Effects of Soil Texture on Belowground C
Effects of Soil Texture on Below ground Carbon and Nutrient Storage in a Low land Amazonian Forest Ecosystem
3 1 Wh en d ee L. Silver, 4 * Jason Neff, Megan McGrod d y, Ed Veld kam p ,
2 Mich ael Keller, 5 an d Raim u n d o Cosm e
1 Dep artm en t of En viron m en tal S cien ces, Policy, an d Man agem en t, Un iversity of Californ ia, 151 Hilgard Hall, Berk eley,
Californ ia 94720, US A ; 2 T h e In tern ation al In stitu te of Trop ical Forestry, US DA Forest S ervice, Call Box 25000 Rio Pied ras, Pu erto Rico 00928, US A ; 3 Dep artm en t of Biological S cien ces, S tan ford Un iversity, S tan ford , Californ ia 94305-5020, US A ;
4 In stitu t fu er Bod en k u n d e u n d Wald ern aeh ru n g, Un iversitaet Goettin gen , Bu esgen w eg 237077 Goettin gen , Germ an y; 5 EMBRA PA A m az oˆn ia Orien tal, S an tare´m , Para´, Braz il
A BSTRACT
Soil textu re plays a key role in below grou n d C en zym e activity. Th e Cen tu ry m odel w as able to storage in forest ecosystem s an d stron gly in flu en ces
predict th e observed tren ds in su rface soil C an d N in n u trien t availability an d reten tion , particu larly in
loam s an d san ds bu t u n derestim ated C an d N pools
h igh ly w eath ered soils. We u sed field data an d th e in th e san ds by approxim ately 45% . Th e m odel Cen tu ry ecosystem m odel to explore th e role of soil
predicted th at total below grou n d C (0–20 cm depth ) textu re in below grou n d C storage, n u trien t pool
in san ds w ou ld be approxim ately h alf th at of th e sizes, an d N flu xes in h igh ly w eath ered soils in an
clays, in con trast to th e 89% w e m easu red. Th is Am azon ian forest ecosystem . Ou r field resu lts
discrepan cy is likely to be du e to an u n derestim a- sh ow ed th at san dy soils stored approxim ately 113 Mg C h a -1
tion of th e role of below grou n d C allocation w ith to a 1-m depth versu s 101 Mg C h a in
low litter qu ality in san ds, as w ell as an overestim a- clay soils. Coarse root C represen ted a large an d
tion of th e role of ph ysical C protection by clays in sign ifican t ecosystem C pool, am ou n tin g to 62%
th is ecosystem . Ch an ges in P an d w ater availability an d 48% of th e su rface soil C pool on san ds an d
h ad little effect on m odel ou tpu ts, w h ereas addin g N clays, respectively, an d 34% an d 22% of th e soil C
greatly in creased soil organ ic m atter pools an d pool on san ds an d clays to 1-m depth . Th e qu an tity
produ ctivity, illu stratin g th e n eed for fu rth er in tegra- of labile soil P, th e soil C:N ratio, an d live an d dead tion of m odel stru ctu re an d tropical forest biogeo- fin e root biom ass in th e 0–10-cm soil depth de-
ch em ical cyclin g.
creased alon g a gradien t from san ds to clays, wh ereas th e opposite tren d w as observed for total P, m in eral
Ke y w o rd s: roots; soil carbon ; cen tu ry m odel; soil N, poten tial N m in eralization , an d den itrification
textu re; biogeoch em istry; tropics.
I NTRODUCTION
ecosystem s by affectin g th e ability of soils to retain
C, w ater, an d n u trien t ion s (Jen n y 1980). For th ese Soil textu re exerts a stron g in flu en ce on m an y
reason s, soil textu re is also a key param eter in
h ydrologic an d biogeoch em ical processes in forest m odels of terrestrial biogeoch em istry, w h ich gen er- ally sh ow th at soil organ ic m atter (SOM) in creases
Received 3 March 1999; accepted 27 Au gu st 1999.
lin early w ith clay con ten t at region al an d global
*Corresponding author; e-mail: w silver@n atu re.b erk eley.ed u
scales (Parton an d oth ers 1993; Sch im el an d oth ers
W. L. Silver an d oth ers
1994). Despite th e w ell-dem on strated im portan ce processes varied in relation to soil textu re w h ile of soil textu re, m an y qu estion s rem ain abou t th e
h oldin g lan dscape position , clim ate, an d cover type role of textu re in below grou n d C allocation , soil C
relatively con stan t. Ou r site occu rred at th e con tact storage, ion exch an ge capacity, an d ecosystem scale
betw een a relic deposition al su rface an d an u plan d processes, su ch as prim ary produ ctivity an d decom -
lan dscape, all of w h ich is n ow u plan d, Tierra Firm e position . Th is is especially tru e for low lan d tropical
forest w ith little or n o topograph ic variation . forests, w h ich exh ibit con siderable spatial h eteroge-
We w ere also in terested in u sin g a m odelin g n eity in soil textu re at both local an d region al scales
approach to test ou r u n derstan din g of th e processes (Cu evas an d Medin a 1986, 1988; Matson an d Vi-
con tribu tin g to pattern s in below grou n d C an d tou sek 1987; Moraes an d oth ers 1995).
n u trien t pools in tropical forests. Soil textu re is Th e in teraction s of soil textu re an d biogeoch em i-
often u sed as a prim ary param eter con trollin g SOM cal cyclin g are com plex. Clay soils can facilitate th e
stabilization in biogeoch em ical m odels. To exam in e form ation of passive C pools w ith slow tu rn over
th e role of textu re in greater detail, an d to test tim es du e to th e ph ysical protection of SOM by clay
m ech an ism s by wh ich textu re in flu en ces th e biogeo- m in erals (Ch risten sen 1992). Clay soils also ten d to
ch em istry of a m oist low lan d tropical forest, w e
h ave h igh er cation exch an ge capacity, n et prim ary u sed th e Cen tu ry biogeoch em istry m odel (Parton produ ctivity (NPP), an d litter decom position rates in
an d oth ers 1987) to sim u late th e forest on th e th e tropics u n der n atu ral con dition s (Ueh ara 1995).
differen t soil textu ral classes at ou r site. Th e Cen tu ry San dy soils are often associated w ith h igh fin e root
m odel origin ally w as developed for u se in tem perate biom ass in tropical forests du e to greater C alloca-
grasslan ds an d agricu ltu ral system s bu t h as sin ce tion to roots for n u trien t an d w ater captu re (Klin ge
been exten ded an d re-param eterized for u se in 1973b, 1975; Cu evas an d Medin a 1988). San dy soils
tem perate an d tropical forests (San ford an d oth ers m ay also h ave slow er litter tu rn over rates du e to
1991; Com in s an d McMu rtrie 1993; Vitou sek an d n u trien t an d w ater lim itation s on decom position
oth ers 1994, Raich an d oth ers 1997), as w ell as for (Cu evas an d Medin a 1986).
