From Traditional Farming in Morocco to Early Urban Agroecology in Northern Mesopotamia

21 Pages·2016·3 MB·English

by  Amy Bogaard

Environmental Archaeology
The Journal of Human Palaeoecology
ISSN: 1461-4103 (Print) 1749-6314 (Online) Journal homepage: http://www.tandfonline.com/loi/yenv20
From Traditional Farming in Morocco to Early
Urban Agroecology in Northern Mesopotamia:
Combining Present-day Arable Weed Surveys
and Crop Isotope Analysis to Reconstruct Past
Agrosystems in (Semi-)arid Regions
Amy Bogaard, Amy Styring, Mohammed Ater, Younes Hmimsa, Laura Green,
Elizabeth Stroud, Jade Whitlam, Charlotte Diffey, Erika Nitsch, Michael
Charles, Glynis Jones & John Hodgson
To cite this article: Amy Bogaard, Amy Styring, Mohammed Ater, Younes Hmimsa, Laura
Green, Elizabeth Stroud, Jade Whitlam, Charlotte Diffey, Erika Nitsch, Michael Charles, Glynis
Jones & John Hodgson (2016): From Traditional Farming in Morocco to Early Urban Agroecology
in Northern Mesopotamia: Combining Present-day Arable Weed Surveys and Crop Isotope
Analysis to Reconstruct Past Agrosystems in (Semi-)arid Regions, Environmental Archaeology,
DOI: 10.1080/14614103.2016.1261217
To link to this article: http://dx.doi.org/10.1080/14614103.2016.1261217
© 2016 The Author(s). Published by Informa View supplementary material
UK Limited, trading as Taylor & Francis
Group
Published online: 21 Dec 2016. Submit your article to this journal
Article views: 86 View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at
http://www.tandfonline.com/action/journalInformation?journalCode=yenv20
Download by: [University of Sheffield] Date: 18 January 2017, At: 05:40
ENVIRONMENTALARCHAEOLOGY,2016
http://dx.doi.org/10.1080/14614103.2016.1261217
From Traditional Farming in Morocco to Early Urban Agroecology in Northern
Mesopotamia: Combining Present-day Arable Weed Surveys and Crop Isotope
Analysis to Reconstruct Past Agrosystems in (Semi-)arid Regions
Amy Bogaarda, Amy Styringa, Mohammed Aterb, Younes Hmimsab, Laura Greena, Elizabeth Strouda,
Jade Whitlama, Charlotte Diffeya, Erika Nitscha, Michael Charlesa, Glynis Jonesc and John Hodgsonc
aSchoolofArchaeology,UniversityofOxford,Oxford,UK;bDepartmentofBiology,UniversitéAbdelmalekEssaâdi,Tétouan,Morocco;
cDepartmentofArchaeology,UniversityofSheffield,Sheffield,UK
ABSTRACT ARTICLEHISTORY
Weintegratefunctionalweedecologywithcropstablecarbonandnitrogenisotopeanalysisto Received9June2016
assesstheircombinedpotentialforinferringarablelandmanagementpracticesin(semi-)arid Accepted3November2016
regions from archaeobotanical assemblages. Weed and GIS survey of 60 cereal and pulse
KEYWORDS
fields in Morocco are combined with crop sampling for stable isotope analysis to frame
Archaeobotany;stable
assessment of agricultural labour intensity in terms of manuring, irrigation, tillage and hand-
isotopes;functionalplant
weeding. Under low management intensity weed variation primarily reflects geographical
ecology;agricultural
differences, whereas under high management intensity fields in disparate regions have intensity;BronzeAge;
similar weed flora. Manured and irrigated oasis barley fields are clearly discriminated from NorthernMesopotamia
less intensively manured rain-fed barley terraces in southern Morocco; when fields in
northern and southern Morocco are considered together, climatic differences are
superimposed on the agronomic intensity gradient. Barley δ13C and δ15N values clearly
distinguish among the Moroccan regimes. An integrated approach combines crop isotope
values with weed ecological discrimination of low- and high-intensity regimes across
multiple studies (in southern Morocco and southern Europe). Analysis of archaeobotanical
samples from EBA Tell Brak, Syria suggests that this early city was sustained through
extensive(low-intensity,large-scale)cerealfarming.
