Pedogenic Thresholds and Soil Process Domains in Basalt-Derived Soils

17 Pages·2013·0.88 MB·English

Ecosystems(2013)16:1379–1395
DOI:10.1007/s10021-013-9690-z
(cid:2)2013TheAuthor(s).ThisarticleispublishedwithopenaccessatSpringerlink.com
Pedogenic Thresholds and Soil
Process Domains in Basalt-Derived
Soils
Peter M. Vitousek1* and Oliver A. Chadwick2
1DepartmentofBiology,StanfordUniversity,Stanford,California94305,USA;2DepartmentofGeography,UniversityofCalifornia,
SantaBarbara,California93106,USA
ABSTRACT
Pedogenic thresholds occur where soil properties and wind erosion (dry young substrate); by
change abruptly and/or nonlinearly with a small weathering and biological uplift of nutrients
increment in environmental forcing; soil process (intermediate rainfall young substrate and dry old
domainsaretheregionsbetweenthresholdswhere substrate); by surface Fe enrichment and nutrient
soils change much more gradually across a large depletion (wet young substrate and intermediate
range of environmental forcing. We evaluated rainfall old substrate); and by Fe mobilization and
thresholds and domains in basalt-derived soils on loss(wetoldsubstrate).Soilsontheoldersubstrate
two rainfall gradients in Hawaii—one from 260 to were more highly weathered, lower in total and
3,540 mm/y precipitation on 150,000-year-old available P, and characterized by more crystalline
substrate, the other from 600 to 3,760 mm/y on clays than otherwise comparable soils on the
4,100,000-year-old substrate. We identified younger substrate. Prior to European contact,
thresholds associated with the initiation of biolog- Hawaiian cultivators developed an intensive rain-
ical uplift of nutrients at about 700 mm/y on the fedagriculturalsystemintheweathering/biological
younger substrate, the depletion of primary min- upliftdomainontheyoungersubstrate;wesuggest
erals at about 2,100 mm/y on the younger and that only this domain could support indigenous
about 900 mm/y on the older substrate, and the agricultural intensification in upland soils.
initiation of anoxic conditions and associated Fe
mobility at about 2,500 mm/y on the older sub- Key words: basalt-derived soil; biological uplift;
strate.Thesethresholdsdelineatedprocessdomains gradient; Hawaii; indigenous agriculture; pedo-
characterized by pedogenic carbonateaccumulation genic threshold; process domain; weathering.
INTRODUCTION
Soil properties and processes may not change lin-
early in response to extrinsic differences in envi-
ronmental forcing such as rainfall; rather their
Received1April2013;accepted28May2013;
publishedonline29June2013 response may be characterized by ‘‘pedogenic
Electronicsupplementarymaterial:Theonlineversionofthisarticle thresholds’’thatrepresentabruptand/ornonlinear
(doi:10.1007/s10021-013-9690-z) contains supplementary material, (but predictable) changes in soil properties and
whichisavailabletoauthorizedusers.
processes (Muhs 1984; Chadwick and Chorover
AuthorContributions:PMVandOACdesignedtheresearch,carried
2001). For example, along a rainfall gradient from
outtheresearchandperformedtheanalyses,andwrotethepaper.
*Correspondingauthor;e-mail:[email protected] 180 to 3,000 mm/y in Kohala, Hawaii, Chadwick
1379
1380 P. M. Vitousek and O. A. Chadwick
and others (2003) demonstrated that base (non- the wetter, arid fringes of the desert. Additional
hydrolyzing) cation saturation of soil exchange thresholds exist along regional scale ecotones such
sites dropped from about 80 to 5% between 1,300 as transitions from tundra to boreal forest (Ugolini
and 1,500 mm of rainfall. This threshold in base and Spaltenstein 1992).
cation saturation was associated with a precipitous Where pedogenic thresholds are important, we
decline in pH and effective cation retention capac- suggest that it would be rewarding to evaluate soil
ity; acid neutralizing capacity (ANC) also de- properties and processes in the regions between
creased, signaling the irreversibility of this change. thresholds, which we term ‘‘soil process domains.’’
