**Organic Vegetable Crops Managed with Agro-Ecological Practices: Environmental Sustainability Assessment by** *DEXi-met* **Decision Support System**

#### **Francesco Montemurro 1,\* , Alessandro Persiani <sup>2</sup> and Mariangela Diacono <sup>2</sup>**


Received: 26 August 2019; Accepted: 1 October 2019; Published: 3 October 2019

**Abstract:** In the last decade, there has been an increasing interest in sustainable agricultural techniques and the environmental evaluation of the effects of agricultural practices. In the present study, we evaluated both the production capacity of organic horticultural systems, and the ex-post sustainability through a new multi-attribute decision model named "*DEXi-met*". This qualitative model is able to estimate the environmental sustainability of cropping systems managed with different agro-ecological approaches. In particular, we compared the following three horticultural systems: (i) ECO, an organic system with full implementation of agro-ecological strategies (agro-ecological services crops (ASC), strip cultivation, and organic amendment); (ii) GM, an organic system with the introduction of the ASC; (iii) NO ASC, an organic system without ASC. The treatments with ASC presence (ECO and GM) showed similar total energy outputs (substantially higher than the NO ASC), indicating the positive effect of this agro-ecological practice. The findings pointed out that the ECO system, which followed the principles of natural ecosystems, can contribute to building up more complex agro-ecosystems, increasing both resilience and biodiversity. This management strategy reached a good compromise between the production of vegetable cropping systems and environmental sustainability achievement. Then, it is possible to optimize the use of natural resources, support climate adaptation, and reduce greenhouse gas emissions.

**Keywords:** qualitative multi-attribute model; total energy output; agro-ecological service crops; ex-post sustainability; organic systems

#### **1. Introduction**

Sustainable development and environmental sustainability are broadly recognized as global and collective goals because of key issues such as limited resources, environmental pollution, and global warming [1]. These increasing challenges, also considering the local and the global legislative changes, have inevitably involved the agricultural sector. Therefore, agronomists, farmers, and researchers should research, design, and experiment with new agricultural systems that are environmentally friendly, economically viable, socially supportive, and efficiently adapted to a climate change context. In different farming systems, a wide range of cultivation techniques and agro-ecological management strategies that enhance biodiversity in crop fields and support the sustainability of the agro-ecosystems are already practiced and should be further promoted [2,3]. Among them, crop rotations, introduction

of agro-ecological service crops (ASC; [4]), utilization of soil amendments [5], and crop/livestock mixing can increase agro-ecosystem diversity and complexity both over space and time [6].

A prerequisite to the systems sustainability implementation is the development, improvement, and/or choice of the best possible assessment methods. In fact, there is a need to determine the reliability of innovative management practices with respect to the conventional ones, and to clarify the benefits and the drawbacks of the full or partial application of an agro-ecological approach. To this end, several approaches to measure, analyze, and assess sustainability have been developed [7]. The different methodologies can be classified on the basis of: (i) the typology of indicators used, from qualitative appraisals to quantitative analytical evaluations [8]; (ii) the scale of analysis, from the single plot to the whole farm or regional scales [9]; (iii) the systems typology, from orchards [10] to arable or horticultural crops [11]; and the timing of analysis, as ex-ante or ex-post evaluation [12].

Within this large number of methodologies, a growing interest focuses on multi-method approaches, which aim at accounting for the complexity of sustainability issues [13]. In this framework, the multi-criteria analysis (MCA) decision-making methods can handle a typical decision-making problem related to sustainability assessment. The MCA are increasingly gaining importance in agriculture, since they can consider multiple and conflicting criteria and, at the same time, they are able to tackle complex decisional problems breaking them down in easily understandable elements [14–16]. In recent years, the scientific community has developed several qualitative MCA tools for the sustainability assessment of different agricultural systems [17,18], based on a computer program for multi-attribute decision-making, defined as "DEXi" by Bohanec et al. [19]. Among them, the *DEXi-met* tool was recently developed, specifically for the ex-post evaluation in organic horticulture, and it was applied to compare different crop rotations in Mediterranean conditions [20]. In this model, a cropping system was considered instead of a single cash crop, in order to have a broader idea of the environmental sustainability of the system.

There is still a lack of knowledge on the sustainability assessment when different levels of the agro-ecological approach are applied, especially in organic horticultural production in the Mediterranean environment. In light of these considerations, the aim of the present research was to evaluate the performance of different cultivation systems managed with agro-ecological practices. To accomplish this aim we evaluated both the production capacity of the systems and the ex-post environmental sustainability by using the *DEXi-met* qualitative multi-attribute model.

#### **2. Materials and Methods**

#### *2.1. Study Site*

The study was performed in the research farm 'Azienda Sperimentale Metaponto' of the Consiglio per la Ricerca in Agricoltura e l'analisi dell'economia Agraria. The farm is located at Metaponto (MT), in southern Italy (lat. 40◦24′ N; long. 16◦48′ E, 8 m above sea level).

The soil, classified as a Typic Epiaquert, has the following properties: low N (1.0 g kg−<sup>1</sup> ) and organic matter (19.0 g kg−<sup>1</sup> ) contents, 759 mg kg−<sup>1</sup> of exchangeable potassium (K), 31.1 mg kg−<sup>1</sup> of available phosphorus (P), pH value of 8.4, clay and silt contents of 60 and 36%, respectively, electrical conductivity of 0.48 mS cm−<sup>1</sup> (at 0 to 30 cm depth), increasing with depth, average bulk density of 1350 kg m−<sup>3</sup> , cation exchange capacity of 27.1 meq 100 g−<sup>1</sup> of dry soil and the soil water content (a percentage of soil-dry weight) of 34.5% and 20.1% at field capacity (−0.03 MPa) and permanent wilting point (−1.5 MPa), respectively.