Introduction
a range of intensity levels in Morocco was conducted
Inarablehabitats,variationinstablecarbonandnitro- in order to assess their potential for investigating past
genisotopevaluesincerealsandpulseshasbeenshown farming practice in (semi-)arid regions (Styring et al.
to reflect water status and soil nitrogen composition, 2016).Thisstudyintegratedtheresultsofisotopicsur-
respectively, with implications for water management vey of traditionally managed barley fields, extending
and manuring (e.g. Araus et al. 1997; Bogaard et al. from the Mediterranean north to the arid south of
2007;Fraseretal.2011;Wallaceetal.2013;Fiorentino Morocco,withthoseofapreviousstudyofplantisoto-
etal.2015).Thusvariationincropstableisotopevalues picvariationwithrainfallintheeasternMediterranean
offersausefulwaytoinvestigatetheecologyofpresent (Hartman and Danin 2010). A working model was
andpastfarmingsystems.Interpretationofcropstable developed for disentangling the effects of aridity and
isotope values is usefully constrained through inte- manuring on cereal δ15N values (Styring et al. 2016,
grationwithfunctionalecologicalanalysisofassociated figure 4). This study opens the way for using cereal
weed flora, which provides complementary insights δ15Nalongsideδ13Cvaluesasevidenceofarablegrow-
into soil productivity and disturbance levels. A model ing conditions in (semi-)arid areas such as northern
for identifying cultivation intensity (i.e. labour and Mesopotamia, where early processes of urbanisation
resource inputs per unit area) that integrates crop have variously been related to agricultural intensifica-
stableisotopeswithfunctionalweedattributesortraits tion(increasinginputsof,forexample,manureormid-
has successfully been developed for temperate Europe den material per unit area – Wilkinson 1982, 1993;
(Bogaard et al. 2016). The aim of this paper is to Wilkinson et al. 1994), extensification (expansion of
develop a similar model for application to (semi-)arid arable with decreasing inputs per unit area – Halstead
regions. in Wilkinson et al. 1994; Halstead 1995) or a combi-
A recent pilot study of stable carbon (δ13C) and nation of the two (Ur 2015; cf. Wilkinson 2003, 118,
nitrogen (δ15N) isotope values in cereals grown under figure 6.16).
CONTACTAmyBogaard [email protected]
Supplementaldataforthisarticlecanbeaccessedatdoi:10.1080/14614103.2016.1261217
©2016TheAuthor(s).PublishedbyInformaUKLimited,tradingasTaylor&FrancisGroup
ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommonsAttribution-NonCommercial-NoDerivativesLicense(http://creativecommons.org/licenses/by-
nc-nd/4.0/),whichpermitsnon-commercialre-use,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited,andisnotaltered,transformed,or
builtuponinanyway.
2 A.BOGAARDETAL.
Asetoffunctionalplanttraitshasbeenshownglob- n
allytooccurinalimitedsetofcombinationsrepeatedly atio Anti-
a2pdvr(faeiiegcri0lit.oacloeqsi0aetnu.uu4tiwsstoinhi,vlrin(ydittee2gGrhsld0ahyisrti1nototiar5ahmsfsetn)ihosdg.eedvserhciF-ass2o-tlcaseuo0olcrtona0wltmiuegmrnc1lriitoad)bcasii.rronoaunellnHiattople,tgawtleespalherr-nnruteetibhotonerdsedaarwrettatieyurhwt,eln,acacelpvstuhvoieiaavtaesannrtyienvtitseaeyedictavrsthuintoaboceelsslnteonneviedmotvtewndiafwonp[etcmifesei(oothdwtDhfennioutceívowantfherehrlzdecaonregtgnrseiifaetoipptmlrcnonrraaacroeeeatllaes---]fl. (1963)andAchhaletal.(1980),veget Agrosystem Traditionalmountainsystem(alluvialplains;MediterraneanslopeofcentralRif) Traditionalmountainsystem(pre-Riffolds,continentalslopeofcentralRif)Traditionalmountainsystem(SWslopesofAtlas) Oasissystem
soil disturbance) in southern Europe (Jones et al. e
acntdT2hol0rhooeon0renut0dhgusg;ittshgtiueBoortfdanoutsdyogslenialmereearnersariagt-sdnsnaigorocoeeienffdtis/anfoarnaitrtonledhi.mdneMas2rivin0otsoyf1uruit6odhibnn)ace-c.cnhMtsoiucWooeomun.retaiohdlc.a/csstWoersa,meeiwstsisa-hlaeswrroreieedlceacodtliiinnnmsfgilaodtthrteioaecr mberger(1955),revisedbySauvagandRankouetal.(2013). Climaxvegetation hyllousforestwithTetraclinisarticulata hyllousforestwithQuercussuber nesianwoodlandformationswithArganniaspinosa) towoodlandformationswithAcaciasp.
