Durum wheat (Triticum durum Desf.) in rotationwith faba bean (Vicia faba var. minor L.): long-termsimulation case study

P. GarofaloA, E. Di PaoloB, and M. RinaldiA,C

ACRA–Unità di Ricerca per lo studio dei Sistemi Colturali degli Ambienti caldo-aridi, Bari, Italy.Email: pasquale.garofalo@entecra.it

BCentro per la Sperimentazione e Divulgazione delle Tecniche Irrigue, Vasto, Italy. Email: dipaolo@cotir.itCCorresponding author. Email: michele.rinaldi@entecra.it

Abstract. The aim of this work was to apply the CropSyst simulation model to evaluate the effect of faba bean cultivationas a break crop in the continuous durum wheat cropping system in southern Italy. The model was previously calibrated andvalidated for durum wheat and faba bean on data derived from experiments carried out in southern Italy (for different yearsand treatments), comparing observed and simulated crop growth, yield, soil water, and nitrogen output variables.

The validation showed good agreement between simulated and observed values for cumulative above-ground biomass,greenarea index, and soilwater content for both crops andgrainyield for durumwheat; a negative correlation forgrainyield infaba beanwas observed due to a reduction in harvest index in thewell-watered crop, which themodel does not simulatewell.

Subsequently, a long-term analysis was carried out to study the effects on durum wheat of introducing a legume crop inrotation with the cereal in 2 and 3-year sequences.

A long-term simulation, based on 53 years of daily measured weather data, showed that faba bean, due to a lower level oftranspiratedwater (on average247mmfor durumwheat and197mmfor fababean), allowed for greater soilwater availabilityat durumwheat sowing for the cerealwhen in rotationwith a legume crop (on average, +84mm/m for durumwheat followingthe faba bean), with positive effects for nitrogen uptake, above-ground biomass, and grain yield of wheat. The yield increaseof wheat when following a faba bean crop was on average +12%, but this effect was amplified in drier years (up to 135%).

In conclusion, the case studyoffered the potential to confirm the positive results previously obtained in long/medium-termfield experiments on the introduction of faba bean in rotation with durum wheat, as well as reduction in the chemicalapplication of nitrogen.

Additional keywords: CropSyst, legumes, cereals, soil water content, soil nitrogen content, Mediterranean environment,grain yield.

Introduction

Continuous durum wheat crop is one of the most commoncropping systems in Mediterranean regions, especially whereirrigationwater is scarce or entirely lacking.This cropping systemis characterisedby low inputs (tillage, fertilisers, pesticides), but itcan produce negative effects on chemical soil properties such as amodification of organicmatter quality andmacro-nutrient content(Blair and Crocker 2000). Consequently, it may be necessary toincrease the level of mineral fertilisation to obtain a satisfactorygrain yield or, alternatively, to use a suitable system of croprotation. Nowadays, in southern Italy, tomato and sugar beet arethe main crops in 3- and 4-year rotations with durum wheat but ifthere is a complete lack of irrigation water, a winter crop could beused more efficiently to alternate the cereal.

Technical progress in soil tillage, fertilisation, and plantprotection has reduced the agronomic significance of croprotation design and consequently the preceding crop effectsof grain legumes have become less and less significant forfarmers than in previous decades, while the competitiveness of

the sole winter cereal has come to the fore. However, even if thevalue of grain legumes as a preceding crop appears to be wellknown, legume–cereal rotations have not been widely adopted.

Leguminous crops induce various benefits for the soil, suchas an increase in soil nitrogen, low nitrate leaching (if in rotationwith cereals), improvements in soil structure, the destructionof pest and disease cycles, stability, and erosion prevention(Senaratne and Hardarson 1988; Chalk 1998). They can alsoplay a meaningful role in enhancing the nitrogen nutrition of thesubsequent crop (Russell and Hargrov 1989). The crop residuescan be further incorporated into the soil by ploughing, while notillage ensures ground mulching. In the first case, nutrients aredirectly supplied to the soil, and in the second, positive benefitsare evident in terms of soil water balance and weed control(Lal et al. 1991; Dou et al. 1994).

Over the last 10 years, grain legumes have aroused newinterest as low-input crops to introduce in rotation with cerealsas they are capable of improving soil fertility and properties andas a consequence of the BSE crises that led to an increased need

� CSIRO 2009 10.1071/CP08208 1836-0947/09/030240

CSIRO PUBLISHING

www.publish.csiro.au/journals/cp Crop & Pasture Science, 2009, 60, 240–250

for plant proteins (EC Regulation no. 1259/EC Council 1999).Furthermore, the revision of the criteria for the assignment of UEfinancial support linked to sustainable agronomical practicesencouraged the cultivation of grain legume crops (ECRegulation no. 1782/EC Council 2003).

The potential of using simulation models such as CropSystto study both different crops in rotation (including legumecrops) and the dynamics of some soil properties in croppingsystems (i.e. organic matter, N content) is of great interest inmodern agronomy. The evaluation of cropping systems needslong-term experiments to obtain significant results, but is notalways easily extended to locations with different pedo-climaticconditions. Mechanistic simulation models can be useful tools toovercome these limitations, representing trends and differencesdue to the effect of cropping systems and agronomic techniqueson crop yield and environment. The faba bean is a marginal cropused for animal feed and greenmanure: it has not been adequatelyparameterised for modelling purposes and is not present in thecrop list of the CropSyst model.

Theprincipal aimsof this study are: (i) to calibrate andvalidatethe CropSyst model for durumwheat (Triticum durumDesf.) andfaba bean (Vicia faba var. minor L.) in southern Italy and (ii) tosimulate durum wheat in continuous cropping and in 2- and3-year rotationswith fababean inorder to compare cereal responseon a long-term basis.

Materials and methodsThe model

CropSyst (Cropping Systems Simulation Model) is a multi-year,multi-crop, daily time-step crop growth simulation model,designed to serve as an analytical tool to study the effect ofcropping system management on crop productivity and on theenvironment (Stöckle et al. 2003). Themodel has been applied tosimulate various crops, cereals (maize, durum wheat, barley,sorghum) and leguminous crops (soya bean, alfalfa, lupin)over different regions, generally with good results (Pala et al.1996; Stöckle et al. 1997; Giardini et al. 1998; Bellocchi et al.2002; Confalonieri and Bechini 2004). In southern Italy it haspreviously been calibrated and validated for durum wheat,sunflowers, sorghum, soya bean and faba bean. (Donatelliet al. 1997; Rinaldi and Ventrella 1997; Ventrella and Rinaldi1999; Di Paolo et al. 2007).