The climate is classified as "accentuated thermo-Mediterranean", considering the UNESCO-FAO classification [21], with mean monthly temperatures of 8.8 ◦C in the winter, and 24.4 ◦C in the summer. Winter temperatures can fall below 0 ◦C, whereas summer temperatures can rise above 40 ◦C. The total rainfall (on average 490 mm year−<sup>1</sup> ) is concentrated mainly during the winter months and the mean annual potential evaporation rate is 1549 mm.

#### *2.2. Experimental Field Trials, Treatments, and Measurements*

The research was carried out during the 2016–2017 cropping season in two different experimental fields. The first one consists in a long-term field trial in organic horticulture, which had been planned to adapt horticultural systems to unfavorable climatic conditions (in particular, extreme rainfall events during autumn and winter periods). In this experimental field, integrated strategies are combined, i.e., soil surface shaping, crop rotations, introduction of agro-ecological services crops (ASC), ASC termination techniques, and fertilization with organic products [4]. The base layer is the soil surface shaping in a "ridge system". Cash crops are planted both on the top of each raised bed 2.5 m wide (ridges) and in the 2.5 m flat areas (strips) between them. The crop rotation is designed to cultivate the cash crop on the ridges and the ASC in the strips during the winter-rainy period of the year. Cover crops are used to prevent soil erosion and provide N to the system via biological fixation, since on the top of the ridges, a leguminous cover crop is intercropped (as living mulch) in the winter as a vegetable crop and maintained as a ground cover. During the winter-rainy period, in the flat soil strips, mixtures of ASC species of different botanical families are cultivated between two consecutive spring–summer cash crops. The used ASC termination methods (before the subsequent cash crops transplant) are green manure (in which the cover crop is chopped and plowed at the end of flowering) vs. cover crop biomass flattening by an in-line roller-crimper, in which the mulch covers the soil surface until the vegetable crop harvest [4]. Finally, the last layer consists of organic fertilization, which is implemented into further horizontal strips, by using commercial and experimental amendments.

The second experimental field was conceived to verify the hypothesis that the use of the in-line roller crimping technology for ASC termination will improve the agronomical performances of the organically managed vegetable cropping systems. A two-year field experiment was carried out to evaluate the effect of ASC termination on tomato, by comparing green manure vs. roller-crimper, NO ASC (control) and plastic mulch (positive control) treatments. Another variability factor consists of three different fertilization treatments (i. on-farm organic fertilizer, ii. commercial organic fertilizer, and iii. unfertilized control).

The experimental design of each field trial was a strip plot with three replications, allowing the ability to calculate the standard deviation of the variables.

In the present research, the following three management systems adopting different agro-ecological approach levels are considered, which have been extrapolated from the two above-defined experimental field trials (Table 1):

1. ECO, an organic system with the full implementation of the described agro-ecological strategies, from the first experimental field. The cultivation area (1 ha) was divided into two parts (0.5 ha ridge furrow and 0.5 ha flat strip) and crops were cultivated both on the ridges and in the strips. On the ridges, with the clover as a living mulch, cauliflower during the winter period (transplanted on 20 October 2016 and harvested on 21 March 2017) and tomato crop during the spring–summer (transplanted on 24 April 2017 and harvested during August 2017) were cultivated. In the strips, the ASC (80% vetch-20% oats) were sown in November 2016 and incorporated as break crops during the following spring. Zucchini, during the spring–summer (transplanted on 27 April 2017 and harvested during July 2017), and lettuce, during the late summer–autumn (transplanted on 31 August 2017 and harvested on 26 October 2017), were then cultivated. A composted anaerobic digestate from cattle manure was used as fertilizer (i.e., on-farm organic fertilizer). The phytosanitary management followed the organic farming rules. The ECO cultivation system is under study in an experimental field in which the adaptation of horticultural systems to extreme climatic events are being tested, since these phenomena are increasing in the Mediterranean area. Consequently, we choose this experimental system to verify the hypothesis that the above-described practices may be used by the farmers as potential adaptation strategies for organic agro-ecosystems.


**Table 1.** Description of the three different systems analyzed. ECO = organic system with the full implementation of the agro-ecological strategies; GM = organic system with the introduction of the agro-ecological service crops (ASC); NO ASC = organic system without ASC.


In ECO, at the cauliflower, zucchini, and lettuce commercial maturity, five randomly selected plants in each plot were collected to determine both "production quantity" attribute and the most important quality parameters for the calculation of "production quality" attribute. Conversely, at harvest, in GM and NO ASC the tomato fruits were collected from two randomly selected plants (center of the 2 rows in each plot) and both marketable and total yields and quality parameters were recorded to calculate production attributes.

#### *2.3. Sustainability Evaluation*

#### 2.3.1. *DEXi-met* Model Application

To assess the sustainability of the agro-ecological practices implemented in the experimental field trials, crops yield and energy outputs were measured. The marketable yields were multiplied by their

own coefficient of equivalent energy taken by the literature, to estimate the energy outputs [22]. The data of each agricultural operation were collected in a standardized procedure. All field practices were recorded (human labor as h ha−<sup>1</sup> , fuels consumption as kg ha−<sup>1</sup> ), during the cover crop management and the cash crop cycles. Moreover, to better understand the systems environmental impact and sustainability, the *DEXi-met* model was used, aiming at assessing the level of sustainability of each considered system [20]. *DEXi-met* was developed for the ex-post assessment in organic horticulture by implementing the original DEXi software, which is utilized in multi-criteria decision analysis [23]. The ex-post assessment carried out with this model includes the basic attributes derived from the field experiment (e.g., productions, organic matter, etc.). In more details, *DEXi-met* is based on a hierarchical decision tree structure that breaks down the sustainability into smaller modules, which can be explained and calculated. Both qualitative and quantitative basic attributes are categorized into a linguistic scale, that is from a three-value scale ("low", "medium", "high"), used for the basic attribute, to a seven-value scale ("very-low", "low", "medium-low", "medium", "medium-high", "high", "very-high") for the "overall sustainability". The evaluation procedure begins with the calculation of the basic attributes, that could be also calculated using a satellite tree [24]. Their homogenization into the rank-ordered qualitative scale and the pyramidal aggregation of attributes contributed to the calculation of the aggregated final sustainability. The aggregation procedure is based on decision rules and relative weightings, that were given to each attribute, according to their alleged significance and contribution to sustainability. The weightings were defined involving both decision analysts and experts (i.e., researchers, agronomists, and farmers) and considering the literature, as indicated in Montemurro et al. [20]. The *DEXi-met* model tree structure is reported in Figure 1. All the attributes (basic and aggregate) from the bottom to the top, their aggregation weights and the corresponding scales are presented, to understand the calculation of the final "overall environmental sustainability".

