Evaluation of CropSyst for studying the effect of mulching with lucerne (Medicago sativa L.) in Austria

Abstract

Conservation of water under semi-arid conditions is of supreme importance to combat forthcoming water scarcity. Crop residue mulches are the key interventions in this regard. Simulation models are emerging as strong analytical tools to study the effect of agronomic interventions but their use to study effect of mulches in Austria is limited. To bridge this knowledge gap, we planned a study to evaluate the efficacy of CropSyst for studying response of lucerne mulches in succeeding lucerne crop. CropSyst is a user-friendly, conceptually simple but sound multi-year multi-crop daily time step simulation model. Above ground biomass and soil water content from field experiments carried out near the city of Vienna, Austria (2007–2008) were compared with simulated results to determine the suitability of CropSyst in regarding mulch. Adequacy of model to regard mulch is demonstrated while pros and cons of disagreement between experimental and modeling results are discussed with a way forward for future research.

Keywords:

• above ground biomass
• CropSyst
• lucerne
• soil water content
• utilization system

1. Introduction

Water is a limiting factor for crop production in many parts of the world. Its efficient utilization remains the key concern for researchers. Different soil and crop management practices can help to obtain better water use efficiency. The aim of management practices shall be to either increase transpiration efficiency or to decrease non-productive water losses. Mulching is regarded as one of the best ways to improve water retention in the soil and to reduce soil evaporation (Steiner 1989; Huang et al. 2005).

Residues and mulches limit evaporation by reducing soil temperature, preventing vapor diffusion, absorbing water vapor on to mulch tissue, and reducing the wind speed gradient at the soil–atmosphere interface (Greb 1966; Lagos et al. 2009). Presence of mulch on soil surface tends to increase water infiltration into the soil and cumulative effect of increase in infiltration and reduction in evaporation is overall better retention of water under mulch (Li & Xiao 1992; Baumhardt & Jones 2002). However, mulch effects depend on the soil type, rainfall, and evaporative demand (Wicks et al. 1994; Tolk et al. 1999; Ji & Unger 2001; Lampurlanes et al. 2002). Incerti et al. (1993) found small gains in water storage attributable to stubble retention in long-fallow periods, with no advantage in crop yield. Greater and more consistent responses to stubble retention were reported for the wetter regions having heavier soils in Australia (Cantero-Martınez et al. 1995). Mulching with legumes improves soil nitrogen balance, adds to soil organic matter that modifies the soil environment, builds up soil fertility, and has positive pre-crop effects (Pietsch et al. 2007; Raza et al. 2013a).

Evaluation of management practices under field conditions involves high cost and time and due to uncertainty and variability in weather and field conditions, results may vary among years. Alternatively, it can be evaluated much more cheaply and quickly using simulations. Models can be used for long-term predictions and their results can be extrapolated if they are calibrated and validated for a management intervention in a given location. Simulation model, CropSyst (Stöckle et al. 2003), is a soil plant atmosphere continuum (SPAC) model which has options for biomass fate where we may opt to harvest the biomass and specify a percentage of biomass to be left in the field as mulch or we may harvest and designate it for beneficial use. These options make the model a suitable choice for comparing results on different lucerne (Medicago sativa L.) utilization system (no mulch versus mulch). The present study was designed with the objectives to evaluate the efficacy of CropSyst under different lucerne utilization systems.

2. Materials and methods

2.1. Experimental details

Field experiments with lucerne cultivar Sitel were established on certified organic fields (Austria Bio Garantie) at Raasdorf, Eastern Austria (during 2007 and 2008) at the experimental farm of the University of Natural Resources and Life Sciences, Vienna, Austria. Lucerne treatments were arranged in a randomized complete block replicated four times. Individual plot dimensions were 3 × 3 m. Lucerne was seeded using pneumatic seed drill (Amazone, Germany) at a seed rate of 25 kg ha⁻¹ in rows 12.5 cm apart during April and received usual management through harvest.

The trial site in Raasdorf is 5 km east of Vienna (48°14′N, 16°35′E) at an altitude of 150–160 m above sea level. The soil can be classified as Calcaric Phaeozem (WRB 1998) or as a Typic Vermudoll in the US soil taxonomy system (Soil Survey Staff 2010) with a particle size distribution of 0.27 kg kg⁻¹ sand, 0.54 kg kg⁻¹ silt, and 0.19 kg kg⁻¹ clay (silty loam). Soil organic carbon content is 1.6% at 0–30 cm soil depth.

