Weed dynamics on Amazonian Dark Earth and adjacent soils of Brazil

12 Pages·2005·0.29 MB·English

Agriculture,EcosystemsandEnvironment111(2005)1–12
www.elsevier.com/locate/agee
Weed dynamics on Amazonian Dark Earth
and adjacent soils of Brazil
Julie Majora,*, Antonio DiTommasoa, Johannes Lehmanna, Newton P.S. Falca˜ob
aDepartmentofCropandSoilSciences,CornellUniversity,Ithaca,NY14853,USA
bCoordenac¸a˜odePesquisasemCieˆnciasAgronoˆmicas,InstitutoNacionaldePesquisasdaAmazoˆnia(INPA),
69060-001Manaus,Amazonas,Brazil
Received27May2004;receivedinrevisedform21April2005;accepted27April2005
Abstract
Field trials were conducted on Amazonian Dark Earth soils in the Manaus region, Amazonas, Brazil to assess the
composition and impact of weedy vegetation on maize yield. Soil fertility among the Dark Earth varied considerably with
differenceslargelyattributabletopast-usehistory.Consequently,maizeyieldandweedpressurevariedamongfieldlocations,
reflectingthesedifferencesinsoilfertilityinadditiontodifferencesinweedreservoirssuchasseedbanks.Maizeyieldinweeded
plotswasasmuchas63timesgreateronDarkEarth(0.55tha(cid:1)1)thanoncorrespondingadjacentsoil(0tha(cid:1)1),andlocation
averagesvariedfrom0to3.15tha(cid:1)1forDarkEarth.Thepercentagegroundcoverofweedsinweedyplotswasupto45times
greateronDarkEarth(65–99%)thanoncorrespondingadjacentsoil(2–89%),andspeciesrichnesswasupto11timesgreateron
DarkEarth(4–14species)thancorrespondingadjacentsoil(1–8species).Therelativeproportionofannualandleguminous
weedswas32and17%greater,respectively,onDarkEarththanadjacentsoil,andvegetativesproutingofplantswasmore
commononsitesthathadbeenusedlessintensivelyinthepast.Ingeneral,asimilarweedcommunitywasobservedonthe
differentDarkEarthsites,includingmanyspeciestypicallyassociatedwithenvironmentsthathavebeenhighlydisturbedby
human activities, such as Cyperus spp., Phyllantus niuri, and Croton lobatus. Seedlings from a greater number of species
emergedfromforestedDarkEarthseedbanks(2.1perflat)thanfromforestedadjacentsoil(1.2perflat).Thetotalnumberof
emergedseedlingswasgreaterforDarkEarthseedbanks(9.1perflat,1,365m(cid:1)2)thanadjacentsoil(2.2perflat,330m(cid:1)2),
however thespeciesobservedwere notlikelytobe problematic forcropping.
#2005Elsevier B.V.Allrights reserved.
Keywords: Derrisamazonica;Seedbanks;Soilfertility;TerrapretadeI´ndio;Tropicalsoil;Zeamays
1. Introduction
* Correspondingauthorat:DepartmentofCropandSoilSciences,
Amazonian Dark Earth (DE; locally called Terra
918 Bradfield Hall, Cornell University, Ithaca, NY 14853, USA.
Preta de I´ndio) patches occur in upland (i.e. non-
Tel.:[phone];fax:[phone].
E-mailaddress:[email protected](J.Major). flooded) environments throughout the Amazon
0167-8809/$–seefrontmatter#2005ElsevierB.V.Allrightsreserved.
doi:10.1016/j.agee.2005.04.019
2 J.Majoretal./Agriculture,EcosystemsandEnvironment111(2005)1–12
Basin, and result from soil formation processes that 2. Materials and methods
occurred in pre-Columbian times in areas settled by
indigenous people. In general, the significantly Field trials were conducted during the first half of
higher organic matter and phosphorus (P) contents, 2003. Maize plantings were established at four
increased cation exchange capacity and pH, and locations near Manaus, Amazonas, Brazil (Fig. 1).
loweraluminum(Al)levels(KernandKa¨mpf,1989; Each location consisted of one planting on DE and
Lehmann et al., 2003) allow more intensive another on AS, for a total of eight plantings. All
agricultural uses of DE soils than adjacent soils planting sites were under secondary forest regrowth
(AS). Based on the Brazilian soil classification and not cropped when first visited in August 2002.