Thesechangeswerecausedbydepletionofprimary Although pedogenic thresholds represent locations
minerals in higher rainfall sites, so that mineral wheresoilpropertiesorprocesseschangemarkedly
weathering could not buffer atmospheric and bio- and/or nonlinearly in response to a small incre-
logicalaciditythere.Insupportofthisexplanation, ment in forcing, soil process domains conversely
Srisotopes(87Sr/86Sr)alsochangedabruptlyatthis represent regions in which soil properties or pro-
threshold, from values characteristic of Hawaiian cesses change much more slowly across a wide
basaltindriersitestothosecharacteristicofmarine rangeofforcing.Inthispaper,weusealargesetof
aerosol in wetter sites (Stewart and others 2001; surface soils, in addition to information from well-
Chadwick and others 2003). characterized soil profiles, to evaluate the preva-
Chadwick and Chorover (2001) identified a sec- lence of pedogenic thresholds alongtwo broad and
ondthresholdinhigherrainfallHawaiiansoilsona continuous gradients of rainfall in basalt-derived
separate gradient; this threshold was associated soils of the Hawaiian Archipelago—one developed
with anaerobiosis and consequent iron reduction inparentmaterialaround150 kyearsold,theother
and mobility in high-rainfall sites. Subsequent re- in older than 4,000 kyear parent material. Where
search there elucidated interactions among iron pedogenic thresholds occur, we evaluate whether
reduction, phosphorus retention, and organic consistent thresholds can be detected on the two
matterdynamicsthatchangedacrossthisthreshold substrate ages—and if so, whether the position of
(Miller and others 2001; Thompson and others these thresholds changes with substrate age.
2011). Chadwick and Chorover (2001) suggested Additionally, we summarize the dominant soil
thattheseandotherpedogenicthresholdscouldbe properties and processes within the soil process
expressed in substantial changes in bulk soil domains that occupy the regions between thresh-
parameters such as mineral crystallinity, nutrient olds. Finally, we explore the significance of these
availability, organic matter retention and release, thresholds and domains for human use of these
and colloidal dynamics. lands.
A strong case for the importance of pedogenic
thresholdshasbeenmadeintheHawaiianIslands—in
part because the Islands support a climate-time METHODS
matriximposedonverysimilarparentmaterialthat
Sites
make thresholds easier to detect, and in part be-
cause rapid weathering of basalt-derived primary TheHawaiianIslandscanbeusedtodeterminethe
minerals and the persistence of metastable and patterns and controls of soil properties and pro-
chemically reactive secondary aluminosilicates and cesses because many of the factors that influence
metal oxyhydroxides there makes pedogenic soils can be held constant there to a much greater
thresholds sharper and more distinct. However, extent than is possible in most continental settings
pedogenic thresholds have been identified in con- (Chadwick and others 2003; Vitousek 2004). The
tinental systems, and on lithologies other than ba- parent material in which soils form is basalt, vary-
salt (Muhs 1984). For example, Dahlgren and ingfromtholeiiticbasalteruptedduringtheshield-
others (1997) analyzed soil properties along a cli- building stage of a volcano’s evolution to alkalic
mate gradient in the Sierra Nevada Range and basalt produced later in its life (MacDonald and
identified a relatively narrow zone where primary others 1983). The constructional surface of shield
mineral weathering and clay mineral formation volcanoessupportslittletopographicvariation,and
were enhanced greatly by a favorable combination remnants of constructional surfaces can be identi-
of temperature and rainfall. Similarly, Ewing and fied on the oldest high islands. Even the dominant
others (2006) demonstrated a strong shift from organisms are (relatively) consistent; the few plant
accumulation of atmospheric salts in the hyperarid species that dispersed naturally to Hawaii have
core of the Atacama Desert to in situ mineral radiated to occupy a very wide range of environ-
weathering and pedogenic carbonate formation on ments. On the other hand, substrate age and
Soil Thresholds and Process Domains 1381
climate vary widely, but generally in well-defined inuplandslopepositionsacrossarainfallgradienton
and continuous ways. Lava flow and other surface the Island of Kauai (substrate age (cid:2) 4,100,000 y),
ages are well constrained, and vary nearly contin- witharainfallrangefrom600to3,760 mm/y.Fiveof
uously from recent deposits to nearly 5 million thesamplesinthecenterofthisgradientwereneara
years from southeast to northwest across the canyon rim, and had evidence of erosion; we based
archipelago. Temperature varies with elevation, ouranalysisontheother28.