2.2. Data collection

Data on yield were recorded at two main harvests in each year. Harvesting was done at 30–40% of the flowering. Lucerne plots were hand clipped with a garden scissor at about 5 cm above the ground level. An area of 1 m² was harvested from each plot at each harvest to determine shoot biomass. Mulching application was done by hand on two 0.5 × 1 m subplots (per plot). The rest of the 3 × 3 m plot was cut and chopped using a cutting machine (Hammerschmeid, Austria). Material was spread as mulch in the entire 3 × 3 m plot in the mulch plot only and lucerne was allowed to re-grow for 9–10 weeks. Every year first harvest was used to apply no mulch and mulch treatments. At second harvest, data on biomass were recorded to compare the effect of treatments. Stubble biomass was determined only on second harvest in each year. Shoot and stubble dry matter yield were determined by oven-drying the sub-sample at 60°C for 48 h. Values of shoot and stubble dry matter yield at second harvest were added to present the data on above ground biomass (AGB) for comparison with model results. Leaf area index (LAI) was measured using LAI-2000 Plant Canopy Analyzer (LI-COR, Lincoln, NE, USA). A summary of management operations during the experimental period is presented in Table 1.

Table 1. Management operations in the utilization system experiment.

Management operation	2007	2008
Sowing	10 April	11 April
Irrigation	12 April (75 mm)	13 April (15 mm)
Irrigation	18 April (50 mm)	20 April (15 mm)
Installation of SENTEK probes	16 May	28 April
Harvest 1	9 July	9 July
Harvest 2	18 September	9 September

Data on roots were collected at the time of final harvest in each year. Maximum rooting depth was determined by visual observations regarding the presence of fine roots in a soil cylinder extracted by a mechanically compressed auger to a depth of 2 m in each replicate. Samples for the analysis of root distribution based on root length density were collected using an auger with 7 cm diameter to a depth of 60 cm, dividing the column into six sub-samples of 10 cm each. After washing and cleaning, root length was determined using WinRhizo 4.1 following Himmelbauer et al. (2004). Root depth distribution to characterize the root system of the cultivar in the CropSyst model was described using the exponential Gerwitz and Page (1974) model.

Precipitation and meteorological data were obtained from weather station of Institute of Agronomy and Plant Breeding, BOKU. Irrigation water in the amount of 125 mm and 30 mm in 2007 and 2008, respectively, was applied for stand establishment using a sprinkler irrigation system. Changes in the water content of soil profile to a depth of 120 cm were calculated from the data on soil water content (SWC) being assessed using SENTEK Diviner 2000 (Sentek Sensor Technologies, Australia) by installing one probe in the center of each lucerne plot. A manual data logger was used to record values of SWC at 7–14 days intervals.

At the end of vegetation period in each year, undisturbed soil samples were collected in two replicates for the determination of soil texture, bulk density, retention curves and saturated hydraulic conductivity. Particle size analyses were based on determination of percentage of sand silt and clay in a soil sample (ONORM L1061 1988). Based on relative proportion of sand, silt, and clay, textural classes were determined following American textural triangle adopted from American Soil Survey Manual (Soil Survey Staff 2010). Bulk density was determined using the relation proposed by Blake and Hartge (1986).

2.3. Model description

CropSyst is a SPAC model. Crop development is simulated based on thermal time requirements to specific growth stages. Daily crop growth is expressed as biomass increase per unit ground area. The model accounts for four growth limiting factors: light, water, nitrogen, and temperature. Modeling biomass production for perennials such as lucerne has been described briefly by Confalonieri and Bechini (2004). Re-growth after clippings can be obtained by enabling the active growth option, and adjusting the minimum reserve biomass and green area index to be retained for a proper re-growth dynamics.

The water budget in the model is driven by precipitation/irrigation and crop evapotranspiration as atmospheric boundary conditions. Crop evapotranspiration is calculated from reference evapotranspiration (ET₀) using a crop coefficient at full canopy and ground coverage determined by canopy LAI. ET₀ is obtained by one of the following models: Penman–Monteith, Priestley–Taylor, or a simplified version of Priestley–Taylor. Water dynamics are simulated by a cascade approach or by numerical solution of Richard's (1931) equation. Based on actual soil water availability, and using threshold values of leaf water potential before the onset of stomata closure and wilting, actual soil evaporation and plant transpiration are obtained (Stockle & Jara 1998). A detailed description of the entire model, its components, mathematical approaches, data requirements and some model evaluations can be found in Stöckle et al. (2003).