The CropSyst model can be considered a generic cropsimulator, simulating different crops from a common set ofparameters. Daily potential crop growth is a function of solarradiation andwater transpiration. The duration of the phenologicalphase is calculated as the sum of heat units, modulated wherenecessary by photoperiod and vernalisation requirements; theaccumulation of thermal time may be accelerated by waterstress. Crop yield production is calculated according to theharvest index and a translocation factor.

The water balance includes rainfall, irrigation, runoff,interception, infiltration, redistribution in the soil profile, croptranspiration, and soil evaporation. Water dynamic in the soil ismodelled by a simple cascading approach or by Richards’equation, the latter solved numerically using the finitedifference technique.

The nitrogen balance includes soil N transformations(mineralisation, nitrification, denitrification, volatilisation),ammonium sorption, symbiotic N fixation, crop N demand,and crop N uptake. This latter is calculated by the modelaccording to the following formula:

Nup1 ¼ UPmax � RL� Navail � PAW2 ð1Þwhere Nup1 is the potential nitrogen uptake (kg/ha.day) for eachsoil layer, UPmax is the maximum N uptake per unit length ofroot (kg/ha.day), RL is root length; Navail (0–1) is a nitrogenavailability factor (model input), andPAWisplant-availablewater.

The potential daily above-ground biomass production(AGBpt) is calculated as the minimum value between theapproach for transpiration use efficiency (Tanner and Sinclair1983) and that for radiation use efficiency (Monteith 1977). Atthis point, water and nitrogen limits are applied to AGBpt tocalculate actual AGB production (AGBa). The biomassproduction affected by a water-limited condition (AGBt) iscalculated by multiplying AGBpt by the ratio of actual topotential transpiration; obviously this ratio increases withavailable soil water. Consequently, daily AGBa production iscalculated by multiplying AGBt by a coefficient based on plantnitrogen concentration (Stöckle and Debaeke 1997), the latterinfluenced by Nupl derived from Eqn 1.

The N2-fixation amount is simulated by CropSyst as thedifference between potential nitrogen demand and potentialnitrogen uptake. N2-fixation is a function of available plantwater, soil mineral nitrogen content in the root zone, and soiltemperature. The user can neither modify the magnitude of theserelationships nor the nitrogen cycle. Few crop parameters can beset to improve the simulation result (mainly N concentration andthe chemical characteristics of residues). Further information anda description of the main processes simulated by CropSystcan be found in Stöckle et al. (2003) and in the user manualdocumentation (www.bsyse.wsu.edu/cropsyst/).

In this study, CropSyst version no. 4.05.5 was used. Potentialevapotranspiration was estimated with the Penman-Monteith orPriestley-Taylor formula as a function of available daily weatherdata: from 1952 to 1978 with the Priestley-Taylor, from 1979to 2006 with the Penman-Monteith formula. The cascadingapproach was used to simulate the water dynamic, using thesoil layer depths and hydrological characteristics as reported inTable 1. To take into account the effects of faba bean residuesincorporatedwithin the soil, the “microbial, stable organicmatterand residue with carbon decomposition” option was checked inthe “Option-Organic matter” menu.

Field experiments

Sites and climates

The experiments for calibration and validation activities werecarried out in two localities of southern Italy, Foggia and Vasto.

In Foggia: (418270N, 158040E; 90m a.s.l.) the soil is aVertisol of alluvial origin, Typic Calcixerert (Soil Taxonomy10th edn, USDA 2006), silty-clay with characteristics reportedin Table 1. The climate is “accentuated thermo-Mediterranean”(FAO-UNESCO 1963), with minimum temperatures below 08Cin the winter and maximum temperatures above 408C in thesummer. Annual rainfall (mean 550mm, considering a 50-year,

Durum wheat/faba bean rotation Crop & Pasture Science 241

long-term period) ismostly concentrated during thewintermonthsand class “A pan” evaporation exceeds 10mm day in summer(average of maximum daily values recorded in July and August).

In Vasto: (428100N; 148380E; 30m a.s.l.) the soil is aVertisol of alluvial origin (Aquic Haploxerert, according to the10th edn of Soil Taxonomy, USDA 2006), silty-clay-loam withcharacteristics reported in Table 1. The climate is “attenuatethermo-Mediterranean” (FAO-UNESCO 1963), with minimumtemperatures below08C in thewinter andmaximum temperaturesof about 34–368C in the summer. Annual rainfall (mean 650mm,considering a 40-year, long-term period) is mostly concentratedduring the autumn and spring months and class “A pan”evaporation fluctuates between 6 and 8mm day on clear daysduring July and August.

In southern Italy, durum wheat is cropped in hilly, but mainlyin flat areas (such as Foggia), where the main limiting factoris water availability, especially during the reproductive phase(April–May).Data derived from long-termvariety trials (AA.VV1975–2008) show a large variation in southern Italy durumwheatgrain yield, from 0.8 to 6.5 t/ha, with an average of 4.0� 1.2 t/ha,principally depending on spring rainfall.

Field crop experiments

In Foggia, the data used were derived from two field experimentscarried out in the 1996–97 and 1997–98 growing seasons onthe comparative response of durum wheat to supplementaryand rainfed irrigation. Common crop management wasimplemented in the model, with sowing at the beginning ofNovember and harvest in the middle of June. Nitrogenfertiliser was applied at a rate of 100 kgN/ha, half at sowingtime, with a fertiliser such as urea, and half at the end of tillering,stage 5 on Feek’s scale (Large 1953), with a fertiliser such asammonium nitrate. Irrigationwas carried out at flowering (beforeheading, stage 10) with a fixed amount of 110mm, using asprinkler method.

In Vasto, in the 2005–06 growing season, the followingirrigation regimes were tested on faba bean:

I0, rainfed;I100, restitution of 100% of the actual evapotranspiration(ETa);I50, restitution of 50% of ETa;IF, irrigation with 40mm at flowering;IR, irrigation with 40mm at seed formation;IFR, irrigation with 40mm at flowering and seed formation.

In the second growing season (2006–07), irrigation regimes IFand IFR were not tested.

Data recorded

Soil water content (m3/m3) was measured monthly in both theexperimental stations in the upper 0–0.60m range using thegravimetric method. The seasonal evapotranspiration (ETa)was estimated from a simplified water balance.

