Massa, D., Incrocci, L., Botrini, L., Carmassi, G., Diara, C., Delli Paoli, P., Incrocci, G., Maggini, R., & Pardossi, A. (2018). Modelling plant yield and quality response of fresh-market spinach (Spinacia oleracea L.) to mineral nitrogen availability in the root zone. Italian Journal of Agronomy, 13(3), 248-259. https://doi.org/10.4081/ija.2018.1120
Spinach is one of the most important green-leafy vegetables, consumed worldwide, and its intake is beneficial for human beings. In this crop, produce yield and quality are closely related to plant nitrogen (N) nutrition. A precise supply of N is also essential for high environmental and economic sustainability. Main aims of the work were: i) to establish relationships between produce yield or quality and mineral N availability in the root zone; and ii) to define an optimal mineral N level to be maintained in the root zone for spinach. Eight experiments were carried out during a four-year-long period under typical Mediterranean climate conditions. Different amounts of N fertilisers were supplied leading to twenty different levels of mineral N in the root zone. Experimental measurements included climate parameters, plant growth, tissue and soil analyses, produce yield and quality indicators. A segmented linear model significantly represented the relationship between crop yield (1.7 to 21.7 t ha–1) and soil mineral N concentration (7.6 to 41.0 mg kg–1). Basing on this model, an optimal mineral N threshold was fixed at 23.4 mg kg–1. Above this threshold, crop yield did not show any significant variations as well as tissue characteristics and produce quality. Plants grown under suboptimal N levels showed reduction in growth, tissue mineral (nutrients) content, and SPAD index. The proposed models could be implemented in fertilisation protocols for the optimization of N supply and the estimation of spinach growth and yield.
Downloads month by month
Download data is not yet available.
In the Mediterranean basin, savoy spinach (Spinacia oleracea L.) represents an important typical produce exported in many countries of northern Europe. For fresh market, savoy spinach is harvested at early growth stage. The most important quality attributes of this leafy vegetable are related to the leaf greenness and morphology (i.e., wrinkledness), to the content of beneficial mineral elements, and to the low content of toxic compounds such as oxalic acid and nitrates (Santamaria et al., 1999; Cavaiuolo and Ferrante, 2014).
Well-balanced nitrogen (N) supply is crucial for high yield and market quality of spinach. Many authors reported a positive relationship between crop yield and increasing N fertiliser rate in spinach cultivated in open field or under greenhouse (Biemond et al., 1996; Wang and Li, 2004; Gülser, 2005; Lefsrud et al., 2007; Stagnari et al., 2007; Rodriguez-Hidalgo et al., 2010). The spinach content in mineral elements and antioxidants, such as lutein and β-carotene, are both positively related to N availability (Lefsrud et al., 2007; Stagnari et al., 2007).
On the other hand, N excess may lead to large leaf accumulation of oxalic acid and nitrates in leaf tissues, especially when N is supplied in the nitric form (Chen et al., 2004; Wang and Li, 2004; Zhang et al., 2005; Stagnari et al., 2007). Similarly to nitrates, oxalic acid and related compounds are harmful molecules for human beings and their continuous intake can induce blood diseases, especially in infants, nutritional disorders, and other health disturbs in human body (Noonan, 1999; Bryan and Loscalzo, 2011; Agnoli et al., 2017). In the European Union (EU) specific limits have been laid down for the nitrate content of some leafy vegetables (The Council of the European Community, 2006; EFSA, 2008). In spinach, these limits are 2000 and 3500 mg NO3– kg–1 (fresh weight basis) for frozen and fresh products, respectively.
Excessive N supply also results in increased water content of leaf tissues (Lefsrud et al., 2007), which may negatively influence plant resistance to pathogens (Dordas, 2008) and its shelf life as well (Lombardo et al., 2016). A correct N fertiliser management is essential for reducing the crop environmental impact associated with nitrate leaching (Robertson and Vitousek, 2009; Zhou and Butterbach-Bahl, 2014), which easily occurs in sandy soils where spinach is typically cultivated. In the EU, the Nitrates Directive (The Council of the European Communities, 1991) was issued to preserve the quality of ground and surface water bodies from the pollution of nitrates produced by agricultural activity, and to promote the adoption of good agricultural practices. According to the Nitrates Directive, growers must follow mandatory rules to tackle nitrate loss from their crops; for example, in the area of Val di Cornia (Tuscany, Italy), where the experiment took place, a maximum N dose of 120 kg ha–1 and well defined (limited) periods of distribution have been ruled for spinach by the local authorities.
Moreover, N waste has negative economic impact on the production costs of field crops (Robertson and Vitousek, 2009). Some authors have related crop yield and N supply to economic parameters, to calculate the N fertiliser rate that maximizes grower’s incomes (Wang and Li, 2004; Milne et al., 2012).