and current precipitation varies predictably from These 30 cm samples allow extensive spatial
less than 250 to greater than 10,000 mm/y. These analyses, but exclude deeper soil properties and
features are well suited to the detection of abrupt, feedbacks between surface and subsurface pro-
nonlinear transitions in soil properties along con- cesses. Accordingly, we drew upon information
tinuous gradients of substrate age or precipitation. from complete soil profiles that had been charac-
Although climate has varied through the history terized, sampled by horizon, and analyzed in 13
ofthearchipelago(Hotchkissandothers2000)and sites along the rainfall gradient on Hawi substrate
Hawaiian soils carry the imprint of past as well as (Chadwick and others 2003); soil pits were exca-
present climate (they are polygenetic, in the sense vated to the depth of unweathered bedrock (about
of Richter and Yaalon 2012), the enduring char- 1 m in low to moderate rainfall areas, but
acteristics determining rainfall patterns in Hawaii increasing to nearly 5 m at 2,500 mm rainfall), ei-
are the interaction of the Northeast trade winds ther by hand or using a backhoe in several sites
with island topography. The direction of the trade (Porder and others 2007; Porder and Chadwick
winds has been consistent for several million years 2009). On Kauai, we sampled and characterized
(Porter 1979), and consequently dry sites on deep soil profiles in 6 sites along the rainfall gra-
modern rainfall gradients have long received less dient; soils were sampled from hand-excavated
rainfall than wet sites, and older sites have accu- pits, supplemented by deeper sampling (to 4 m
mulatedmorerainfall(andleaching)thanyounger depth) with an auger, and in one case sampling to
sites (Hotchkiss and others 2000). 17 m from a cliff face exposed by recent erosion.
In this analysis, we evaluate soils sampled along Soils on the dry extreme of the Hawi gradient
precipitationgradientsontheIslandsofHawaii(the classified as either Typic Haplotorrands or Sodic
youngest island in the archipelago) and Kauai (the Haplocambids; soils of intermediate rainfall sites
oldest high island). Two distinct types of soil sam- were Andic Haplustands, Humic Haplustands, and
ples were included in our analysis—numerous Pachic Haplustands, giving way to Hydric Fulvud-
depth-integrated surface samples to 30 cm depth, ands and ultimately Hydric Pachic Placudands and
andwholesoil-profilesamples.Initially,thesurface Alic Epiaquands in the wettest sites (Sato and
samples were collected on the Island of Hawaii as others 1973, Chadwick and others 2003). Kauai
part of an analysis of the distribution and limits of soils ranged from Haplotorrox, Typic Eutrotorrox,
indigenousHawaiianagriculturepriortoEuropean and Tropeptic Haplustox in the driest sites through
contact (Vitousek and others 2004; Palmer and Haplustox and Plinthic Acrorthox to Typic Gib-
others 2009); equivalent samples later were col- bsihumoxandultimatelyaHumicEpiaqueptinthe
lectedontheIslandofKauaiforcomparison.These highest rainfall site (Foote 1972).
sampleswerecollectedalongtransectsusingatiling
shovel, normally at around 200-m intervals within
Soil Analyses
or near ancient agricultural systems and 500-m
intervals away from them; GPS coordinates were Soils were air-dried, passed through a 2 mm sieve,
obtained for each sample,and rainfall at that point and divided into three homogenous subsamples.
