ARTICLE
Impac s o de i i o e di e si y loss on ins eam
decomposi ion a e g ea es in he opics
Luz Boye o 1,2✉, Naia a López-Rojo 1, Alan M. Tonin 3, Ja ie Pé ez 1, F ancisco Co ea-A aneda 4,
Richa d G. Pea son 5,6, Jaime Bosch 7,8, Rica do J. Alba iño 9, Sanka appan Anbalagan10,
Leon A. Ba mu a11, Ana Basagu en 1, F ancis J. Bu don 12, Ad iano Caliman13, Ma cos Callis o 14,
Adol o R. Calo 15, Ian C. Campbell16, B adley J. Ca dinale17, J. Jesús Casas 18, Ana M. Cha á-Se na 19,20,
E ic Chau e 21, Szymon Ciapała22, Checo Colón-Gaud 23, Aydeé Co nejo 24, Aa on M. Da is5,
Monika Degeb od 25, Eme son S. Dias26, Ma ía E. Díaz27,28, Michael M. Douglas 29, And ea C. Encalada30,39,
Rica do Figue oa 28, Alexande S. Flecke 31, Tadeusz Flei uch 32, E ica A. Ga cía 33, Gab iela Ga cía 34,
Pa el E. Ga cía35,36, Ma k O. Gessne 25,37, Jesús E. Gómez38, Se gio Gómez31, Jose F. Gonçal es J 3,
Manuel A. S. G aça 39, Daniel C. Gwinn 40, Robe O. Hall J 41, Neusa Hamada42, Cang Hui 43,44,
Daichi Imazawa45, Tomoya Iwa a 46, Samuel K. Ka iuki47, And ea Landei a-Daba ca 30,40, Kelsey Laymon23,
Ma ía Leal48, Richa d Ma chan 49, Rena o T. Ma ins 42, F ank O. Masese 50, Megan Maul 51,
B endan G. McKie12, Ad iana O. Medei os 15, Cha les M. M’E imba47, Jen A. Middle on 29, Sil ia Mon oy1,
Timo Muo ka52, Junji o N. Negishi53, Alonso Ramí ez 54, John S. Richa dson 55, José Rincón48,
Juan Rubio-Ríos 18, Gisele M. dos San os 14,56, Romain Sa emejane 52, F an Sheldon57, Augus ine Si a i50,
Na halie S. D. Tenkiano58, Sco D. Tiegs 51, Janine R. Tolod 53, Michael Vena sky57, Anne Wa son11 &
Ca he ine M. Yule 59
The ela ionship be ween de i i o e di e si y and decomposi ion can p o ide in o ma ion on
how biogeochemical cycles a e a ec ed by ongoing a es o ex inc ion, bu such e idence has
come mos ly om local s udies and mic ocosm expe imen s. We conduc ed a globally dis-
ibu ed expe imen (38 s eams ac oss 23 coun ies in 6 con inen s) using s anda dised
me hods o es he hypo hesis ha de i i o e di e si y enhances li e decomposi ion in
s eams, o es ablish he ole o o he cha ac e is ics o de i i o e assemblages (abundance,
biomass and body size), and o de e mine how pa e ns a y ac oss ealms, biomes and
clima es. We obse ed a posi i e ela ionship be ween di e si y and decomposi ion, s on-
ges in opical a eas, and a key ole o abundance and biomass a highe la i udes. Ou
esul s sugges ha li e decomposi ion migh be al e ed by de i i o e ex inc ions, pa i-
cula ly in opical a eas, whe e de i i o e di e si y is al eady ela i ely low and some
en i onmen al s esso s pa icula ly p e alen .
h ps://doi.o g/10.1038/s41467-021-23930-2 OPEN
A ull lis o au ho a filia ions appea s a he end o he pape .
NATURE COMMUNICATIONS | (2021) 12:3700 | h ps://doi.o g/10.1038/s41467-021-23930-2 | www.na u e.com/na u ecommunica ions 1
1234567890():,;
Akey ques ion in con empo a y ecology is whe he changes in
biodi e si y lead o al e a ions in he unc ioning o ecosys-
ems and associa ed biogeochemical cycles1,2. In e es in his
opic eme ged in he 1990s, mo i a ed in pa by he ema kable
inc ease in global biodi e si y loss3, and led o hund eds o expe i-
men s ha manipula ed biodi e si y a di e en le els (species, genes
o unc ional ai s) in di e en g oups o e es ial and aqua ic
o ganisms, o examine possible e ec s on ecosys em p ocesses4,5.
While his la ge body o p ima y esea ch and subsequen syn heses
ha e demons a ed a s ong, posi i e ole o di e si y o p ima y
p oduce s on biomass p oduc ion6–8, he pa e ns o decomposi ion
ha e p o en o be weake and less consis en 6,9. This con as may
occu because decomposi ion can be simul aneously a ec ed by he
di e si ies o plan li e , mic obial decompose s and animal con-
sume s, wi h consequen ly mo e complex ela ionships10.
Plan li e decomposi ion is a key p ocess in he biosphe e, as
90% o he annual plan p oduc ion escapes he bi o y11 and
e en ually becomes li e , which is ul ima ely decomposed o
seques e ed in e es ial o aqua ic ecosys ems10. S eams play a
pa icula ly impo an ole in ecei ing and p ocessing li e om
hei ca chmen s12, con ibu ing significan ly o global ca bon
and nu ien fluxes13–15. Li e en e s s eams mainly in he o m
o lea es, and i is decomposed by mic oo ganisms (mos ly
aqua ic hyphomyce es) and specialised in e eb a es (li e -con-
suming de i i o es) ha can ob ain ca bon and nu ien s om
he li e and associa ed ungi16,17.