Crop sampling focussed on above-ground dry biomass (kg/ha),green leaf area index (GAI, m2/m2) with a Delta T devices areameter (Decagon Devices Inc., WA, USA), main phenologicalphases (seedling emergence, peak of leaf area development,flowering, and physiological maturity) and, at harvest, dry plantbiomass and grain yield (kg/ha). Furthermore, N content (%) wasmeasured for the faba bean plant using the Kjeldhal methodduring the crop cycle and referred to a base area (kg N/ha).

Calibration and validation

Individual simulation runs were prepared for all“crop� year� treatment” interactions. Weather inputs –

rainfall, maximum and minimum air temperatures, solarradiation, maximum and minimum relative humidity, and windspeed – were measured in weather stations located near theexperimental sites. Soil data, initial soil water content, initial soilnitrogen content and management schedules were also recorded.

Table 1. Soil input information used by the CropSyst model for the two locationsMethodologies: sand, clay, and silt content with the hydrometer method (or Bouyoucos method); permanent wilting point (PWP) and field capacity (FC) withRichard’s plates at –15 and –0.03MPa pressure; bulk density (BD) with undisturbed core soil sampled in Kopechy rings; total N with Kjeldhal; organic matter(OM) with the Walkley-Black method; pH with water extraction at 2 : 1 ratio. Ksat, Saturated hydraulic conductivity; AEP, air entry potential; Sat., saturation

Layer Thickness Sand Clay Silt PWP FC BD Tot. N KsatA AEPA OM Sat.A pH

no. (m) (%) (m3/m3) (t/m3) (g/kg) (m/day) (J/kg) (%) (m3/m3)

Foggia [Typic Calcixerert (Soil Taxonomy, USDA 2006)]1 0.1 12.9 43.7 43.4 0.22 0.44 1.20 1.3 0.091 –3.82 2.0 0.55 8.332 0.1 12.9 43.7 43.4 0.22 0.44 1.20 1.3 0.091 –3.82 3.0 0.55 8.453 0.2 12.9 43.7 43.4 0.22 0.44 1.20 1.3 0.091 –4.16 2.7 0.55 8.514 0.3 9.6 54.6 35.8 0.25 0.44 1.45 1.0 0.004 –11.99 2.7 0.45 8.655 0.4 21.5 34.6 43.9 0.19 0.35 1.29 0.7 0.083 –2.91 1.6 0.51 8.576 0.2 34.4 27.7 37.9 0.16 0.30 1.35 0.7 0.108 –1.65 1.6 0.49 8.42

Vasto [Aquic Haploxerert (Soil Taxonomy, USDA 2006)]1 0.1 5.8 44.7 49.5 0.15 0.41 1.22 1.6 0.074 –5.24 2.0 0.54 8.332 0.1 6.2 44.7 49.1 0.15 0.41 1.22 1.6 0.073 –5.22 2.0 0.54 8.103 0.2 5.4 44.3 50.3 0.18 0.40 1.22 1.7 0.076 –5.18 2.0 0.54 8.204 0.2 4.9 42.3 52.8 0.18 0.38 1.23 1.6 0.082 –4.89 1.8 0.54 8.315 0.2 4.3 42.4 53.3 0.18 0.38 1.23 1.5 0.083 –4.91 1.7 0.54 8.226 0.2 3.0 42.0 55.0 0.18 0.38 1.23 1.5 0.087 –4.86 1.7 0.54 8.25

AComputed by the model from user-specified values.

242 Crop & Pasture Science P. Garofalo et al.

Calibration was carried out with a tool implemented in theCropSyst model, CropCalibrator, comparing simulation resultsand observed values of phenology, cumulative above-ground drybiomass, green area index, specific leaf area, plant nitrogenconcentration (only for faba bean), above-ground dry biomassat harvest, and grain yield. After calibration, some parametersestimated by the model were further manually tuned to obtain thebest fit between simulated and observed values.

For the calibration of durumwheat (cv. Simeto), the data usedwere for irrigated treatment (110mm at flowering stage with theaim of replacing potential evapotranspiration) derived from theexperiment carried out in 1997–98 (cv. Simeto).

The calibration of faba bean (cv. Chiaro Torre Lama) provedto be more difficult because this crop was not present in themodel default documentation and few parameters are reported inliterature. It was performed using observed values derived from a2-season experiment considering the optimal irrigation scheduling(I100) treatment for both seasons (2005–06 and 2006–07).

Model validationwasperformedusing an independent data setof the same variables used for calibration.

For durum wheat, the data set used for validation was rainfedtreatment in 1997–98 and both rainfed and irrigated treatments inthe 1996–97 season. Additional data used were those for plantbiomass and grain yield of durum wheat derived from long-termexperiments (comparisons between cropping systems and strawsoil incorporation methods) carried out in Foggia in the 1978–2001 period (Rizzo et al. 1990;Maiorana et al. 1992, 1993, 1997,2004; Di Bari et al. 1993).

In the validation phase of faba bean we used the 2-year dataset (2005–06 and 2006–07) from the irrigation schedulingexperiment, excluding the I100 treatment, previously used inthe model calibration.

The validation processwas carried out calculating a number ofstatistical indices: root mean square error (RMSE), modellingefficiency (ME) and residual mass coefficient (RMC) (Loagueand Green 1991). Pearson’s correlation coefficient and graphicaljudgements were also used.

Simulation of long-term rotations

The CropSyst model was applied to a seasonal analysis from1953 to 2006 using daily weather data derived from theagrometeorogical station of the CRA experimental farm inFoggia. The aim was to compare wheat cropped as a“continuous crop” (CC) and in sequence with faba bean in2-year (R2) and 3-year rotations. In the 3-year rotations, wheatfollowing faba bean was labelled “R3_F1” and wheat followingwheat was named “R3_F2”. For the 2-year and 3-year rotations,the simulations were performed starting with the different crops(all the phases of the rotation for every year), for a total of 6experimental runs.

In the simulations, durum wheat was fertilised with100 kgN/ha in two applications, split half at sowing and halfat 90 days after emergence. No nitrogen application was used forfaba bean. Straw was removed in the case of wheat and was soil-incorporated for faba bean. Nitrogen balance components andsoil moisture at sowing were examined.

Theyearly output of the two simulations for the 2-year rotationand the three runs for the 3-year rotation were merged, so the endresult was 53 yearly output values of durum wheat in the three

compared rotations. Average values and differences expressed aspercentages are shown in tables and displayed as figures.