Literature on leafy vegetables mostly focuses on the effects of N on yield and quality (Chen et al., 2004; Wang and Li, 2004; Zhang et al., 2005; Lefsrud et al., 2007; Stagnari et al., 2007) while less attention is paid to effective N management (Canali et al., 2014).
However, most works attempt to describe growth and yield response curves as a function of fertiliser doses instead of the actual N availability in the root zone. For optimal N management different authors have therefore introduced the concept of minimum optimal concentration (Heckman et al., 2002; Cui et al., 2008; Bottoms et al., 2012); this would represent the reference value to be maintained in the root zone to minimize crop environmental impact and to support high yield and quality (Incrocci et al., 2017). To the best of our knowledge, no previous study addresses the above issue for spinach, thus an N optimal concentration for this crop has not been yet determined.
The paper reports experimental data collected in eight different experiments carried out during a four-year-long period. Main aims of the work are: i) to assess the effects of soil N concentration on spinach yield and quality throughout a medium-long observation period under different climate conditions; ii) to define an optimal soil mineral N concentration for effective N management; and iii) to test an optical sensor for the quick monitoring of N nutritional status in fresh-market spinach grown under open-field Mediterranean climate conditions.
Materials and methods
Growing conditions and treatments
Experimental data were collected in eight different experiments (E), on spinach (Spinacia oleracea L.) crops, during a fouryear- long period (from September 2007 to April 2011). Experimental fields were located in Val di Cornia (Tuscany, Italy), a coastal area with sandy-loam soils. The area is intensively cultivated with vegetables under typical Mediterranean climate conditions (Figure 1) with mild winters and 650 mm annual rainfall (tenyear average).
The physico-chemical characteristics of the soil (Table 1) were determined in the root zone of spinach (5-40 cm) prior to sowing. The quantity of fertilisers containing P, K, Ca, Mg and micronutrients was then calculated through a soil nutrient balance aimed to: i) replace the nutrients taken up by a crop grown under optimal conditions; and ii) restore the initial soil fertility if necessary. Only N was supplied in a variety of different doses (Table 2) based on the following criteria. During the first two experiments (E1 and E2), N was supplied at the fixed rate of 0, 80, 120 or 160 kg ha–1. In the other experiments (E3-E8), N was supplied following growers’ fertilisation practice. This is based on the standard fertilisation rate of 120 kg N ha–1, according to the blueprint laid down for spinach production in the Val di Cornia area. The above quantity is usually increased by growers up to 175 kg ha–1 in relation to the rainfalls occurred in the growing period, which may increase the risk for N shortage due to nitrate leaching. Each experiment always included a dose of N equal to 0 as untreated control (Table 2). In all treatments, the total amount of N was split over three periods: i) 40% blended with the soil, before sowing (4-6 days), using a mixed organic-mineral fertiliser (1% organic, 5% ammonia and 30% ureic); ii) 30% distributed as top-dressing fertilisation, at 1/3 of the cultivation cycle (roughly 4-5 true leaves), using ammonium nitrate; iii) 30% distributed as top-dressing fertilisation, at 2/3 of the cultivation cycle (roughly 10-11 true leaves), using calcium nitrate. In the case of rainy periods (i.e., when total rainfall in the 8-14 true-leaf phase was 50% higher than the average of the previous ten years), with high N leaching, a third top-dressing fertilisation (40-50 kg N ha–1) was applied when spinach had 14-15 true leaves (E3, E4 and E8).
Soil preparation of seedbed included ploughing, harrowing and levelling for bringing the soil into the better tilt for water drainage. Spinach (cv. Spitfire, Seminis®, Monsanto Company, USA) was sown in order to have a plant density of 30 plants m–2 taking into account the percentage of emergency. Each treatment was applied in a completely randomized experimental design on an area of 800-1000 m2.
Irrigation was applied rarely (only once at germination in E1 and E3), using traveling sprinklers, to restore the field capacity when rainfall events were not sufficient to preserve the quantity of easily available water (50-60% of the available water) in the root zone (Table 1). Soil moisture was monitored by using a tensiometer (Delta-T SWT 4, Delta-T Device Ltd, Cambridge UK) with the ceramic cup positioned within 20-30 cm depth in the driest area of the field. Crop protection was accomplished following the standard protocol used by local growers that includes treatment against insects, fungi and weeds.
Climate parameters were monitored hourly using a meteorological station (Pessl Instruments GmbH, Weiz, Austria) located in the experimental area. Air and soil temperature, radiation, wind speed, rainfall and air humidity data were collected and summarized in Figure 1 and Table 2.