system, these adjacent soils are considered to be Paired plantings at a given location were established
primarily nutrient-impoverished, acidic Latossolos approximately200mapartfromeachotherexceptat
(SalgadoVieira,1988),whichcorrespondtoOxisols RP, where this distance was 5km. In the absence of
and Ultisols in the USDA soil classification system soil analysis data, the location of experimental plots
(Soil Survey Staff, 1997). Where markets can be wasbasedlargelyonsoilcolor;asdarkaspossibleand
easilyaccessed,DEsoilscanbeusedforpermanent, containingpotshardsforDE,andasredoryellowas
mechanized crop production requiring high external possible for AS. After slash-and-burn, maize was
inputs. In more remote areas, swidden subsistence seededbyhandinallfieldsbetween30Januaryand24
agriculture is commonly practiced on these soils, February, 2003. The open-pollinated maize cultivar
with a greater range of crops produced than on AS ‘‘Solimo˜es’’ BR 5110, obtained from EMBRAPA
including maize (Zea mays L.), vegetable, and fruit (Empresa Brasileira de Pesquisa Agropecua´ria) was
crops. German (2003) reported that farmers con- used. Weeds present at the time of seeding were
sistently rated the performance of a wide variety of removed by hand or with a hoe. Maize rows were
cropstobebetteronDEthanAS.Thisdifferencein spaced 75cm apart with 20cm separating individual
rating was especially striking for maize, with plants within the row, to give a density of
farmers rating its performance at 2.00(cid:2)0.00 on 66,500plantsha(cid:1)1. The size of each planting was
DEsoiland0.29(cid:2)0.62onAS,onascaleof0to2. approximately 12 by 13m. A randomized complete
Fallowperiodsarereducedbymorethan50%onDE blockdesignwasapplied,withthreeblocksarranged
compared with AS, and this results in fewer new perpendiculartothegradientjudgedmostimportantat
swiddenareasclearedfrommatureforestonDEthan each site, for example slope, distance from road or
AS (German, 2003). forest edge. The three treatments evaluated were: (i)
In 2003, swidden plots planted to maize on both weeds only, where no maize was seeded, (ii) weedy
DEandASwereusedtodeterminetherichnessand maize,whereweedswereallowedtocompetewiththe
diversity of the weed flora, as well as maize and maize until harvest, and (iii) weeded maize, where
weed biomass production on these two contrasting weeds were removed from plots monthly with a hoe.
soils. Working hypotheses included that weed and Ateachplantingsite,nine1.5m2permanentquadrats
maize biomass would be greater on DE, and that a wereestablishedandconsideredsamplingunits.Each
different, more diverse weed community would be quadrat spanned two rows of maize with 1.5m
found above and belowground on these soils perpendiculartotherowsand1malongtherows,and
compared with AS. Surveys were carried out on included 10 maize plants at seeding. The treatments
16 DE farms to obtain additional data on weed were applied to an area of approximately 2.25m(cid:3)
diversity. The seedbanks of DE and AS soils under 1.6m,withthepermanentquadratinthecenterofthis
forest were sampled to assess the initial weed area,andanunweededrowofmaizeservedasabuffer
pressure when swidden agriculture is initiated. between all treated areas.
Differences in seedbank size between AS and DE After seeding, each sitewas visited monthly for 4
werenotexpected,becausesamplingbothsoiltypes months to collect crop and weed data. At each
under a forest environment would mask any effects sampling, all weed individuals in the 1.5m2 perma-
ofindigenousdisturbanceandmanagementincurred nent quadrats were identified and counted, and the
centuries ago. general physiological stage and percentage ground
J.Majoretal./Agriculture,EcosystemsandEnvironment111(2005)1–12 3
Fig.1. LocationofexperimentalsitesnearManaus,Amazonas,Brazil.MaizetrailswereconductedatlocationsCA,RP,RU1andRU2,and
weedssurveysonplotsatthelocationsindicatedby(~).Numbersindicatethenumberofplotssurveyedateachlocation.Seedbanksamples
weretakenatlocationsdenotedby( ,DE)and( ,AS).
coverofeachspeciesrecorded.Totalgroundcoverof intoonecompositesamplepersiteandsamplingtime.