The homogeneity of variance was checked by means of theLevene test and a rejection of the assumption was countered bymeans of the Welch ANOVA test. Subsequently, the Shapiro-Wilk test analysis was applied to verify assumptions about thenormality of the data.After several attempts to normalise the data,a non-parametric test (Wilcoxon/Kruskal-Wallis) was used toevaluate differences between means of treatment.

Results and discussion

Calibration and validation

Table 2 shows the values of crop parameters of durum wheat andfaba bean derived from calibration, direct measurement, or bydefault. Faba bean is not present in the crop list of the model; forthis reason, we modified the pea crop parameters, taking intoaccount the different crop habit and physiology.

Wheat crop parameters were similar to those reported foranother cultivar by Ventrella and Rinaldi (1999) in the sameenvironment, especially for crop phenology. The most sensitiveparameter to biomass yield in CropSyst is transpiration-useefficiency (TUE) (Confalonieri and Bechini 2004) for whichwe obtained a value greater than those reported by Ventrellaand Rinaldi (1999) (6.5 v. 3.7 kPa kg/m3). However, there arewidely varying results in the literature for this coefficient; Bechiniet al. (2006) obtained values of TUE equal to 5.8 kPa kg/m3 innorthern Italy. In general, water-limited conditions force theplant to adopt a more efficient use of water. For RUE, the valueof 3.5 kg/MJ was similar to those reported by the authorsmentioned previously.

Specific crop parameters for the faba beanwere also estimatedand calibrated. The large TUE is due to the very large vegetativehabit of this cultivar (maximum LAI equal to 7m2/m2 and SLAequal to 28m2/kg), which the CropCalibrator estimated to fitexperimental data.

The validation process showed a good agreement betweensimulated and observed values of above-ground biomass, greenarea index, and soil water content during the entire crop cycle,both for wheat and faba bean (Tables 3 and 4). For grain yieldand plant dry matter at harvest, we used several data sets oncontinuous cropping and the overall RMSE obtained was verygood (21.4% and 16.4%, respectively). Moreover, modellingefficiency, the coefficient of residual mass, and Pearson’scoefficient correlation confirmed the suitability of the modelfor durum wheat.

For faba bean we observed a good agreement betweensimulated and measured values for most of the parametersanalysed (Table 4), as well as for phenology: average values(relative to the 2005–06 growing season) in terms of GDDfor plant emergence, initial flowering and grain filling, werewell in accordance with observed values of 195, 848, and 990degree-days, respectively, compared with 199, 858, and 1060degree-days obtained by CropSyst.

As regards the seed yield, a negative correlationwas observed,caused by a reduction of the harvest index in the well-wateredcrop, which the model did not take properly into account by notconsidering the time when the water stress occurred. In fact, thefaba bean is a typical semi-arid environment crop and optimal

Durum wheat/faba bean rotation Crop & Pasture Science 243

Table 2. CropSyst model crop parameters for durum wheat (cv. Simeto) and faba bean (cv. Chiaro Torre Lama)C, Calibrated values; M, measured values; D, default values

Parameter Unit Durum wheat Faba bean

Growth Above-ground biomass transpiration coefficient kPa kg/m3 6.5 C 8.5 CUnstressed light to above-ground biomass g/MJ 3.5 C 6.6 COptimum mean daily temperature for growth 8C +18 D +20 D

Leaf Maximum expected LAI m2/m2 5 D 7 CSpecific leaf area m2/kg 17 M 28 CStem/leaf partition coefficient 3.1 M 3.7 CLeaf duration Degree-days 720 M 600 CLeaf duration sensitivity to water stress 1 M 1 DFraction of maximum LAI at maturity 0.8 D 0.8 D

Root Maximum rooting depth m 1.6 D 1.2 D

Transpiration Extinction coefficient for solar radiation 0.48 D 0.45 DET crop coefficient at full canopy 1.05 D 1.05 DMaximum water uptake mm/day 10 D 9 DCritical leaf water potential J/kg –1500 C –1000 DWilting leaf water potential J/kg –2200 C –1500 D

Phenology Emergence Degree-days 120 C 197 CPeak LAI Degree-days 1400 C 1259 CBegin flowering Degree-days 1500 C 854 CBegin filling Degree-days 1756 C 1066 CPhysiological maturity Degree-days 2316 C 2183 CBase temperature 8C 0 D +2 CCut-off temperature 8C +30 C +25 DUnstressed harvest index 0.30 C 0.26 C

Grain sensitivity to water and nitrogen stressDuring flowering 0.10 D 0.05 CDuring grain filling 0.05 D 0.05 CTranslocation to yield factor 0.30 D 0.26 D

Nitrogen Nitrogen availability adjustment 1 D 1 DAmount of residual nitrogen per soil layer kg/ha 1 D 10 CMax N concentration during early growth kg N/kg DM 0.05 D 0.05 CMax N concentration at maturity kg N/kg DM 0.015 D 0.07 CMin N concentration at maturity kg N/kg DM 0.007 D 0.02 CMax N content of standing stubble kg N/kg DM 0.007 D 0.03 C

Crop residues Top and root carbon fraction % 0.70 C 0.46 CResidue top biomass fast cycling % 0.1 C 0.5 DResidue top biomass slow cycling % 0.3 C 0.4 DLignified biomass % 0.6 C 0.1 DResidue root biomass fast cycling % 0.1 C 0.5 DResidue root biomass slow cycling % 0.3 C 0.3 DLignified biomass % 0.6 C 0.2 D

Table 3. Indices of agreement between observed and CropSyst simulated values for some output variables in durum wheatGAI, Green leaf area index during crop cycle; TDM, total drymatter during crop cycle; FDM, total drymatter at harvest; ETa, seasonal actual evapotranspiration;SWC, soil water content; RMSE, root mean square error; ME, modelling efficiency; CMR, residual mass coefficient; r, Pearson’s correlation coefficient

Variable Unit No. of Obs. s.d.obs Sim. s.d.sim Diff. RMSE ME CMR rdata (%)

GAI m2/m2 19 2.03 1.06 2.14 1.17 5.46 24.19 0.79 0.00 0.90TDM kg/ha 20 5449 3728 5552 4117 1.90 17.63 0.93 –0.02 0.97FDM kg/ha 13 8949 2617 9299 3088 3.90 16.46 0.66 –0.04 0.88Yield kg/ha 25 2843 863 2733 653 –3.88 21.38 0.48 0.04 0.71ETa mm 3 427 80 396 85 –7.19 8.08 0.73 0.07 0.97SWC 0–0.20m m3/m3 23 0.26 0.05 0.27 0.06 1.80 14.70 0.40 –0.02 0.80SWC 0.20–0.40m m3/m3 23 0.27 0.06 0.29 0.07 5.00 14.46 0.52 –0.05 0.85SWC 0.40–0.60m m3/m3 23 0.28 0.06 0.30 0.06 8.32 16.53 0.44 –0.08 0.81