Plant and soil analysis
Plant samples were collected at different crop stages, i.e., at: i) 4-5 true leaves; ii) 10-11 true leaves; and iii) harvest time; for each treatment, three replicates were collected in the first and second sampling, and four replicates at harvest. Each replication unit consisted of an area of 3.5 m2 that corresponded roughly to 100 plants. Sample units were collected randomly in each treatment. Plants were harvested by hand, stored in plastic bags to limit water loss, and moved rapidly to the laboratory for growth and tissue analyses. The growth analysis consisted in the measurement of fresh (FW) and dry weight (DW, obtained in a forced-air oven at 80°C for 96 h), number of true leaves, and leaf area, determined by a planimeter (Delta-T Device, Cambridge, UK) for the calculation of leaf area index. Leaf chlorophyll was assessed through SPAD index (SPAD-502, Konica Minolta Optics, 2970 Ishikawa-machi, Hachioji, Tokyo, Japan). Plant dry matter was analysed for its mineral nutrient content. In more detail, total N was determined as the sum of reduced N (by the Kjeldhal method) and N-NO3; the latter was determined in the aqueous extract of dry matter (1:300, w/w) using a colorimetric method (Cataldo et al., 1975). After nitric-perchloric acid digestion of dried samples (90 min at 150°C), K, Ca, and Mg were determined by atomic absorption spectroscopy (Spectra-AA240 FS, Varian, Australia), while P was measured through a colorimetric method (Olsen and Sommers, 1982).
In occasion of plant destructive analyses, soil samples were also collected in each replication unit and analysed for mineral N content (Nmin). Nitrate was measured in the aqueous extract of dry soil (soil-water ratio 1:2 w/w) using the Cataldo’s method (1975). Ammonium was extracted from soil using 1 M KCl (soil-KCl ratio 1:2 w/w) and quantified spectrophotometrically through the indophenol method (Kempers and Kok, 1989).
Data analysis and modeling
Most of the variables analysed in the work refer to the averaged concentration of total (N-NH4+ plus N-NO3–) mineral N (Nmin) in the root zone. Nitrogen concentration in this work is mostly expressed as mg of element per kg of dry soil. The conversion from mg kg–1 N to N expressed as kg per hectare can be computed using the following equation:
where BD is the soil bulk density (1.45 t m–3, on average; Table 1), RD is the root depth (0.40 m), and 10 is a multiplicative factor for unit conversion (from mg kg–1 to kg ha–1); then in our growing conditions the product of BD, RD and 10 was 5.8.
To compare the yield (Y) of different experiments carried out under different temperature and radiation levels (Figure 1 and Table 1) the photothermal use efficiency of each single treatment (YPTU; mg MJ–1 °C–1) was calculated according to Thornley and Johnson (1990). For each experiment, Y was divided by the photothermal units (PTU; MJ m–2 °C) accumulated during in the growing period from the appearance of the first true leaf to harvest, (Eq. 2):
where Radi and Tai are the radiation and the mean air temperature measured at the ith day, respectively, and Tb represents the base temperature below which plant development does not occur. The base temperature was estimated by empirical methods, as suggested in other works (Wolfe et al., 1989; Jenni et al., 1996). Specifically, Tb was calculated as the value (ranging between –4°C and 8°C, with steps of 1°C) that maximized the determination coefficient of the linear regression (Eq. 3) between the number of true leaves (nLeaves) and growing degree (Σni =1(Tai-Tb); GDD):
In our case, the best fit was obtained with the following equation, using Tb=3°C: nLeaves = –6.47 + 0.030 • GDD, (n=260; R2=0.81, P<0.001).
For each experiment, normalized values of YPTU (Y*PTU) were obtained as the ratio between YPTU and its maximum value (YmaxPTU ):
Following the approach proposed by Magán et al. (2008), (YmaxPTU ) was represented by the average yield of those treatments that did not differ statistically (i.e., following ANOVA results) from the maximum yield obtained among all treatments.
A segmented linear-plateau model was adopted to fit normalized data of dry and fresh YPTU (Y*DW and Y*FW, respectively). With this model, biomass production is assumed to be zero if Nmin is less than or equal to a minimum threshold value (N0); afterward, Y starts to increase linearly with Nmin up to the optimal N concentration (Nopt) that represents the level of Nmin above which Y reaches its maximum value (Eq. 5):
where Y*PTU is the ratio YPTU/(YmaxPTU ). Following the approach proposed by Magán et al., (2008), the coefficients a and b were determined by consecutive linear regression analyses run fitting the complete group of treatments and step by step subtracting the treatments at the right of a possible Nopt threshold till achieving the highest coefficient of determination (R2). Finally, Nopt and N0 were obtained solving the equation Y*PTU=a+b•Nmin when Y*PTU was equal to 1 and 0, respectively.