#### 2.3.2. *DEXi-met* Sensitivity Analysis

In order to identify the most significant variables that affected the sustainability of the systems, a sensitivity analysis (SA) of the *DEXi-met* model was also performed. According to the suggestions of Iocola et al. [17], the SA was performed utilizing the IZIEval tool (http: //wiki.inra.fr/wiki/deximasc/Interface+IZI-EVAL/Accueil). The IZIEval is an interface shaped to facilitate the multi-criteria sustainability assessment of cropping systems based on models developed with the DEXi software, supplementing the existing features of DEXi. The Algdesign and XML packages, of the open-source R software [25], were used for the SA.

Through the IZIEval interface, both the sensitivity indexes (SI) and Monte Carlo (MC) analyses were performed, to gain the SA. In particular, according to Carpani et al. [15] we used the same basic attributes utilized for the "overall environmental sustainability" in the sensitivity indexes computation. The software automatically attributed an equal weight or probability of occurrence to all possible values of each variable. The SI highest values corresponded to the most important effect for a specific variable within the "overall environmental sustainability". The SI used the hierarchical model tree structure to obtain the results. Aggregation weights and number of the basic variables at the same level, aggregation weights of the aggregated variables, and depth levels influenced the findings.

To model the probability of different outcomes when random variables are involved, the Monte Carlo simulations are a possible tool. They allow obtaining the relative frequency distribution of the output values of an aggregated variable. In our study, according to Iocola et al. [17] this analysis was carried out by using IZIeval, randomly sampling and simulating a large number of values (5000) of each variable, to obtain the frequency distribution of the overall sustainability values and their main components.


"high"), used for the basic attribute, to a seven value scale ("very low", "low", "medium low", "medium", "medium high", "high", "very high") for the "overall sustainability". The evaluation

calculation of the final "overall environmental sustainability".

value scale ("low", "medium",

**Figure 1.** The *DEXi-met* model decision tree. The model includes 30 basic attributes, aggregate attributes at different levels, four nodal attributes, and the overall sustainability. The numbers between attribute levels represent the default aggregation weights (expressed in %). For each attribute level (basic, aggregate, and overall) the scale is reported at the bottom of the figure.

#### **3. Results**

#### *3.1. Yields Performance and Energetic Outputs*

The highest absolute value of tomato marketable yield was found in GM, whereas the ECO treatment showed the lowest one with a reduction of −58 and −24% in comparison with GM and NO ASC systems, respectively (Table 2).

**Table 2.** Effects of management strategies on marketable yields (Mg ha −1 , values ± standard deviation). ECO = organic system with the full implementation of the agro-ecological strategies; GM = organic system with the introduction of the agro-ecological service crops (ASC); NO ASC = organic system without ASC.


The ECO treatment also showed a very low marketable yield in cauliflower cultivation. On the whole, considering that in ECO the cultivation area of each crop was 0.5 ha, while in GM and NO ASC it was doubled, the GM treatment determined higher total energy output by 70.3% and 14.4% as compared to NO ASC and ECO treatments, respectively (Table 3). Furthermore, the difference between GM and ECO was due to the low energy output occurred in cauliflower. In any case, all values of energy output were characterized by a high variation.

**Table 3.** Crop ((MJ ha <sup>−</sup><sup>1</sup> values ± standard deviation) and total energy output divided by the management strategies. ECO = organic system with the full implementation of the agro-ecological strategies; GM = organic system with the introduction of the agro-ecological service crops (ASC); NO ASC = organic system without ASC.


#### *3.2. Environmental Sustainability Evaluation*

The overall environmental sustainability of the tested cropping systems varied considering the different crop management (Figure 2). In particular, the "high" score was reached by ECO treatment, while a "medium-high" and "medium-low" score was obtained for GM and NO ASC, respectively. different crop management (Figure 2). In particular, the "high" score was reached by while a "medium high" and "medium low" score was obtained for GM and NO ASC, respectively.

**Figure 2.** Comparison among the different crop management strategies: evaluation results of the multi-criteria decision model *DEXi-met* on the overall sustainability. ECO = organic system with the full implementation of the agro-ecological strategies; GM = organic system with the introduction of the agro-ecological service crops (ASC); NO ASC = organic system without ASC.

ate attribute "production capacity" resulted in "medium" in ECO and GM treatments and "medium low" in NO ASC (Figure 3). The nodal aggregate attribute "production capacity" resulted in "medium" in ECO and GM treatments and "medium-low" in NO ASC (Figure 3).

The ECO treatment scored "high" value for the aggregate attributes "soil and water preservation" and "resource preservation", while GM and NO ASC reached "medium-high" and "medium" scores, respectively, for these same aggregates. The "biodiversity conservation" ranged from "high" in ECO to "low" in NO ASC.

The sustainability evaluation of all the components (from the overall sustainability to the basic attributes) for the three scenarios is reported in Figure 4, as a comparison among the different tested systems. It is also reported the level of sustainability (from "sustainable" to "not sustainable") of each item, according to the specific linguistic scale (from three to seven) described in Figure 1.