an alyses of global biogeoch em ical dyn am ics (Sch im el Soil textu ral properties vary in respon se to w eath -
an d oth ers 1997). Wh ereas th e applicability of erin g rates an d th e in itial m in eralogy of th e paren t
Cen tu ry h as been tested exten sively for tem perate m aterial or th e deposition of m aterial from stream s,
ecosystem s, less testin g of th e m odel h as been don e ru n off, an d erosion . In gen eral th erefore, soil tex-
for tropical ecosystem s (bu t see Vitou sek an d oth ers tu re ten ds to ch an ge at local scales alon g topo-
1994; Gijsm an an d oth ers 1996; Raich an d oth ers graph ic gradien ts an d at lan dscape or region al scales
1997). Ou r m otivation for u sin g Cen tu ry is tw ofold. associated with ch an ges in paren t m aterial or weath -
First, w e exam in e th e ability of th e m odel to captu re erin g rates. Th e n atu re of soil-form in g processes
th e tren ds w e observe alon g ou r soil textu re gradi- th at lead to spatial h eterogen eity in soil textu re are
en t. Th is exercise is h elpfu l both becau se it allow s u s also gen erally associated w ith differen ces in m icro-
to evalu ate ou r u n derstan din g of textu ral con trols an d m acroclim ate, vegetation , an d h ydrologic in -
on tropical forest biogeoch em istry, an d becau se it pu ts an d exports m akin g it difficu lt to con trol for
represen ts a validation test for a key param eter textu re w h ile h oldin g several oth er en viron m en tal
in flu en cin g th e beh avior of th is m odel. Secon d, w e variables con stan t.
exam in e th e poten tial role of feedbacks am on g In th is stu dy, w e u se both an em pirical an d a
textu re, n u trien ts, an d w ater in th e m odel by m odelin g approach to exam in e th e relation sh ip of
exam in in g m odel sen sitivity to ch an ges in N, P, an d soil textu re to pattern s in below grou n d C an d
w ater availability.
n u trien t pools an d N tran sform ation rates in a low lan d Am azon ian forest. Th e Am azon basin is a geologically old region th at is diverse w ith regards to
M ETHODS
soil textu re (Fu rch an d Klin ge 1978; Moraes an d oth ers 1995). Pattern s in biogeoch em ical an d eco-
Site Descrip tion
logical processes w ith soil textu re h ave been de- Th e stu dy w as con du cted in th e Tapajos Nation al scribed alon g topograph ic gradien ts in Am azon ian
Forest (TNF) located 50 km sou th of San tarem , Para, forests (Cu evas an d Medin a 1986, 1988; Matson
Brazil. Th e region h as a m ean an n u al tem peratu re an d Vitou sek 1987; Livin gston an d oth ers 1988;
C, an d receives approxim ately 2000 m m of Vitou sek an d Matson 1988; Medin a an d Cu evas
of 25 o
rain per year w ith a dry season lastin g from May 1989), an d larger, region al-scale gradien ts (McKan e
th rou gh October (Parrotta an d oth ers 1995). Th e an d oth ers 1995; Moraes an d oth ers 1995). In th is
1000-h a stu dy site w as located on an old, n early flat, stu dy, w e w ere in terested in h ow below grou n d
erosion al rem n an t plateau (plan alto) w ith a very
Soil Textu re Effects on Below grou n d C in a Tropical Forest
lim ited drain age n etw ork form ed on sedim en ts of th e Barreiras form ation . Th e u n derlyin g bedrock geology con sists of slan ted beds of sedim en tary rocks deposited as a flu vial-lacu strin e sequ en ce. Th e soil textu ral differen ces at th is site m ost likely reflect th e u n derlyin g sedim en tary bedrock th at con sists of altern atin g san dy an d clayey beds.
Soil textu ral classification of th e larger stu dy area w as determ in ed by field in spection of sam ples from 551 con trol poin ts spaced at 50-m in tervals in a regu lar pattern th rou gh ou t th e 1000-h a area (5 x 2 km ). In total, tran sects covered approxim ately 28 km . A lim ited n u m ber of poin ts w as skipped be- cau se of th e difficu lty of access in clu din g a sm all area of poorly drain ed soil. Soils w ere divided in to tw o categories: (a) th ose h avin g clay an d clay loam textu re (fou n d at 375 poin ts); an d (b) th e rem ain der
h avin g coarser textu res (m ain ly san d an d san dy loam fou n d at 176 poin ts). Assu m in g th e regu lar tran sect pattern is represen tative of th e area, approxi- m ately 68% of th e area su rveyed is on clays or clay loam s (u ltisols an d oxisols) an d 32% on san ds an d san dy loam s (u ltisols).
Field Sam p lin g We sam pled su rface soils (0–10 cm depth ), forest
floor, an d fin e root biom ass alon g six 10 x 60–m tran sects located perpen dicu larly to a 400-m gradi- en t from san d to clay soils. Tran sects w ere located at
Figu re 1. Soil textu re m ap of th e stu dy site in th e Tapajos
0 an d 50 m (san ds), 150 an d 200 m (san dy loam to Nation al Forest, Para, Brazil. Th e m ap w as gen erated from clay loam ), an d 350 an d 400 m (clays) alon g th e textu re sam ples collected every 50 m alon g 0.5-km tran sects th rou gh th e research area. Th e textu re tran sect
textu ral gradien t (Figu re 1). Addition al sam ples of w as located at th e con tact betw een san ds an d clays in forest floor an d fin e root biom ass w ere taken on
block 6. Addition al sam ples w ere collected in h igh clay san ds an d clays to facilitate com parison of th ese tw o
an d h igh san d soils.
textu ral extrem es (see below ). To estim ate soil
ch em ical an d ph ysical properties an d root C, N, an d P to 1 m depth , w e took sam ples from five large 3 x
w ith h eat an d H 2 O 2 to rem ove organ ic m atter an d
1 x 1–m qu an titative soil pits ran dom ly located at 0, w ith NaHMP as a dispersal agen t to m in im ize
50, 200, 350, an d 400 m alon g th e textu re gradien t foam in g. Soil pH w as determ in ed on fresh soils in a an d 18 addition al 1 x 1 x 1–m soil pits w ith n in e
slu rry of 4 g in 8 m L deion ized H 2 O. each on san d an d clay soils.