wasobtainedfromtheonlineHawaiiRainfallAtlas All of the 30 cm samples were analyzed as de-
(Giambelluca and others 2012). For this analysis, scribed in the supplemental material to Vitousek
we use only sites in upland slope positions in sites and others (2004). Briefly, one subsample was
without clear evidence of erosion or deposition. analyzed for resin-extractable phosphorus (P) and
Soils were collected as a continuous sample from for total carbon (C) and nitrogen (N) at Stanford
thesurfaceto30 cmonmultipletransectsacrossthe University; P was determined using the anion-ex-
rainfall gradient on Kohala Volcano, the oldest vol- change resin method of Kuo (1996) and analyzed
cano on the Island of Hawaii. Here, we include 160 on an Alpkem RFA/2 AutoAnalyzer, whereas total
samples from across a rainfall range from 260 to CandNwereanalyzedusingaCarloErbaNA1500
3,540 mm/y on the Hawi Volcanic Formation (sub- elemental analyzer. A second subsample was ana-
strateage (cid:2) 150,000 y)(SpenglerandGarcia1988). lyzed for pH and cation exchange capacity (CEC)
Inaddition,wecollected33integrated30 cmsamples and exchangeable Ca, Mg, Na, and K at the
1382 P. M. Vitousek and O. A. Chadwick
University of California, Santa Barbara, using the ER ¼100 (cid:3) ðX =Nb Þ=ðX =Nb Þ;
x s s pm pm
NH OAc method at pH 7.0 (Lavkulich 1981). We
4
where ER is the percentage of element X
measured CEC buffered at pH 7 and used it in the x
remaining in the soil, relative to its abundance in
calculation of base saturation as a way to stan-
parent material; X and X are the concentrations
dardize our measurements among mineralogically s pm
of element X in the soil and parent material,
diverse soils. When CEC and effective cation ex-
respectively, and Nb and Nb are the concentra-
change capacity (unbuffered ECEC) are compared s pm
tionsofNbinsoilandparentmaterial.Weusedthe
along the Hawi climate gradient, CEC increases
average of 12 rock samples to characterize parent
linearly with increasing rainfall whereas ECEC in-
material on Kohala (Chadwick and others 2003),
creases up to a point where soluble Al begins to
andanaveragevalueforHawaiianalkalicbasaltfor
outcompete the base cations for exchange sites
Kauai (Chadwick and others 1999).
(Chadwick and others 2003). By calculating base
Onechallengewiththisapproachisthatitcanbe
saturation using CEC, we assume that the inverse
difficult to define the composition of the parent
of % base saturation is % acid saturation, which is
composedofAl3+(anditsdivalentandmonovalent material, becausesoilmaterial can be derived from
counterparts) and H+ (Ross and others 2008). The a mixture of sources in addition to the underlying
rock substrate. In Hawaii, this additional material
third subsample was shipped to ALS Chemex
can include volcanic tephra from younger erup-
(Sparks, Nevada, USA) and analyzed for total con-
tions in the archipelago, the deposition of material
centrations of 11 elements (Si, Al, Fe, Ca, Mg, Na,
that was eroded nearby by water or especially
K, Ti, P, Nb, Zr) using lithium borate fusion fol-
wind, and the long-distance transport of dust from
lowed by X-ray fluorescence spectrometry. The
continental sources (Kurtz and others 2001). Te-
depth profile samples were analyzed similarly, ex-
phra from other volcanoes or eruptive phases is
cept that total N and resin P measurements were
likelytobe most important asa source of variation
not made on all samples.
in parent materialon a volcanicarchipelago, while
Calculations the eolian deposition of material from very differ-
ent sources is much greater in most continental
We summarized results of horizon-based soil pro- situations than it is in Hawaii (Simonson 1995).
file sampling on the Hawi substrate by calculating The relative contribution of these sources can vary
integrated element concentrations and other soil at different positions on a single rainfall gradient.
properties on a depth- and bulk density-weighted We estimated the contribution of two distinct
basis for the entire profile from the surface to the eruptive phases and continental dust to surface
top of C horizon (generally hard rock or saprolite). soils on the Hawi gradient to illustrate the chal-
No Kauai profiles (even the 17 m deep one) lenges involved (Electronic supplementary mate-
reachedtotheChorizon,accordinglywecalculated rial); the strong and well-characterized distinction
integrated soil element concentrations and other between the widespread alkalic Hawi Formation
propertiesonadepth-weightedbasistoaconsistent and tholeiitic Pololu Formation substrates on Ko-
depth of 1.3 m—the shallowest profile sampled on halaVolcano(SpenglerandGarcia1988;Wolfeand
Kauai. Morris 1996) makes this calculation feasible. A
Element concentrations alone can give a mis- similar calculation for the parent material under-
leadingpictureofelementmobilityandloss,inthat lying the Kauai gradient would be less easy to
hydrationandaddedorganicmatterineffectinflate justify. Accordingly, we focus on soil properties
the denominator of any concentration, whereas rather than provenance here, although we make
loss of more mobile elements reduces the denom- use of calculations of elements remaining from
inator for concentrations of less-mobile elements. parent material where they are useful, and we
(The volume of soil can be altered similarly.) To discuss the likely significance of mixed parent
correct for these changes, we used Nb as an index materials as appropriate.