Mul iple s udies ha e manipula ed de i i o e di e si y and
assessed i s e ec on decomposi ion locally in s eams o in labo a-
o y mic ocosms, wi h inconsis en esul s10. These inconsis encies
ha e been a ibu ed o he exis ence o di e en species in e ac ions
d i ing ei he posi i e18,19 o nega i e e ec s20,21,whichcancom-
pensa e o each o he and some imes esul in o e all neu al
e ec s22. Howe e , he e has been no global assessmen o he ela-
ionship be ween de i i o e di e si y and decomposi ion in s eams,
which would help accoun o local and egional en i onmen al
con ingencies in he di e si y–decomposi ion ela ionship23.Ame a-
analysis o e es ial and aqua ic s udies e ealed s ong e ec s o
de i i o e di e si y on decomposi ion, bu he e was no sepa a e
assessmen o ins eam decomposi ion9. Se e al s eam s udies ha e
sugges ed a di ec link be ween as e decomposi ion24 and g ea e
de i i o e di e si y25,26 in empe a e s eams, bu did no explo e he
ela ionship explici ly. A la ge-scale s udy demons a ed ha
decomposi ion in s eams was enhanced when de i i o e assem-
blages we e mo e complex (la ge- and medium-sized o ganisms as
opposed o medium-sized only), al hough i did no examine de i-
i o e di e si y27.
He e, we desc ibe esul s om a global-scale decomposi ion
expe imen conduc ed by pa ne s o he GLoBE collabo a i e
esea ch ne wo k (www.globene wo k.es) in 38 s eams dis ibu ed
ac oss 23 coun ies in all inhabi ed con inen s. We use a s anda -
dised design and me hodology o examine global-scale ecological
ques ions, which educes he numbe o con ounding ac o s ha
need o be s a is ically con olled o in a me a-analysis28,29.Ou
main wo king hypo hesis is ha de i i o e di e si y has a majo
posi i e e ec on decomposi ion9, al hough we also expec an
influence o o he de i i o e assemblage cha ac e is ics such as
abundance, biomass, and body size18,22,27. Mo eo e , we p edic
ha bio ic d i e s o decomposi ion a y ac oss si es a di e en
la i udes, possibly because o he a ying in e play be ween posi i e
and nega i e species in e ac ions22. We also explo e de i i o e
a ia ion ac oss la i udes, biogeog aphic ealms, biomes and cli-
ma es, o u he explain hei global dis ibu ion and he po en ial
consequences o educed di e si y o decomposi ion in di e en
a eas o he wo ld. Unlike p e ious la ge-scale decomposi ion s u-
dies using 1 o 2 li e ypes24,30, we use se e al mix u es ep e-
sen ing a a ie y o li e ai s o maximise he gene ali y o ou
esul s. Ou global expe imen suppo s he expec ed posi i e ela-
ionship be ween de i i o e di e si y and decomposi ion, and
e eals ha de i i o e species loss may ha e i s g ea es con-
sequences on s eam ecosys em unc ioning in he opics.
Resul s
The model ha bes explained global a ia ion in o al decom-
posi ion explained 73% o he a ia ion and e ealed a significan
influence o de i i o e di e si y, abundance, biomass, la i ude,
and in e ac ions be ween di e si y and la i ude, abundance and
la i ude, and biomass and la i ude (Table 1and Supplemen a y
Table 1). The model ha bes explained global a ia ion in
de i i o e-media ed decomposi ion explained 82% o a ia ion in
he da a, and showed ha he in e ac ions be ween di e si y and
la i ude, abundance and la i ude, and biomass and la i ude we e
significan (Table 1and Supplemen a y Table 1). As hese esul s
indica ed ha he h ee de i i o e a iables we e impo an
p edic o s o decomposi ion, bu hei influence a ied wi h
la i ude, we explo ed he in e ac ions wi h a second ype o model
whe e la i ude was a ca ego ical a iable (Supplemen a y Table 2).
These models e ealed ha he ela ionship be ween de i i o e
di e si y and decomposi ion was s onge in opical a eas han in
empe a e a eas and absen in bo eal a eas; and ha abundance
and biomass we e impo an in empe a e and bo eal a eas, bu
no in opical a eas (Fig. 1and Supplemen a y Table 2).
All de i i o e a iables a ied significan ly among ealms, biomes
and clima es, and so did assemblage composi ion (Figs. 2–4,Table2
and Supplemen a y Table 3). Di e si y and abundance we e highes
in he Palea c ic ealm, und a and empe a e b oadlea and con-
i e ous o es s, and wa m empe a e and snow clima es; and lowes in
Neo opical, A o opical and Indomalayan ealms, opical we
o es s and sa annas and xe ic sh ublands, and equa o ial clima es.
Biomass and mean body size we e highes in Palea c ic and Nea c ic
ealms, empe a e b oadlea and coni e ous o es s, and again wa m
empe a e and snow clima es, wi h he lowes alues in he Indo-
malayan ealm, opical sa annas and xe ic sh ublands, and
Table 1 Resul s o he bes addi i e models explaining
a ia ion in o al and de i i o e-media ed li e
decomposi ion based on de i i o e di e si y, abundance,
biomass, mean body size, la i ude, and in e ac ions be ween
de i i o e a iables and la i ude.
E ec ed Fp
To al decomposi ion
Di e si y 4.00 6.94 <0.001
Abundance 3.14 6.34 <0.001
Biomass 1.00 2.00 0.159
Mean body size 1.86 2.10 0.102
La i ude 1.00 3.01 0.085
Di e si y × la i ude 14.56 6.17 <0.001
Abundance × la i ude 1.00 8.67 0.004
Biomass × la i ude 7.91 4.20 <0.001
De i i o e-media ed decomposi ion
Di e si y 4.00 0.53 0.716
Abundance 1.05 0.01 0.912
Biomass 1.00 0.04 0.843
Mean body size 1.08 1.00 0.843
La i ude 1.71 0.27 0.763
Di e si y × la i ude 14.14 4.74 <0.001
Abundance × la i ude 8.76 3.30 <0.001
Biomass × la i ude 7.99 4.36 <0.001
All p edic o s we e fi ed as enso p oduc in e ac ion smoo hs. We show e ec i e deg ees o
eedom (ed ) and alues o Fand p o each ac o . Models explained 69% and 78% o a ia ion
in he da a, espec i ely.
ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-021-23930-2
2NATURE COMMUNICATIONS | (2021) 12:3700 | h ps://doi.o g/10.1038/s41467-021-23930-2 | www.na u e.com/na u ecommunica ions
To al decomposi ion De i i o e-media ed decomposi ion
Tem pe a e
0123456
0.000
0.002
0.004
0.006
Di e si y (no. amilies)
Bo eal
0123456
T opical
0123456
Tempe a e
0123456
0.000
0.002
0.004
0.006
Bo eal
0123456
T opical
0123456
p = <0.001 p = 0.001 p = 0.765 p = 0.008 p = 0.054 p = 0.444
Tempe a e
012345
0.000
0.002
0.004
0.006
Bo eal
012345
T opical
012345
Tem pe a e
012345
0.000
0.002
0.004
0.006
Bo eal
012345
T opical
012345
p = 0.553 p = <0.001 p = 0.146 p = 0.247 p = <0.001 p = 0.038
Abundance [log (no. indi iduals)+1]
Biomass [log (mg+1)]
Tem pe a e
0246
0.000
0.001
0.002
0.003
Bo eal
0246
T opical
0246
Tempe a e
0246
0.000
0.002
0.004
0.006
Bo eal
0246
T opical
0246
p = 0.482 p = <0.001 p = 0.026 p = 0.515 p = 0.002 p = <0.001
Decomposi ion (dd-1)Decomposi ion (dd-1)Decomposi ion (dd-1)
a
b
c
Fig. 1 Gene alised addi i e models explo ing he influence o de i i o e di e si y, abundance and biomass on decomposi ion in di e en la i udinal
zones ( opical: ≤23°; empe a e: 24–60°; and bo eal: >60°). Va ia ion in o al and de i i o e-media ed decomposi ion (measu ed as he p opo ion o
li e mass loss pe deg ee day, dd; mean ± SE) wi h ade i i o e di e si y (numbe o amilies pe li e bag), blog- ans o med abundance (numbe o
indi iduals pe li e bag) and clog- ans o med biomass (mg pe li e bag), in di e en la i udinal zones. Lines ep esen he smoo he s and shading he
95% confidence in e als om gene alised addi i e models o significan ela ionships (p- alue < 0.05); whole-model esul s a e gi en in Supplemen a y
Table 3.
NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-021-23930-2 ARTICLE
NATURE COMMUNICATIONS | (2021) 12:3700 | h ps://doi.o g/10.1038/s41467-021-23930-2 |www.na u e.com/na u ecommunica ions 3
-0.25
0.00
0.25
-0.25 0.00 0.25 0.50
MDS1
MDS2
Pa Na Au N A Im
0
2
4
6
8
10
De i i o e di e si y (no. amilies)
0
50
100
150
200
250
De i i o e abundance (no. indi iduals)
Pa Na Au N A Im
0
500
1000
1500
2000
2500
De i i o e biomass (mg)
Pa Na Au N A Im
0
10
20
30
40
50
60
70
De i i o e mean body size (mm)
Pa Na Au N A Im
abcddd ab bccc
abbccabc bbabaabc
Fig. 2 Global dis ibu ion o s udy si es in di e en biogeog aphic ealms (Pa, Palea c ic; Na, Nea c ic; Au, Aus alasian; N , Neo opical; A ,
A o opical; Im, Indomalayan); n=38. Box plo s show he median, in e qua ile ange and minimum-maximum ange o li e -consuming de i i o e
di e si y (numbe o amilies pe li e bag), abundance (numbe o indi iduals pe li e bag), biomass (mg pe li e bag) and mean body size (mm) in each
ealm (o de ed om highes o lowes di e si y); di e en le e s indica e significan di e ences. The NMDS o dina ion o li e -consuming de i i o es
wi h ealms is ep esen ed by polygons o di e en colou s as in maps and box plo s. Significan di e ences in assemblage s uc u e we e: Pa s. Na, A ,
Au, Im; Na s. N , Au; N s. Au.
ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-021-23930-2
4NATURE COMMUNICATIONS | (2021) 12:3700 | h ps://doi.o g/10.1038/s41467-021-23930-2 | www.na u e.com/na u ecommunica ions
Tu TeBF TeCF MeF XeS T WF T S
0
2
4
6
8
10
De i i o e di e si y (no. amilies)
0
50
100
150
200
250
Tu TeBF TeCF MeF XeS T WF T S
De i i o e abundance (no. indi iduals)
0
500
1000
1500
2000
2500
Tu TeBF TeCF MeF XeS T WF T S
De i i o e biomass (mg)
0
10
20
30
40
50
60
70
Tu TeBF TeCF MeF XeS T WF T S
De i i o e mean body size (mm)
ab ab a c c cbbdbacddbc
bc a ab c bc cbc bd a b bd bc cdd
-0.25
0.00
0.25
-0.25 0.00 0.25 0.50
MDS1
MDS2
Fig. 3 Global dis ibu ion o s udy si es in di e en biomes (Tu, und a; TeBF, empe a e b oadlea o es ; TeCF, empe a e coni e ous o es ; MeF,
Medi e anean o es ; XeS, xe ic sh ubland; T WF, opical we o es ; T S, opical sa anna); n=38. Box plo s show he median, in e qua ile ange
and minimum-maximum ange o li e -consuming de i i o e di e si y (numbe o amilies pe li e bag), abundance (numbe o indi iduals pe li e bag),
biomass (mg pe li e bag) and mean body size (mm) in each biome (o de ed om highes o lowes di e si y); di e en le e s indica e significan
di e ences. The NMDS o dina ion o li e -consuming de i i o es wi h biomes is ep esen ed by polygons o di e en colou s as in maps and box plo s.
Significan di e ences in assemblage s uc u e we e: T WF s. TeBF, TeCF, MeF.
NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-021-23930-2 ARTICLE
NATURE COMMUNICATIONS | (2021) 12:3700 | h ps://doi.o g/10.1038/s41467-021-23930-2 |www.na u e.com/na u ecommunica ions 5
equa o ial clima es. Assemblage composi ion mos ly di e ed be ween
he Palea c ic/Nea c ic (wi h many amilies o Lau asian o igin) and
o he ealms ( amilies o Gondwanan dis ibu ion); be ween opical
we o es s and se e al o he biomes; and be ween equa o ial and
o he clima es.
Discussion
Ou s udy demons a es a posi i e influence o de i i o e
di e si y on decomposi ion, suppo ing p e ious sugges ions ha
la i udinal g adien s in de i i o e di e si y and ins eam
decomposi ion a e linked24,25 and ag eeing wi h esul s o a
De i i o e di e si y (no. amilies)
De i i o e abundance (no. indi iduals)
0
10
20
30
40
50
60
70
Csa D b D c C a C b Csb Am A As Aw
De i i o e biomass (mg)
De i i o e mean body size (mm)
b ab bc ac de cd e e e a a ac ab c cacbcc
ad ab c a bd dabc d d cd bd a cd ab a abcabc d ad d
0
2
4
6
8
10
Csa D b D c C a C b Csb Am A As Aw Csa D b D c C a C b Csb Am A As Aw
0
50
100
150
200
250
0
500
1000
1500
2000
2500
Csa D b D c C a C b Csb Am A As Aw
-0.25
0.00
0.25
-0.25 0.00 0.25 0.50
MDS1
MDS2
ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-021-23930-2
6NATURE COMMUNICATIONS | (2021) 12:3700 | h ps://doi.o g/10.1038/s41467-021-23930-2 | www.na u e.com/na u ecommunica ions
me a-analysis o con olled expe imen s pe o med in e es ial
and aqua ic ecosys ems9. Ou esul also ag ees wi h esul s o
con olled expe imen s ha ound a e age inc eases in decom-
posi ion o 12–30% pe de i i o e species added18,19,31, sug-
ges ing ha posi i e in e ac ions (i.e. esou ce pa i ioning and
acili a ion) a e p e alen in de i i o e assemblages. Clea ly, ou
field s udy does no demons a e causali y among hese a iables
o he sugges ed mechanisms, bu he finding o a consis en
ela ionship ac oss 113° o la i ude indica es ha de i i o e
di e si y, a leas a he amily le el, is indeed a d i e o
decomposi ion. Whe he his ela ionship would change by
conside ing species di e si y canno be cu en ly asce ained due
o limi ed axonomic knowledge in many egions32.
The ela ionship be ween de i i o e di e si y and decom-
posi ion, when da a we e g ouped acco ding o la i udinal zone,
was mos e iden in opical a eas, less impo an in empe a e
a eas and unimpo an in bo eal a eas (al hough he la e
we e unde ep esen ed in ou da ase ). O he s ha e demons a ed
a posi i e ela ionship be ween de i i o e di e si y and decom-
posi ion in some s eams o bo eal a eas33, bu ou global da ase
indica es a ela i ely weak ela ionship when compa ed o o he
la i udinal zones. Impo an ly, he s onge ela ionship be ween
de i i o e di e si y and decomposi ion in he opics sugges s
ha species losses in hese a eas, whe e de i i o e di e si y is
al eady lowe han a highe la i udes as shown he e and
elsewhe e25,26, may cause he g ea es impac on decomposi ion.
De i i o es in opical a eas a e pa icula ly ulne able, because
o he p e alence o mul iple en i onmen al s esso s. Fo
example, concen a ions o ag icul u al pes icides ha e limi ed
egula ion in many opical coun ies34 and a e known o cause
mo ali y in many de i i o es35–37. Clima e wa ming is also
likely o cause mo e ex inc ions in he opics because mo e
de i i o e species a e close o hei he mal maxima han
elsewhe e25 and a e likely o su e g ea e physiological changes,
despi e he smalle changes in empe a u e occu ing in his
la i udinal zone38. Ne e heless, o he clima ic changes such as
inc eased d ough s can be mo e impo an a highe la i udes39.
We ound ha he influence o de i i o e abundance and
biomass on decomposi ion also a ied wi h la i ude, bu wi h
negligible e ec in he opics and mo e impo an a highe
la i udes. These a iables ha e p e iously been ound o be
impo an p edic o s o decomposi ion in some opical
s eams40, bu he e hei impo ance was lowe in he opics han
elsewhe e. In empe a e a eas, bo h ela ionships we e non-linea
and complex (wi h decomposi ion fi s dec easing and hen
inc easing wi h highe abundance o biomass), which impedes
p edic ions abou how decomposi ion migh be al e ed by
changes in hese a iables. Mo eo e , esponses o abundance and
biomass o en i onmen al s esso s a e no as s aigh o wa d as
di e si y loss, because los species can be eplaced by mo e ol-
e an ones ha h i e unde s ess ul condi ions and can cause an
o e all inc ease in numbe s41,42. Smalle de i i o es a e o en
mo e sensi i e o s esso s han la ge ones42, al hough his a -
ia ion could be due o axonomic di e ences a he han o size.
Ou esul s sugges ha species eplacemen s unde en i on-
men al s ess could esul in an o e all inc ease in biomass, bu
his possibili y needs confi ma ion.
The dis ibu ion o mos de i i o e amilies co esponded o
b oad ealms (Fig. 5), wi h 26 amilies showing a Lau asian dis-
ibu ion (i.e. being p esen in he Palea c ic and/o Nea c ic
ealms) and 14 amilies a Gondwanan dis ibu ion (Neo opical,
A o opical, Aus alasian, and/o Indomalayan ealms).