A linear model was used to describe the relationship between the N nutrition index (NNI) and the SPAD index of spinach. Both parameters were calculated by the average of data collected during the entire cultivation cycle in each treatment. The NNI was calculated as the ratio between the tissue total N concentration for each treatment and the tissue average N concentration of those treatments with optimal Nmin levels in the root zone (i.e., Nmin ≥ Nopt).
Other data (i.e., crop Y and N uptake at harvest, and tissue nutrient concentrations) were analysed through one-way ANOVA and Tukey test (HSD) for the separation of the means.
The programs Statgraphics Centurion XV (Statpoint Technologies, Inc., Warrenton, Virginia, USA) and Prism 5 (GraphPad Software, Inc., La Jolla, California, USA) were used for data analysis.
Figure 1 and Table 2 report the main climate variables monitored during the whole experimental period. The average values of Ta did not differ much among the different trials, showing the lowest CV as compared with other climate variables (Table 2). Mean daily Ta was used to calculate GDD (see Eq. 2 for details), which averaged 799.2°C among all trials, with the minimum and maximum values recorded for E8 (652.6°C) and E3 (907.3°C), respectively (Table 2).
Mean daily Rad varied more than Ta during the experiment (Table 2) with the minimum daily-averaged value recorded for E5 and E6 (0.44 MJ m2 day–1) in December 2009 (Figure 1), and the maximum value recorded for E6 in March 2010 (25.0 MJ m2 day–1). The combination of Ta and Rad in Eq. 2 resulted in different PTU values that ranged between 2297.7°C MJ m–2 for E7 and 5550.3°C MJ m–2 for E6 with an average of 3404.5°C MJ m–2 (Table 2).
The accumulated rainfall, among the other climate variables (Table 2), showed the highest variability with the minimum and maximum value recorded for E8 (219.4 mm) and E4 (829.8 mm), respectively. In general, rainfall events were regularly distributed during the growing seasons, excluding some exceptional precipitations above 60 mm day–1 recorded for E3, E4 and E5 (Figure 1).
Due to diverse climate conditions, the duration of different crop cycles varied over the entire experimental period (Figure 1 and Table 2). The shortest culture was recorded in 2009, when E5 lasted 84 days, while the longest one (E6) was registered in 2009-2010 and lasted 132 days (Table 2).
The different N treatments (Table 2) and growing conditions produced a variety of Nmin levels in the root zone as summarized in Figure 2. Data reported in Figure 2A represent the averages of soil samples collected during the experimentation in each treatment (5 samplings). To evaluate the time-dependent variability of collected data, CV and SD were also calculated and averaged at each sampling time. The coefficient of variation averaged 30% and the maximum values were generally recorded for those treatments with no supply of N fertiliser (N0). High CV values in these treatments were therefore due to high variability in Nmin that generally tended to decrease with time. This effect could mainly be related to N depletion in the root zone, which occurred because of the absence of N fertilisers supply. On the contrary, the lowest variability for Nmin concentration was observed in those treatments that underwent N fertilisations and scarce rainfall with limited N-NO3– leaching. However, the different experiments differed significantly (ANOVA, P<0.001) for the actual mean values of N-NO3–, N-NH4+ and as a consequence Nmin concentration in the root zone (Figure 2A); the latter ranged from 7.6 to 41.0 mg kg–1 that corresponded to 44.3 and 237.6 kg ha–1 Nmin available in the soil profile explored by the root system (see Eq. 1 for conversion coefficients). Mineral N was strongly correlated to the concentration of N-NO3– (R=1.00, n=300, P<0.001), which accounted for roughly 81% Nmin on the average of all treatments. A significant albeit weak correlation was also found between N-NH4+ and Nmin (R=0.49, n=300, P=0.02).
On the other hand, it appears crucial to highlight that a very poor relationship was instead found between the actual Nmin and the nominal N dose supplied by fertilisers (Figure 2B).
Biomass accumulation and biometric parameters
At harvest, plant DW and FW significantly varied among treatments (Table 3). The dry weight accumulated in the aboveground biomass ranged between 0.25 (E6-0) and 2.43 (E1-160) t ha–1 while FW varied from 1.72 (E6-0) to 21.69 (E1-120) t ha–1.
Leaf area index ranged between 0.3 and 2.3 and was significantly correlated to both the accumulated DW (R=0.75, n=80, P<0.001) and FW (R=0.71, n=80, P<0.001). Similar results were also found for the total N uptake calculated on a DW basis, which was significantly correlated to plant biomass accumulation at harvest (R=0.95, n=80, P<0.001); N uptake increased from 6.28 (E6-0) to 87.37 kg ha–1 (E1-160) depending on treatments (Table 3).