244 Crop & Pasture Science P. Garofalo et al.

Table 4. Indices of agreement between observed and CropSyst simulated values for some output variables in faba beanGAI, Green leaf area index during crop cycle; TDM, total drymatter during crop cycle; FDM, total drymatter at harvest; ETa, seasonal actual evapotranspiration;Nitrogen, plant N content during crop cycle; SWC, soil water content; RMSE, rootmean square error;ME,modelling efficiency; CMR, residualmass coefficient;

r, Pearson’s correlation coefficient

Variable Unit No. of Obs. s.d.obs Sim. s.d.sim Diff. RMSE ME CMR rdata (%)

GAI m2/m2 24 3.28 2.26 3.43 1.97 4.83 24.45 0.87 –0.05 0.94TDM kg/ha 49 5863 6556 6048 6699 3.16 28.59 0.93 –0.03 0.97FDM kg/ha 8 16 790 3845 16 451 2778 –2.02 15.09 0.51 0.02 0.72Yield kg/ha 8 4584 501 4518 1010 –1.44 28.49 –6.78 0.01 –0.66ETa mm 4 379 65 361 64 –4.75 5.24 0.88 0.05 0.99Nitrogen kg/ha 13 296 200 252 133 –14.62 28.87 0.80 0.15 0.97SWC 0–0.10m m3/m3 46 0.32 0.06 0.33 0.05 3.96 12.48 0.56 –0.04 0.79SWC 0.10–0.20m m3/m3 51 0.31 0.08 0.33 0.05 7.05 16.26 0.61 –0.07 0.84SWC 0.20–0.40m m3/m3 51 0.33 0.06 0.34 0.05 1.78 10.97 0.68 –0.02 0.83SWC 0.40–0.60m m3/m3 51 0.33 0.06 0.34 0.05 1.06 9.26 0.71 –0.01 0.84

Durum wheat

0

5000

10 000

15 000

20 000

25 000

0

2

4

6

8

Faba bean

0 50 100 150 200 250

0.0

0.1

0.2

0.3

0.4

0.5

0 50 100 150 200 250

Days after sowing

Wat

er c

onte

nt (

m3 /

m3 )

Gre

en a

rea

inde

x (m

2 /m

2 )B

iom

ass

(kg/

ha)

(a)

(b)

(c)

Fig. 1. Comparison between observed and CropSyst simulated data of (a) total dry biomass, (b) green area index, and (c) soil water content(0–0.6m) during the crop cycle of wheat in 1996–97 season (rainfed treatment), and of faba bean in 2005–06 season (IR treatment), in thevalidation process.

Durum wheat/faba bean rotation Crop & Pasture Science 245

water availability during the February–April period can causeexcessive vegetative growth.When seed formation andfinal yielddo not increase proportionally, this leads to a lowering of theharvest index.

Figure 1 shows a comparison between simulated and observedvalues of above-ground biomass, green leaf area index, and soilwater content during the crop growth of wheat and faba bean inthe validation process. The model agrees quite well with thebehaviour of the measured temporal data.

Cropping systems simulation

Durum wheat yield and growth

The CropSyst model simulated a positive effect on durumwheat when in rotation with faba bean, especially if thecereal followed the legume crop (Table 5). Indeed, significant

differences between wheat in continuous cropping (CC) andwheat following the faba bean were observed: for CC the totaldry biomass was 8186 kg/ha with a yield equal to 2377 kg/ha,statistically similar to those obtained in R3_F2, while greateryields were observed for wheat in R2 and in R3_F1 for above-ground biomass and grain yield (an average long-term value of9115 kg/ha and 2684 kg/ha, respectively). The enhancementwas equal to 11% for total biomass and 13% for yield. Despitethe fact that this increase was below the RMSE calculated in thevalidation process for grain yield, in dry years this increase wassignificantly higher.

Figure 2 shows the cumulative rainfall during the crop cycleand during the fallow period (July–October). The greatestincreases in grain yield were evident during the growingseasons when the rainfall was lowest, as shown in Fig. 3. Forexample, in the years 1961, 1972, and 1976, the driest years of thesimulated period, the grain yield improvement for R2 and R3_F1cropping systems was 140%, 106%, and 88%, respectively, forthe 3 years, on average larger than the model error in grain yieldsimulation. This positive rotation effect is explained by differentvalues of soil water content at wheat sowing and it is amplifiedduring drought years (Fig. 4). In fact, Table 5 shows an averagevalue of soil water content (0–0.6m depth) of about 0.37m3/m3

(about 85% of crop-available water) for R2 and R3_F1, while forCC and R3_F2, water content at sowing was 0.28m3/m3 (about45% of crop-available water), with the soil significantly drier inCC and R3_F2 than in R2 and R3_F1. The reasons for less waterdepletion from the faba bean are a shallower root system of fababean (1.2m v. 1.6m of wheat) and its greater transpiration-useefficiency compared with wheat (8.5 v. 6.5 kPa kg/m3).

The temporal variation of wheat biomass and grain yield wasstrongly affected by the introduction of faba bean in rotation, witha coefficient of variation that decreased from about 25% in CCwheat to15%inwheat in rotationwith the fababean.Weobserved(Fig. 3) that in 2001 there was a complete failure of crops in CCandR3_F1 caused by particularly scarce rainfall during the initial

Table 5. Averages and standard deviations of durum wheat simulatedfor 53 years by CropSyst in different rotations

For each row, P< 0.001*** and P< 0.01** of analysis of variance (WelchANOVA test). Within rows, values followed by the same letter are not

significantly different at P= 0.05 (Wilcoxon/Kruskal-Wallis test)

CC R2 R3_F1 R3_F2

Plant biomass atharvest (kg/ha)**

8186b 9106a 9123a 8462b

s.d. 1924 1335 1357 1791

Grain yield (kg/ha)** 2377b 2681a 2687a 2467bs.d. 592 388 380 552

Nitrogen uptake (kg/ha)*** 194b 218a 213a 198bs.d. 35 21 20 29

Nitrogen leaching (kg/ha)*** 9.0b 4.3a 3.1a 0.7bs.d. 35.8 6.0 4.9 2.2

Soil water contentat sowing (m3/m3)***

0.28b 0.37a 0.36a 0.28b

s.d. 0.06 0.06 0.06 0.06

0

200

400

600

800

1950 1960 1970 1980 1990 2000

Rai

nfal

l (m

m)

Fig. 2. Rainfall duringwheat crop growth seasons (Nov.–June, open bars) and between harvest andfollowing sowing date (July–Oct., solid bars). Long-term average of Nov.–June rainfall (continuousline) and July–Oct. (dotted line) is also displayed.