We m easu red den itrification en zym e activity (DEA) accordin g to Tiedje (1994) w ith som e m in or
Su rface Soil Ph ysical an d Ch em ical
ch an ges becau se of th e h igh activity m easu red. A Prop erties alon g th e Tran sects
10-g sam ple of field m oist soil w as placed in a Soils w ere sam pled from th e 0–10-cm depth by
250-m L Erlen m eyer flask an d 25 m L of a solu tion u sin g a 2.5-cm diam eter soil corer. We took m u ltiple
con tain in g 1 m M glu cose, 1 m M NaNO 3 an d 1 g L-1 core sam ples at each of five ran dom ly selected
ch loram ph en icol w as added. Th e flasks th en w ere poin ts alon g each tran sect yieldin g 30 aggregate soil
evacu ated an d flu sh ed six tim es w ith He to produ ce sam ples. Sam ples w ere refrigerated im m ediately
an an aerobic en viron m en t. Acetylen e (m ade from after collection an d u n til an alyses cou ld be con -
calciu m carbide) w as added to a fin al con cen tration du cted (approxim ately 3–7 days). Roots an d litter
of 10 kPa. Th e flasks w ere in cu bated du rin g a w ere m an u ally rem oved from field m oist sam ples,
30-m in an aerobic assay on a rotary sh aker (100 w h ich th en w ere passed th rou gh a 2-m m sieve. Soil
rpm ). Head space gas w as sam pled after 10, 20, 30, textu re was determ in ed u sin g a Bou you cos h ydrom -
an d 40 m in u tes an d an alyzed for N 2 O on a gas
ch rom atograph w ith electron captu re detector (Gee an d Bau der 1986). Sam ples w ere pretreated
eter in a soil su spen sion of 50 g of soil in 1 L of H 2 O
(Keller an d Rein ers 1994). Th e DEA w as calcu lated
W. L. Silver an d oth ers
from th e lin ear in crease in N 2 O takin g th e dissolved
Fin e Root Biom ass
N 2 O in to accou n t by u sin g th e Bu n sen relation sh ip. Fin e root stan din g stocks (0–10 cm depth ) w ere Poten tial n et N m in eralization an d n et n itrifica-
sam pled from five ran dom location s alon g each tion were m easu red u sin g aerobic laboratory in cu ba-
tran sect (n 5 30) by u sin g a root corer of 4.1 cm tion s (Hart an d oth ers 1994). Tw o 15-g su bsam ples
in side diam eter (Vogt an d Perrson 1991; Silver an d of field m oist soils w ere w eigh ed in to plastic Nal-
Vogt 1993). Cores w ere refrigerated u n til th ey w ere gen e bottles. On e set of su bsam ples w as im m edi-
processed by sortin g live an d dead roots by size class ately extracted w ith 100 m L of 2 M KCl; th e oth er
(less th an or equ al to 2 m m diam eter; greater th an set w as covered w ith perforated parafilm an d in cu -
2–5 m m diam eter) from w ash ed sieves. Th is tech - bated at am bien t tem peratu res (approxim ately 25 o C) n iqu e yielded few sam ples in th e greater th an for 7 days before extraction . Net n itrification an d N 2–5-m m size class, so th ese data are n ot reported m in eralization were an alyzed an d calcu lated accord-
in g to Hart an d oth ers (1994).
h ere. In stead, w e u se data from th e large qu an tita- Sam ples for oth er ch em ical an alyses w ere air-
tive pits described below to estim ate th e coarse root
dried an d grou n d to pass th rou gh a 2-m m sieve. fraction . Root sam ples w ere dried at 65
C an d Approxim ately 5-g sam ples w ere extracted sepa-
w eigh ed to determ in e m ass. We took 10 addition al rately w ith 55 m L 1 M NH
4 Cl for exch an geable
ran dom ly located sam ples for fin e root stan din g
cation con cen tration s an d w ith 50 m L of NH
F for
stocks (less th an or equ al to 2 m m diam eter) on
extractable P (Olsen an d Som m ers 1982) by u sin g a both san d an d clay soils (n 5 20). To ch aracterize vertical vacu u m extractor (Joh n son an d oth ers
pattern s in fin e root stan din g stocks by depth , w e 1991). Sam ples w ere an alyzed for exch an geable Ca,
took th ree root cores from both th e 20–30-cm an d Mg, K, Al, Mn , Fe, an d P on a Direct Cu rren t Plasm a
th e 30–40-cm depth s each on san d an d clay soils Spectraspan V spectroph otom eter, Fu llerton , CA
(n 5 12). All cores w ere processed as above. Root C USA at th e In tern ation al In stitu te of Tropical For-
an d N w ere m easu red on a C.E. In stru m en ts CN estry (IITF). For total C an d N, sam ples w ere
an alyzer at U.C. Berkeley, an d root P w as m easu red regrou n d to a fin e pow der an d an alyzed u pon
at IITF after a H 2 O 2 -H 2 SO 4 digest (Parkin son an d com bu stion by u sin g a Fison s CN an alyzer, Su ffolk,
Allen 1975).
UK. Stan dard referen ce m aterial, procedu ral blan ks, an d replicate sam ples w ere ru n for qu ality assess-
Coarse Roots
m en t at th e rate of 1 per 10 sam ples. Su bsam ples of all soils w ere dried at 105 o
A total of 23 large qu an titative pits w ere excavated
for coarse root biom ass. We sam pled th e five pits con ten t. All data reported h ere are on an oven dry
C to determ in e m oistu re
located at 0, 50, 200, 350, an d 400 m alon g th e soil soil basis. Bu lk den sity of th e 0–10-cm depth w as
textu re gradien t, as w ell as 18 1 x 1 x 1–m pits w ith determ in ed at th ree ran dom poin ts on each of th e
n in e each on san d an d clay soils. Each pit w as six tran sects alon g th e gradien t (n 5 18) by u sin g a
sam pled for root biom ass at th e 0–10-cm , 10–40- kn ow n volu m e bu lk den sity corer. Sam ples w ere
cm , an d 40–100-cm depth s. In th e 0–10-cm depth , dried at 105 o
C to a con stan t w eigh t an d w eigh ed for w e separately w eigh ed th e m ass of fin e roots th at bu lk den sity determ in ation s.
cou ld be easily separated from soil in th e field. Th is tech n iqu e u n dou btedly exclu ded a proportion of
Forest Floor Mass an d Elem en tal Con ten t th e very fin e (less th an 1 m m ) roots an d root Th e forest floor w as sam pled at five ran dom loca-
fragm en ts, w h ich are better sam pled u sin g th e tion s alon g each tran sect (n 5 30), an d at 10
corin g tech n iqu e m en tion ed above. At th e tw o addition al ran dom location s each on clay an d san dy
deeper depth s, fin e root biom ass w as low, so w e soils by u sin g a 15 x 15–cm tem plate (in side area).
took sam ples of total root biom ass com bin in g all size Th e forest floor con tain ed recen t litter an d h u m ified
classes togeth er. Roots w ere clean ed of adh erin g soil organ ic m atter. Sam ples were dried at 65 o
an d w eigh ed in th e field by u sin g a sprin g balan ce to determ in e m ass, grou n d in a Wiley m ill, an d
C, weigh ed
(650 g), an d su bsam pled for m oistu re determ in a- redried. Grou n d sam ples w ere predigested in H 2 O 2 tion s. Sam ples w ere dried at 65 o
C an d w eigh ed to
determ in e m ass, an d to estim ate a con version factor block digestor (Lu h Hu an g an d Sch u lte 1985) an d
an d th en digested in con cen trated HNO 3 by u sin g a
from field w et w eigh t to oven dry w eigh t. Grou n d an alyzed at Boston Un iversity for Ca, Mg, K, Al, P,
su bsam ples w ere predigested in H 2 O 2 an d digested Fe, an d Mn at IITF an d for total C an d N. Stan dard
in H 2 SO 4 for total P an d an alyzed on a DCP at IITF referen ce m aterial (citru s an d apple leaves), proce-
(Parkin son an d Allen 1975). Addition al su bsam ples du ral blan ks, an d replicate sam ples w ere ru n for
w ere an alyzed for C an d N on a CN an alyzer at qu ality assessm en t at th e rate of 1 per 10 sam ples.