2.3.1. Model set-up and calibration

CropSyst version 4.09.00 was used. Crop growth was simulated using a default lucerne parameter set as implemented in the CropSyst crop database. Table 2 gives an overview of all relevant crop parameters used for the simulations. Calibration of model parameters that were not directly measured was based on minimizing differences between measured and modeled crop biomass and water content as state variables. Sensitivity analysis showed that the two parameters mean daily temperature for optimum growth as well as maximum water uptake were critical for proper crop growth simulation. Optimal temperature for growth and biomass accumulation in lucerne can vary from 12°C to 30°C according to diversity of germplasm in different agro-ecologies (Arbi et al. 1979; Confalonieri & Bechini 2004). Mean daily temperature was calibrated by minimizing the deviation in the experimental (measured) AGB and the simulated biomass. The value obtained from calibration and used throughout the simulation was 15°C. Also for maximum water uptake, a common value was set by minimizing deviation between measured mean and simulated water content. The value obtained was 12 mm day⁻¹, again being close to the default value (10 mm day⁻¹).

Table 2. Crop model parameters for lucerne.

Parameter	Determination	Value	Units
Photosynthetic pathway	–	C3	–
AGB transpiration coefficient	D	5	kPa kg m^−3
Unstressed light to AGB conversion (radiation use efficiency)	D	3	g MJ^−1
Actual to potential transpiration ratio that limits leaf are growth	D	0.8	–
Actual to potential transpiration ratio that limits root growth	D	0.5	–
Optimum mean daily temperature for growth	C	15	°C
Maximum water uptake	C	12	mm day^−1
Leaf water potential at the onset of stomatal closure	D	−1300	J kg^−1
Wilting leaf water potential	D	−2100	J kg^−1
Maximum rooting depth	M	1.5	M
Maximum expected leaf area index	M	4.5	–
Root length per unit root mass	M	30.92	km kg^−1
Surface root density	M	5.56	cm cm^−3
Curvature of root density distribution	M	0.029	–
Fraction of maximum LAI at physiological maturity	D	0.8	–
Specific leaf area	C	24	m^2 kg^−1
Stem/leaf partition coefficient	D	3	–
Extinction coefficient for solar radiation	D	0.5	–
Et crop coefficient at full canopy	C	0.8	–
Degree days emergence	D	100	°C days
Degree days flowering	C	800	°C days
Base temperature	L	5	°C
Cutoff temperature	L	30	°C
Adjustment factor for phenological response to water stress	D	0	–

C, calibrated; D, default; M, measured; L, literature; E, local experience.

Value used for specific leaf area was based on measurements of leaf area and leaf weight for Sitel variety. Specific leaf area was 24 m² kg⁻¹ which is close to the default value of 22 m² kg⁻¹. Maximum expected LAI was 4.5. For specific root weight (root length per mass), root depth distribution, and maximum rooting depth, a variety specific measured value was used based on previous field experiments under non-water-limiting growth conditions (Raza et al. 2013b).

Reference evapotranspiration was calculated with Penman–Monteith equation (Monteith 1985). Soil water dynamics were simulated using Richards (1931) equation. Run off was assumed to be zero as the experimental sites were flat. Soil hydraulic parameters of the Campbell (1985) model used by CropSyst were obtained from measured retention curves, while measured values of saturated hydraulic conductivity were used for upper part of soil profile (0–60 cm). Saturated hydraulic conductivity was determined using method of rising head soil core – a modified form of falling head soil core/tank method (Reynolds & Elrick 2002). Values for lower part of soil profile (60–120 cm) were obtained by calibration starting from an initially texture-based estimate.

The model was calibrated with 2 years of data from field experiments (2007–2008). This allowed us to set reliable values for the lucerne variety description (crop files) at the level of non-water-limited growth including measured data on leaf and root traits for parameterization (Raza et al. 2013b). The files for soil (hydraulic properties) and meteorology were also based on measurements.

The state variables used to evaluate the model performance were cumulative biomass and profile water content. The agreement between measured and simulated values was evaluated following Loague and Green (1991) by root mean squared error (RMSE), coefficient of determination (CD), modeling efficiency (EF) index, and coefficient of residual mass (CRM). All indices were calculated using statistical software IRENE (Fila et al. 2003).