Al hough we did no pe o m phylogene ic analyses, his
dicho omy, oge he wi h he obse a ion ha di e si y and
abundance o de i i o es we e highe in he Palea c ic and
Nea c ic (and hei p edominan biomes and clima es), sugges s
ha pa e ns o a ia ion in di e si y and abundance we e a leas
pa ly de e mined by biogeog aphy. Ou findings con as wi h
hose o angiospe ms, cu en dis ibu ions o which do no
co espond o ec onic his o y, possibly because o he exis ence
o high ansoceanic dispe sal43; howe e , hey suppo
pa e ns o o ganisms wi h lowe dispe sal, such as li e wo s
and coni e s44, which show clea Lau asian–Gondwanan
disjunc ions45.
The s ong influence o biogeog aphy on de i i o e di e si y
and abundance, and he ac ha hese wo a iables a e key
Table 2 Resul s o linea mixed e ec s models explo ing
a ia ion in de i i o e and o al in e eb a e di e si y,
abundance, biomass and mean body size, and
PERMANOVAs explo ing a ia ion in assemblage
composi ion, among ealms, biomes and clima es.
E ec d Fp
Di e si y
Realms 6, 1090 387.33 <0.001
Biomes 7, 1089 251.67 <0.001
Clima es 10, 1086 196.78 <0.001
Abundance
Realms 6, 1090 109.38 <0.001
Biomes 7, 1089 64.70 <0.001
Clima es 10, 1086 58.46 <0.001
Biomass
Realms 6, 1090 44.16 <0.001
Biomes 7, 1089 60.57 <0.001
Clima es 10, 1086 31.64 <0.001
Mean body size
Realms 6, 1090 472.25 <0.001
Biomes 7, 1089 472.33 <0.001
Clima es 10, 1086 363.65 <0.001
Composi ion
Realms 5, 37 2.30 0.002
Biomes 6, 37 1.54 0.015
Clima es 9, 37 1.32 0.029
We show deg ees o eedom (d ) o nume a o and denomina o , and alues o Fand p o
each ac o . Realms: Pa, Palea c ic; Ne, Nea c ic; Au, Aus alasian; N , Neo opical; A ,
A o opical; and In, Indomalayan. Biomes: Tu, und a; TeBF, empe a e b oadlea o es ; TeCF,
empe a e coni e ous o es ; MeF, Medi e anean o es ; XeS, xe ic sh ubland; T WF, opical
we o es ; and T S, opical sa anna. Clima es: A, equa o ial (A , ully humid; Am, monsoon; As,
wi h d y summe ; Aw, wi h d y win e ); C, wa m empe a e (C a, ully humid wi h ho summe ;
C b, ully humid wi h wa m summe ; Csa, wi h d y and ho summe ; Csb, wi h d y and wa m
summe ); D, snow (D b, ully humid wi h wa m summe ; D c, ully humid wi h cold summe ).
Fig. 4 Global dis ibu ion o s udy si es in di e en clima es [A, equa o ial (A , ully humid; Am, monsoon; As, wi h d y summe ; Aw, wi h d y win e );
C, wa m empe a e (C a, ully humid wi h ho summe ; C b, ully humid wi h wa m summe ; Csa, wi h d y and ho summe ; Csb, wi h d y and wa m
summe ); D, snow (D b, ully humid wi h wa m summe ; D c, ully humid wi h cold summe )]; n=38. Box plo s show he median, in e qua ile ange
and minimum-maximum ange o li e -consuming de i i o e di e si y (numbe o amilies pe li e bag), abundance (numbe o indi iduals pe li e bag),
biomass (mg pe li e bag) and mean body size (mm) in each clima e (o de ed om highes o lowes di e si y); di e en le e s indica e significan
di e ences. The NMDS o dina ion o li e -consuming de i i o es wi h biomes is ep esen ed by polygons o di e en colou s as in maps and box plo s.
Significan di e ences in assemblage s uc u e we e: Aw s. C b, C a, D b; A s. C a, C b, D b.
NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-021-23930-2 ARTICLE
NATURE COMMUNICATIONS | (2021) 12:3700 | h ps://doi.o g/10.1038/s41467-021-23930-2 |www.na u e.com/na u ecommunica ions 7
d i e s o decomposi ion, sugges ha he spli o Pangea in he
La e Ju assic (≈200 Ma ago) had a c ucial legacy e ec on he
cu en unc ioning o s eam ecosys ems and he influence o
ongoing en i onmen al change. The lowe de i i o e di e si y o
opical s eams25 and he highe suscep ibili y o hei auna o
ex inc ion38 make hese s eams mo e ulne able o educ ions in
decomposi ion a es ha a e associa ed wi h impai ed ecosys em
unc ioning46,47. This obse a ion, oge he wi h he o e -
exploi a ion o na u al esou ces ha se e ely a ec s opical
s eam ecosys ems48, indica es ha opical de i i o e species
should be o high conse a ion conce n globally.
Me hods
S udy si es. We conduc ed ou s udy in 38 headwa e s eams loca ed in di e en
egions in 23 coun ies (Figs. 2–4). A andom dis ibu ion o si es was un easible,
so some egions we e unde ep esen ed (mos ly A ica and no he n Asia), which
is usually he case o globally dis ibu ed expe imen s28,49,50. S eams we e simila
in size (mean ± SE: we ed channel wid h, 3.9 ± 0.1 m; wa e dep h, 28.7 ± 0.4 cm;
1s –3 d o de ) and physical habi a (al e na ing i fles and pools). Mos had ocky
subs a e and we e shaded by a dense ipa ian ege a ion (64 ± 1%) ep esen a i e
o he egion. They we e loca ed in 6 ealms, 7 biomes, and 10 Köppen clima e
classes51. In each s eam we selec ed a ca. 100-m long each wi h 5 consecu i e pool
habi a s in which o conduc he expe imen . Fu he in o ma ion on si e physi-
cochemical cha ac e is ics is gi en in Supplemen a y Table 4.