Tgehoegrparpehseicnatl-dzaoynfeasr,miningthseitensobrethlonagndtostowuothseopfartahtee ollowE(2000) Sclerop Sclerop Macaro(Arga Steppe
country (Table 1). The northern zone is situated in fd
trsp2hao0aneur0Ttet0chah;eoneenRfrtnenartcnoahsorlkle--toowrhpueSeeegarseinhtooteafnzrarnto;lah.tnnRhe2ee0iAefi1sscna3oolnt)uo-i.d-rcthAeafgetoteirlroadnmnsiznsmo(pnFaoaeiu‘grhniutsotroatsefiisnttpu1hsoa)eatte’n(MdBodfeeofndMonairtebtmehdirdes-- oclimateclassificationsnsaccordingtoBenabi Vegetationzone Thermo-mediterranean Infra-mediterranean Saharan
weitteitrahrlai.nn1e9tah9ne8b)t.ihoTedrhimveeorn-somitrytehd(eMitrenérdrfaaanirlemeatinnQgvuesegtzeuetdlayt1i9os9nit7e;zsoMonocecouirner orocco;biformatio Rainfalla(mm) 597665665703 272 194 222
Mn
aonfdAaarrokuubn,dCthhemRaaiflamaonudnEtasisnafcohuarienn. Tarheessittuuadtyedsitoens esinetatio ate oarid
tcstRbtmeihtlyniiunaefnedQtciMnfeesyoutwlaesfaedadilrirtstrctisehmtiulswecoswuriopinrstlofeaahugonBtboadaoeeeflalfralts(nohVnuaLtebsd.aatl-wrhopvhaapleoen)udrgeodmiieMtpdoitAhilofaadantegtnsihrdaotcdyeellanir(omcmsacBhfrhaoe(aetatBnuhienrleaneoanbaticcRnaniaatdiiebtdnfaeri1iddcscst9heheyo8m1adpa2ni9reni)b8a-t.,yaic2hSsirt)eTnuei.pdrebcprTitosscarrihenlaceside------ zonesofthefarmingstudysitdeco-regionsandclimaxveg SiteBioclim AarkubSemi-aridChmaalaEssafourenAgdaSub-humidBellotaAgniOuramSemi-aridtIssiliZemmourenIdLahsoOmarIdAïssaAridAguelouyTougrare uastat/climateinfotool/index.stm).
taminsuedlde-Aaocterrrocsas2t0ttl0he8-e)d.sriaTtweislnlaopgfelothuwegahsn(owgretohnoeedrrenanllzyoornbemy(edHtaomlnckihmeaysr-a-, egetation2000)an Commune SteihatBniBouzra Bellota Tighirt Amtoudi Tagmoute nr/water/aq
rraifniuenelgeldad,)tnsiwv(oeTteelieayrndbrdliilongeewadgt2iato()ol.nnitd)Mtblteesoaontshommaiggenaheolmil(rmhrbeianuganttandtuvii-onradnitrunei)egngd;s(oiTintriyantbehnlraseaisnznid2veg)e,-e.wmdfrAeoaefrmndraoiubnmrlcge-. ofclimaticandvoBenabid(1982, Province Chefchaouen Ouezzane SidiIfni Guelmim Tata http://www.fao.org/
1Atlhoh4agwa1ddnaamb,hte2duebensoopbnteahennsatoddvafirebnyrtlga:ihsgemhpesereaidamntsaidatBgeereeisclml.aylodMeotenanoststlatoiongocotcafe.lnot1thsr7oiet2plyo7ofingwemgrladae2sspr,ahrstebuylnurarvtetaieatvtyrhwebeeldoyyr Table1.Summarystagesaccordingt Eco-region NorthMediterranean SouthSaharan aDatafromAquastat(
ENVIRONMENTALARCHAEOLOGY 3
Figure1.Mapshowingthepresent-dayfarmingstudysitesinMorocco.
ofthefieldsatAgdahadjustbeencleared(byburning) hulled6-rowbarley;inthesouththiswasalocalland-
ofmaturewoodlandthathadgrownfor30+years;both race, rather than a commercial variety, used to make
werebeinginitiallycultivatedwithpulsesatthetimeof couscous. Other cereals included bread wheat (Triti-
the survey. cum aestivum L.), durum wheat (Triticum durum
The southern zone is situated at the limit of the Desf.) and einkorn wheat (Triticum monococcum L.),
Mediterraneaneco-regionunderstrongSaharaninflu- while pulse crops were broad bean (Vicia faba L.),
ence, and can be considered part of the Saharan eco- chickpea(CicerarietinumL.),bittervetch(Viciaervilia
region (Benabid 2000; Rankou et al. 2013). This eco- (L.) Willd.) and lentil (Lens culinaris Medikus). All of
regionischaracterisedbyclimaticaridityandlowpre- the cereals were autumn(-winter) sown, while the
cipitation. Aside from Tougrare (Tata), which occurs pulses varied from winter- to spring-sown. Harvest
withinaSaharanvegetationzone(steppic,witharbor- times ranged from late April (southern oasis fields) to
eal vegetation typically of Acacia species), the other June (northern mountain system).