246 Crop & Pasture Science P. Garofalo et al.

stage of crop growth, as confirmed by De Vita et al. (2007) in afield experiment on durum wheat in the Apulia region.

The different crop growth dynamic of wheat (GAI and TDM)in the two rotations is shown in Fig. 5 with the wheat cropped in1975–76, a dry year (277mm of rainfall during the growingseason v. 324mm long-term average). It can be seen that themaximum GAI value is higher in R2 than in CC (+0.6m2/m2).The strong effect of soil moisture is principally evident in thewheat reproductive period, from April to June, when thegreater soil water content allows for a longer green area indexduration in R2 than in CC wheat. The grain yield followedthe biomass yield (approximately twice) in the 4 driest yearsand actual evapotranspiration was about 100mm greater.Water availability, especially during spring, is crucial forcereal production to ensure spikelet fertility and seed ripening.The results obtained in a long-term experiment in the same area(Ventrella et al. 1996) confirmed this positive rotation effect,

which can be explained by a different soil moisture content atsowing for wheat productivity.

Water and nitrogen results

Figure 4 shows that in 50 out of 53years, CropSyst simulated asoil moisture content at wheat sowing in R2 and R3_F1 that wasgreater than that for wheat CC, with similar behaviour in the firsttwo cases; however, no difference was observed between CC andR3_F2. Table 5 shows similar results, with an average increase ofsoil water content at sowing of about 7% (corresponding to about84mm/mof soil depth). This is an important advantage in awater-limited environment in order to avoid final water stress in latespring during the grain-filling stage.

As can be seen in Table 5, greater soil water availability forwheat following faba bean allowed for greater nitrogen uptake.Figure 6 shows howmorewater availability during the crop cycle

–20

0

20

40

60

80

100

1950 1960 1970 1980 1990 2000 2010

Soi

l wat

er c

onte

nt in

crem

ent (

%)

R2

R3_F1

R3_F2

Fig. 4. Soilwater content at wheat sowing: percentage of variationwith respect towheat continuouscropping. For legend, see Fig. 3.

–40

0

40

80

120

160

200

1950 1960 1970 1980 1990 2000 2010

Yie

ld in

crem

ent (

%)

R2

R3_F1

R3_F2

Fig. 3. Wheat yield increment with respect to the wheat continuous crop in the 53 years ofsimulation. R2, Wheat after faba bean in 2-year rotation; R3_F1, wheat after faba bean in 3-yearrotation; R3_F2, wheat after wheat in 3-year rotation.

Durum wheat/faba bean rotation Crop & Pasture Science 247

resulted in greater plant growth and a lengthening of the cropcycle, and also allowed for a greater nitrogen uptake with anotable increase in terms of biomass and yield, particularly inthe less rainy years. The CropSyst model also simulated nitrogenleaching, a variable that was not easy to measure. It proved to bevery low, as it is a function of soil water content and waterdrainage, which were correspondingly low in this environmentdue to scarce rainfall. Consequently, as a result of greater soilmoisture, wheat following faba bean simulated an N-leaching

higher than that of CC and R3_F2. This was despite the fact thatin CC cropping, a limited number of extremely large summerrainfall events produced a higher average value (Table 5), whichwas significantly lower if analysed with a non-parametric test.

For wheat in continuous cropping with yearly nitrogenfertilisation, CropSyst first simulated an increase andsubsequently a progressive reduction of mineral soil nitrogencontent at wheat sowing in the period between 1957 and 1980,which finally reached values of 50–250 kg/ha (Fig. 7). Thedifferent behaviour of CC compared with other rotations can beexplainedbydifferentNmineral supplies (higher inCC) andby theprevalence of organic v. mineral N in rotations with faba bean.

Faba bean and legume crops in general are large nitrogenconsumers: Ranalli (2001) reported values of nitrogen consumedbyVicia fabaofup to450 kg/ha.Although fababeandeterminedarapid reduction of the initial soil nitrogen content in the first yearsof rotation, the crop residue nitrogen cycle allowed for a moreefficient use of nitrogen in the soil by the plant compared withwheat. Indeed, as shown in Fig. 8a, the higher organic matter

0

50

100

150

200

250

N u

ptak

e (k

g/ha

)

0.00

0.10

0.20

0.30

0.40

0.50

CC N uptake R_2 N uptake CC soil water R_2 soil water

0

50

100

150

200

250

0 50 100 150 200 250

Days after sowing

Pla

nt N

(kg

/ha)

0

100

200

300

400

Soi

l N (

kg/h

a)

CC plant N R_2 plant N CC N in the soil R_2 N in the soil

Soi

l wat

er c

onte

nt (

m3 /

m3 )(a)

(b)

Fig. 6. (a) Wheat nitrogen uptake and soil water content, and (b) plant andsoil nitrogen content in the top 0.6m soil depth simulated by CropSyst:comparison between CC and R2 cropping during a dry year (1976).

0

150

300

450

600

750

1950 1960 1970 1980 1990 2000 2010

Soi

l N (

kg/h

a)

CCR2R3_F1R3_F2

Fig. 7. Soil nitrogen content (0–1.3m) at wheat sowing, simulated byCropSyst. For legend, see Fig. 3.

0

2000

4000

6000

8000

10 000

0 50 100 150 200 250

Days after sowing

Dry

mat

ter

(kg/

ha)

0

1

2

3

4CC TDM

R_2 TDM

CC GAI

R_2 GAI

Gre

en a

rea

inde

x (m

2 /m

2 )

Grain yield CC = 1344 kg/ha Grain yield R2 = 2884 kg/ha

Fig. 5. Wheat total dry matter and green area index simulated by CropSyst: comparison between CC and R2cropping during a dry year (1976). For legend, see Fig. 3.

248 Crop & Pasture Science P. Garofalo et al.

mineralisation of faba bean crop residues allows for an increase innitric nitrogen (depending on soil moisture), which is easilyabsorbed by the plant, as well as a reduction in crop residueimmobilisation (wheat residues were mainly removed from thesoil surface), as shown in Fig. 8b and c. This is due to the differentquality of faba bean crop residues, richer in nitrogen andwith lesscarbon content than the durum wheat straw. In large amounts,these both produce large quantities of mineralised N for thesubsequent crop, as observed by Mayer et al. (2003).