weeds was calculated as the sum of percentages for Potassium(K),calcium(Ca),magnesium(Mg)andP
individual species. The number of live maize plants were extracted using the Mehlich 1 solution (HCl
was determined in each quadrat. At harvest, maize 0.05moll(cid:1)1+H SO 0.0125moll(cid:1)1),withavailable
2 4
plants were cut at soil level, separated into ears with P determined colorimetrically using the ammonium
husk and stalks, and fresh weights were recorded. molybdate with ascorbic acid method, K determined
Weedswereseparatedintothetwodominantspecies, by flame emission photometry, and Ca and Mg
seedlings were separated from sprouts, and the fresh determined by atomic absorption spectrophotometry.
weight of all groupings recorded. When possible, all Exchangeablealuminumwasextractedwith1moll(cid:1)1
plant material was brought back to the laboratory, KClandthefiltratetitratedwith0.025moll(cid:1)1NaOH.
dried at 608C for at least 24h, and weighed. Total carbon and nitrogen were determined by dry
Weeds were identified with the collaboration of combustion with an automatic Leco CN-2000 CN
INPA(InstitutoNacionalde Pesquisas daAmazoˆnia) analyzer (Leco Corporation, St. Joseph, MI, USA).
and according to Ribeiro et al. (1999) and Lorenzi Particle size analysis was carried out using the
(2000). hydrometer method (Bouyoucos, 1927).
Soilsamplingtoadepthof20cm wascarriedout Transectswereestablishedon16croppedplotson
usingaDutch auger.Eightsamples were collectedat tenfarms(Fig.1),andweedsidentifiedinten1.5m2
seeding from each 12m(cid:3)13m site, and combined quadrats,separatedby1malongthetransect.Thus,a
4 J.Majoretal./Agriculture,EcosystemsandEnvironment111(2005)1–12
total of 15m2 were sampled on each of the 16 plots. and 1208C for 30min were also included. Species
Theplotssurveyedwereplantedtoanumberofcrops, emerginginthecontrolflatswereconsideredtohave
whichwereatdifferentstagesofgrowthatthetimeof beencarriedintothescreenhouse,whichwasnotfully
sampling. A variety of crop management and weed screenedfromtheoutside,andwereexcludedfromthe
controlstrategieswereusedonthedifferentplots.The analysis. Flats were watered as required, but not
lastweedingpriortothesurveyrangedfrom2weeks fertilized. Emergence was assessed twice weekly for
to 2 months. The quadrats sampled for the four DE approximately 59 days. Emerged seedlings were
sites in the maize field experiment, with a total of removed and transplanted into separate pots until
13.5m2 surveyed per site, were also included in the theycould be identified. Due to high temperatures in
survey. the screenhouse, shallow soil depths in the flats and
Fortheseedbankstudy,soilsampleswerecollected logistical limitations, excessive soil drying occurred
on 16–17 April 2003 along the Urubu´ River, on several occasions, resulting in 37% of observed
Amazonas, Brazil (Fig. 1), in the latter part of the seedlings dying before they could be identified.
rainy season that typically lasts from December to Hereafter, these seedlings are categorized as being
May.Allsamplinglocationswereunderoldforest,and ‘‘unaccounted’’.
based on knowledge from local inhabitants had not Analyses of variance were performed using the
been disturbed. For AS sites, this likely meant that GLMprocedureinSAS(SASInstitute,2001),withall
these areas were under ‘‘virgin’’ forest, whereas the termstreatedasfixedvariablesexceptlocationinthe
DE sites had probably not been cropped since seedbank study, which was treated as a random
abandonment by Indigenous people when their variable nested within soil type. After examining
populations were decimated between 1600 and 1700 residual plots, weed percentage ground cover data
A.D.Sixlocationsweresampled:Z,A,andSonDE werearcsine-squareroottransformed,countsofweed
and D, I, and L on AS. At each site, a transect was individuals as well as seedbank emergence counts
established approximately 10m into the forest and weresquareroottransformed,andbiomassdatawere
paralleltotheedgeoffalloworcroppedfields.Atone log transformed to stabilize variance. In analyses,
meter intervals for a distance of 10m, sampling was ramets of species reproducing by rhizomes or root
carried out along a 1m line perpendicular to the sprouts were considered to be separate individuals.
