 74 (l982ì.
3. 8. \. \a,, |n||os. |/ans. 8. 5oc. |ond. 8 8|o|. 5c|. 359,
733 (2004ì.
4. 8. \. \a,, 5c|ence 329, 4l (20l0ì.
5. A. ¦. haulltuu e| a|., 4m. ha|. 176, 90 (20l0ì.
6. C. \ura, 0. l. !ltteusur, S. Adl, A. 6. ß. Slupsuu,
ß. wuru, ||o5 8|o|. 9, el00ll27 (20llì.
7. ¦. A. Cuddluqtuu, l. Aquarssuu, ¦. A. \lller, \. Kuutuer,
6. hurulqa, j. 4n|m. /co|. 78, 573 (2009ì.
8. Y. ßasset, ||an| /co|. 153, 87 (200lì.
9. !. \. LeWlusuhu, !. 8usllu, /co|. |e||. 11, 398 (2008ì.
l0. v. Nuvutu, e| a|., conse/t. 8|o|. 18, 227 (2004ì.
ll. v. Nuvutu, e| a|., j. 4n|m. /co|. 79, ll93 (20l0ì.
l2. l. 0deqaard, 8|od|te/s. conse/t. 15, 83 (2006ì.
l3. ¦. !. Luuqluu, ¦. Cuddluqtuu, 8. K. CulWell, /co|o¡, 83,
689 (2002ì.
l4. Y. ßasset, v. Nuvutu,, S. L. \lller, 8. L. Kltchluq, Lds.,
4/|n/oµods oj |/oµ|ca| /o/es|s. 5µa||o·|emµo/a| u,nam|cs
and 8esoa/ce Jse |n |ne canoµ, (Caubrldqe uulv. lress,
Caubrldqe, 2003ì.
l5. ¦. !. Luuqluu, 8. K. CulWell, /co|. 4µµ|. 7, l263 (l997ì.
l6. 8. K. CulWell, ¦. A. Cuddluqtuu, |n||os. |/ans. 8. 5oc.
|ond. 8 8|o|. 5c|. 345, l0l (l994ì.
l7. 0. 8. Struuq, ¦. h. LaWtuu, !. 8. L. SuuthWuud, |nsec|s on
||an|s. comman||, |a||e/ns and Vecnan|sms (ßlaclWell,
0tíurd, ed. l, l984ì.
l8. L. Sleuauu, 0. !lluau, ¦. haarstad, \. 8ltchle, 4m. ha|.
152, 738 (l998ì.
l9. v. Nuvutu,, Y. ßasset, |/oc. 8. 5oc. |ond. 272, l083
(2005ì.
20. \aterlals aud uethuds are avallable as suppleueutar,
uaterlals uu 5c|ence 0ullue.
2l. v. Nuvutu, e| a|., ha|a/e 416, 84l (2002ì.
22. !. C. ßuuebrale, L. C. luulslu, C. L. ßuqqs, l. 8. Lhrllch,
8|o|. conse/t. 143, l83l (20l0ì.
23. v. Nuvutu, e| a|., ha|a/e 448, 692 (2007ì.
24. A. ¦. haulltuu e| a|., 4m. ha|. 177, 544 (20llì.
25. 8. L. Kltchluq e| a|., Vem. 0aeens|. Vas. ha|. 55, 235 (20llì.
AcknoWledgments: lßlSCA-lauaua ls au lultlatlve uí
lru-Natura luteruatluual, 0céau vert, the uulversltles ßlalse
lascal aud uí lauaua, aud the Sulthsuulau !ruplcal 8esearch
lustltute (S!8lì, Wlth cure íuudluq íruu Sulvlu-Sulva, SA, S!8l,
the uulted Natluus Luvlruuueut lruqrauue, the Sulthsuulau
lustltutluu (walcutt luudì, the Lurupeau Scleuce luuudatluu,
aud the 6lubal Cauup, lruqrauue. ¦. herrera, L. Audrade,
\. Sauaulequ, S. ¦. wrlqht, N. ßalbeu, S. ßechet, ¦. ßellequlc,
!. Aubert, K. ¦urdau, 6. Lbersult, 0. Cle,et-\arrel, L. l,ut,
0. lascal, l. ßasset, aud L. ßauhaus helped Wlth luqlstlcs
lu the íleld. A. ßarba, 8. Cabrera, A. Curue|u, l. 0íaì,
A. l. 8. du Caruu, l. C. du Nasclueutu, L. A. dus Sautus,
\. 6uuìaleì, A. heruaudeì, \. \auuubur, \. \uqla,
S. lluìuu, ß. léreì, L. S. 8auus-Lacau, aud 0. valdeì helped
Wlth lultlal surtluq uí the arthrupud aud plaut uaterlal. 0ata
(as uí l0 \a, 20l2ì have beeu depuslted lu the 0r,ad
repusltur,: http://dt.dul.urq/l0.506l/dr,ad.í3p75.
Supplementary Haterials
WWW.scleuceuaq.urq/cql/cuuteut/íull/338/6ll3/l48l/0Cl
\aterlals aud \ethuds
Suppleueutar, !ett
llqs. Sl tu S3
!ables Sl tu S5
8eíereuces (26151ì
29 ¦uue 20l2, accepted l Nuveuber 20l2
l0.ll26/scleuce.l226727
llg. 3. Average sanple coverage |SLV, error bars, see nethods (ì| plotted agaìnst the cunulatìve nunber of sìtes surveyed, for the naìn (Aì arthropod
guìlds and orders and (8ì beetle guìlds. lor the sake of clarìty, SLVs are onìtted ìn (Aì.
14 DECEMBER 2012 vÒL 338 5CIENCE www.sciencemug.org 1484