Stan ford Un iversity.
Soil Textu re Effects on Below grou n d C in a Tropical Forest
Soil Ch em ical an d Ph ysical Prop erties su m m ed by depth to produ ce pool estim ates for to 1 m Dep th
specific depth in tervals. We h ad n o bu lk den sity
Soil textu ral properties, bu lk den sity, soil d 15 N, an d
estim ates for su bsu rface loam soils, an d so th ese soils w ere exclu ded from th is an alysis.
soil cation , P, C, an d N con ten t w ere determ in ed in Soils w ere sam pled for d 15 N from vertical profiles 1-m deep qu an titative soil pits located alon g th e soil at 0 m (san d), 200 m (loam ), an d 350 m (clay) alon g textu re gradien t an d in addition al pits located on th e gradien t. We took on e sam ple from th e su rface san ds an d clays. An aggregate soil sam ple w as to 2-cm , 2–5 cm , 5–10 cm depth s an d su bsequ en t collected at th e 0–10-cm , 30–40-cm , an d 90– 10-cm -depth in crem en ts to 1 m in depth . Addi- 100-cm depth s from pits located at 0, 50, 200, 350, tion al sam ples (n 5 4–6) w ere taken from th e an d 400 m alon g th e gradien t (n 5 15) for textu ral su rface an d 90–100-cm depth s in th e san ds (0 m ) properties, exch an geable cation con cen tration s, ex- an d clays (350 m ). Soils w ere air-dried im m ediately
tractable an d total P, an d pH. Soil pH w as deter- after field sam plin g an d again at 60 o
C to a con stan t m in ed in w ater as above, an d in a slu rry of 4 g of soil
w eigh t. Soils th en w ere grou n d to a pow der an d in 8 m L of 1 M KCl. Soil textu re, exch an geable
an alyzed for d 15 N at th e Natu ral Resou rce Ecology cation s, an d extractable P w ere m easu red as above.
Lab at Colorado State Un iversity on a VG Isoch rom To estim ate total P con cen tration s, w e digested
Mass Spectrom eter, Fran klin , MA, USA w ith a
approxim ately 5 g of soil in H 2 O 2 an d con cen trated
dedicated Carlo Erba sam ple preparation system .
H 2 SO 4 by u sin g a block digestor (Parkin son an d Th e 15 N abu n dan ce is expressed in delta u n its Allen 1975). Solu tion s w ere an alyzed at IITF. Stan -
relative to th e 15 N/ 14 N ratio (R) of atm osph eric N dard referen ce soils, replicate sam ples, an d blan ks
( 15 N air/ 14 N air 5 0.0036765). Th e d 15 N is calcu lated w ere u sed for qu ality con trol.
as d 15 N 5 [(R sam ple /R stan dard )-1] x 1000]. Total C an d N w ere estim ated from su bsam ples
collected at 10-cm -depth in crem en ts (0–100 cm )
Mod el Ap p lication
from th e pits at 0 m (san d), 200 m (loam ), an d 350 We u sed th e Cen tu ry biogeoch em istry m odel (Par- m (clay) alon g th e textu re gradien t. We also took ton an d oth ers 1987) for ou r m odelin g exercise. sam ples from six addition al pits each on clay an d Cen tu ry h as a th ree-com partm en t SOM m odel con - san d soils at 7 cm , 40 cm , an d 100 cm depth . We fit a sistin g of active, slow an d passive pools of organ ic logarith m ic cu rve to th ese valu es an d in terpolated m atter. Tem peratu re, m oistu re, an d n u trien t con - th e soil C an d N con cen tration s for th e in term ediate
strain ts con trol th e m ovem en t of C from plan t pools 10-cm -depth in crem en ts. Th e m ean squ ared residu -
to SOM pools, as w ell as betw een SOM pools. In als for fitted C cu rves w ere 0.98 (60.01) for clays
addition , th e m ovem en t of C from th e slow to an d 0.94 (60.03) for th e san ds, an d for N th ey w ere
passive organ ic m atter pools in creases proportion -
0.95 (60.02) for both san ds an d clays. All sam ples ally w ith clay con ten t. We in itialized Cen tu ry by w ere an alyzed on a CN an alyzer at U.C. Berkeley
u sin g param eters from tropical forest sim u lation s an d in clu ded replicate sam ples an d referen ce soils.
don e by Vitou sek an d oth ers (1998) an d Raich an d We in terpolated total an d extractable P con cen tra-
oth ers (1997) w ith rain fall an d tem peratu re data tion s by depth by u sin g th e valu es from th e 0–10-
from th e TNF. Th e tropical forest param eterization s cm , 30–40-cm , an d 90–100-cm sam ples from pits
differ from th e origin al grasslan d an d forest m odels alon g th e textu re gradien t. Total P did n ot vary
in th e stoch iom etry of C:N ratios an d th e relatively sign ifican tly w ith depth in th e clays, so w e u sed a
h igh er N in pu ts. On e of th e m ain biogeoch em ical m ean valu e for all depth s. Total P follow ed a
differen ces between tem perate system param eteriza- decreasin g pattern w ith depth in th e san ds (r 2 5 tion s an d th e TNF is th e lack of paren t m aterial P in
0.84 6 0.01). Total P data reported h ere represen t a Brazil. For th ese ru n s, w e in itialized th e m odel w ith sm all sam ple size (n 5 4) an d th u s sh ou ld be view ed
n o paren t m aterial P an d h igh P sorption capacity to w ith cau tion .
sim u late th e very low P con dition s of th is low lan d Bu lk den sity by depth w as sam pled on san ds
tropical forest.
(50 m ) an d clays (350 m ) at 10-cm -depth in cre- Ou r m ain objective w as to evalu ate th e sen sitivity m en ts by u sin g a kn ow volu m e bu lk den sity corer.