element to calculate the accumulation or loss of
elements in soils. Although Zr is commonly used
for this purpose elsewhere, earlier research dem-
RESULTS
onstrated that Nb (with Ta) is the least mobile
elementinHawaiiansoils(Kurtzandothers2000). We consider multiple soil properties in this analy-
Accordingly, we calculated element gain or loss sis, including: (1) concentrations of the abundant
following Chadwick and others (1990) and Brim- elements Al, Fe, and Si, which together make up
hall and others (1992), using nearly 80% of the initial parent material (when
Soil Thresholds and Process Domains 1383
Figure1. Concentrationsofrock-derivedelementsinsoilson150,000-year-oldHawisubstrate.AConcentrationsofthe
abundantelementsSi,Al,andFein thesurface 30cm ofsoil.BConcentrations oftherelativelyimmobileelementsNb
andTiinsurfacesoil.CConcentrationsofthebiologicalnutrientsCaandPinsurfacesoil.DConcentrationsofSi,Al,and
Fecalculatedonadepth-andbulkdensity-integratedbasisfromthesurfacetothetopoftheChorizon.ENbandTifrom
surface toC. FCaandPconcentrations fromsurface to C.
calculated as oxides); (2) concentrations of the soil properties from the surface 30 cm, and com-
immobileelementsNbandTi;(3)concentrationsof parethesesurfacepropertieswithintegratedvalues
the biological nutrients Ca and P, to illustrate the fromthesurfacetothetopoftheChorizonforthe
dynamics of relatively mobile and relatively Hawi gradient, or to 1.3 m depth for Kauai.
immobile biologically essential elements, respec-
tively; and (4) a set of more dynamic measures of
Kohala-Hawi Rainfall Gradient
soil properties relating to soil fertility and nutrient
availability, including pH, base saturation, resin- Concentrationsoftherock-derivedelementsSi,Al,
extractable P, the ratio of exchangeable Ca to Fe, Ti, Nb, Ca, and P across the Hawi rainfall gra-
exchangeable Na, soil C, and C:N ratio. We report dientaresummarizedinFigure 1;thepanelstothe
1384 P. M. Vitousek and O. A. Chadwick
there, as illustrated in Figure S1 in Electronic sup-
plementary material.
Concentrations of the biologically important
elements Ca and P follow a different pattern. Both
areintermediateinconcentrationinthedriestsites,
higher (and more variable) in concentration in
intermediate rainfall sites, and low in concentra-
tion inthe wettest sites (Figure 1C). This pattern is
not an artifact of the accumulation or mobility of
other elements; the overall pattern of enrichment
atintermediaterainfallissimilarwhenCaandPare
expressed in terms of the percentage of element
remaining from parent material (Figure 2A). In
fact, there is an absolute increase in P (relative to
Nb, and so relative to parent material) in many
surface soils in this intermediate rainfall zone.