southernstudysitesbelongtotheinfra-mediterranean
vegetation zone. Among the latter, the Tighirt study
sites (Table 1) experience oceanic influences, and cli- Materials and methods
max vegetation is characterised by Macaronesian
Field methods
elements and trees such as Argania spinosa (L.) Skeels
(Benabid 2000). The agrosystems encountered in the InMarch–May,2014,floristicsurveyswerecarriedout
southernzoneareoftwotypes:mountainagrosystems of the weeds in 60 cereal and pulse fields, using
(Tighirt sites) on the south-west slopes of the Anti- methodscomparabletothoseemployedinsimilarsur-
Atlas, and oasis systems (Amtoudi sites). The rain- veys of the weeds associated with different husbandry
fedterracedfieldsofthemountainsystemweremana- regimes in France (Bogaard et al. 2016), Spain (Jones
ged with relatively low intensity (tractor-ploughed, et al. 1995; Charles et al. 2002), Jordan (Palmer 1998;
biennial manuring, variable weeding) whereas the Charles and Hoppé 2003) and Greece (Jones et al.
oasis fields were very intensively tilled (by animal- 1999). The weed species present in five 1m2 quadrats
drawn plough plus hand-digging), irrigated, manured were recorded along a linear transect from one end
and weeded (Table 2). Oasis fields were frequently of each field to the other.
shadedtovaryingdegreesbydatepalmandothertrees. GPS data locating the boundary of every field sur-
A total of 60 crop fields was included in this study, veyedandthespecificlocationofeveryquadratwithin
with one additional field sampled for stable isotope each field were recorded using a handheld GPS unit.
analysis only (Table 2). The most frequent crop The data were analysed using ArcGIS 10.3.1 in order
among these fields, encountered in all zones, was to detect any associations between topographical
4
A
.
B
O
G
A
A
R
D
E
T
A
Table2.Summaryofagriculturalpracticesatthefarmingstudysites. L.
North South
Zone
Chefchaouen Ouezzane SidiIfni
Province Guelmim Tata
BniBouzra Bellota Tighirt Amtoudi
Commune Steihat Tagmoute
Agni Issili IdLahso
Ouram Zemmouren Omar IdAïssa Aguelouy Tougrare
Site Aarkub(AAR) Chmaala(CHM) Essafouren(ESA) Agda(AGD) Bellota(BEL) (AGU) (ISS) (OMA) (IDA) (AGE) (TGM)
Fieldtype Terracedfields Alluvialplainfields Terracedfields Terracedfields Oasisplots
Numberoffields 3 1 1 13 10 5 5 5 11a 5 1
Surveyedhectares 0.38 0.02 0.27 2.24 0.14 0.25 0.13 1.05 0.13 0.03 0.09
Averagefieldhectares 0.13 – – 0.17 0.01 0.05 0.03 0.21 0.01 0.01 –
Soils Darkbrowndung- Darkbrowndung-enrichedalluvialsoils Deepcoarsebrownsoilsonschist Shallow,stonyandrichincalcium Darkbrowndung-enrichedalluvial
enrichedon carbonate(lithosols) soils
serpentine
Rotationregime Cereal-pulse/summervegetables Cereal-pulse(-fallow) Continuouscereal Wintercereal-summermaize/
vegetables
Cropssurveyed Breadwheat;pea Barley;oat Breadwheat Hulled6-rowbarley;durum Hulled6-rowbarley, Hulled6-rowbarley Hulled6-rowbarley
wheat;chickpea,lentil,bitter einkorn;chickpea,
vetch,broadbean broadbean
Sowingtime Oct CerealsOct–Dec;pulsesNov-May Oct Oct
Tillage Donkey-drawn Donkey-drawnplough(charrue) Cattle-ormule-drawnplough(charrue) Tractororanimal-drawnplough(charrue) Donkey-drawnplough(charrue)plus
ploughorhand- hoe
dugonly
Manuring(ruminant Annualspreadingofmanuringfromstables/pens Nomanurespread;summerstubblegrazing Bienniallyorlessfrequent,dependingon Spreadatrateofc.100t/ha,larger
dung) availability piecesbrokenupbyhoe
Irrigation n/a Everytwoweeksas Occasional,to n/a n/a Every15daystopointofsaturation,
needed,topointof pointof 28×peryear(forwinterplus
saturation saturation summercrops)
Weeding Frequent,foranimalfodderetc. Frequenttonone None Frequencyvariable Frequent,foranimalfodderetc.
Harvesttime May June May LateApril–May
a1ofthesefieldswassampledforcropisotopeanalysisonly(weedfloranotsurveyed).