Field and labo a o y wo k. A each si e, we incuba ed 6 di e en 3-species li e
mix u es, which included 9 species in o al (Supplemen a y Table 5). The species
and mix u es we e chosen o ep esen di e en le els o unc ional di e si y o a
companion s udy52, bu he e ou in e es was o use a a ie y o mix u es and hus
inc ease he gene ali y o ou esul s (as opposed o wo king wi h a single o a ew
species). The 9 species we e collec ed a di e en loca ions a ound he wo ld and
dis ibu ed among pa ne s52; we conside ed he possible home-field-ad an age
e ec o using li e om di e en o igins negligible based on a ailable
li e a u e53,54.
Li e mix u es we e enclosed wi hin pai ed coa se-mesh (5 mm) and fine-mesh
(0.4 mm) li e bags con aining he same amoun and ype o li e . The wo ypes o
li e bag espec i ely quan ified o al and mic obial decomposi ion, and allowed he
calcula ion o de i i o e-media ed decomposi ion (see below). The e we e 60
li e bags pe s eam (n=5 pe li e mix u e and mesh size), each con aining 3 g
o senescen li e (1 g pe species), which had been collec ed eshly allen om he
o es floo , ai -d ied and dis ibu ed among esea ch pa ne s52. Li e bags we e
deployed in each s eam (one li e bag pe li e mix u e ype and mesh size in a
di e en s eam pool, wi h all 5 pools consecu i e) in 2017–2019 a he local ime
o he yea wi h he g ea es li e inpu and we e e ie ed a e 23–46 d,
depending on wa e empe a u e in each s eam, he eby hal ing he decomposi ion
p ocess a a compa able s age (mean ± SD: 32 ± 17% li e mass loss on a e age o
all he li e mix u es, 41 ± 18% o he as es decomposing mix u e52; mean alues
o each biome a e gi en in Supplemen a y Fig. 1). Li e bags we e anspo ed o
he labo a o y on ice enclosed indi idually in zip-lock bags and insed wi h fil e ed
s eam wa e o emo e a ached sedimen and in e eb a es. Li e was o en-d ied
(70 °C, 72 h) and a subsample weighed, incine a ed (500 °C, 4 h) and e-weighed o
calcula e he final ash- ee d y mass (AFDM). In e eb a es we e so ed, and li e -
consuming de i i o es we e coun ed and iden ified unde a binocula mic oscope
o he highes axonomic le el possible (mos ly species o genus, and amily in
some cases), using a ailable li e a u e and local expe knowledge.
Calcula ion o a iables. We quan ified li e decomposi ion in each li e bag as
he p opo ion o li e mass loss (LML) pe deg ee day (dd), o accoun o
di e ences in empe a u e ac oss si es; LML =[ini ial AFDM (g) –final AFDM
(g)]/ini ial AFDM (g), whe e ini ial AFDM was p e iously co ec ed by leaching,
d ying and ash con en , which we e es ima ed in he labo a o y55. We calcula ed
de i i o e-media ed decomposi ion as he di e ence in LML be ween pai ed
coa se-mesh and fine-mesh li e bags30. To al and de i i o e-media ed decom-
posi ion we e s ongly co ela ed ( 2=0.90, p< 0.001), bu we used bo h as
esponse a iables in he analyses because he o me is mo e ele an a he
ecosys em le el and he la e eflec s pa e ns media ed solely by de i i o es.
We quan ified de i i o e di e si y in each coa se-mesh li e bag as axon and
amily ichness; as hey we e s ongly co ela ed ( 2=0.90, p< 0.0001), we used
amily ichness o analyses o a oid axonomic inconsis encies among si es. We
quan ified abundance as he numbe o indi iduals pe li e bag. We es ima ed o al
biomass based on mean body size using published equa ions o each amily, and
mean body size based on abundance and he mean o a body size ca ego y (2.5–5.0,
5.0–10.0, 10–20, 20–40 and 40–80 mm) ha was assigned o each amily using
a ailable li e a u e56–63.
Da a analyses. We examined he influence o de i i o e di e si y, abundance,
biomass, mean body size, la i ude and he in e ac ions be ween de i i o e a iables
and la i ude on decomposi ion, using gene alised addi i e models (GAMs, gam
unc ion, ‘mgc ’package . 1.8.3164,65) and a model selec ion (d edge unc ion,
Lep oce idae*
Nemou idae Gamma idae Leuc idae Tipulidae*
G ipop e ygidae
Asellidae Limnephilidae
Calamoce a idae Palaemonidae Po amonau idae Hyallelidae Blabe idae P ilodac ylidae
LAURASIA
GONDWANA
Eu asia
No h
Ame ica
A ica
Sou h
Ame ica
India Aus alia
An a c ica
Lepidos oma idae Se icos oma idae
Fig. 5 Dis ibu ion o de i i o e amilies in ou s udy, which was p edominan ly Lau asian (blue) o Gondwanan (g een); inse indica es o igins o
hose wo egions (≈200 Ma). Pho og aphs ep esen a subse o amilies (o de ed le o igh om he mos o he leas abundan in ou s udy) and
as e isks deno e amilies ha we e globally dis ibu ed bu mo e abundan in one o he wo a eas. A comple e lis o amilies is p o ided in Supplemen a y
Table 1. Pho og aph c edi s: L. Boye o, A. Co nejo, R. Figue oa, N. López-Rojo, F. Masese, J. Pé ez, J. Rubio-Ríos, J. Viei a and C. M. Yule.