transect. Using a Dutch auger, four equidistant soil Sorensen’s coefficient of similarity (S) was used to
cores were taken along this line on either side of the compareweedcommunitiesatdifferentsites,accord-
transect,toadepthof15cm.Coarsedebrisandloose ing to the equation
surfacerootswereremovedbeforesampling.Thefour
2c
samplescollectedforeachofthe10mlocationswere S¼
s þs
combinedtoyield10compositesamplespersampling 1 2
location. The sampling area of 667cm2 is within the where c is the number of species common to both
range used in other similar studies, such as Uhl and communities, s is the number of species in the first
1
Clark, 1983 and Benoit et al., 1989. community and s is the number of species in the
2
Samples were air-dried for one week in the dark second community. Pearson Moment Correlation
beforeprocessing.Afterbreakinglargesoilaggregates coefficients were calculated for the various weed
byhand,1lsub-samplesofsoilwereplacedinplastic and maize parameters.
germination flats. Sixty flats, 10 for each of the six
locations, were placed in a screenhouse receiving
natural light with flats arranged in a completely 3. Results and discussion
randomized design. The photoperiod during the trial
was approximately12h,andprevious datashowthat 3.1. Soil fertility
temperatures inside the screenhouse during the study
periodaveraged318Cbutcouldhavereachedadaily DE soils at CA and RU1 were more fertile than
maximumof418C.Twentycontrolflatscontaining1l theircorrespondingASareas,asevidencedbyhigher
of DE soil of unknown origin autoclaved at 101kPa pH, P content and base saturation, and lower Al
J.Majoretal./Agriculture,EcosystemsandEnvironment111(2005)1–12 5
Table1
KeysoilpropertiesofsamplestakeninFebruary2003afterslash-and-burnofsecondarygrowthonfourAmazonianDarkEarth(DE)andfour
adjacentsoils(AS)inthecentralBrazilianAmazon
Parameter Unit CA-DE CA-AS RP-DE RP-AS RU1-DE RU1-AS RU2-DE RU2-AS
pH (CaCl) 5.02 4.07 4.71 5.03 4.62 4.48 4.51 4.47
2
P (mgkg(cid:1)1) 200.1 5.1 44.8 2.8 15.0 3.4 7.5 4.0
BSa (%) 65 2 12 24 20 2 2 4
ALSb (%) 0 19 14 12 10 19 17 17
a Basesaturation.
b Aluminumsaturation.
saturation(Table1).DEsoilsatRPandRU2,however, (P<0.0001). While yields were low, maize dry ear
exhibitedhighAlsaturation,lowPcontentinthecase biomasswasgreatestonthemorefertileDEsitesCA
of RU2-DE, and lower base saturation than the andRU1(1.4and3.2tha(cid:1)1,respectively).Lowerthan
corresponding AS areas (Table 1). Past management expectedyieldsattheCA-DEversustheRU1-DEsite
mayhavenegativelyaffectedfertilitylevels,resulting were likely due to soil compaction from a road
in a fertility gradient among DE sites. crossingtheCA-DEsiteinthepast.AttheCA-ASand
RU2-ASsitesnoearswereproduced.TheCA-DEsite,
3.2. History of study sites although subjected to high disturbance and soil
compaction in the past, did not experience large soil
Important differences in past-use history charac- fertility declines as might be expected.