The patterns of variation in multiple element
concentrations in surface soils (to 30 cm depth)
generally are not reflected in element concentra-
Figure2. Elements in the upper 30cm of soil (A) and tionsintegratedthroughsoilprofiles(Figure 1D–F),
integratedfromthesurfacetotopoftheChorizon(B)as though there is a progressive decrease in Si and
a percentage of the quantity of each element in parent particularly Ca concentrations in profiles from
material, for sites on the Hawi rainfall gradient. Calcu- wettersites. Correctingforthegain or lossof other
lated as described in the text, using Nb as an immobile
material using Nb as an index element, there has
index element and the mixture of parent material de-
beenlittle gain orlossofAlorFealong therainfall
scribedin theelectronic supplementary material.
gradient from the profiles as a whole (Figure 2B);
the differences observed in surface soils reflect
left show results for the upper 30 cm of soil from withinprofiletranslocationoftheseelementsrather
160 sites (Palmer and others 2009; Kagawa and thanabsoluteloss.ThemoremobileelementsSiand
Vitousek 2012), whereas the panels to the right Ca—and surprisingly P—have been lost from the
show results integrated from the surface to the top profiles,especially(forSiandCa)inthewettestsites.
of the C horizon from 13 deep soil pits along the Enriched P in surface soils (Figures 1C, 2A) does
gradient (Chadwick and others 2003; Porder and not carry over to the profiles as a whole (Fig-
Chadwick 2009). Concentrations of Si, Al, and Fe ures 1F, 2B), and observations of the depth distri-
in surface soils decrease with increasing rainfall bution of P in soils show that it is depleted at
fromthedriestsitestoapproximately1,700 mm/y; intermediate depth as well as enriched at the sur-
thereafter, Fe concentrations increase whereas Si face soil in sites with intermediate rainfall (Fig-
and Al concentrations continue to decline, sharply ure 3), as Vitousek and others (2004) and Porder
so for Al in the interval from 1,700 to 2,100 mm and Chadwick (2009) observed. This pattern of
(Figure 1A). Correcting for changes in soil mass surface enrichment reflects biological uplift from
andvolume using Nbasan indexelement,there is deeper in the soil to the surface (Jobbagy and
netlossofFeaswellasSiandAlfromsurfacesoils Jackson 2001), over the tens of thousands of years
above 1,700 mm annual rainfall (Figure 2A); the that these sites havebeenvegetated.Drier sites are
increased Fe concentration there (Figure 1A) re- either too unproductive to lift up much P, or any
flects loss of Al (and other mobile elements) rather enriched layer has been removed by wind erosion;
thananyabsolutegaininFe.Concentrationsofthe in wetter sites, intense leaching prevents the
immobile elements Nb andTiinitiallydecline from maintenance of an enriched surface layer.
dry sites toaround 1,500 mm/y, reflecting dilution More dynamic soil properties (pH, base satura-
by hydration of soil minerals and increasing soil tion, exchangeable Ca, resin-extractable P, soil C
organicmatterconcentrations.Above1,500 mm/y, andC:Nratios)alsovarysubstantiallyinsurfacesoils
Nb and Ti concentrations increase substantially, acrosstheHawigradient(Figure 4).Basesaturation
reflectingthelossofmoremobileelementsinthese andpHarevariablebutrelativelyhighatlowrain-
sites (Figure 1B). The relatively greater concentra- fall, decline (remarkably consistently in the case of
tionsofTicomparedtoNbindriersites(Figure 1B) base saturation) to about 2,100 mm/y, and then
reflect an admixture of tholeiitic Pololu substrate remainlowtothewettestsitessampled(Figure 4A).
Soil Thresholds and Process Domains 1385
Figure3. DepthprofilesofthepercentageofPremainingfromparentmaterial(calculatedusingNbasanimmobileindex
element)for5sitesarrayedalongtherainfallgradientonHawisubstrate.A227mm/y.B722mm/y.C1,323mm/y.D
1,745mm/y. E2,397mm/y.
Resin-extractable P (an index of biologically avail- 1,200 mm/yprecipitation(Figure 4B).Theseresults
able P) has a distinct hump-shaped pattern, with reinforcethesuggestionderivedfromoverallCaand
intermediate concentrations in surface soils in dry Pconcentrationsthatthereisaverystrongsignalof
sites, variable but often extremely high concentra- biological uplift—probably as a legacy of the sea-
tionsinintermediaterainfallsites(withapeaknear sonally dry forests that occupied these sites before
1,200 mm/y),and very low concentrations inwet- humansarrived(Chadwickandothers2007)—that
ter sites (Figure 4B). The ratio of exchangeable Ca shapes P and Ca pools and biological availability in
(which is strongly accumulated biologically) to surface soils in this intermediate rainfall zone.