of th e Cen tu ry m odel to soil textu re an d to com pare Roots an d rocks w ere rem oved, an d th e sam ples
th e m odel sim u lated textu ral effects w ith pools of C w ere dried at 105 o
an d elem en ts th at w e cou ld m easu re at ou r site. To w eigh ed to determ in e m ass. To estim ate th e m ass of
C to a con stan t w eigh t an d
perform th is sen sitivity stu dy, w e ‘‘tu n ed’’ th e m odel soil C, N, an d P to a depth of 1 m , w e m u ltiplied th e
to ach ieve good correspon den ce betw een m odeled con cen tration s by depth w ith th e bu lk den sity
an d m easu red SOM carbon for ou r clay site. Th ere estim ation s in san d an d clay soils. Valu es w ere
are a n u m ber of w ays to adju st Cen tu ry, bu t on e of
W. L. Silver an d oth ers
Table 1. Exch an geable Cation s, Extractable P, Total C an d N, an d Soil Ph ysical Properties a in th e Tapajos Nation al Forest, Para, Brazil
M eters alon g th e Textu re Grad ien t
Soil Prop erty
400 Bu lk den sity (g cm 23 )
1.23 (0.09) ac 1.02 (0.02) c San d (% )
37 (1.0) c 38 (1.1) c Clay (% )
60 (1.1) c 60 (1.3) c Silt (% )
3 (0.5) 2 (0.5) pH (H 2 O)
4.26 (0.18) bc 4.19 (0.13) c Total C (% )
2.66 (0.11) 2.26 (0.06) Total N (% )
0.18 (0.01) 0.17 (0.01) Extractable P (µ g g 21 )
3.96 (0.25) c 3.01 (0.18) d Ca (cm ol 1 kg 21 )
0.07 (0.01) 0.06 (0.01) Mg (cm ol 1 kg 21 )
0.11 (0.01) b 0.08 (0.01) a Mn (µ g g 21 )
1.72 (0.20) b 1.74 (0.29) b Al (cm ol 1 kg 21 )
3.25 (0.13) d 2.56 (0.25) e
a 0–10 cm depth. Values are means 6 1 SE (in parentheses). Rows with different lowercase letters identify statistically significant differences among sites at the 95% level. Magnesium
concentrations were below detection (,) by the analytical instrumentation (0.013 cmol 1 kg 21 ).
th e m ost sen sitive aspects of th e m odel is th e stron g in terpolated NH 4 F-P an d total soil P con cen tration s poten tial N lim itation to prim ary produ ctivity (Vi-
(see above). For th e loam soil, w e u sed th e 0–10-cm tou sek an d oth ers 1998). By adju stin g N in pu ts, w e
bu lk den sity m easu rem en t becau se th ere w ere n o cou ld alter th e steady state con ten t of SOM for th e
data for th e deeper soil depth s. On ly clay an d san dy clay soil to ach ieve a good fit to ou r observed data.
soils w ere m easu red for root C, N, an d P. Coarse root Th e equ ilibriu m valu es of N addition s requ ired to
data w ere collected from th e 0–10-cm , 10–40-cm , obtain reason able SOM C an d N valu es for th e clay
an d 40–100-cm depth s. We estim ated th e m ass of site w ere addition s of 2.2 kg N h a -1 y -1 . Wh en th e
coarse root C, N, an d P for th e 0–20-cm depth by m odel in pu ts w ere set for th e clay site, w e altered
addin g on e-th ird of th e valu es for th e 10–40-cm on ly th e bu lk den sity an d textu re to reflect th e
depth to th e valu es for th e 0–10-cm depth . Becau se con dition s in th e loam an d san dy soils.
root biom ass ten ded to decrease w ith depth , th is is On ce Cen tu ry w as param eterized for th e clay site,
likely to be a sligh t u n derestim ation of th e pool sizes w e th en com pared th e equ ilibriu m soil an d root C,
of root C, N, an d P.
N, an d P con ten ts from th e m odel to m easu red valu es for san d an d loam soils. To system atically
Statistical An alyses
explore th e m odels sim u lated lim itation s to plan t Statistical an alyses w ere perform ed u sin g Systat grow th , an d con cu rren tly exam in e th e sen sitivity of
(Wilkin son 1990). Data w ere log tran sform ed w h en Cen tu ry to w ater an d n u trien t availability in th e
n ecessary to m eet th e assu m ption s for an alysis of san ds, w e ran experim en ts w h ere w e dou bled N
varian ce (ANOVA). On e-w ay an d tw o-w ay ANO- in pu ts (from 2.2 to 4.4 kg h a -1 y -1 ), P in pu ts (from
VAs w ere u sed to determ in e sign ifican t differen ces
1.1 to 2.2 kg h a -1 y -1 ), an d rootin g depth (60–120 cm w ith in variables alon g th e gradien t, an d w ith tex- depth ). Becau se w e cou ld n ot adju st w ater u se by
tu re an d depth . Pairw ise com parison s u sin g th e vegetation in th e m odel, w e u sed rootin g depth as a
Least Sign ifican t Differen ces protocol w ere per- rou gh su rrogate for w ater access an d u se, becau se
form ed to determ in e w h ere sign ifican t differen ces rootin g depth h as been sh ow n recen tly to correlate
occu rred. Pearson correlation s w ere u sed to exam - w ith ecosystem -level w ater u se in season ally dry
in e relation sh ips am on g soil textu re an d soil an d tropical forests (Nepstad an d oth ers 1994).
root properties. We also u sed sim ple an d stepw ise Th e Cen tu ry m odel ou tpu t u ses a fixed depth of
m u ltiple lin ear regression s to iden tify tren ds w ith 0–20 cm . To com pare th e Cen tu ry m odel ou tpu t
soil textu re. We perform ed t-tests to exam in e differ- w ith th e field data, w e u sed ou r m easu red bu lk
en ces in soil, root, an d forest floor C, N, an d P den sity, C an d N valu es, an d th e m easu red an d
betw een san d an d clay. Residu als from all an alyses
Soil Textu re Effects on Below grou n d C in a Tropical Forest
w ere ch ecked for n orm ality an d h om ogen eity of Table 2. Correlation Coefficien ts of th e varian ces (Steel an d Torrie 1980). We report sign ifi-
Relation sh ip of Soil Ch em ical an d Ph ysical can t differen ces at th e 95% level u n less oth erw ise
Properties, Fin e Root Biom ass, an d Forest Floor n oted. Valu es reported in th e text are m ean s fol-
Biom ass to Clay Con ten t in th e Tapajos Nation al low ed by stan dard errors.
Forest, Para, Brazil Variab le
Clay (% )
R ESULTS
Ca cm ol 1 kg 21 20.32
Ph ysical an d Ch em ical Prop erties
K cm ol 1 kg 21 0.47
of Su rface Soils
Mn µ g g 21 20.57 Th e percen t san d an d clay varied sign ifican tly in th e 1 Al cm ol kg 21 0.91*
0–10-cm depth alon g th e textu re gradien t (Table 1 Pµgg 21 20.84* an d Figu re 2) ran gin g from 80% (61) san d to 60%
C (% )
(61) clay. Extractable soil P con cen tration s w ere 0.49*
N (% )
0.35 sign ifican tly n egatively correlated w ith clay con ten t
Live roots (g m )
(r 2
20.56* of clay con ten t an d exch an geable K con cen tration s
5 0.70, P , 0.01; Table 2), an d th e com bin ation
Dead roots (g m 22 )
20.06 (r 2 5 0.81, P , 0.01). Th e correlation w ith K m ay be
Forest floor (kg h a 21 )
20.73* du e to som e K occlu sion by th e 1:1 clays, alth ou gh
C:N
0.88* K is likely to be sorbed less stron gly th an P (Grah am
20.26 also occu rred with P an d exch an geable Al con cen tra-
NH -N (µ g g an d Fox 1971). A sign ifican t n egative relation sh ip 21 4 )
0.60 tion s (r 2 5 0.57, P , 0.01). Extractable soil P w as
NO 3 -N (µ g g 21 )
Net n itrification (µ g g 21 d 21 )
0.70* Net N m in eralization (µ g g 21 d 21 )
Bu lk den sity (g cm 23 )
*P , 0.01.
very low in clays soils (3.0 µ g g -1 ), an d in creased by
a factor of 3 in th e san ds (Table 1). Th ere w ere n o sign ifican t tren ds in total C or N
con cen tration s alon g th e soil textu ral gradien t in th e 0–10-cm soil depth , bu t th e C:N ratio decreased sign ifican tly from 18.4 in th e san ds to 13.6 in th e clays (Figu re 2). Total C an d N in th e soil w ere positively an d sign ifican tly correlated alon g th e
gradien t (r 2 5 0.73; P , 0.01). Th e soil C:P extractable an d N:P extractable ratios both in creased dram atically from san ds to clay alon g th e textu re gradien t (Fig- u re 2), an d w ere sign ifican tly positively correlated w ith clay con ten t (Table 2). Total C pools in th e su rface soils ran ged betw een 23 (60.6) an d 38 (67.2) Mg C h a -1 an d did n ot follow a sign ifican t tren d w ith soil textu re.