ARTICLE NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-021-23930-2
8NATURE COMMUNICATIONS | (2021) 12:3700 | h ps://doi.o g/10.1038/s41467-021-23930-2 | www.na u e.com/na u ecommunica ions
‘MuMIn’package . 1.43.17) based on Akaike weigh s66. A model selec ion
app oach was used o iden i y which ac o s and in e ac ions we e included in he
models wi h he highes condi ional p obabili ies (i.e. Akaike weigh s; Supple-
men a y Table 2). Models we e fi ed using enso p oduc in e ac ion smoo hs ( i)
wi h a no mal o gamma dis ibu ion (depending on model fi and esiduals) and
he iden i y-link unc ion67. We used his ype o model ins ead o a linea model
because p elimina y da a explo a ion showed he exis ence o non-linea
pa e ns68. To al o de i i o e-media ed decomposi ion was he esponse a iable,
and de i i o e di e si y, abundance, biomass, mean body size, absolu e la i ude
and he in e ac ions be ween de i i o e a iables and la i ude we e p edic o s,
fi ed as smoo h e ms. Explo ing di e ences among li e mix u es was beyond he
scope o his s udy (bu see Boye o e al.52, whe e li e di e si y e ec s on
decomposi ion we e examined based on he same expe imen desc ibed he e), so
we a e aged alues o di e en mix u es a he han including he mix u e as a
andom ac o in a gene alised addi i e mixed model, which would be highly
complex and would no con e ge when using in e ac ions and a iance unc ions
(see below). Spa ial co ela ion among si es was es ed using he au oco ela ion
unc ion (ACF) wi h esiduals o he final model; all alues we e <1 as ecom-
mended by Zuu e al.67. Abundance and biomass da a we e log (x +1)- ans-
o med o a oid he disp opo iona e influence o ou lying da a obse a ions on
model es ima es68. As in e ac ions o de i i o e a iables wi h la i ude we e sig-
nifican , we explo ed he ela ionships o opical (≤23° o la i ude), empe a e
(24–60°) and bo eal zones (>60°) h ough a model ha was simila o he one
desc ibed abo e, bu wi h la i ude as a ca ego ical a he han con inuous p edic o .
This was done o acili a e he ep esen a ion and in e p e a ion o complex non-
linea ela ionships be ween wo con inuous p edic o s.
We explo ed di e ences in de i i o e a iables ac oss ealms, biomes and
clima es wi h linea mixed-e ec s models (lme unc ion, ‘nlme’package .
3.1.15169) whe e ealm, biome o clima e we e fixed ac o s and li e mix u e ype
was a andom ac o , ollowed by pai wise compa isons using adjus ed P- alues
(glh and mcp unc ions, ‘mul comp’package . 1.4.1370). The a iance was
allowed o di e among ealms and biomes using he Va Iden s uc u e.
No malised esiduals o he final model we e inspec ed wi h plo s o esiduals s.
each p edic o , and no pa e n was obse ed. Va ia ion in assemblage composi ion
was explo ed wi h non-me ic mul idimensional scaling (NMDS, monoMDS
unc ion, ‘ egan’package . 2.5.6)71 calcula ed on Hellinge ans o med
abundance da a and pe mu a ional analysis o a iance (PERMANOVA) based on
aB ay–Cu is dissimila i y ma ix. We compa ed ealms, biomes and clima es
(adonis unc ion, ‘ egan’package), ollowed by pai wise compa isons (pai wise.
adonis unc ion), and de e mined which we e he mos ep esen a i e amilies in
each assemblage (simpe unc ion). All analyses we e un on R . 4.0.2.
Repo ing summa y. Fu he in o ma ion on esea ch design is a ailable in he Na u e
Resea ch Repo ing Summa y linked o his a icle.
Da a a ailabili y
Da a suppo ing he findings o his s udy a e a ailable a h ps://doi.o g/10.6084/m9.
figsha e.14245538. 1.
Recei ed: 29 Janua y 2021; Accep ed: 25 May 2021;
Re e ences
1. an de Plas, F. Biodi e si y and ecosys em unc ioning in na u ally assembled
communi ies. Biol. Re . 94, 1220–1245 (2019).
2. Isbell, F. e al. Quan i ying e ec s o biodi e si y on ecosys em unc ioning
ac oss imes and places. Ecol. Le . 21, 763–778 (2018).
3. Naeem, S., Thompson, L. J., Lawle , S. P., Law on, J. H. & Woodfin, R. M.
Declining biodi e si y can al e he pe o mance o ecosys ems. Na u e 368,
734–737 (1994).
4. Ca dinale, B. J. Impac s o biodi e si y loss. Science 336, 552–553 (2012).
5. Caliman, A., Pi es, A. F., Es e es, F. A., Bozelli, R. L. & Fa jalla, V. F. The
p ominence o and biases in biodi e si y and ecosys em unc ioning esea ch.
Biodi e s. Conse 19, 651–664 (2010).
6. Ca dinale, B. J. e al. The unc ional ole o p oduce di e si y in ecosys ems.
Am. J. Bo . 98, 572–592 (2011).
7. Po e R. J., an de We W., De Deyn G. B., S omph T. J. & Ho fland E. Is
li e decomposi ion enhanced in species mix u es? A me a-analysis. Soil Biol.
Biochem. 145, 107791 (2020).
8. Kou, L. e al. Di e si y-decomposi ion ela ionships in o es s wo ldwide. eLi e
9, e55813 (2020).
9. S i as a a, D. e al. Di e si y has s onge op-down han bo om-up e ec s on
decomposi ion. Ecology 90, 1073–1083 (2009).
10. Gessne , M. O. e al. Di e si y mee s decomposi ion. T ends Ecol. E ol. 25,
372–380 (2010).
11. Ceb ian, J. Pa e ns in he a e o p oduc ion in plan communi ies. Am. Na .
154, 449–468 (1999).
12. Vanno e, R. L., Minshall, G. W., Cummins, K. W., Sedell, J. R. & Cushing, C.
E. The i e con inuum concep . Can. J. Fish. Aqua . Sci. 37, 130–137 (1980).
13. Raymond, P. A. e al. Global ca bon dioxide emissions om inland wa e s.
Na u e 503, 355–359 (2013).
14. Ho chkiss, E. R. e al. Sou ces o and p ocesses con olling CO2 emissions
change wi h he size o s eams and i e s. Na . Geosci. 8, 696–699 (2015).
15. Yao, Y. e al. Inc eased global ni ous oxide emissions om s eams and i e s
in he An h opocene. Na . Clim. Change 10, 138–142 (2019).