exchangeable Na (which is not) also has a hump- Concentrations of C in surface soil increase consis-
shapedpatternalongtherainfallgradient.Although tently from 10 to over 100 g/kg up to about
CaandNaarepresentinHawibasaltataratioof1.05 1,700 mm/y precipitation; at higher rainfall and
(by mass), and in marine aerosol at a ratio of higher elevation, they average over 100 g/kg (but
less than 0.04, the ratio exchangeable Ca to arehighlyvariable)(Figure 4C).SoilC:Nratiosare
exchangeableNainthetop30 cmofsoiliselevated strikinglyvariableatlowrainfall(from4to16),and
substantiallybetween750and2,100 mm/yprecip- consistentlyaround12–13insurfacesoilsinwetter
itation, peaking at an average >40 (by mass) near sites(Figure 4C).
1386 P. M. Vitousek and O. A. Chadwick
Figure4. Selectedsoilpropertiesonarainfallgradienton150,000-year-oldHawisubstrate.ApHandbasesaturationin
thesurface30cmofsoil.BResin-extractablePandtheratioofexchangeableCatoexchangeableNainsurfacesoil.CSoil
CandC:Nratioinsurfacesoils.DBasesaturationintegratedfromthesurfacetothetopoftheChorizon.EExchangeable
CatoNa, integrated fromthe surface to thetopof theChorizon.
Base saturation and exchangeable Ca:Na ratios gradientbyChadwickandothers(2003),nowshif-
vary similarly along the rainfall gradient in whole tedtoahigherrainfallvaluebecausethenewrain-
soilprofiles(integratedfromthesurfacetothetopof fallatlasforHawaii(Giambellucaandothers2012)
the C horizon) as they do in the top 30 cm of soil. reportshigherrainfallinthistransitionregionthan
Thetransitionforbasesaturationinwholeprofilesis did the previous atlas (Giambelluca and others
sharper than that in surface soils; base saturation 1986)usedbyChadwickandothers(2003).
decreases from close to 100% to near zero by
2,100 mm/y, whereas exchangeable Ca:Na ratios
Kauai Rainfall Gradient
are elevated to similar levels in the intermediate
rainfall zone in both surface and profile samples Concentrationsoftherock-derivedelementsSi,Al,
(Figure 4D, E). The transition in base saturation Fe, Ti, Nb, Ca, and P across the Kauai rainfall gra-
represents the threshold described for this rainfall dientaresummarizedinFigure 5;thepanelstothe
Soil Thresholds and Process Domains 1387
Figure5. Concentrationsofrock-derivedelementsinsoilsonapproximately4,100,000-year-oldsubstrateontheIslandof
Kauai. A Concentrations of the abundant elements Si, Al, and Fe in the surface 30cm of soil. B Concentrations of the
relativelyimmobileelementsNbandTiinsurfacesoil.CConcentrationsofthebiologicalnutrientsCaandPinsurfacesoil.
D Concentrations of Si, Al, and Fe, integrated from the surface to around 1.3m depth. E Concentrations of Nb and Ti,
integratedfromsurface to 1.3m. FConcentrations of CaandP, integratedfromsurface to1.3 m.
left show results from 28 samples of the top 30 cm mostKauaisites,higherthanthosealongtheHawi
ofsoil,whereasthepanelstotherightshowresults gradient (Figures 5A, 1A). However, Fe drops to
integrated from the surface to a depth of 1.3 m very low concentrations in the wettest Kauai site.
from 6 soil profiles along the gradient. Concentra- The increase in Al concentration in wetter sites
tions of Si in surface soil are relatively constant occurs because the loss of Fe leads to residual
across the gradient, and substantially lower in the enrichment of Al. Concentrations of the relatively
drier Kauai sites than the drier Hawi sites (Fig- immobile elements NbandTiincreasemarkedlyin
ures 5A, 1A). Concentrations of Al decline with higher rainfall surface soils (Figure 5B), reflecting
increasing rainfall to around 2,500 mm/y, then residual enrichment associated with the loss of
increase slightly; concentrations of Fe are high in most other elements.