Nitrogen pools an d flu xes varied sign ifican tly w ith soil textu re. Am m on iu m -N con cen tration s w ere sign ifican tly low er in th e loam soils th an in th e
Figu re 2. Soil ch em ical an d ph ysical properties in th e san ds (P 5 0.05), w h ereas NO 3 -N w as sign ifican tly 0–10-cm depth alon g a soil textu re gradien t in th e Tapajos
h igh er in th e loam soils th an in th e san ds (Figu re 3). Nation al Forest, Brazil. (A) Textu re; (B) C:N ratio; (C) N:P
Net n itrification rates in creased sign ifican tly from ratio; (D) C:P ratio. To calcu late ratios, w e u sed total C an d
0.68–0.72 µ g g -1 d -1 in th e san ds to 1.91–2.29 µ g N an d extractable P.
g -1 d -1 in th e clays (Figu re 3). Net n itrification rates
W. L. Silver an d oth ers
Forest Floor an d Root Biom ass Forest floor m ass alon g th e textu re gradien t aver-
aged 6.9 (60.7) Mg h a -1 (Table 3). Alon g th e gradien t, forest floor C con cen tration s w ere greater on th e san ds th an on th e oth er soil types, an d forest floor N an d P con cen tration s decreased sligh tly bu t sign ifican tly from san ds to clays. Wh en com parin g ju st th e tw o textu ral extrem es (n 5 40), forest floor
C, N, an d P con ten t w ere sign ifican tly greater on san ds th an on clays (Table 4). Forest floor C:N an d C:P ratios w ere very sim ilar on san d an d clays soils.
Stan din g stocks of total fin e root biom ass (live plu s dead) in th e 0–10-cm depth decreased sign ifi- can tly from approxim ately 4.5 (60.4) Mg h a -1 in th e san dy soils to less th an 1.9 (60.4) Mg h a -1 in clays alon g th e textu re gradien t (Figu re 4). Both live an d dead fin e root biom ass follow ed th e sam e tren d w ith soil textu re; live fin e root biom ass w as very low in all soil types (0.05–0.4 Mg h a -1 ). San dy soils h ad sign ifican tly greater live an d dead fin e root biom ass to a depth of 40 cm th an th e clays (Figu re 4). Total fin e root biom ass w as sign ifican tly n egatively corre- lated w ith N m in eralization rates an d positively
correlated w ith extractable P pools (r 2 5 0.48; P , 0.05). Fin e root C, N, an d P con ten t w ere sign ifi- can tly greater on san dy soils th an on clays in th e top
Figu re 3. In itial m in eral N con cen tration s A, poten tial
10 cm of soil (Table 4).
n et m in eralization an d n itrification B, an d den itrification en zym e activity (DEA) C in th e 0–10-cm soil depth alon g
Total root biom ass to a depth of 1 m ran ged from
11 to 188 Mg h a a soil textu re gradien t in th e Tapajos Nation al Forest, -1 in th e five pits alon g th e textu re Brazil. Clay con ten t in creased alon g th e gradien t.
gradien t (Figu re 4) an d w as greatest in th e 0–10-cm depth at all sites. Th ere w as a gen eral pattern of in creasin g total root biom ass alon g th e gradien t
explain ed m ost of th e tren d in n et N m in eralization from san ds to clays. Th e exception w as at m eter 50 rates, w h ich also gen erally in creased sign ifican tly
from san ds to clays, w ith th e exception of th e last (san ds) w h ere w e en cou n tered an extrem ely h eavy clay soil site alon g th e gradien t (Figu re 3). Den itrifi-
root (u p to 15 kg m -2 w et w eigh t) th at pen etrated cation en zym e activity in creased sign ifican tly from
from n ear th e su rface to 1 m in depth . Com parin g all
77 (623) n g g -1 h -1 in th e san ds to 317 (624) n g g -1
22 pits on san d an d clay soils (in clu din g th e tw o clay
h -1 in th e clays an d w as sign ifican tly positively an d tw o san d pits alon g th e textu re gradien t), th ere
w as sign ifican tly m ore coarse root C, N, or P in clays clay con ten t (Figu re 3 an d Table 2).
correlated w ith in itial NH 4 -N con cen tration s an d
to a depth of 10 cm , an d greater root P in 40–100-cm Con cen tration s of exch an geable Al in creased
depth on clays. Th e ratios of root C:N an d C:P w ere sign ifican tly from san ds to clays (Table 1), an d
h igh er on th e san ds th an on th e clays for all soil clay con ten t alon e explain ed 84% of th e variabil-
depth s (Table 4). Th e total root C pool to 1 m depth ity in exch an geable Al (Table 2). Exch an geable
w as approxim ately 26 6 8 Mg h a -1 on san ds an d n u trien t cation con cen tration s w ere very low in
17 6 2 Mg h a -1 on clays.
th e su rface soils alon g th e textu re gradien t (Table 1). Exch an geable Mg w as below th e detection lim it
Soil Prop erties to 1 m Dep th of th e an alytical in stru m en tation (1.6 µ g g -1 or
0.013 cm ol 1 kg -1 ). Pool sizes (kg h a -1 ) of m ost Total soil C pools to 1 m depth w ere very sim ilar on elem en ts follow ed sim ilar tren ds as th e elem en tal
clays (80 Mg C h a -1 6 3) an d san ds (81 Mg C h a -1 6 con cen tration s alon g th e gradien t. Soil pH w as
4; Table 4). Th ere w as sign ifican tly m ore C in th e sign ifican tly low er in th e loam soils th an in th e clays
su rface 20 cm of th e clays bu t greater C in low er 50 (Table 1).