"soil and water preservation", "resources preservation", and "biodiversity conservation"). ECO =

on the four main aggregate attributes ("production capacity",

ECO GM NO ASC

**Management strategy**

**−**

**Overall sustainability**

very high high medium high medium medium low low very low

**Figure 3.** Comparison among the different crop management strategies: evaluation results of the multi-criteria decision model *DEXi-met* on the four main aggregate attributes ("production capacity", "soil and water preservation", "resources preservation", and "biodiversity conservation"). ECO = organic system with the full implementation of the agro-ecological strategies; GM = organic system with the introduction of the agro-ecological service crops (ASC); NO ASC = organic system without ASC. "medium" scores, respectively, for these same aggregates. The "biodiversity conservation" ranged from "high" in ECO to "low" in NO ASC. systems. It is also reported the level of sustainability (from "sustainable" to "not sustainable") of each

**Figure 4.** Evaluation results from *DEXi-met* model from the overall environmental sustainability to the basic attributes. ECO = organic system with the full implementation of the agro-ecological strategies; GM = organic system with the introduction of the agro-ecological service crops (ASC); NO ASC = organic system without ASC.

only basic attribute (N balance) with "low" value. Conversely, in NO ASC, the most frequent score was " low", while GM showed intermediate values between the other two treatments.

The "production capacity" aggregate attribute is generated from the first order aggregate

The "production capacity" aggregate attribute is generated from the first order aggregate attributes "control of pests and diseases", "physical-chemical fertility", and "production". In the ECO management, most of these attributes scored "high", "medium-high" or "medium-low", with one only basic attribute (N balance) with "low" value. Conversely, in NO ASC, the most frequent score was "medium-low", while GM showed intermediate values between the other two treatments. The "soil and water preservation" aggregate attribute showed small differences among water management options, while the first order attribute "soil" was "high", "medium " and "low" for ECO, GM, and NO ASC, respectively. These differences are generated by the basic attributes "tillage

The "soil and water preservation" aggregate attribute showed small differences among water management options, while the first order attribute "soil" was "high", "medium-low" and "low" for ECO, GM, and NO ASC, respectively. These differences are generated by the basic attributes "tillage diversification" and "tillage typology and depth" ("high" in ECO and "medium" in GM and NO ASC), "soil erosion control ("high" in ECO and "medium-high" and "medium-low" for GM and NO ASC, respectively) and "organic matter balance" ("high" in ECO and GM and "low" in NO ASC). diversification" and "tillage typology and depth" ("high" in ECO and "medium" in GM and NO ASC), "soil erosion control ("high" in ECO and "m high" and "medium low" for GM and NO ASC, respectively) and "organic matter balance" ("high" in ECO and GM and "low" in NO ASC). The "resources preservation" aggregate attribute differed for the attributes related to the "energy" and "fertilization", which scored frequently "high" and "medium" in ECO treatment, "medium" in GM and "medium low" and "low" in NO ASC. No differences were found in the first

The "resources preservation" aggregate attribute differed for the attributes related to the "energy" and "fertilization", which scored frequently "high" and "medium" in ECO treatment, "medium" in GM and "medium-low" and "low" in NO ASC. No differences were found in the first order "phytosanitary management" attribute and in the basic attributes. order "phytosanitary management" attribute and in the basic attributes. fferences were recorded in the "biodiversity conservation" component. In particular, the ECO treatment scored "high" and "medium high" in most of the basic attributes, GM showed frequently "medium high" values, while NO ASC scored "medium low" and

Finally, a large number of differences were recorded in the "biodiversity conservation" component. In particular, the ECO treatment scored "high" and "medium-high" in most of the basic attributes, GM showed frequently "medium-high" values, while NO ASC scored "medium-low" and "low" values. The results of the SI calculation for the basic attributes are reported in Figure 5. "low" va

**Figure 5.** Sensitivity index values obtained with IZIEval tool for each basic variable of *DEXi-met* referred to the overall sustainability. The vertical line distinguishes the more sensitive variables (right side of the line) from the others.

The "microorganism preservation" and the "macrofauna preservation" reached the highest (0.12 ely) SI values, being the most influential variables of the first order "biodiversity conservation" attribute. Within the "resources preservation" aggregate attribute, the "fertilizer C/N" basic attribute reached the highest SI value, while in the "production capacity" attribute, the "insects and pest diseases" and the "weeds" and "production quantity" showed higher values compared with the other basic attributes. Within the "soil and water preservation" component, the only basic The "microorganism preservation" and the "macrofauna preservation" reached the highest (0.12 and 0.08, respectively) SI values, being the most influential variables of the first order "biodiversity conservation" attribute. Within the "resources preservation" aggregate attribute, the "fertilizer C/N" basic attribute reached the highest SI value, while in the "production capacity" attribute, the "insects and pest diseases" and the "weeds" and "production quantity" showed higher values compared with the other basic attributes. Within the "soil and water preservation" component, the only basic attribute "soil erosion control" overtakes the 0.2 sensitivity index.

(MC = 0.543) than the other values. Among the nodal attributes, the "production capacity" recorded

attribute "soil erosion control" overtakes the 0.2 sensitivity index.

Table 4 reports the frequency distributions of the 5000 simulated outputs of the Monte Carlo (MC) analysis for the overall sustainability and for the main model components (nodal attributes). The "overall environmental sustainability" showed the qualitative value "medium" more frequently (MC = 0.543) than the other values. Among the nodal attributes, the "production capacity" recorded the value "medium" (MC = 0.496) that occurred more frequently than the other modalities. The "medium-low" and "medium" values were the most frequent for the "soil and water preservation" (MC = 0.571 and MC = 0.275, respectively), while the "resources preservation" principally scored the "medium" and "medium-high" values (MC = 0.475 and MC = 0.399, respectively). In the nodal attribute "biodiversity conservation", the "medium-low" and "medium-high" were the most frequent values.