1388 P. M. Vitousek and O. A. Chadwick
ConcentrationsofCaarehighestinthetwodriest
Kauai sites, and then decline to very low concen-
trations—buteventhehighestCaconcentrationson
Kauai are little more than a tenth of those on the
youngerHawisubstrates(Figures 5C,1C).Concen-
trationsofParerelativelyconstantacrosstheKauai
gradientandagainsubstantiallylowerontheKauai
than the Hawi gradient (Figures 5C, 1C). The frac-
tion of Ca and P remaining from parent material
(Figure 6A) also is substantially less on Kauai than
on Hawi substrates, at all rainfall levels along the
gradient.
The concentrations of most elements integrated
from the soil surface to 1.3 m depth vary substan-
tially (if often not linearly) across the Kauai gra-
dient (Figure 5D–F); in contrast to the Hawi
gradient(Figure 1D–F),manyofthechangesalong
the Kauai gradient represent loss from the soil
profile rather than translocation within it. In
particular, concentrations of Fe are enriched in
intermediate rainfall sites, reflecting residual
enrichment from the whole-profile loss of Al and
Si;inthewettestsite,Feismobilizedandlostfrom Figure6. The percentage of elements remaining from
the profile as a whole, causing a relative enrich- parent material across the Kauai rainfall gradient, cal-
ment in Al and Si (compared to intermediate culated using Nb as an immobile index element. A Ele-
rainfallsites)(Figure 5D).Inbothcases,thereisan ments remaining in the top 30cm of the soil profile. B
absolute loss of elements (in comparison to their Elements remainingto adepth of1.3m.
abundance in parent material, calculated using Nb
as in index element) moving from drier to wetter More dynamic soil properties (pH, base satura-
sites (Figure 6B). tion, exchangeable Ca, resin-extractable P, soil C,
The element remaining calculation (Figure 6) and C:N ratios) across the Kauai gradient are
yields the expected pattern of decreasing retention summarized in Figure 7. All of these properties
of multiple elements with increasing rainfall along vary substantially in surface soils. Base saturation
the gradient—but the absolute values are too high and pH are relatively high in the driest sites; base
for the less-mobile elements in some of the whole saturation declines rapidly to around 10% around
profiles from drier sites (Figure 6B), with values 900 mm/y precipitation and pH declines below 4.5
reachingover200%forAl,Fe,andTi.Onepossible byaround1,500 mm/y(Figure 7A).Thesedeclines
explanation for the high values for less-mobile occuratlowerrainfallthandocomparabledeclines
elements in the drier profiles is that the Kauai on Hawi substrate (Figure 4A). The ratio of
parent material represents a mixture of sources, as exchangeable Ca to exchangeable Na varies simi-
described for the Hawi gradient in the Electronic larly to base saturation,reaching very low levelsat
supplementarymaterial.Theinitialparentmaterial rainfall greater than 900 mm (Figure 7B). Resin P
Nb concentration we use for Kauai—24 ppm—is concentrations are much lower everywhere on the
appropriate for alkalic basalt, which certainly was Kauai gradient than on the Hawi gradient (peak
produced in later phases of Kauai’s eruptive his- concentrations < 2 mg/kg on Kauai versus
tory, and it appears to be about the right value for >500 mg/kg on Hawi substrate—Figure 7B, 4B).
surface soils (Figure 6A), where later eruptions Soil C concentrations vary across a similar range
could have deposited alkalic tephra on older sur- and in a similar pattern on the Kauai and Hawi
faces. However, the bulk of the profiles (at least in gradients, although the wettest Kauai site has a
the dry sites) may represent tholeiitic basalt from higher soil C concentration than any Kohala site
an earlier eruptive phase. Tholeiitic basalt typically (Figures 7C, 4C). However, soil C:N ratios are
has Nb concentrations near 10 ppm; using this higher in surface soils on Kauai, especially in the
value for parent material Nb would yield near the wetter sites—reaching a surprisingly high ratio of
expected value of 100% retention of less-mobile 55inthewettestsite(Figures 7C,4C).Neitherbase
elements in the driest Kauai sites. saturation nor the exchangeable Ca:Na ratio varies