The cultivation of legumes led to a greater exploitation of soilN by the wheat crop. Several authors have shown that yieldincreases observed in wheat following legumes compared withwheat following cereals were due to an N-conserving effect, acarry-over of N from the legume residue, and a greater uptake ofsoil N by subsequent crops when they had been previouslycropped with legumes (Senaratne and Hardarson 1988).

The presence of faba bean in 2-year and 3-year rotations withdurum wheat allowed us to save on chemical fertiliser by takinginto consideration the atmospheric N-fixation capability of thefaba bean. In fact, the introduction of a legume crop in rotation led

to a saving of about 50 kg of N/ha.year in 2-year rotations and33 kg of N/ha.year in 3-year rotations compared with CC wheat.This can be considered an important result within the context ofsustainable and low-input agriculture.

Conclusions

From this research study on the CropSyst model, a completeset of crop parameters for durum wheat and faba bean wasobtained, even though the latter crop had never previouslybeen parameterised.

The calibration and subsequent validation of durumwheat in asemi-arid environment, underlining the efficiency coefficient ofthe crop to convert solar radiation to plant biomass (not a limitingfactor in this environment), are similar to those reported by otherauthors, although the efficiency of the water conversion has stillto be reviewed for the durum wheat parameters in the model.

Wesupplieda set ofparameters for fababean, a cropwhichhadnot previously been implemented in the model, but furtherexperiments are required to confirm the validity of this work.The CropSyst model gave good results in the durum wheatsimulation. For faba bean, some difficulties were observed inthe results of the vegetative-reproductive organs as a function ofwater stress. The greater efficiency for water and light conversionto dry biomass and the reduced sensitivity to water stress andnitrogen stress, without N fixation, underline the potential ofinserting this crop even in environmentswith a poorwater supply.

The case study on the application of the CropSyst modelto evaluate the possibility of introducing a legume crop, fababean, in rotationwithdurumwheat in southern Italy, gavepositiveresults. An increase in biomass and grain productivity comparedwith continuous cropping was simulated on a long-term basis,with +16% and +21% for R2, +17% and +21% for R3_F1, and+4% and +4% for R3_F2, respectively. However, in thedriest years these increases rose to 76 and 100% on averagefor all cropping systems.

The rotation effect on wheat productivity was analysed anddifferences were explained by the better conditions of soil watercontent at wheat sowing. This had a positive influence on cropgrowth during the reproductive phenological stage and allowedfor a more efficient use of fertiliser input, which needed to begreater in the rotation with faba bean (both for residueincorporation and nitrogen fixation).

CropSyst proved to be a useful tool in comparing croppingsystems over a long-term basis, saving both time and money incarrying out the comparison and in analysing output variableswhich were not easy to measure.

Acknowledgments

This work was supported by the Italian Ministry of Agriculture and ForestryPolicies under contract no. 209/7393/05 (AQUATER Project).

References

AA. VV (1975–2008) Rete nazionale di prove di confronto tra varietà difrumento duro. L’Informatore Agrario, supplemento “Grano duro” 35.

Bechini L, Bocchi S, Maggiore T, Gonfalonieri R (2006) Parametrizationof a crop growth and development simulation model at sub-modelcomponents level. An example for winter wheat (Triticumaestivum L.). Environmental Modelling & Software 21, 1042–1054.doi: 10.1016/j.envsoft.2005.05.006

–30

–15

0

15

30

45

60

–40

0

40

80

120

160

Nitr

ifica

tion

(%)

R2 R3_F1 R3_F2

–60

0

60

120

180

1950 1960 1970 1980 1990 2000 2010

R2 R3_F1 R3_F2

R2 R3_F1 R3_F2

OM

min

eral

isat

ion

(%)

OM

imm

obili

satio

n (%

)

(a)

(b)

(c)

Fig. 8. Net (a) mineralisation, (b) nitrification, and (c) immobilisation oforganic matter in the soil simulated by CropSyst: percentage of variation withrespect to wheat continuous cropping. For legend, see Fig. 3.

Durum wheat/faba bean rotation Crop & Pasture Science 249

Bellocchi G, Silvestri N, MazzoncinM,Menini S (2002) Using the CropSystmodel in continuous rainfed maize (Zea mais L.) under alternativemanagement options. Italian Journal of Agronomy 6, 43–56.

Blair N, Crocker GJ (2000) Crop rotation effects on soil carbon and physicalfertility of two Australian soils. Australian Journal of Soil Research 38,71–84. doi: 10.1071/SR99064

Chalk PM (1998)Dynamics of biologicallyfixedN in legume-cereal rotation:a review. Australian Journal of Agricultural Research 49, 303–316.doi: 10.1071/A97013

Confalonieri R, Bechini L (2004) A preliminary evaluation of the simulationmodel CropSyst for alfalfa.European Journal of Agronomy 21, 223–237.doi: 10.1016/j.eja.2003.08.003

De Vita P, Di Paolo E, Fecondo G, Di Fonzo N, Pisante M (2007) No-tillageand conventional tillage effects on durum wheat yield, grain quality andsoilmoisture content in southern Italy. Soil&TillageResearch 92, 69–78.doi: 10.1016/j.still.2006.01.012

Di Bari V, Rizzo V,MaioranaM, DeGiorgio D, RinaldiM (1993) Variazioniproduttive del frumento duro in avvicendamenti annuali e biennalisottoposti a differenti livelli agrotecnici con e senza intercalari.Agricoltura e Ricerca 151/152, 15–22.

Di Paolo E, Garofalo P, Rinaldi M (2007) Il Favino (Vicia faba var.minor L.)per uso zootecnico sottoposto a regimi irrigui. Calibrazione e validazionedel modello CropSyst, Atti del XXXVIII Convegno Nazionale dellaSocietà Italiana di Agronomia “Il contributo della ricerca agronomicaall’innovazione dei Sistemi colturali mediterranei”, 13–14 Settembre2007, Catania, pp. 231–232.

DonatelliM, Stöckle CO,Ceotto E, RinaldiM (1997) Evaluation of CropSystfor cropping systems at two locations of northern and southern Italy.European Journal of Agronomy 6, 35–45.