Significant differences among N treatments were also observed for the specific leaf area that increased with Nmin availability. With respect to the thermal time (GDD), the specific leaf area significantly changed during the crop cycle, decreasing linearly from emergence to harvest (157.8 to 88.1 cm2 g–1 DW, calculated as the average of all treatments).
Modelling crop response to Nmin
Figure 3A clearly shows that N nominal doses, supplied by fertilisers, and plant biomass accumulation were poorly correlated. This was consistent with: i) the poor correlation observed between N nominal dose and actual Nmin in the root zone (Figure 2B); and ii) the high variability of Y in relation to the different growing conditions (i.e., Ta and Rad, Table 2 and Figure 1). Yield values were therefore divided by the PTU accumulated in each experiment, thus obtaining YPTU, and then plotted versus Nmin (Eq. 2 for details). This expedient allowed the standardization of Y in terms of different growing seasons with significant improvement in data analysis (Figure 3B). YPTU response to increasing Nmin was zero below a minimum threshold; then, it began to increase linearly with Nmin up to a maximum YPTU (YmaxPTU), after which no significant variation was observed. Data analysis produced (YmaxPTU ) values of 73.0 and 675.5 mg MJ–1 °C–1, respectively for DW (Figure 3B) and FW (data not shown). These quantities were finally used for the normalization of the biomass datasets. The normalized YPTU (Y*PTU), which represents YPTU as a proportion of (YmaxPTU ), was fitted using a segmented linear model (Figure 3C). Equation 5 significantly fitted Y*PTU, calculated for both YDW (Figure 3C) and YFW (data not shown), explaining 89 and 91% of the experimental variability, respectively (P<0.001). Model parameterization produced the following coefficients: i) an intercept (a; in Eq. 5) equal to –0.40 for Y*DW and –0.47 for Y*FW; ii) a slope (b; in Eq. 5), which represents the relative yield increase per Nmin unit, equal to 0.06 for both Y*DW and Y*DW; iii) N0 equal to 6.85 mg kg–1 for Y*DW and 7.36 mg kg–1 for Y*FW; iv) Nopt equal to 23.80 mg kg–1 for Y*DW and 23.08 mg kg–1 Y*FW. Since no significant difference was found for the parameterisation of Eq. 5 using the two datasets, it could be concluded that the averaged values of N0 (i.e., 7.11 mg kg–1) and Nopt (i.e., 23.44 mg kg–1) were representative for spinach crops. Basing on Eq. 1, the quantity of Nmin per surface unit corresponded to Nopt=140.0 kg ha–1 and N0=41.2 kg ha–1.
Plant tissue analyses and N uptake
Nmin in the root zone significantly affected dry matter percentage in plant tissues. A one-phase exponential decay equation was fitted to experimental data, explaining 62% and 66% of the measurement variability for data averaged over the whole crop cycle (data not shown) or only at harvest (Figure 4A), respectively. At harvest, DW percentages were 9.1 and 14.6%, for treatments with Nmin above and below Nopt, respectively. Data averaged during the whole crop cycle showed a more restricted range, from 9.6 to 12.9%.
An opposite pattern was observed for tissue total N concentration. In this case, a one-phase exponential growth equation was fitted to experimental data explaining 63% and 69% of the measurement variability for data averaged over the whole crop cycle (data not shown) or only at harvest (Figure 4B).
At harvest, a positive relationship was found between N-NO3– accumulation in plant tissues and Nmin in the root zone. However, in all treatments, N-NO3– concentration did not exceed the limits suggested by the European Food Safety Authority (EFSA) for this crop (Figure 4C). A one-phase exponential equation fitted the experimental data, showing a plateau value at 1030 g kg–1 FW (Figure 4C).
Plant DW and tissue N content (Figure 4C) were combined to calculate the crop total N uptake at harvest (Table 3) and over the cultivation cycle for each treatment. The latter data for spinach grown under optimal N conditions (i.e., excluding the treatments with Nmin below Nopt in Figure 3C) are reported in Figure 5 as a function of the GDD accumulated from sowing. The number of true leaves, which is also reported in Figure 5, is a further parameter that could be used to estimate the N uptake rate for spinach.
Data on leaf mineral content (Figure 6) and SPAD index (Figure 7) were pooled into two groups corresponding to Nmin values above Nopt (N+opt) or below Nopt (N–opt). The concentrations of N, P and Mg in plant tissues analysed at harvest were significantly affected by suboptimal Nmin availability in the root zone (Figure 6). Nitrogen was reduced by 15% from 3.9 to 3.3 g 100 g–1 DW while P and Mg were more severely affected with a reduction of 21 and 23%, respectively. Conversely, no significant difference was observed at harvest for K and Ca tissue concentrations (Figure 6).