ENVIRONMENTALARCHAEOLOGY 5
variables and weed floristic composition or crop iso- for weeds) and 14 pulse. Ten ripe ears or pods were
tope values. randomly selected from each field, or were selected
Ecological attributes of the most commonly occur- from the five weed quadrat locations per field. Half
ring weed species identified in this survey were of the collected ears/pods were then threshed, and a
measured during the same study season. Species were random subset of 50 grains/seeds was homogenised
selected for the measurement of functional attributes using a Spex 2760 FreezerMill to give an average iso-
if they occurred in seven or more of the 60 fields tope value representative of the growing conditions at
(12%offields),athresholdprovidingareasonablebal- each location (cf. Bogaard et al. 2016).
anceofspecies’richnessandfieldnumberssuitablefor Pairs of isotopic measurements (δ13C and δ15N)
the exploratory multivariate statistical technique used were also obtained from 28 charred cereal grain and
(see below). These species are listed in Supplementary 5 charred pulse seed samples (comprising 4–10 indi-
Table 1. The attributes measured are summarised in vidual grains/seeds) from the archaeological site of
Table 3 (see Charles, Jones, and Hodgson 1997; TellBrak.Cerealgrainsandpulseseedswereexamined
Bogaard et al. 1999; Jones et al. 2000; Bogaard et al. at ×7–45 magnification for visible surface contami-
2001; Charles et al. 2003 for fuller explanations). nants, such as adhering sediment or plant roots; these
Well-grown, fully established specimens were were removed by gentle scraping. Around 10% of the
selected for measurement of functional attributes so total number of samples from the site were scraped
thatspeciespotentialwas assessedratherthanindivid- clean, crushed and analysed using Fourier transform
ual plant performance under variable conditions. infraredspectroscopy with attenuated totalreflectance
Detailed protocols for the measurement of each of (FTIR-ATR): Agilent Technologies Cary 640 FTIR
these attributes are given in Jones et al. (2000) and instrument with a GladiATR™ accessory from PIKE
Bogaard et al. (2001). Canopy height and diameter Technologies. Each sample was measured once, the
were converted to a log scale. Data on flowering dur- background was subtracted and a baseline correction
ation were extracted from Maire (1952–1987), sup- was carried out using Agilent Resolution Pro to look
plemented by other Floras where necessary. for the presence of carbonate, nitrate and/or humic
contamination (cf. Vaiglova et al. 2014). Peaks
characteristic of carbonate contamination (870 and
Laboratory analysis 720cm−1)wereobservedinthreeoftheFTIRspectra.
Pairs of isotopic measurements (δ13C and δ15N) were Itwasthereforedecidedtoacidpre-treatallofthecer-
obtained from 47 of the fields, 33 cereal (including eal grain samples to dissolve any carbonate (cf. Bronk
one Amtoudi oasis barley field that was not surveyed Ramsey 2008). This procedure consists of treatment
Table3.Thefunctionalattributesmeasuredandtheirpossibleecologicalsignificancewithinanarablecontext.
Functionalattribute Ecologicalattributeforwhich
measured measurementisasurrogate Relationshiptohabitatconditions References
Attributesrelatingtothedurationandqualityofthegrowthperiod
Maximumcanopy Maximumplantsize,the Positivelycorrelatedwithpotentialproductivityand Grime(2001)
heightanddiameter productofgrowthrateand negativelywithdisturbanceofhabitat
periodofgrowth
Leafareapernode/leaf Plantgrowthrate Positivelycorrelatedwithpotentialproductivityof Jackson(1967);Dale(1982);
thickness habitat Givnish(1987);Jonesetal.(2000)
Meanspecificleafarea Plantgrowthrate Positivelycorrelatedwithpotentialproductivityof Reich,Walters,andEllsworth
(SLA:leafarea/dryleaf habitat (1992);Reich(1993);Wrightetal.
weight) (2004)
Attributerelatingtothecapacitytoregenerateunderconditionsofhighdisturbance
Lengthofflowering Durationoflifecycleand Positivelyassociatedwithdisturbance Grime,Hodgson,andHunt(1988);
period potentialtoregeneratefrom SansandMasalles(1995);
seed Bogaardetal.(2001)
Vegetativespread Abilityofperennialsto Positivelyassociatedwithdisturbance
regeneratevegetativelyfrom
rootfragments
Attributerelatingtodroughtavoidance
Timingofflowering Abilitytocompletelifecycle Latecompletionoffloweringassociatedwithmoist Parker(1968)
beforedryseason conditions
Attributerelatingtodroughttolerance
Meanstomatalsize Capacitytorestrict Positivelycorrelatedwithmoist,productiveconditions Salisbury(1928);Carpenterand
transpirationalwaterloss andnegativelywithdroughtedconditions Smith(1975);Donselmanand
Flint(1982);Raven(2014)
Meanstomataldensity Capacitytorestrict Negativelycorrelatedwithmoist,productiveconditions (asabove)
transpirationalwaterloss (giventhenegativerelationshipbetweensizeand
density)andpositivelywithdroughtedconditions
Epidermalcellsize Capacitytominimiselossof Negativelycorrelatedwithdroughtedconditions Cutler,Rains,andLoomis(1977)
turgor
Epidermalcellwall Providesstructuralsupportin Positivelycorrelatedwithmoistconditions Linsbauer(1930);Watson(1942)