DouZ, FoxRH, Tot JD (1994) Tillage effect on seasonal nitrogen availabilityin corn suppliedwith legumegreenmanures.Plant andSoil162, 203–210.doi: 10.1007/BF01347707

EC Council (1999) EC Council Regulation, no. 1259/1999, of 17 May 1999.Establishing common rules for direct support schemes under the commonagricultural policy. O. J. E. C. Law no. L 160, 26 June 1999.

EC Council (2003) EC Council Regulation, no. 1782/2003, of 29 September2003. Establishing common rules for direct support schemes under thecommon agricultural policy and establishing certain support schemes forfarmers and amending Reg. no. 2019/93, no. 1452/2001, no. 1453/2001,no. 1454/2001, no. 1868/94, no. 1251/1999, no. 1254/1999, no. 1673/2000, no. 2358/71 and no. 2529/2001. O. J. E. C. Law no. 270 of 21October 2003.

FAO-UNESCO (1963) Bioclimatic map of the Mediterranean Zone,explanatory notes. Paris, France.

Giardini L, Berti A,Morari F (1998) Simulation of two cropping systemswithEPIC and CropSyst models. Italian Journal of Agronomy 1, 29–38.

Lal R, Regnier E, EckertDJ, EdwardsWM,HammondR (1991) Expectationsof cover crops for sustainable agriculture. In ‘Cover crop for clean water’.(Ed.WLHargrove) pp. 1–11. (Soil andWaterConservationSocietyPubl.:Ankey, IO)

Large EC (1953) Growth stages in cereals. Illustrations of the Feek’s scale.Plant Pathology 3, 128–129. doi: 10.1111/j.1365-3059.1954.tb00716.x

Loague K, Green RE (1991) Statistical and graphical methods for evaluatingsolute transport models: overview and application. Journal ofContaminant Hydrology 7, 51–73. doi: 10.1016/0169-7722(91)90038-3

Maiorana M, Convertini G, Di Bari V, Rizzo V (1992) Yield and quality ofdurum wheat (Triticum durum Desf.) under continuous cropping afternine years of straw incorporation. European Journal of Agronomy 1,11–19.

Maiorana M, Convertini G, Fornaro F (2004) Gestione del suolo nellaomoosuccessione di grano duro. L’Informatore Agrario 33, 79–82.

MaioranaM,DiBari V, ConvertiniG (1993) Interramento dei residui vegetalidi frumento duro in monosuccessione con dosi crescenti di azoto fosforo.Effetti sulle componenti quantitative e qualitative della produzione.Agricoltura e Ricerca 151–152, 69–76.

Maiorana M, Rizzo V, Ventrella D, Convertini G, Ferri D, Colucci R (1997)Interramento e bruciatura dei residui colturali di frumento duro inmonosuccessione: effetti di diverse modalità di lavorazione del terrenoe di somministrazioni dell’azoto. Agricoltura e Ricerca 168, 49–56.

Mayer J, Buegger F, Jensen ES, Schloter M, Heb J (2003) Residual nitrogencontribution from grain legumes to succeedingwheat and rape and relatedmicrobial process. Plant and Soil 255, 541–554. doi: 10.1023/A:1026081015076

Monteith JL (1977) Climate and the efficiency of crop production in Britain.Philosophical Transactions of the Royal Society of London. B 281,277–294. doi: 10.1098/rstb.1977.0140

Pala M, Stöckle CO, Harris HC (1996) Simulation of durum wheat (Triticumdurum Desf.) growth under differential water and nitrogen. AgriculturalSystems 51, 147–163. doi: 10.1016/0308-521X(95)00043-5

Ranalli P (2001) ‘Leguminose e agricoltura sostenibile, specie da granella ecover crops.’ pp. 628. (Gruppo Calderini – Edagricole s.r.l.: Bologna,Italy)

Rinaldi M, Ventrella D (1997) Uso dei modelli EPIC e CropSyst in sistemicolturali del Sud Italia. Agricoltura e Ricerca 171, 47–58.

Rizzo V, Di Bari V, Maiorana M, Convertini G, Rinaldi M, De Giorgio D(1990) Effects of the previous crops and continuous cropping on theproduction of durum wheat and some chemical characteristics of soil. In‘Proceedings of 1st Congress of European Society of Agronomy’. Paris,5–7 December, session 5, (Ed. A Scaife) pp. 20–21. (European Society ofAgronomy: Colmar Cedex, France)

Russell MP, Hargrov WL (1989) Cropping systems: ecology andmanagement. In ‘Developments in agricultural and managed-forestecology. 21. Nitrogen management and groundwater protection’.(Ed. RF Follett) pp. 277–317. (Elsevier: Amsterdam, The Netherlands)

Senaratne R, HardarsonG (1988) Estimation of residual N effect of faba beanand pea on two succeeding cereals using 15Nmethodology.Plant and Soil110, 81–89. doi: 10.1007/BF02143543

Stöckle CO, Cabelguenne M, Debaeke P (1997) Comparison of CropSystperformance for water management in southwestern France usingsubmodels of different levels of complexity. European Journal ofAgronomy 7, 89–98. doi: 10.1016/S1161-0301(97)00033-6

Stöckle CO, Debaeke P (1997) Modelling crop nitrogen requirements:a critical analysis. European Journal of Agronomy 7, 161–169.doi: 10.1016/S1161-0301(97)00038-5

Stöckle CO, Donatelli M, Nelson R (2003) CropSyst, a cropping systemssimulation model. European Journal of Agronomy 18, 289–307.doi: 10.1016/S1161-0301(02)00109-0

Tanner CB, Sinclair TR (1983) Efficient water use in crop production:Research or re-search? In ‘Limitations to efficient water use in cropproduction’. (Eds HM Taylor, WR Jordan, TR Sinclair) pp. 1–27.(ASA, CSSA, SSCA, Inc.: Madison, WI)

USDA (2006) ‘Keys to Soil Taxonomy.’ 10th edn, pp. 333. Available at:ftp://ftp-fc.sc.egov.usda.gov/NSSC/Soil_Taxonomy/keys/keys.pdf.

Ventrella D, Rinaldi M (1999) Comparison between two simulationmodels to evaluate cropping systems in Southern Italy. Yield responseand soil water dynamics. Agricoltura Mediterranea 129, 99–110.

Ventrella D, Rinaldi M, Rizzo V, Carlone G (1996) Disponibilità idrica delsuolo ed efficienza nell’uso dell’acqua in nove avvicendamenti a sussidioidrico limitato. Rivista di Agronomia 30, 1–8.

Manuscript received 23 June 2008, accepted 19 December 2008

250 Crop & Pasture Science P. Garofalo et al.

http://www.publish.csiro.au/journals/cp