**Table 4.** Relative frequency distributions of the results of 5000 Monte Carlo simulations among the seven different qualitative values ("very low", "low", "medium-low", "medium", "medium-high", "high", "very high") for the overall sustainability and among the five different qualitative values (low", "medium-low", "medium", "medium-high", "high") for the main aggregate attributes ("production capacity", "soil and water preservation", "resources preservation", and "biodiversity conservation") obtained with *DEXi-met.*


#### **4. Discussion**

#### *4.1. Yield Performances and Energetic Output*

The values of the tomato marketable yields in GM were higher by 70 and 123% compared with NO ASC and ECO treatments, respectively. This result was probably due to higher availability of readily available N, which derives from decomposition of the aboveground biomass of the ASC plowed into the soil [4,11]. Conversely, in the ECO plots, in which the clover was used as living mulch [26,27], and in the NO ASC, the tomato yield did not reach the standard level of organic production [28]. Moreover, in the ECO system, the cauliflower marketable yield was very low, likely because adverse climatic conditions occurred. In fact, during the growing season there was an extreme adverse event, unusual in the experimental area, showing low mean month temperature (4.5 ◦C) and several days of values below −4 ◦C on January 2017, associated with high rainfall intensity (117 mm). The zucchini and lettuce marketable yields, which benefited from the residual fertility of the ASC, were comparable with other experimental results on organic crops production [29–31].

The total energy output per hectare was higher in GM by 14% and 70% than ECO and NO ASC, respectively (Table 3). In particular, the treatments with the presence of ASC showed similar total energy output productions (about 20,000 MJ ha−<sup>1</sup> ), which were substantially different in comparison with the management without cover crops, thus indicating the positive effect of such agro-ecological practice [32]. This result was probably due to the large difference generated in the total tomato output for GM and NO ASC treatments. Conversely, the ECO treatment showed a slightly reduced energy output in comparison with GM, because of the very low productions in the cauliflower cropping cycle [33], even if the four different crops contributed to the overall production by half a hectare. According to the USDA indications [22], the coefficient of equivalent energy of the cauliflower was the highest (value equal to 1), therefore, the low production of this crop influenced the total energy output of the ECO treatment. In any case, we must take into account that the differences among treatments showed a high standard deviation, which was generated by the huge variability of the data.

#### *4.2. Environmental Sustainability Evaluation by DEXi-met*

The "overall environmental sustainability" of the cropping systems varied among the three evaluated management systems, passing from "medium-low" showed in NO ASC to the "high" in ECO. This response can be explained by the scores both of the aggregate nodal attributes (Figure 3) and the basic attributes (Figure 4), and it was a consequence of the intensification of the agro-ecological strategies adopted. In particular, the *DEXi-met* output showed that the ECO strategy was the most sustainable one, mainly due to the differences detected in the nodal attribute "biodiversity conservation". In fact, the study of Depalo et al. [34] pointed out a general positive influence of the living mulch techniques on arthropods in plant/soil systems, as shown by a high level of soil biodiversity and a lack of negative impacts on the density of canopy insects. Also, the presence of ASC in the rotation as break crops enhances the "biodiversity conservation" and, at the same time, it may have impact on occurrence of weeds, diseases, and pests [4,35]. Our results confirmed these findings, being the ECO system characterized by ASC presence both as break crops and living mulch, compared with the other two treatments (Table 1).

For the "production capacity", differently from the yield performance and the energetic output, the *DEXi-met* model considered not only the crop productions, but also the physico-chemical fertility and the systems control on pests and disease. The differences between ECO and GM were not perceivable by the model and scored "medium" value in both systems (Figure 3). Conversely, the NO ASC scored "medium-low", due to the "low" value of pests and disease control (Figure 4).

The systems with the introduction of the agro-ecological service crops (ECO and GM) scored high soil erosion control, in agreement with the study of De Benedetto et al. [36]. This result was generated by a better soil cover, in particular during the winter/heavy rainy period. Therefore, the nodal attribute "soil and water preservation" was the highest in ECO, followed by the GM system. The differences in "resources preservation" was mainly due to the different fertilizers used. In fact, the composted anaerobic digestate, which was utilized in the ECO and GM plots, is a renewable, more sustainable fertilizer than the commercial organic one, and its application did not compromise the systems production capacity, thus confirming the findings of previous studies [11,37].

Finally, the *DEXi-met* output showed substantial differences in the aggregate nodal attribute "biodiversity conservation" among the systems, as explained above. In particular, the introduction of the ASC increased the score both in ECO and GM ("high" and "medium-high", respectively), compared to the NO ASC management strategy. Similarly, other studies indicated that the presence of ASC enhances the insect and arthropods communities [38], as well as the soil microbial activities [39].

Even if the *DEXi-met* presents some aspects that should be improved, it showed some strengths and, therefore, its application gave us the possibility to analyze in detail the general structure of the overall sustainability, as well as the components and the single variables of the systems considered. We should also take into account that *DEXi-met* model is one of the new ex-post tools, which considers some attributes derived from the field experiment. However, to better understand how the model tree structure affects the results and to find the most significant variables that contributed most to the output variability, a sensitivity analysis was necessary. The sensitivity index results were affected by the level of complexity of each component and by the number of variables. Carpani et al. [15] indicated that a simpler component structure has a greater influence on the overall sustainability, whereas a higher number of variables, that individually could have no significant impact, become more sensitive if they are considered together. In our study, the *DEXi-met* produced both the "microorganism preservation" and the "macrofauna preservation" attributes as the most influential variables of our sensitivity analysis. This last result was due to the difference in the systems analyzed. In particular, it is a consequence of using agro-ecological practices, especially in the ECO system. Except for "strip cultivation with agro-ecological functions", all the other basic attributes in the nodal aggregate "biodiversity conservation" showed the SI higher than 0.2, indicating the positive influence of the systems on the environmental sustainability [40]. The basic attributes "fertilizer C/N" (within the "resources preservation" nodal attribute), the "insects and pests diseases", "weeds", and "production

quantity" (within the "production capacity"), and "soil erosion control" (within the "soil and water preservation") reached the highest values. According to Carpani et al. [15], when the SI is high, the effect of each variable on the overall sustainability is more relevant.