16. Gessne , M. O., Chau e , E. & Dobson, M. A pe spec i e on lea li e
b eakdown in s eams. Oikos 85, 377–384 (1999).
17. Ma ks, J. C. Re isi ing he a es o dead lea es ha all in o s eams. Annu.
Re . Ecol. E ol. Sys . 50, 547–568 (2019).
18. Tonin, A. M. e al. In e ac ions be ween la ge and small de i i o es influence how
biodi e si y impac s li e decomposi ion. J. Anim. Ecol. 87,1465–1474 (2018).
19. Jonsson, M. & Malmq is , B. Mechanisms behind posi i e di e si y e ec s on
ecosys em unc ioning: es ing he acili a ion and in e e ence hypo heses.
Oecologia 134, 554–559 (2003).
20. Bas ian, M., Pea son, R. G. & Boye o, L. E ec s o di e si y loss on ecosys em
unc ion ac oss ophic le els and ecosys ems: a es in a de i us-based
opical ood web. Aus al. Ecol. 33, 301–306 (2008).
21. McKie, B. G., Schindle , M., Gessne , M. O. & Malmq is , B. Placing
biodi e si y and ecosys em unc ioning in con ex : en i onmen al
pe u ba ions and he e ec s o species ichness in a s eam field expe imen .
Oecologia 160, 757–770 (2009).
22. McKie, B. G. e al. Ecosys em unc ioning in s eam assemblages om
di e en egions: con as ing esponses o a ia ion in de i i o e ichness,
e enness and densi y. J. Anim. Ecol. 77, 495–504 (2008).
23. Tylianakis, J. M. Resou ce he e ogenei y mode a es he biodi e si y- unc ion
ela ionship in eal wo ld ecosys ems. PLoS Biol. 6, e122 (2008).
24. Boye o, L. e al. A global expe imen sugges s clima e wa ming will no
accele a e li e decomposi ion in s eams bu may educe ca bon
seques a ion. Ecol. Le . 14, 289–294 (2011).
25. Boye o, L. e al. Global pa e ns o s eam de i i o e dis ibu ion: implica ions o
biodi e si y loss in changing clima es. Glob. Ecol. Biogeog . 21,134–141 (2012).
26. Boye o, L. e al. Global dis ibu ion o a key ophic guild con as s wi h
common la i udinal di e si y pa e ns. Ecology 92, 1839–1848 (2011).
27. Handa, I. T. e al. Consequences o biodi e si y loss o li e decomposi ion
ac oss biomes. Na u e 509, 218–221 (2014).
28. Bo e , E. T. e al. Finding gene ali y in ecology: a model o globally
dis ibu ed expe imen s. Me hods Ecol. E ol. 5,65–73 (2014).
29. F ase , L. H. e al. Coo dina ed dis ibu ed expe imen s: an eme ging ool o
es ing global hypo heses in ecology and en i onmen al science. F on . Ecol.
En i on. 11, 147–155 (2013).
30. Woodwa d, G. e al. Con inen al-scale e ec s o nu ien pollu ion on s eam
ecosys em unc ioning. Science 336, 1438–1440 (2012).
31. Jonsson, M. & Malmq is , B. Ecosys em p ocess a e inc eases wi h animal species
ichness: e idence om lea -ea ing, aqua ic insec s. Oikos 89,519–523 (2000).
32. Boye o, L., Ramí ez, A., Dudgeon, D. & Pea son, R. G. A e opical s eams
eally di e en ? J. No h Am. Ben hol. Soc. 28, 397–403 (2009).
33. Jonsson, M., Malmq is , B. & Ho s en, P. O. Lea li e b eakdown a es in
bo eal s eams: does sh edde species ichness ma e ? F eshw. Biol. 46,
161–171 (2001).
34. Co nejo, A. e al. E ec s o mul iple s esso s associa ed wi h ag icul u e on
s eam mac oin e eb a e communi ies in a opical ca chmen . PLoS ONE 14,
e0220528 (2019).
35. Co nejo, A. e al. A common ungicide impai s s eam ecosys em unc ioning
h ough e ec s on aqua ic hyphomyce es and de i i o ous caddisflies. J.
En i on. Manag. 263, 110425 (2020).
36. Zub od, J. P. e al. Long- e m e ec s o ungicides on lea -associa ed
mic oo ganisms and sh edde popula ions-an a ificial s eam s udy. En i on.
Toxicol. Chem. 36, 2178–2189 (2017).
37. Rasmussen, J. J. e al. E ec s o a iazole ungicide and a py e h oid
insec icide on he decomposi ion o lea es in he p esence o absence o
mac oin e eb a e sh edde s. Aqua . Toxicol. 118-119,54–61 (2012).
38. Dillon, M. E., Wang, G. & Huey, R. B. Global me abolic impac s o ecen
clima e wa ming. Na u e 467, 704–706 (2010).
39. Dai, A. D ough unde global wa ming: a e iew. Clim. Change 2,45–65 (2011).
40. Tonin, A. M., Hepp, L. U., Res ello, R. M. & Gonçal es, J. F. Unde s anding o
coloniza ion and b eakdown o lea es by in e eb a es in a opical s eam is
enhanced by using biomass as well as coun da a. Hyd obiologia 740,79–88 (2014).
41. Pé ez, J., Basagu en, A., Descals, E., La añaga, A. & Pozo, J. Lea -li e
p ocessing in headwa e s eams o no he n Ibe ian Peninsula: mode a e
le els o eu ophica ion do no explain b eakdown a es. Hyd obiologia 718,
41–57 (2013).
42. F ibe g, N. e al. Biomoni o ing o human impac s in eshwa e ecosys ems:
he good, he bad and he ugly. Ad . Ecol. Res. 44, 211–278 (2011).
NATURE COMMUNICATIONS | h ps://doi.o g/10.1038/s41467-021-23930-2 ARTICLE
NATURE COMMUNICATIONS | (2021) 12:3700 | h ps://doi.o g/10.1038/s41467-021-23930-2 |www.na u e.com/na u ecommunica ions 9