cm in th e san ds (P , 0.10). Total N follow ed a
Soil Textu re Effects on Below grou n d C in a Tropical Forest
Table 3. Forest Floor Mass an d Elem en tal Con cen tration s alon g a Textu re Gradien t in th e Tapajos Nation al Forest, Para, Brazil
M eters alon g th e Textu re Grad ien t
Forest Floor Prop erty
400 Mass (g m 22 ) 839 (165)
42 (1.6) b 47 (1.6) ab N (% )
1.66 (0.1) c 1.77 (0.3) c C:N
27 (2.1) 31 (8.0) P (m g g 21 )
0.53 (0.03) a 0.37 (0.05) b Ca (m g g 21 )
5.45 (0.82) 4.49 (0.79) Mg (m g g 21 )
1.67 (0.18) 1.74 (0.45) K (m g g 21 )
1.62 (0.25) 1.16 (0.12) Mn (m g g 21 )
0.25 (0.02) 0.33 (0.13) Al (m g g 21 )
6.31 (0.82) d 3.66 (0.66) cd
Values are means 6 1 SE (in parentheses). Rows with different lowercase letters identify statistically significant differences among sites at the 95% level; n 5 30.
differen t pattern th an C (Table 4). Th ere w as sign ifi-
Mod el Resu lts
can tly m ore total soil N in clays to 1 m (8.5 Mg N Th e Cen tu ry m odel sim u lation of th e soil textu re
ha -1 6 0.4) th an in th e san ds (6.6 Mg N h a -1 6 0.4), gradien t predicted greater soil C, N, an d P in clays
du e prim arily to th e sign ifican tly greater pools of N th an in loam s or san ds, respectively (Table 7). in th e top 50 cm in th e clays. Total C an d N Cen tu ry u n derestim ated th e soil C an d N pools by con cen tration s in th e su rface soils w ere low er w h en 45% an d 44% , respectively, on th e san ds, bu t sam pled from th e large qu an titative pits th an from captu red th e tren ds of low er C an d N in th e san ds th e core sam ples. It is u n clear w h at cau sed th ese th at w e fou n d w ith ou r sam ples from th e large differen ces, bu t th ey cou ld resu lt from th e sm aller
sam ple size an d differen ces in th e area sam pled. We qu an titative pits. Cen tu ry h as several P pools in clu d- estim ated a total below grou n d C pool (forest floor
in g m in eral P, SOM P, secon dary P, occlu ded P, an d plu s roots plu s soil) of 113 Mg h a -1
paren t P. It is som ew h at difficu lt to com pare th ese 101 Mg h a -1
C on th e san ds to
C on th e clays (Table 4). m odel pools directly to data from P fraction ation Total soil P con cen tration s an d pools in th e 0–10
sch em es (Gijsm an an d oth ers 1996), an d w e lack cm depth follow ed th e reverse tren d of extractable
th e data for a com preh en sive evalu ation of th e P. Alon g th e soil textu re gradien t, total soil P
Cen tu ry P m odel; h ow ever, th ere are several in ter- con cen tration s in creased by a factor of 4 to 5 from
estin g pattern s w orth n otin g. We report th e m od- san ds to clays (Table 5). Total P con cen tration s
eled SOM P fraction from Cen tu ry in com parison to ten ded to be greater at th e deeper soil depth s in th e
ou r NH 4 F extractable P, forest floor P, an d total P san ds bu t follow ed n o tren d in th e loam or clays
valu es. In terestin gly, th e vast m ajority of th e P in soils (Table 5).
th e m odel sim u lation s w as located in th e SOM P Th ere w ere few tren ds in soil textu re w ith depth
pool rath er th an in secon dary or occlu ded P. Field alon g th e textu ral gradien t (Table 5). San d con ten t
m easu rem en ts of NH 4 F extractable P an d forest floor decreased sligh tly (-10% ) w ith depth in th e san d
P sh ow ed approxim ately th ree tim es m ore labile P pits w ith a correspon din g in crease in clay con ten t.
in th e san ds th an in th e clays. In con trast, m odeled P Exch an geable n u trien t cation an d extractable P
m in eralization (an in dex of labile P) follow ed th e con cen tration s gen erally decreased w ith depth in
opposite pattern an d ran ged from 0.15 g P m 2 y -1 in th e soil (Table 5).
th e san ds to 0.2 g P m 2 y -1 in th e clays. Th e m odel Delta 15 N valu es ran ged from 6.2 to 12.7. San ds
sim u lation predicted a th reefold in crease in SOM P
from san ds to clay, sim ilar to th e 2.7-fold in crease su rface soils (Table 6). Th ere w ere n o statistically
h ad sign ifican tly low er d 15 N th an clays in both th e
w e m easu red for total P in san ds versu s clays at th is
sign ifican t pattern s in d 15 N betw een th e su rface an d
depth . Th e average pool size of total P m easu red on
1 m depth an d n o stron g pattern s w ith depth in th e th e clays (398 kg h a -1 ) w as con siderably greater san d or loam soils. In th e clays, th e top 10 cm of
th an th e m odel SOM P ou tpu t (129 kg h a -1 ), an d
in clu sion of paren t, occlu ded, or secon dary P pools deeper soil depth s (Table 6).
m in eral soil w ere sligh tly depleted in d 15 N relative to
in th e m odel do n ot ch an ge th is pattern .
W. L. Silver an d oth ers
Table 4. Forest Floor, Root, an d Total Soil C, N, an d P Pools by Depth in San d an d Clay Soils in th e Tapajos Nation al Forest, Para, Brazil
Forest floor Litter C
Litter N
Litter P
Litter C:N
Litter C:P 1030 (102)
0–10 cm Soil C
Soil N
Soil P
Soil C:N
Fine root C
Fine root N
Fine root P
Coarse root C a 6.00 (3.25)
Coarse root N
Coarse root P
Root C:N
Root C:P 1888 (116.3)
10–40 cm Soil C
Soil N
Soil P 277.5 (5.30)
Soil C:N
Figu re 4. Th e distribu tion of fin e an d coarse root biom ass Total root C
in th e Tapajos Nation al Forest, Brazil. A Th e proportion of Total root N
live an d dead fin e roots in th e su rface soils alon g th e soil Total root P
textu re gradien t. Error bars represen t 61 SE. B Th e Root C:N
distribu tion of live an d dead fin e root biom ass by depth to Root C:P
40 cm in san d an d clay soil types. C Total root biom ass by 40–100 cm
depth in five 3 x 1 x 1–m qu an titative soil pits alon g th e Soil C
soil textu re gradien t. Th e h igh apparen t biom ass at m eter Soil N
50 w as h eavily w eigh ted by a sin gle large an d h eavy root Soil P
th an passed th rou gh all th ree depth s. Soil C:N
Total root C
Th e m odel gen erally estim ated fin e an d coarse Total root P
Total root N
root C con ten t w ith in 20% of th e m easu red valu es, Root C:N
except for overestim atin g fin e root C in th e clays by Root C:P
67% . Th e m odel closely predicted fin e root N an d P 0–100 cm
con ten t in clays, bu t u n derestim ated th ese in th e Total below-
san ds. We m easu red greater coarse root N an d P in ground C
Total below- both clay an d san dy soils th an Cen tu ry predicted,
bu t Cen tu ry accu rately captu red th e tren ds of greater Total below-
ground N
coarse root N an d P in clays th an in san ds (Table 7). ground P
Cen tu ry u n derestim ated th e total below grou n d C (root plu s soil) in th e san ds by 53% . Abovegrou n d
a P 5 0.05. C and N values are in Mg ha 21 ; values for P are in kg ha 21 . Fine roots
biom ass C on san ds w as predicted to be 20% low er
are #2 mm diameter; coarse roots are .2 mm diameter. Asterisks signify statistically significant differences between soil texture types by using a 2 sample
th an on clays, w h ereas th e total forest C pool w as
t-test. Standard errors are in parentheses.