Suboptimal Nmin levels also affected the SPAD index that provides an indirect estimation of chlorophyll content. For this parameter, no significant difference was found between data averaged at harvest or during the whole crop cycle (ANOVA, P>0.05) while a significant reduction was observed in N–opt treatments compared with N+opt treatments. Figure 7 shows the relationship between NNI and SPAD index. The linear regression model significantly fitted the measured data, explaining 78% of the experimental variability. The collected data suggest that values of SPAD index between 64.0 and 68.5 would be optimal for savoy spinach to avoid N deficiency or excess. This range was calculated on the basis of the confidence interval resulting from the linear regression analysis (Figure 7).
Yield response curve to Nmin concentration in the root zone
Very poor correlations between the nominal dose of N, applied through mineral fertilisers, and the actual Nmin concentration in the root zone or the harvested biomass were found in this work. In some cases (i.e., treatments E1-0 and E2-0; Table 3) high Y occurred without any N supply, since a sufficient (optimal) level of Nmin was already present in the root zone before sowing, which supported adequately plant N nutrition throughout the whole cultivation cycle. Similar results have been obtained in previous studies with spinach (Gülser, 2005; Stagnari et al., 2007). Defining the relationship between crop Y and Nmin appears therefore of fundamental importance for a balanced N supply whereby both economic and environmental sustainability of the crop can be maximized (Schroder et al., 2000; Cui et al., 2008; Bai et al., 2013).
The harvest time varied widely among the different experiments. Spinach was harvested when the plants achieved 16 to 22 leaves, according to the market requirement and the weather (for example, rainfalls can delay the harvest). Furthermore, the quite different climate conditions (mainly Ta and Rad) during the years of observation caused a large variability in terms of plant growth. Crop Y was therefore standardized by PTU thus obtaining comparable data (i.e., YPTU) collected over different growing seasons. The use of YPTU was successfully applied by other authors to assess differences in the theoretical Y of several crops grown in different regions (Hou et al., 2012).
For modelling crop response to soil nutrient concentration, several authors fitted crop Y using linear-plateau models (Reid, 2002; Cui et al., 2008; Bai et al., 2013; Magán et al., 2008) as was done in the present work by Eq. 5. On the other hand, many other authors described the response of different crops to Nmin or N fertiliser dose using non-linear equations (Thornley and Johnson, 1990; Van Noordwijk and Wadman, 1992; Reid, 2002; Milne et al., 2012). Therefore, the preliminary data analysis in the present work included non-linear models consisting in quadratic and hyperbolic functions as suggested by Thornley and Johnson (1990). However, the resulting determination coefficients (in the range of 0.83-0.86), or other error performance indices, were close to the ones achieved with the proposed linear model or even worse. The same was obtained for the other observed statistical parameters.
With respect to non-linear models, Eq. 5 has the advantage of producing an unequivocal value for Nopt; in contrast, the use of non-linear equations may lead to higher uncertainty in the identification of Nopt. The Nopt value found in the present study (23.44 mg kg–1) for spinach was quite similar to others reported in literature for different crops. In fact, optimal Nmin in the root zone have been found in the range of 16-30 mg kg–1 dry soil for corn (Cui et al., 2008; Peng et al., 2013), 20-21 mg kg–1 for wheat (Bundy and Andraski, 2004), 20-30 mg kg–1 for potato (Doll et al., 1971), 24 mg kg–1 for cabbage (Heckman et al., 2002), and 20-24 mg kg–1 for celery and lettuce (Hartz et al., 2000; Bottoms et al., 2012).
Crop quality response to Nmin concentration in the root zone
Nitrate accumulation in spinach leaves is due to many factors depending on both environmental growth conditions (e.g., temperature, radiation, N fertilisation management) and plant-specific characteristics (e.g., nitrate reductase activity, leaf age) (Lasa et al., 2001; Chen et al., 2004; Gülser, 2005; Stagnari et al., 2007). High availability of Nmin in the root zone indeed promotes nitrates and total N accumulation in leaves. Similar results have been observed in other leafy vegetables, such as romaine and red-oak leaf lettuce (Di Gioia et al., 2017). However, the tissue nitrate content observed in this work was within the limits suggested by EFSA (2008) for spinach. It was likely due to the quite low plant density and to the N level in the soil that was not too much exceeding Nopt (maximum N-NO3 value was 36 mg kg–1, recorded in the treatment E2-160).