terizedthefieldsitesusedinthisstudy.Forthisreason, Weeding did not have a significant effect
thefieldsitesweredividedintotwogroups,basedon (P>0.05)onoverallmaizeperformance.Ingeneral,
the intensity ofpast-use history.The ‘‘intensive-use’’ maize did not respond to weed removal, however at
sites include locations CA and RP (Fig. 1, Table 2), the CA-DE site both ear and stalk biomass were, on
whichareclosertomarketsinManausandRioPreto average, more than double in weeded plots
daEva,inthecaseofRP,andwheremostDElandis (1.9tha(cid:1)1 ear) versus non-weeded plots (0.9tha(cid:1)1
underpermanentcropproduction.AttheCA-DEsite, 1ear).Incontrasttoothersites,weedpressureatthe
landusepriortofallowingwasintensiveasremnants CA-DE site occurred early but decreased rapidly
ofaroadandasuspecteddwellingwerenoticedonthe following the removal of weeds 1 month after
experimentalplot.Similarly,theRP-DEsitewasused seeding when maize was at the six-leaf stage. Yield
as a cattle pasture from 1988 to 1996. The intensive losses in maize ranging from 2 to 28% have been
past use of these sites contrasts with the ‘‘light-use’’ reported in temperate regions, when weed removal
sites and may have influenced weed-crop dynamics. was delayed until four weeks after seeding (Wilson
The‘‘light-use’’sitesincludeRU1andRU2.These and Westra, 1991; Hall et al., 1992; Hellwig et al.,
sites are further away from the city of Manaus and, 2002). Higher soil fertility may increase the period
therefore, market access is substantially reduced. duringwhichmaizecropscantolerateweedssuchas
Although some horticultural crops are produced for in Nebraska, where the initiation of the critical
marketonasmallscale,subsistenceagricultureisthe periodforweedcontrolinamaizefieldwasdelayed
dominant practice on these sites. Experimental plots until the six-leaf stage with a nitrogen (N)
selectedatthese‘‘light-use’’sitesreportedlyhadbeen application of 120kgha(cid:1)1, and resulted in an
underfallowforshorterperiodsthanplotsinthemore acceptable yield loss of 5% (Evans et al., 2003).
intensive-usesites,althoughcroppinghadnotbeenas Whennofertilizerwasadded,thecriticalperiodfor
intensive. weedcontroloccurredatthetwo-leafstageofmaize.
Similarly, maize was better able to tolerate weed
3.3. Maize performance pressure with a 220–250kgNha(cid:1)1 application as
opposedtoa100–130kgNha(cid:1)1applicationinfield
The effects of soil type, DE or AS, site, and their trials in southwestern Ontario, Canada (Tollenaar
interaction on maize yield were all highly significant et al., 1994).
6
Table2
Fieldsitedescription,methodoflandpreparation,andintensityofburnforfourAmazonianDarkEarth(DE)andfouradjacentsoil(AS)sitesinthecentralBrazilianAmazon
Locationname Locationcode Standingvegetationa Methodoflandpreparation Intensityofburnb
CostadoAc¸utuba CA-DE Morethan10yearoldregrowth,dominated SlashedwkofJan.21,2003 Notburned
bySennasp.trees15cmdiameteratbase.
J
Suspectedlocationofadwelling. .
M
CA-AS Virginforest,trees25mhigh SlashedinFeb.2002,burnedinJuly2002, Intense a
jo
hoedbeforeplanting r
e
t
RioPretodaEva RP-DE 7-Yearoldregrowthafterpastureabandonment, SlashedwkofJan.28,2003, Light a
l.
treessparselydistributed,5mhigh burnedwkofFeb.10,2003 /
A
RP-AS Virginforestclearedin1995andplantedtorice, Slashed,andburnedandcleanedwithahoe Light gr
thenmaniocandfruittreesplantedin1996, wkofJan.10,2003 icu
sitesomewhatabandonedattime ltu
re
experimentestablished ,
E
c
RioUrubu1 RU1-DE 6-Yearoldregrowthdominated UnderstoryslashedwkofJan.28,2003, Moderate os
y
byOcoteasp.,10mhigh treescutFeb.4,burnedwkofFeb.5, ste
branchesremovedbyhandbeforelightly m
s
re-burningtheleaflitterandplanting an
d
RU1-AS 16-Yearoldregrowthdominated ClearedSept.2002,burnedwkofFeb.5, Moderate
E
byVismiasp.,10mhigh branchesremovedbyhandbeforelightly nv
re-burningtheleaflitterandplanting iro
n
RioUrubu2 RU2-DE 3-Yearoldregrowthdominated ClearedandburnedNov.2002 Moderate m
e
byVismiasp.,10mhigh, nt
1
treediameterbetween3and10cmatbase 1
1
RU2-AS 8-Yearoldregrowthslashedin2001, Re-cutNov.2002,burned7Feb.2003,plot Light (2
notpreviouslyplantedtocrops cleanedwithahoe 00
5
a Asobservedandreportedbylocalinhabitants. )1
b Light:treetrunksandmostbranchesleftunburned,moderate:mosttreebranchesburned,intense:mostplantmaterialburned. –1
2
J.Majoretal./Agriculture,EcosystemsandEnvironment111(2005)1–12 7
3.4. Weed population weedings,weedcoverwasgreateronweededDEplots
atRP(28%)andCA(19%)sitesthanonsimilarplots
Weed populations measured at the different sites at the other two sites (on average 6.5%) (P<0.05).