The detailed analysis of the *Dexi-met* model structure through the distribution of frequencies in the overall sustainability and the nodal attributes (showed by the Monte Carlo analysis) revealed that the model adequately represents the diversity of the systems evaluated. In fact, the frequencies obtained in the "overall sustainability" showed the highest value for the "medium" modality, following a normal Gaussian pattern. This behavior is due to the use of till to seven qualitative classes at the "overall sustainability" level, allowing to distinguish the different scenarios. However, the number of the qualitative classes (from "very low" to "very high") was not so large to generate unnecessary complications in the use of the model, and to reduce its ability to distinguish differences between systems. Besides, to avoid further complications, in agreement with Craheix et al. [24], the aggregate attributes were composed by five qualitative classes (from "low" to "high") and the basic attributes were composed by only three classes ("high", "medium", and "low").

#### **5. Conclusions**

Sustainability in agriculture is a complex concept and there are no common viewpoints among scientists about its dimension. Nonetheless, various parameters for measuring agricultural sustainability have been proposed, since the measure of the mere production capacity of an agro-system is not enough to evaluate it. This study clearly highlighted the relevance of considering different criteria, when we assess the advances in sustainability achievement that could be obtained introducing agro-ecological management practices and innovations. The findings also demonstrated that applying principles and practices which tend to follow the natural ecosystems can contribute to building up more complex agro-ecosystems, increasing resilience, and optimizing and maintaining biodiversity. In particular, the agro-ecological approach (ECO) both reduces the use of and dependency on external synthetic inputs by enabling to control pests, weeds, and improving fertility with ecological management. This management strategy could optimize and close resource loops (nutrients, biomass, etc.), by recycling nutrients and biomass in the farm. It may also support climate adaptation and resilience and contribute to greenhouse gas emissions mitigation, through lower use of fossil fuels and higher carbon sequestration in soils.

As revealed by our results, the introduction of the agro-ecological management practices such as ASC, use of on-farm produced fertilizers (composts), intercropping, etc., is an interesting way to improve the sustainability of the system. In any case, the results found here could not be fully generalized, since the *Dexi-met* model did not take into account some other aspects (e.g., the economic sustainability, the length of the study period, etc.). Moreover, when these strategies are applied, agronomic and productions difficulties should be kept in mind, at least in the short transition period between conventional and agro-ecological systems.

The proposed modeling approach provides a simple method of decisional support to farmers to efficiently select different crop management strategies, by assessing the environmental sustainability of the cultivation systems. An interesting topic of further research could be testing the considered agro-ecological management practices in different environmental conditions.

**Author Contributions:** Conceptualization, F.M., A.P., and M.D.; formal analysis, A.P.; data curation, A.P.; writing—original draft preparation, F.M., A.P., and M.D.; writing—review and editing, F.M., A.P., and M.D.; funding acquisition, F.M.

**Funding:** This paper is a result of the research projects RETIBIO (*Attività di supporto nel settore dell'agricoltura biologica per il mantenimento dei dispositivi sperimentali di lungo termine e il ra*ff*orzamento delle reti di relazioni esistenti a livello nazionale e internazionale*), funded by the Organic Farming Office of the Italian Ministry of Agriculture, and SOILVEG (*Improving soil conservation and resource use in organic cropping systems for vegetable production through introduction and management of Agro-ecological Service Crops (ASC)*) funded by ERA-Net CORE Organic Plus Funding Bodies partners of the European Union's FP7 research and innovation programme under the grant agreement No. 618107.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Stomatal Response of Maize (***Zea mays* **L.) to Crude Oil Contamination in Soils**

#### **Chaolan Zhang 1,2, He Huang <sup>1</sup> , Yongxin Zhou <sup>3</sup> , Haiying Lin <sup>2</sup> , Tian Xie <sup>3</sup> and Changjun Liao 3,\***


Received: 9 August 2019; Accepted: 27 September 2019; Published: 29 September 2019

**Abstract:** In this study, maize plant was cultured in soil contaminated with different levels of crude oil. The purpose was to investigate the change of soil properties, leaf physiological and chemical parameters, and phenanthrene content in the leaf. Results showed that soil water content significantly increased when the levels of total petroleum hydrocarbons were 3700–17,800 mg/kg in soil, and soil electrical conductivity significantly increased compared with the control. In maize leaf, stomatal length and density, as well as K and Na contents decreased in contaminated treatments compared with the control. Stomatal length has a significant positive correlation with K content in leaf (r = 0.92, *p* < 0.01), while stomatal density was negatively correlated to the crude oil level in soil (r = −0.91, *p* < 0.05). Accumulation of phenanthrene in maize leaf was mainly through the foliar uptake pathway. Phenanthrene concentrations of maize leaf in oil-treated soil were less than that of the control, which exhibited a significant positive relationship with stomatal length (r = 0.98, *p* < 0.01). This study demonstrated that the stomata structure of maize could be influenced by crude oil and thus possibly controlling the accumulation of polycyclic aromatic hydrocarbons in aerial tissues. Based on these results, controlling stomata movement will be beneficial to phytoremediation of contaminated soil.

**Keywords:** maize; stomata; soil; phenanthrene; remediation

#### **1. Introduction**

Petroleum oil is the main energy source and plays an important role in modern society. Soil contaminated with petroleum hydrocarbons is an increasing environmental concern as oil consumption increases dramatically around the world [1]. Among the numerous components of petroleum hydrocarbons, polycyclic aromatic hydrocarbons (PAHs) are persistent and carcinogenic in the environment, and thus threaten human health through contaminated food chain [2]. In general, compared with inhalation or skin contact, ingestion of contaminated food is the primary pathway for human to exposure to PAHs [3,4].