undulation conditionsofhighturgor
6 A.BOGAARDETAL.
with 10mlof0.5 Mhydrochloricacidat70°Cfor 30– few large seeds with potentially higher seedling survi-
60 minutes, then rinsing in distilled water three times vorship,ormanysmallwidelydispersedseeds,depends
before freeze drying. The samples were then crushed uponecologicalcircumstance(Grime2001).Thequan-
using an agate mortar and pestle. titative data from the modern weed survey studies are
Thehomogenisedpowdersofeachplantsamplewere thus not directly comparable with archaeobotanical
weighedintotincapsulesforIRMSanalysisonaSerCon data. Analysis was therefore conducted in two ways:
EA-GSL mass spectrometer, with δ13C and δ15N quantitatively (using the number of quadrats in
measured separately. An internal alanine standard was which each taxon was found); and semi-quantitatively
usedtocalculaterawisotopicratios.Forδ13Ctwo-point (using presence/absence of species in each cultivated
normalisation to the VPDB scale was obtained using field). The latter is more directly comparable between
four replicates each of IAEA-C6 and IAEA-C7, while modern and archaeobotanical datasets but involves
for δ15N the standards were caffeine and IAEA-N2. some loss of information (Charles et al. 2002).
Reported measurement uncertainties are the calculated Discriminant analysis was used to distinguish crop
combineduncertaintyoftherawmeasurementandrefer- fields cultivated under relatively high- and low-inten-
ence standards, after Kragten (1994). The average sity regimes in southern Morocco, and in southern
measurement uncertainty for δ13C was ±0.09‰, and and northern Morocco combined. For this purpose,
±0.23‰ for δ15N. These calculations were performed an average score for each attribute was determined
using the statistical programming language R. (3.0.2). for each cultivated field as follows:
(cid:1)
The δ13C and δ15N values of carbonised crop remains nak
(cid:1)i i i
were corrected for the effect of charring by subtracting nk
0.11‰ and 0.31‰, respectively, from their determined i i
δ13Candδ15Nvalues(Nitsch,Charles,andBogaard2015). Forthequantitativemeasure:k =numberofquadratsin
i
The Δ13C values of modern cereal grains and pulse whichtheithspecieswasrecorded,a =valueofattribute
i
seedswerecalculatedfromthedeterminedδ13Cvalues fortheithspecies,andn=numberofspeciesrecordedin
(δ13C ) and an average δ13C value of atmospheric eachfield.Forthesemi-quantitativemeasure,kisalways
plant
CO (δ13C ) determined from air sampled at weekly equalto1andsothenumeratorissimplythesumofthe
2 air
intervalsduringthemonthsthatthecropsweregrow- attributevaluesforthespeciesinthecultivatedfieldand
ing (White et al. 2015), using the equation below. We thedenominatoristhenumberofspeciesineachfield.
have therefore used more up to date measurements The success of the discriminations was measured in
of the δ13C value of atmospheric CO than in Styring terms of the percentage of fields correctly reclassified
2
etal.(2016),withtheresultthatthemodernMoroccan as‘lowintensity’or‘highintensity’,usingthediscrimi-
cereal Δ13C values are slightly lower than those nantfunctionextractedineachanalysis.
reported in Styring et al. (2016). The Δ13C values of Semi-quantitative data from southern Morocco
archaeologicalcerealgrains/pulseseedswerecalculated were combined with those from previous studies in
from the determined δ13C values (δ13C ) and a southern Europe (Jones et al. 1999, 2000; Bogaard
plant
δ13C value approximated by the AIRCO2_LOESS et al. 2016) in order to discriminate betweenrelatively
air
system (http://web.udl.es/usuaris/x3845331/AIRCO2_ high- and low-intensity regimes across different cli-
LOESS.xls; Ferrio et al. 2005) to enable comparison mate zones. Archaeobotanical samples from mid-late
with the modern values. The equation was defined by third-millennium Tell Brak, Syria (Charles and
Farquhar, Ehleringer, and Hubick (1989): Bogaard 2001) were entered into the classification
phaseofthisanalysisascasesof‘unknown’cultivation
d13C −d13C
D13C = air plant regime. SPSS version 20.0 was used to perform discri-
1+d13Cplant/1000 minant analyses, using the ‘leave-one-out’ option.