Tissue content of other plant mineral nutrients responded differently to Nmin treatments depending on the nutrient element. Literature on spinach is quite heterogeneous with regard to the relationship between tissue content of plant mineral nutrients and Nmin in the root zone. In several studies, positive correlations between N, Ca, or Mg and Nmin have been reported (Lefsrud et al., 2007; Staganari et al., 2007). However, contrasting findings have been reported for P or K that were found to increase (Stagnari et al., 2007), to be constant (Lefsrud et al., 2007, only K) or even to decrease (Gülser, 2005, only P) by increasing Nmin in the root zone. A significant positive correlation between P or Mg and Nmin in the root zone (P<0.001, R=0.70 or 0.71, respectively) was observed in this work, as previously reported by other authors (Lefsrud et al., 2007; Stagnari et al., 2007). On the other hand, no significant correlation was found between K or Ca tissue content and N treatments, in agreement with Gülser (2005).
Increasing Nmin in the root zone led to an enhanced blade colour. This is a relevant reference extrinsic characteristic, much appreciated by consumers of fresh-market spinach. The SPAD index was significantly lower in N–opt treatments, in which plants were grown under suboptimal nutritional conditions. SPAD index is significantly correlated to leaf N and leaf chlorophyll concentration (Schepers et al., 1992), on which spinach colour depends mostly. Leaf chlorophyll concentration has been found positively correlated to Nmin in the root zone in spinach (Lefsrud et al., 2007), and to leaf N and Mg concentration in many species (Shaahan et al., 1999) as also observed in this work.
The use of chlorophyll meters, such as the SPAD, could be not reliable to estimate on-time crop N requirement because of possible delays between the nutrient status of the root zone and its effects on plants (Westerveld et al., 2003; Wu et al., 2007). Nevertheless, the information provided by these tools and other optical sensors may be valuable if coupled with data on soil composition obtained through either laboratory analyses or easy-to-use and rapid methods available at farm level (Hartz et al., 2000; Maggini et al., 2010; Thompson et al., 2017).
Implementation of the results for improving spinach N fertilisation management
Growers usually adopt very simple N fertilisation plans for fresh-market spinach. In Tuscany (Italy), where the experiment took place, they usually apply from 120 to 175 kg ha–1 N for each growing cycle, often as fixed reference dosages without pre-sowing soil analyses. This fertilisation approach is quite robust and effective from the growers’ point-of-view. In common practice, the application of 120 kg ha–1 N to a crop like savoy spinach, which takes up from roughly 60 to 75% of the above quantity, consists in a sort of insurance against possible low soil N concentration and rainy periods that could hinder N top dressing applications. Obviously, this approach contributes to N loss phenomena related to excess N in the root zone and may reduce produce quality. In the above scenario, the risk for high crop environmental impact and production costs increases drastically (Massa et al., 2013).
According to our results, the presence of roughly 23 mg kg–1 of Nmin in the soil is enough for a regular spinach growth and development without any yield and/or quality reduction. An ideal management of N fertilisation would consist in the supply of small fertiliser amounts at high frequency to keep Nmin concentration close to Nopt, as in fertirrigated crops. Nevertheless, a similar fertiliser application plan would not be sustainable for winter spinach under open-field operative conditions from both the agronomic and the economic point of view. Based on the crop N uptake curve reported in Figure 5, 50% of the total N absorbed by spinach is concentrated in the last part of the growing cycle (the last three to five weeks between the 13-15 true-leaf phase and harvest time). Similar results have been reported for other leafy vegetables such as lettuce (Bottoms et al., 2012). Therefore, when N fertilisers have to be applied in advance, data reported in Figure 5 must be carefully taken into account in terms of doses and distribution frequency to avoid excess N supply when plants are not ready to take this element up.
The results obtained in this work could be implemented for advanced N fertilisation strategies, which are based on a pre-sowing Nmin soil analysis followed by Nmin monitoring in the root zone (Thompson et al., 2017). At the pre-sowing stage, if the soil Nmin is lower than Nopt, a base fertilisation is necessary. After sowing, topdressing fertilisations will then be necessary only if the level of Nmin in the root zone drops below Nopt. Nowadays, soil Nmin can be easily monitored at farm level by using quick tests that are faster, simpler, and cheaper than the conventional laboratory analyses (Hartz et al., 2000; Maggini et al., 2010; Incrocci et al., 2017). For example, this approach has been validated for the fertilisation plan of lettuce and celery in California (Hartz et al., 2000), and for cabbage in North America (Heckman et al., 2002) or in The Netherlands (Everaarts and de Moel, 1998).
Furthermore, the model calibrated in the present work (Eq. 5) can be implemented in decision support systems for the precise nutrient management of spinach. In addition to the estimation of Nopt, the implementation of Eq. 5 in decision support systems can be useful to simulate spinach growth, as a function of Nmin in the root zone, and eventually calculate the dose and distribution frequency of N fertilisers based on real plant needs.