2.3.2. Scenario analysis

Simulation scenarios for two different utilization systems were created separately in each year to predict the model behavior towards the utilization system (no mulch versus mulch). Under the biomass fate option for clipping file of first harvest of each year, 95% of the clipped biomass was left as residue on the surface to implement mulch treatment using the model. In case of no mulch treatment, model was set to remove 95% of the accumulated AGB from the field and designate it for beneficial use.

3. Results

3.1. Comparison of experimental results with model results

The model does not seem to produce an effect on the accumulation of AGB under different utilization systems in both years of experimentation. Results from simulations revealed that there were no differences among treatments regarding their effect on accumulation of AGB. Based on experimental results, minor differences were found in AGB under different treatments of utilization system. Results on AGB under different utilization systems are compared (measured versus simulated) in each year separately and presented in Figures 1 and 2.

Figure 1. AGB under different utilization systems in 2007 (data from second harvest).

Figure 2. AGB under different utilization systems in 2008 (data from second harvest).

Values of AGB from second harvest under both treatments in both years were compared with simulated values to assess the efficacy of model for prediction of biomass on statistical grounds. Results revealed the EF value of 0.83, CD value of 0.36, and CRM value of −0.0184. As the EF and CD values are below the optimum value of 1, it is an indicative of inadequacy of model to accurately predict biomass accumulation under different utilization systems.

Results on profile SWC (0–120 cm) were also compared for each treatment separately in each year and statistical indices are presented in Table 3 and data are shown in Figure 3. Model tends to apply the mulching treatment and it seems to conserve more moisture under mulching compared to no mulching as is evident from Figure 3 where in both years profile SWC becomes higher following the application of treatments on July 10. The model slightly over estimated the profile SWC as is evident from Figure 3. However, only in 2008 in mulch treatment, the measured values of profile SWC were found higher than simulated values and it could be due to the plot effect. A good agreement was not found among measured and simulated profile SWC that is also reflected in indices of agreement where values of EF and CD are not close to 1 (Table 3). This indicates the inadequacy of model to simulate soil water distribution in the profile under different utilization systems. Mulching usually had effect on retention of water in upper 10 cm soil layer. An increase in SWC was observed in both years for simulated values of SWC in upper 10 cm soil layer as is evident from data shown in Figure 4. Based on measurements of SWC, it was difficult to clearly establish an effect of mulch on conservation of water in upper 10 cm soil layer following the application of treatments. While comparing the effects of treatment in upper 10 cm soil layer, we found that SWC was slightly higher in mulched plots even before the application of treatments.

Figure 3. Profile SWC (0–120 cm) under different utilization systems.

Figure 4. SWC under different utilization systems in 0–10 cm.

Table 3. Indices of agreement for profile SWC.

Year	Treatment	RMSE	EF	CRM	CD
2007	No mulch	39.5	0.17	−0.16	1.84
	Mulch	61.5	0.88	−0.28	2.61
2008	No mulch	45.7	0.24	−0.17	1.38
	Mulch	37.7	0.24	0.09	0.77

Mulching was effective in lowering soil temperature by 1–6°C in the top 5 cm of soil but did not improve biological nitrogen fixation and plant productivity in the present study (Raza et al. 2013a). Use of CropSyst for assessment of effect of mulching on soil temperature and biological nitrogen fixation offers interesting research avenues for future.

4. Discussion

Principal effect of mulching with crop residues is to reduce soil evaporation and this effect usually takes place if residues are applied during a wet period for a longer duration of at least few months. Reduction in evaporative losses of water under mulching tends to improve soil water retention with consequences of improved plant growth and increase in overall accumulation of biomass. Mulch will reduce evaporation most effectively early in the drying cycle when the surface soil is wet and early in the growing season when the leaf area is small (Ji & Unger 2001; Tao et al. 2006).

In the present study, mulch was applied in July when soil surface was becoming drier and crop re-growth after first harvest was relatively faster. This might be the reason that smaller amounts of water are conserved under mulch treatment in the present study and these smaller amounts of water conserved at the time when soils were becoming drier under hot summer conditions were unable to bring a significant increase in AGB accumulation due to mulching under both years of experimentation.