Theidentificationofprobablearableweedtaxaassoci-
atedwithcrops,andof cropprocessingstages,is dealt
Data analysis
withinCharlesandBogaard(2001).Sincecropproces-
Inthemodernweedstudiesdiscussedhere,thenumber singdoesnotappeartobiasfunctionalweedecological
of quadrats out of five in which each taxon occurred inferences of management intensity (Bogaard, Jones,
was recorded. By contrast, archaeobotanical data, to and Charles 2005), samples representing various pro-
whichtheweedecologicalmodeldevelopedhereisulti- cessing (by-)products are considered here.
mately applied, were quantified on the basis of num- Canoco for Windows 4.5 and CanoDraw for Win-
bers of seeds per sample. The relationship between dows (ter Braak and Smilauer 2002) were used to
numberofseedsproducedandnumberofplantsreach- carry out correspondence analyses of the weed survey
ing maturity is complex. For a given allocation of datafromMoroccoinordertoexplorevariationinflor-
resources a species may produce either many small isticcompositionamongfields.Theweeddatawereused
seeds or fewer larger ones (Shipley and Dion 1992; intheformofnumberofquadratsoutoffiveperfieldin
Henery and Westoby 2001) and the optimal strategy, which each taxon occurred. In the figures, axis 1 is
ENVIRONMENTALARCHAEOLOGY 7
plotted horizontally and axis 2 vertically.Statistical cal- species occurring in at least seven of the fields presents
culations related to the isotopic data were performed a clear contrast on axis 1 (horizontal) between fields in
using the statistical programming languageR. (3.0.2). the northern Rif region (right) and other regions (left).
For the Morocco study, topographic analysis was In Figure 2(b), with fields coded by husbandry, axis
based on the SRTM 1 Arc-Second Global digital 2 (vertical) distinguishes high-intensity systems in
elevation model (URL: https://lta.cr.usgs.gov/ both northern and southern Morocco, concentrated
SRTMVF). A wide range of topographic variables near the positive (top) end of the axis, from relatively
was analysed, using the Hydrology, Solar Radiation low-intensitymanagementespeciallyinsouthernMor-
and Surface toolsets in ArcGIS 10.3.1. occan rain-fed terraces towards the negative (bottom)
end(Figure2(b)).Croptaxonomy(Figure2(c))clearly
doesnotdrivethispatterning:cropvariationalongaxis
Results
1merelyreflectsgeography(Figure2(a)),andhulled6-
Causes of floristic variation in the weed flora rowbarley–thedominantcropinthisstudy–occurs
under all of these regimes (Figure 2(c)). In sum,
Correspondenceanalysisofall60quadrat-surveyedcrop
Figures 2(a–b) show that geographical differences
fields in Morocco (Figure 2(a)) on the basis of 72 weed
between northern and southern Morocco dominate
wheremanagementintensityislow,butareminimised
or eliminated under high-intensity management fea-
turing intensive manuring, weeding and, especially in
the south, irrigation (Table 2).
InFigure3weedtaxaareclassifiedtoshowtheextent
ofpatterninginselectedfunctionalattributesthatrelate
to soil productivity. Specific leaf area or SLA (Figure 3
(a)), which is positively correlated with productivity
(Table3),showssomepatterninginrelationtomanage-
ment intensity on axis 2 (Figure 2(b)), with a few taxa
havingthehighestvaluesassociatedwithhighintensity,
and others having the lowest values with low intensity.
Taxawithintermediatevaluesaregenerallydistributed.
Similarly, Figure 3(b), coded by values for leaf area per
node: thickness, shows an association of low values
withlow-intensity management.
Intermsofspecies’abilitytorecoverfrommechan-
ical soil disturbance, duration of flowering (Figure 3
(c)),whichreflectsthelengthofthegerminationperiod
(Table 3), shows some weak patterning: species with
long-flowering periods are generally distributed, per-
haps reflecting the practice of hand-weeding in most
regimes and all regions (Table 2), but those which
flowerfor<3monthsaremostlyconcentratedinfields
withrelativelylow-intensitymanagement.Figure3(d),
in which fields are represented as pie-charts showing
thenumberofspeciesperfloweringdurationcategory,
shows that long-flowering species are especially con-
centrated in the high-intensity regimes, whereas
short-flowering species tend to be best represented in
low-intensity systems. Figure 3(e) shows the classifi-
cation of perennial species (only) with and without
vegetative spread (the ability to regenerate from root
fragments under high disturbance – Table 3). Though
thepatternisbasedonasmallgroupofspecies,peren-
nialsthatdonotregeneratefromfragmentsareassoci-
atedwithlowerintensitymanagement.Theassociation
of the high-intensity regimes with long-flowering
annuals and perennials that can spread vegetatively
Figure2.CorrespondenceanalysisofMoroccanfieldsshowing by rhizomes, bulbils or stolons reflects the fact that
(a)geographicalzones;(b)farmingregimes;(c)crop. these plots were weeded very frequently (e.g. every
8 A.BOGAARDETAL.
Figure3.Continued
morning by women in the oasis). The weeds removed
were generally viewed by farmers as a resource rather
than simply as plants lowering crop yield, and were
often used, for example, as salad greens or as fodder
for penned goats. The kinds of palatable plants with
fast-growing leaves (e.g. high SLA – see Wright et al.