Oil residuals can cause some major changes in the soil chemical properties, such as decreased total nitrogen content, and increased pH value to some extent [5], which influences the growth of plants in soil. The toxic effects of crude oil contamination prevent germination from occurring and provide unsatisfactory soil conditions [6,7]. Such poor soil conditions may result from insufficient aeration caused by decreased air-filled pore space, low water content, as well as a reduction of available nutrients [8]. Therefore, it is necessary to fully understand the changed properties of petroleum hydrocarbon contaminated soils that are closely related to plant growth.

Plants adapt to environmental stress by adjusting their external morphology, internal structure, and physiological and ecological characteristics [9]. Crude oil is likely to directly affect plants after contact and provide a resulting intake of contaminants [10,11]. Plants take in PAHs mainly through soil-to-plant and air-to-plant, i.e., root uptake and atmospheric deposition from gaseous or particulate forms [12,13]. In the former pathway, PAHs can be taken in the aerial plant tissues from the root through the transpiration stream within the xylem, while in the latter one, they can be diffused into plant leaves via the cuticle or the stomata [14]. Since stomatal closure or opening is vital to the transpiration and gas exchange, it should be considered in the study of PAHs uptake by plants.

In botany, a stoma is a pore found in the epidermis of leaves, stems, and other organs controlling gas exchange between the atmosphere and plants [15]. Stomatal density and aperture (length of stomata) vary under many environmental factors such as atmospheric CO<sup>2</sup> concentration, light intensity, air temperature [16,17], potassium and sodium concentration [18], air pollution, and environmental stress [19]. For example, stomatal size obviously decreases with water deficit, and stomatal density is positively correlated with stomatal conductance, net CO<sup>2</sup> assimilation rate, and water use efficiency [20]. Thus, the stomata can adapt to local and global changes on all time scales from minutes to millennia [15].

The edible plants grown in contaminated soils are of great concern to human health for its potential risk. For instance, maize plant has been reported to be a candidate for phytoremediation of hydrocarbon-contaminated soil [21,22]. In our previous work, maize plant was also applied for phytoremediation of crude oil contaminated soil, in which several PAHs had been detected in maize plants, and phenanthrene (PHE) had the highest level [23,24]. Stomata play an important role in the air-to-plant uptake pathway of PAHs. However, the information about the changes in leaf stomata of maize plants grown in crude oil contaminated soils is scarce. Therefore, it is necessary to investigate the changes of leaf stomata of plants grown in contaminated soil and the influence factors. The main objectives of this study were (1) to understand the changes of stomata structure and the concentrations of potassium (K) and sodium (Na) in maize leaf in response to crude oil contaminated soil; (2) to evaluate the relationships between the PHE (a representative of PAHs) leaf concentration and leaf parameters/soil properties.

#### **2. Materials and Methods**

#### *2.1. Chemicals, Seed of Maize and Soil*

Crude oil without refining was obtained from Guangzhou Department, Sinopec Corporation, China. All other agents used in this study were analytical grade. Seed of CT 38 was purchased from Research Institution of Crop, Guangdong Academy of Agricultural Sciences, China.

The soil in the experiment was collected from the upper layer (0–20 cm) of an abandoned farm in Guangzhou Higher Education Mega Center, Guangzhou, China. After stones and roots were removed, the soil was air-dried, smashed, and passed through a 4 mm sieve. The organic matter content and pH of the soil were 1.3% and 6.54, respectively. Nutrient levels were 24.5 g/kg ammoniac nitrogen, 4.32 g/kg total phosphorus (P), and 0.40 g/kg total K.

#### *2.2. Experimental Design and Management*

The soil (1.5 kg) was placed in a plastic crate, spiked with different amounts of crude oil, and stirred for homogeneity with a wood spoon. The soil was then put into plastic pots and placed outdoors for 4 months to adequately evaporate the volatile fractions of crude oil. And then the total petroleum hydrocarbon (TPH) levels were measured to be 0, 2600, 3700, 6500, 17,800, and 48,800 mg/kg, respectively, using the method of our previous work [23]. Each treatment was replicated three times. Three maize seeds were placed into soils at 2 cm depth in each pot. After the maize seedlings grow out of the soils with three expanded leaves, one seedling was left in each pot. The pots were placed outdoors at the top of our laboratory building. The experiment was started in September. The average

temperature was 23.2 ◦C during the experiment. Water was added into potholders for soil moisture. After two months, soil properties and maize leaf parameters were determined.

#### *2.3. Analytical Methods*

#### 2.3.1. Soil Water Content and Soil Electrical Conductivity

To understand the water content of soil contaminated with different levels of crude oil, soil samples were collected in pots with a core sampler when water sufficiently infiltrated into the soil from a potholder. Soil water content was determined gravimetrically by weighing, after drying in an oven at 105 ◦C for 12 h according to the method described by Liu et al. [25]. Soil electrical conductivity was measured by a portable electrical conductivity meter (Hanna HI-993310D).

#### 2.3.2. Determination of Stomatal Traits

Stomatal traits in maize leaf were determined according to the method described by Zheng et al. [26]. The first leaf fully expanded on the main stem was sampled for each plant. Colorless nail polish was carefully smeared on leaf samples for about half an hour. Then the thin film was immediately covered with a cover slip and pressured lightly with a fine-point tweezer. Leaf stomatal length and density were measured from the base, middle, and tip sections on leaves of maize. Three slides were prepared for each taxon. Stomatal length was determined by micro-morphological observations carried out on 1 cm<sup>2</sup> portion per leaf (excised from similar areas) with a microscope (Carl Zeiss Micro-imaging, GER) equipped with a spot insight color camera (Diagnostic Instruments, Sterling Heights, Sterling Heights, MI, USA). Stomatal density (NO/cm<sup>2</sup> ) was calculated on 10 representative fields of leaves according to the method described by Orsini et al. [27].