variedwidely.Forexample,aweedcoverof100%was Theseedbank at theCAsitewasprobablyverylarge
obtainedonplotsattheCA-DEsite,whilesomenon- since within 1 month of the start of the trial, annual
weeded plots at the CA-AS site had a weed cover of weedsandespeciallyEuphorbiaheterophyllaL.were
0%. Consequently, therewas a significantinteraction thedominantgroupandcomprisednearly80%ofthe
(P<0.05) between site and soil type for the number groundcover(Fig.2).AttheRP-DEsite,however,the
ofweeds,percentageground coverandbiomass.The dominant species on weeded plots at harvest were
percentage ground cover of weeds was strongly members of the Cyperaceae such as Cyperus cf.
correlated to their dry biomass at harvest (Pearson’s aggregatus Endl., C. cf. diffusus Vahl, and Rhynch-
correlation coefficient=0.8, P<0.0001), whereas ospora nervosa (Vahl) Boeck. subsp. ciliata T.
the total number of individual weed plants in the Koyama, which reproduce primarily vegetatively.
1.5m2quadratswaspoorlycorrelatedwiththeircover OnmorefertileDEsoilsatsitesCAandRU1,the
and biomass. Positive weed responses on DE soils better performance of maize reduced weed cover in
were most apparent on the more fertile DE sites CA the weedy maize plots on average by 32 and 58% at
(98.5%cover)andRU1(77.3%cover),incomparison CA and RU1, respectively, compared with the plots
to AS at these same sites (2.2 and 26.7%, respec- whereonlyweedsgrew,whereasthiswasnotthecase
tively).WeedcoveronASwasgreatestattheRPsite at the less fertile DE sites. Early weed emergence at
where Solanum subinerme Jacq. was found, which the CA-DE site likely explains why differences in
reproduces vigorously by root sprouts. Biomass was weed coverage between the weed-only and weedy
generally greater in the plots with only weeds maizeplotswerenotasdrasticasfortheRU1-DEsite,
compared to those with weedy maize. As expected, where weed emergence was more gradual. Similar
theweededmaizeplotshadsignificantlyfewerweeds, findingswereobtainedbyJornsgardetal.(1996)ina
lowerweedgroundcoverandbiomassatharvestthan cereal field experiment in Denmark. At lower N
othertreatmentplots.Atharvest,after4monthlyhand applications of 30–54kgNha(cid:1)1, crop growth and
Fig.2. ProportionofgroundcovercomprisedofannualweedsandperennialsproutsatfourAmazonianDarkEarth(DE)andfouradjacentsoil
(AS)sitesinthecentralBrazilianAmazon.The‘‘other’’categoryincludessedges,perennialsreproducingbyseed,speciesthatcanbehaveas
annualsorperennials,andspeciesthatcouldnotbeidentified.
8 J.Majoretal./Agriculture,EcosystemsandEnvironment111(2005)1–12
development were reduced and weed growth favored E. heterophylla, Phyllanthus niuri L., Cyperus cf.
because a greater proportion of light reached the aggregatus, C. cf. diffusus, and Microtea debilis Sw.,
weeds. At higher N application rates of 90– allofwhicharenativetotheAmericas,andAcalypha
162kgNha(cid:1)1,cropgrowthwasimprovedandplants arvensisPoepp.andEndl., whoseoriginis unknown.
were able to overtop and shade out the weeds. On some sites, a single species represented more
The greater weed pressure observed early in the than 60%of theweed coverage at harvest. Examples
growing season on the CA-DE site was likely due to includeSpermacocelatifoliaAubl.attheRU1-DEsite
the high disturbance history of the site and to it not and P. niuri at the CA-DE site. However, no single
beingburned(Table2).Firehasbeendemonstratedto weed species was dominant across all field sites.