The simulation model, CropSyst, does not seem to produce an effect on the accumulation of AGB under the scenario of different utilization systems in both years of study. This can be explained based on the behavior of model toward estimation of profile SWC after the application of mulch and no mulch treatments on 10th July in each year. Ideally, the model shall show an increase in profile SWC in mulching scenario following the application of treatment, but the model does not show any major change in profile SWC (Figure 3). This effect is also translated into the accumulation of biomass and model predicts the same amounts of biomass under both scenario of no mulch and mulch under both years of study (Figures 1 and 2). Results from previous studies with CropSyst indicate that effect of mulching varies among years. Donatelli et al. (1997) found from a 6-year study in Italy that only in some years, mulching with barley residues had positive effect on yield of subsequent soybean crop due to reduced evaporation. Crop response to mulching is influenced by a complex interaction that could vary from year to year (Unger 1978; Zhang et al. 2007). Besides effect of year, other factors such as mulch mass, irrigation frequency, rainfall, evaporative potential, and soil texture can also affect the response to mulching (Tolk et al. 1999; Ji & Unger 2001; Lampurlanes et al. 2002; Mupangwa et al. 2007). Mulches may affect productivity of grasses under dry site conditions through their effect on radiation use efficiency by modifying soil water storage and soil temperature (Bat-Oyun et al. 2012).

In the present study, the masses of mulch used were 4311 and 6253 kg ha⁻¹ in 2007 and 2008, respectively. Although these amounts are reasonable but timing of mulch application, drier soil conditions and short duration of mulching may be the probable reasons for no effect of mulch on soil water conservation and biomass accumulation. Chen et al. (2007) compared effect of different mulch masses of chopped maize straw on the reduction of evaporation during subsequent wheat season in northern China. They found from a 5-year study that mulch reduced soil evaporation by 21% under less mulching (3000 kg ha⁻¹) and 40% under more mulching (6000 kg ha⁻¹) compared with control (no mulching). In India, Kar and Kumar (2007) compared the effect of number of irrigations (1–4) with and without rice straw mulch applied at the rate of 6 tones ha⁻¹ on the yield of potato. Based on 2 years pooled data, mulch application increased the potato tuber production by 24–42% depending on the irrigation treatments. The variation in yield of potato under different number of irrigations indicates how the effect of same mass of mulch may vary with the irrigation frequency. In the present study, one mass of mulch was applied and no additional irrigations were applied during the treatment period, so the magnitude of variation among treatments of mulch and no mulch was expected to be less.

Previous modeling studies to evaluate the effect of mulching indicate that duration of experiments usually lasts for 4–6 years on the same piece of land and residues are left over in the field either for entire duration of fallow period or subsequent growing season (Donatelli et al. 1997; Zhang et al. 2007). This enables the mulch material to settle in the field with consequences of increase in the amount of water entering the soil. The residue cover reduces non-productive water losses through evaporation and increases SWC to enhance crop growth and yield of subsequent crop (Mupangwa et al. 2007). In the present experiment, location of experiment was changed every year that might have led to production of no major effect on water content and yield.

The effects of residue mulching on water retention through reduced evaporation are usually smaller in drier or semi-arid environments than in wetter or humid environments. Previous long-term studies using CropSyst has already demonstrated that in environments with low rainfall and coarse-textured soil, contribution of stubble to gains in water storage is often smaller than in wetter environments with heavier soil (Monzon et al. 2006). The present study was carried out in a semi-arid environment where mulch is left over the soil surface for a short duration of 2 months. This might be another reason for no big effect of mulch on SWC and AGB accumulation. Modeling results demonstrated that mulching tended to improve SWC in upper 10 cm soil layer as well as slightly improved profile SWC but the amount of water conserved was not so big to bring a drastic change in AGB accumulation under two different treatments. This can be the reason for no differences in AGB accumulation under mulch and no mulch treatments.

5. Conclusion

A 2-year experimental and modeling study with lucerne utilization system in eastern Austria revealed that mulching has no pronounced effect on biomass accumulation and moisture conservation. The potential of CropSyst in studying response of mulches is demonstrated. The study provided useful insights into the short-term effect of mulches and strengthened the view point that to be effective, mulching needs to be carried out in long-term field experiments with supplemental irrigation. We propose further modeling studies with variable duration of mulching, higher mulch masses with varying irrigation events to identify the interventions that can save water and enhance crop yields under prevailing site conditions.

Acknowledgments

The corresponding author acknowledges Higher Education Commission, Government of Pakistan for providing fellowship for doctoral studies in Austria as work presented in the manuscript is part of doctoral thesis. Authors also appreciate the cooperation provided by various laboratories at University of Natural Resources and Life Sciences, Vienna, Austria.