The relationship between NNI and SPAD index provides additional information for the precise management of N nutrition in spinach using optical sensors, as previously reported for other vegetable crops (Padilla et al., 2015; Thompson et al., 2017; Incrocci et al., 2017). Spinach for fresh market is harvested at early growth stage. We observed that, within this growing period, N tissue concentration did not vary significantly; it was therefore possible to find a significant correlation between SPAD index and NNI using data averaged over the whole cultivation cycle. On the other hand, much higher variability was observed for other vegetable crops in different growing periods (Padilla et al., 2014, 2015).
As observed in the present work, SPAD values lower than 64.0 would imply N deficiency with reduced Y for fresh-market spinach. Lower values were observed for open-field processing (Canali et al., 2014) and greenhouse-grown spinach (Liu et al., 2006; Muchecheti et al., 2016). However, the threshold values reported in Figure 7 (in the range of 64.0-68.5) were assessed in this work for the first time for NNI-based quick N monitoring in fresh-market spinach grown under open-field Mediterranean climate conditions. The combination of soil analyses and optical leaf sensors, like the SPAD used in this study, indeed appears of great interest for the optimized management of N fertilisation in vegetable crops (Incrocci et al., 2017).
Conclusions and remarks
Meaningful relationships between crop yield or quality and mineral nitrogen availability in the root zone are reported in this work for fresh-market spinach. A linear-plateau model significantly represented the yield response to N concentration in the root zone. The adopted model was successful in determining an optimal threshold value (i.e., Nopt) of 23.44 mg kg–1 DW (140.0 kg ha–1 in our experimental conditions) to be maintained in the root zone for efficient N fertilisation plans of this crop.
Plant tissue analyses supported the hypothesis that there is no reason to exceed Nopt in the root zone since above this threshold most of plant tissue characteristics tend to be unaffected. In contrast, plants grown under suboptimal levels of N in the root zone (i.e., below Nopt) show a reduced yield and tissue content of mineral nutrients and SPAD index, which imply that Nopt must be maintained in the root zone to obtain high yield and quality. Maintaining the suggested conditions in the root zone may therefore significantly improve the environmental and economic sustainability of the crop.
The use of optical sensors (e.g., the SPAD by Minolta used in the present work) can be helpful for growers to check quickly the N nutritional status of the crop; SPAD index higher than 64.0 would ensure optimal N nutrition in fresh-market spinach. The combined monitoring of the root zone and crop canopy therefore appears a strategy worth exploring for the correct management of N fertilisers in this crop. The results reported in this paper can be implemented in protocols, algorithms and decision support systems for the optimized N fertilisation of fresh-market spinach.
Coefficient of variation
Growing degree days or thermal time
Leaf area index
Soil mineral nitrogen concentration below which yield is zero mg kg–1, kg ha–1
Number of true leaves
Pool of treatments below Nopt
Pool of treatments above Nopt
Total mineral nitrogen in the root zone
mg kg–1 DW or kg ha–1
Optimal nitrogen concentration to be maintained in the root zone for spinach
mg kg–1, kg ha–1
Nitrogen nutrition index
MJ m–2 °C–1
MJ m2 day–1
Base air temperature
Photothermal use efficiency or potential yield
t MJ–1 m–2 °C–1
Maximum photothermal use efficiency or potential yield
t MJ–1 m–2 °C–1
Table 1.: Chemical and physical characteristics in the 0-40 cm depth layer of the different experimental fields (n=8) used for spinach cultivation.
Bulk density (t m–3)
Field capacity (%, v/v)
Wilting point (%, v/v)
Organic matter (%)
Total N (mg kg–1)
N-NO3– (mg kg–1)
N-NH4+ (mg kg–1)
P2O5 (mg kg–1)
K2O (mg kg–1)
CaO (mg kg–1)
MgO (mg kg–1)
CEC (meq 100g–1)
EC (dS m–1)
Table 2.: Period of cultivation (dates), experiment duration (days after sowing, DAS), and nominal doses of N supplied in each experiment (E) are reported in the table as single values. Mean air temperature (Ta), growing degree days (GDD), mean and cumulative daily global radiation (Rad and Cum. Rad, respectively), photothermal units (PTU) and cumulative rainfall (Rain) recorded in each experiment (E).
Dose of N applied (kg ha–1)
Rad (MJ m–2 day–1)
Cum. Rad (MJ m–2)
PTU (°C MJ m–2)
Table 3.: Effect of different N supplies on dry weight and fresh weight accumulated at harvest. Each value represents the average of replicates ±SD.
*The abbreviations represent Experiment number-Dose of N supplied with fertilisers (kg ha–1, see Table 2); P-value for one-way ANOVA is reported. Different letters in each column represent significant differences according to Tukey’s (HSD) test (P≤0.05).
2008-2019 • PAGEPress is a registered trademark property of PAGEPress srl, Italy. • VAT: IT02125780185 • Privacy