reduce seedbanks in proportion to its intensity Sixty-two percent or 48 out of 77 weed species were
(Monaco, 1998). found on only one site, including E. heterophylla L.,
Sida rhombifolia L., and S. subinerme Jacq. which
3.5. Weed species richness and dominance were each dominant on the site they colonized.
Interestingly,S.rhombifoliahasbeenreportedtobean
A total of 77 plant species were identified at the indicatorspeciesofhighsoilfertilityinCostaRica(de
eight experimental sites. Mean species richness for la Cruz, 1994)and we found this species only onthe
individualpermanentquadratswasgreateronDEthan CA-DEsite,themostfertileofthesites.Dominanceof
AS soils, 8.0 and 4.2 respectively, except at the RU2 grasses and sedges was reported on intensively
site,whichhadsimilarlylowchemicalfertilityonboth managed soils of Nigeria and Uganda where fertility
soiltypes.Thegreatestnumberofspecieswasfoundat declines had occurred (Ugen and Wortman, 2001;
the CA-DE site, while the lowest number of species Akobundu and Ekeleme, 2002). In this study, sedges
was at the CA-AS site, 29 and 5 respectively. and grasses were each dominant with >40% of total
Speciesrichnesshasoftenbeenobservedtofollow weed cover on four sites: RP-DE, which was
aunimodalpatternalongaproductivitygradient,with previously used for pasture, and three AS sites: RP-
the highest diversity occurring at moderate produc- AS, RU1-AS, and RU2-AS. The leguminous woody
tivity levels and lowest diversity found at range vine,DerrisamazonicaKillip,waspresentonallsites
extremes (Grime, 1979; Stevens and Carson, 1999). exceptCA,althoughsometimesatverylowdensities.
Other factors may also be critical, with disturbance
playing an especially important role in facilitating 3.6. Leguminous weeds
species colonization (Grace, 1999 and references
therein).InDEcropproductionsystems,fertilityand Although differences varied significantly between
disturbance are often closely linked. In general, DE sites,thenumberofleguminousindividualsandtheir
soilsarelocatedclosertodisturbedoropenareassuch percentage ground cover was greater on DE than AS
as rivers and dwellings, where plant dispersal and (average 21 and 4%, respectively, P<0.0001).
migration are facilitated. In addition, DE soils are Legumes occupied the greatest proportion of ground
subjected to more frequent disturbance, with shorter cover on ‘‘light-use’’ sites RU1 and RU2 where the
fallow periods (German, 2003). Weed dispersal and woodyvine D.amazonica Killip was dominant.This
colonizationarealsoassistedbytheuseofmanureand species relies largely on vegetative sprouting for
weedseed-contaminatedsoilamendments,which are propagationand,forthisreason,isexpectedathigher
most often applied to DE soils and not AS in small- densitieswhereburnsarelessfrequentandintense(de
holder agricultural production systems. Rouw, 1993) and where weeding is less intensive
WeedcommunitiesonDEsiteshadamoresimilar (Kellman, 1980).
species composition (P=0.069) than on AS sites. High availability of P, Ca, and micronutrients as
Many weed species present on sites CA and RP well as low N availability stimulate the biological
characterized as having an ‘‘intensive-use’’ history fixation of atmospheric N , which gives legumes a
2
werealsoobservedinurbanareasofManausincluding competitiveadvantageonDE(Lehmann etal.,2003)
roadsides and abandoned lots. Examples of these explaining the greater proportion of legume weeds
‘‘cosmopolitan species’’ include Euphorbia hirta L., observed on DE than AS. In fact, N availability has
J.Majoretal./Agriculture,EcosystemsandEnvironment111(2005)1–12 9
beenfoundtobereducedinDEsoilsbecauseofhigh Interestingly, weed species found on both soil types
C:N ratios. didnotdifferintheirtimingoffloweringandfruitset.
The proportion of biomass from woody sprouting
3.7. Weed reproductive strategy vegetation was greater at sites having ‘‘light-use’’
histories,anddidnotvarysignificantlywithsoiltype
Annual species constituted a greater proportion of or weed treatment. Sprouting, as a regenerative
weeds and vegetative cover on DE soils than AS strategy, is facilitated on sites with low weeding
(P<0.05), although differences varied significantly intensity, less frequent or lighter burns (Kellman,
betweensites(Fig.2).Thisfinding isconsistentwith 1980; Uhl et al., 1982; de Rouw, 1993). The
workbyTilman(1987)andDielemanetal.(2000)in substantially lower woody sprout biomass at the
temperate regions showing that annuals are most RP-DE site was likely due to the site being used for
abundant on more fertile soils. The greatest differ- pasturefor8years,whileatCA-DEthepresenceofa
encesintheproportionofgroundcovercomprisedby dwelling in the past could explain the lower sprout
annuals occurred at locations CA and RU1, where biomass obtained (Fig. 2).
fertilitydifferencesbetweenDEandASweregreatest.