30% low er on san ds th an on clays. Cen tu ry pre- dicted th at total forest C w ou ld in crease from san ds to loam s to clays (Table 7).
Soil Textu re Effects on Below grou n d C in a Tropical Forest
Table 5. Soil pH, Textu re, Exch an geable Cation s, Table 6. Mean Delta 15 N by Depth in Th ree Soil Extractable (P ex ) an d Total P by Depth in Five Large
Pits a alon g a Soil Textu re Gradien t in th e Tapajos Qu an titative Soils Pits a Nation al Forest, Para, Brazil
Loam s Clays Soil
M eters alon g th e Textu re Grad ien t
San d s
M eter 200 M eter 350 Prop erty
Dep th
M eter 0
8.59 (0.49) b 0–10 cm depth
9.08 10.02 8.38 pH (KCl)
9.55 10.11 9.18 pH (H 2 O)
9.57 12.49 12.64 Total P (µg g )
6.58 9.54 Mg (cmol 1
9.62 kg )
Ca (cmol 1 21
6.62 6.72 12.74 K (cmol 1 kg 21 )
9.45 (0.20) a Mn (µg g )
Al (cmol 1 kg 21 )
0.99 1.10 2.57 2.20 3.28 a 3 3 1 3 1 m.
10–40 cm depth
Replicate samples were taken from the 0–2-cm and 90–100-cm depths. For these
pH (KCl) 4.07 4.16 3.93 3.96 3.93 depths, standard errors are in parentheses. Different lowercase letters indicate pH (H 2 O)
4.28 4.18 4.91 4.25 3.91 statistically significant differences between sand and clay by using replicate samples within depths. There were no statistically significant patterns with depth within Sand (% )
75.22 80.48 51.52 44.79 37.12 sites (P , 0.05). nd, no data.
Total P (µg g 21 )
Pex (µg g 21 ) ,
21 Ca (cmol on th e clays th an on th e oth er tw o soil types (Table 1 kg ) 0.01 0.01 0.08 0.00 0.01 Mg (cmol 1 kg 21 )
7). Th ese m odel resu lts agree w ith ou r experim en tal K (cmol 1 kg 21
data th at sh ow h igh er DEA, m in eralization , an d )
m ore en rich ed d N w ith depth on th e clays th an on Al (cmol 1 kg 21 )
1.53 1.27 1.87 2.40 2.31 th e san ds.
40–100 cm depth Dou blin g th e rate of N in pu ts in th e san ds in - pH (KCl)
4.23 4.18 3.96 3.88 3.93 creased th e m odel soil C con ten t from 12.7 to 26.3 pH (H 2 O)
4.21 4.35 4.28 4.50 4.21 Mg h a -1 , w h ereas dou blin g th e P in pu ts in creased Sand (% )
75.27 73.32 50.14 44.91 39.31 th e soil C by on ly 0.4 Mg h a -1 (Table 7). Dou blin g Clay (% )
23.77 22.38 49.86 52.69 61.69 th e fu n ction al rootin g depth from 60 to 120 cm Silt (% )
0.95 4.30 0.00 2.40 0.00 depth also h ad on ly a m in im al im pact on soil C Total P (µg g 21 )
pools in san ds w ith an in crease of on ly 0.5 Mg h a -1 Pex (µg g 21 )
79.42 78.25 128.36 n a
(Table 7).
Ca (cmol 1 kg 21
Mg (cmol 1 kg 21 ) ,
K (cmol 1 kg 21 )
0.01 0.01 0.01 0.01 0.01 D ISCUSSION
Mn (µg g 21 ) ,
Th e Effects of Soil Textu re on Soil
Al (cmol 1
kg 21 )
1.07 1.40 1.50 1.77 2.54 Nu trien t Pools
a 3 3 1 3 1 m.
Th e pool sizes an d distribu tion of soil P an d N varied
Values represent one pooled sampled per depth. Samples differ from those in Table 1. , signifies below detection of the analytical instruments. na 5 not
sign ifican tly alon g th e 400-m soil textu re gradien t
available.
in th is forest. San dy soils h ad m ore extractable P th an th e loam an d clay-rich soils. In w ell-aerated soils su ch as th ese, P is easily com plexed w ith exch an geable Al an d Fe in th e m in eral soil, sh ow n
Th e Cen tu ry m odel also in clu des param eteriza- by th e sign ifican t n egative correlation of exch an ge- tion s of losses du e to leach in g an d volatilization .
able Al an d P alon g th e gradien t as clay con ten t Nitrogen losses gen erally follow ed th e sam e tren ds
in creased. Th ere w as sign ifican tly greater total P as for total SOM N pools w ith h igh er rates of N
w ith in creasin g clay con ten t alon g th e gradien t, leach in g on th e clays th an on th e loam s or san ds
probably du e to th e form ation of Fe an d Al ph os- (data n ot sh ow n ), an d h igh er rates of trace gas loss
ph ates an d organ ically bou n d P, n eith er of w h ich
W. L. Silver an d oth ers
Table 7. Cen tu ry Model Sim u lation an d Measu red Valu es of th e Below grou n d C, N, an d P Pools an d N Trace Gas Flu x in th e Tapajos Nation al Forest, Brazil
Clays Pools or
San d s
San d y Loam s
Flu xes
CENTURY M easu red Soil organ ic m atter C
CENTURY
M easu red
CENTURY
M easu red
Mg h a 21 12.7 22.9 24.0 31.4 31.2 29.8 Soil C 2 3 N
na na Soil C 2 3 P
Mg h a 21 26.3 na
na
na
na na Soil C 2 3 w ater
Mg h a 21 13.1 na
na
na
na na Soil organ ic m atter N
Mg h a 21 13.1 na
na
na
Mg h a 21 0.96 1.70 2.12 2.44 2.90 2.79 Soil organ ic m atter P
129.1 5; 398 Fin e root C
kg h a 21 44.1 5; 148
2.24 1.34 Fin e root N
Mg h a 21 1.78 1.94 2.11 na
0.03 0.03 Fin e root P
Mg h a 21 0.02 0.08 0.03 na
1.06 1.01 Coarse root C
kg h a 21 0.84 2.07 0.99 na
10.9 8.89 Coarse root N
Mg h a 21 8.63 10.63 10.2 na
0.07 0.24 Coarse root P
Mg h a 21 0.06 0.16 0.07 na
1.58 6.23 Below grou n d C
kg h a 21 1.25 4.72 1.48 na
44.3 40.0 Abovegrou n d C
Mg h a 21 23.1 35.5 36.3 na
88.6 na Total forest C
Mg h a 21 69.8 na
83.0 na
132.9 na Trace gas loss
Mg h a 21 92.9 na
na
kg N h a 21 y 21 1.07 na
1.31 na
1.42 na