2004) valued for these uses are indeed those expected
to grow under highly productive conditions.
None of the drought-related attributes showed the
predictedpatterninginthecorrespondenceanalysis,per-
haps in part because each of the regimes studied was
located in a different rainfall zone, complicating further
potentialrelationshipsamongspecies’droughttolerance,
drought avoidance and seasonality (cf. Charles et al.
2003). Thus, late-flowering species, previously shown in
Figure 3. Correspondence analysis of Moroccan weed taxa
a single study region in Jordan to distinguish artificially
showing(a)SLA;(b)leafareapernode:thickness;(c)flowering
period; (d) fields as pie-charts showing number of weed taxa watered fields (Charles et al. 2003), were ubiquitous
presentperfloweringdurationcategory;(e)vegetativespread. acrosstheMoroccanregimes(plotnotshown).
ENVIRONMENTALARCHAEOLOGY 9
Overall, the weed flora of the Moroccan fields vary and rain-fed fields was achieved on the basis of fully
primarily by geography (north versus south), but quantitative (Figure 4) and semi-quantitative data
these differences are overwritten where management (Figure 5). In both of these analyses, functional attri-
intensity is high. Contrasts between high- and low- butes discriminate between high-intensity oasis and
intensity management show patterning in relation to low-intensity rain-fed fields as predicted. Species that
functional traits reflecting species’ potential growth grow rapidly (high SLA, leaf area per node: thickness)
rate,andhencetheirabilitytorespondtohighsoilpro- andhavetall,broadcanopiesareassociatedwithinten-
ductivity, as well as functional attributes reflecting sively maintained oasis conditions; a long-flowering
species’responsetodisturbance:thelengthoftheflow- period is also associated with greater disturbance
ering period and (for perennials) vegetative spread. (more thorough tillage and weeding) in oasis fields
comparedtorain-fedterracedfields.Thus,arestricted
setof functionalattributesrelating tosoil productivity
Ecological comparison of Moroccan regimes on
and disturbance has been shown to distinguish con-
the basis of weed functional attribute values
trastingintensitylevels,whetherbasedonfullyquanti-
Discriminantanalysiswasusedtodistinguishbetween tative or semi-quantitative data, as in previous studies
the weed flora of low- and high-intensity agrosystems in southern Europe (Jones et al. 2000; Charles et al.
inMorocco,usingfivefunctionalattributespreviously 2002; Bogaard et al. 2016).
shown to distinguish between high- and low-intensity Further discriminant analyses performed on the
cultivation regimes in southern Europe (Jones et al. basis of semi-quantitative data – i.e. the form that is
2000; Bogaard et al. 2016) as discriminating variables. applicablearchaeologically–showedthatthesouthern
Additional drought tolerance/avoidance attributes Moroccanregimescanbeseparated(witha100%cor-
(Table3)werealsoincludedasdiscriminatingvariables rectreclassification)onthebasisof a singlefunctional
but contributed little to the regime comparisons con- attribute,SLA(plotnotshown).Thisresultreflectsthe
sideredbelow,likelyduetotheconflationofagronomic extremity of the productivity contrast between oasis
with climatic differences in water availability among and rain-fed terraced fields (Table 2). Previous discri-
the study regions. Analyses including drought toler- minant analyses of relatively high- and low-intensity
ance/avoidance attributes are therefore not shown. fieldsin Evvia,Greece (Joneset al. 2000) and Asturias
Discriminant analysis was carried out initially on versus Haute Provence (Bogaard et al. 2016) involved
the southern Moroccan fields only, to distinguish more subtle ecological contrasts requiring a broader
high-intensity management of oasis fields from lower suite of functional attributes (reflecting soil fertility
intensity management of rain-fed terraced fields. A and disturbance) to be distinguished successfully. It is
100% correct reclassification of both southern oasis also worth noting that a relatively high degree of
Figure4.(a)Therelationshipofsouthernrain-fedterracedfields(opencircles,n=15)andoasisfields(filledcircles,n=16)tothe
discriminant function extracted to distinguish these two groups on the basis of fully quantitative data (larger symbols indicate
groupcentroids);(b)correlationsbetweenthediscriminatingvariablesandthediscriminantfunction.