WeedsenteredthereproductivephaseearlieronDE 3.8. Weed survey
than AS sites. During the experiment, an average of
50%ofthespeciesoccurringonDEsitesand31%of Ninety-eight species were identified in the weed
those occurring on AS sites flowered. These differ- surveyof20DEplotsincludingthosesurveyedforthe
ences could be due to the contrasting phenologies of maizetrial.Awiderdiversityofweedswasrecorded,
the species observed on each soil type rather than compared to that found in the controlled maize
effects of the soil per se. More annual species were experiment,with44ofthespeciesfoundononlyone
foundonDEsoilsthanAS,andtheseareexpectedto surveyed plot. The most frequent species were A.
flowerandsetfruitearlierthanmostperennialspecies. arvensis Poepp. and Endl. (frequency (f)=0.65), P.
Fig.3. Frequencyratingsbylifeformforweedssurveyedon16croppedAmazonianDarkEarth(DE)plotsinthecentralBrazilianAmazon.
Shrubsandepiphytesarenotincludedsinceeachhadafrequencyof0.6%.
10 J.Majoretal./Agriculture,EcosystemsandEnvironment111(2005)1–12
niuri L. (f =0.60), and Mollugo verticillata L. The mean seedling density for AS corresponds to
(f =0.55). All three may be considered ‘‘cosmopo- densities and associated seedbank sizes of 344–862
litan’’ species, as they were frequently observed seedlings m(cid:1)2 reported by Guevara and Gomez-
growing in highly disturbed, urban areas. Thirteen Pompa (1972) for tropical forest soils. The seedling
species were in the Euphorbiaceae family and densityforDEsoilsanditsassociatedseedbanksizeis
accounted for 24% of frequency recordings. Forbs within the 2,300 seedlings m(cid:1)2 range reported by
werethemostfrequentlyrecordedlifeform,followed Garcia(1995)foragriculturalsoilsinthetropics,even
by trees (Fig. 3). Overall, 57% of species surveyed though the sample area was not under agricultural
were perennials, 24% were annuals and 19% not production. The rate of seedling emergence was
known. greatest between days 21 and 31 and was generally
consistent across locations. Emergence rates
3.9. Seedbanks decreased gradually thereafter (Fig. 4).
The observed differences between DE and AS are
Seedlingsfromatotalof15specieswererecorded: likelyduetodifferentlevelsofdisturbanceassociated
one member each of the Cyperaceae and Legumino- with the areas adjacent to sampling locations. In a
seae,threemembersofCecropia,onemembereachof slash-and-burnagriculturalsystemwherenotillageis
Vismia and Fittonia, Solanum appressum K.E. Roe, practiced, increased disturbance leads to increased
Trema micrantha Blume, and six species groups that weed biomass and seedbanks when compared with
could not be identified by name. In general, more fallowing (Akobundu et al., 1999; Ekeleme et al.,
seedlingsemergedinDEsoilsthanAS(9.1versus2.2 2004).DisturbanceisincreasedonDEbecauseshorter
per flat, respectively, P<0.05), but there were fallows are used (German, 2003), and these soils are
significant differences (P<0.05) between sampling oftencroppedmoreintensively,atleastwhenmarkets
locations. Similarly, seedlings of more species areaccessible.Furthermore,whileswiddensonASare
emerged from DE soils than AS (2.1 versus 1.2, sometimesclearedfarfrompermanentlivingareasor
respectively, P<0.1), with no significant effect of evenonlandnotdirectlyadjacenttothem,dwellings
samplinglocation.Thesevaluesconvertto1,365and areusuallylocatedclosetoswiddenplotsonDE,and
330seedlingsm(cid:1)2forDEsoilsandAS,respectively. thereforedisturbanceisincreased.Infact,DElocation
Fig.4. Cumulativenumberofemergedseedlingsoveran8-weekperiodinascreenhousefor667cm2soilsamplesfromundisturbedsoils
collectedatthreeAmazonianDarkEarth(DE)andthreeadjacentsoil(AS)locationsinthecentralBrazilianAmazon.