The gravimetric measurement of water loss after leaf excision is a rapid method to evaluate the transpiration rate. The initial fresh weight (FW) and the weight after 5 min (W) were recorded. Water loss in 5 min was the difference between FW and W, which were used to calculate the transpiration rate [28]. The leaf water content was also determined gravimetrically by the method of soil water content mentioned above. The leaves were cut into pieces and dried in an oven at 105 ◦C until they reached a constant weight.

#### 2.3.3. Determination of K and Na Concentration in Maize Leaves

To determine the concentration of K and Na in maize leaves, samples were collected and dried, followed by digestion with HNO<sup>3</sup> and oxidation by H2O<sup>2</sup> with a heating plate. The residual was dissolved in 5% (*V*/*V*) HNO<sup>3</sup> solution, the concentrations were measured by atomic absorption spectrophotometry (AAS, Z-2000, Hitachi, Tokyo, Japan) as previously described by Cicek and Cakirlar [29].

#### 2.3.4. Determination of PHE in Maize Leaves

Concentration determination of PHE in maize leaf was conducted according to the previous method described by Tao et al. [30] with some modification. Maize samples (1.00 g) homogenized with about 1 g of anhydrous sodium sulfate were put in glass tubes. The samples were extracted with 10 mL hexane/dichloromethane (1:1) under ultrasonic conditions for 30 min. Then the extract was collected in a beaker. This process was replicated three times. The collected extract was purified by passage through a silica gel column and vacuum concentrated with a rotary evaporator at 40 ◦C. The samples were re-suspended in *n*-hexane to a final volume of 1 mL for further analysis by gas chromatography mass spectrometry (GC-MS).

Analysis of plant samples was conducted using a GC–MS with Thermo Trace GC Ultra instrument coupled to a Thermo DSQ II mass spectrometer (Thermo Electron Corporation, Waltham, MA, USA). Compounds were separated in a 30 m 0.25 mm id capillary column coated with 0.25 µm film (HP-5MS, Agilent, USA). GC temperature was programmed from an initial 80 ◦C before commencing at 10 ◦C/min

up to 290 ◦C, with a final holding time of 10 min. Helium was used as carrier gas. A 1.0 µL aliquot of the extract was injected while the injector port was held at 280 ◦C and operated in a splitless mode at a flow rate of 1.0 mL/min. The head column pressure was 30 kPa. The mass spectrometer was operated in scan mode with an electron impact ionization of 70 eV and an ion source temperature of 230 ◦C. Solvent delay was set at 4 min. Selective ion monitoring model was used. The target ions and retention time was 178 and 14.84 min for PHE, respectively [31].

#### *2.4. Statistical Analysis*

Statistical Product and Service Solutions statistic software 17.0 (SPSS company, Chicago, IL, USA) was used for the statistical evaluation of the results designed as completely randomized with three replicates of each parameter. Mean values followed by the same letter were not significantly different, as determined by an analysis of variance (ANOVA). The differences were compared by Duncan's range at a significance level of *p* < 0.05. The relationships between parameters were evaluated by Pearson correlation analysis.

#### **3. Results**

#### *3.1. Changes in Soil Properties*

The changes in soil water content and soil electrical conductivity in different treatments were shown in Figure 1. Soil water content was significantly increased when the TPH levels rose from 3700 to 17,800 mg/kg, but dramatically decreased at the extremely high level of 48,800 mg/kg, compared to the control soil (Figure 1A). At the low-level contaminated soil (2600 mg/kg), water content was similar to that of the control. The values of soil electrical conductivity in the contaminated treatments were significantly higher than that of the control (Figure 1B), but it exhibited no regular tendency.

**Figure 1.** Effect of soil contamination with crude oil on soil water content (**A**) and soil electrical conductivity (**B**). (*p* < 0.05). Different letters on top of the bar indicate they are significantly different at *p* < 0.05.

#### *3.2. Leaf Growth and Stomatal Density and Length*

Stomatal length of maize leaf in contaminated treatments significantly decreased in comparison with that in the control (Figure 2A), but there were no significant differences among 3700–48,800 mg/kg treatments. Stomatal density decreased with increasing TPH levels in soil (Figure 2B). In the highest-level contaminated soil, stomatal density was decreased by 46% compared with the control. In comparison, the downtrend showed that stomatal length was more sensitive than stomatal density to contaminated soil. 70

a

(A)

**Figure 2.** Effect of soil contamination with crude oil on stomatal length (**A**) and stomatal density (**B**) of maize leaf. (*p* < 0.05). Different letters on top of the bar indicate they are significantly different at *p* < 0.05.

#### *3.3. Water Content and Transpiration Rate*

Water content and transpiration rate are important physiological functions of plants, which may be influenced by soil conditions. As shown in Figure 3A, leaf water contents in all samples grown in contaminated soil were similar, suggesting the crude oil contaminated soil with different concentration did not have a remarkable effect on the water transport from soil to plant tissues, but did change the water content in the maize leaf. As shown in Figure 3B, transpiration rates of maize leaf in contaminated soil did not exhibit a significant difference, but were slightly higher than the control. This indicated the transpiration rate of maize plant could be affected by crude oil contaminated soil to some extent.

Concentration of total peroleum hydrocarbon in soil (mg/kg) **Figure 3.** Effect of soil contamination with crude oil on water content (**A**) transpiration rate (**B**) of maize leaf (*p* < 0.05). Different letters on top of the bar indicate they are significantly different at *p* < 0.05.

0 2600 3700 6500 17800 48800

#### *3.4. Concentrations of K and Na in Maize Leaf*

As the major mineral elements in plant tissues, K and Na play important roles in maintaining the physiological functions, especially in regulating the opening and closure of stomata. It is necessary to investigate the effect of crude oil contaminated soil on the K and Na assimilation of maize. As shown in Figure 4, both K and Na concentrations in maize leaf significantly decreased with increasing TPH levels of soils, indicating the crude oil contaminated soil had a significant effect on the assimilation of K and Na in maize plant. Additionally, K and Na concentrations in maize leaf at high levels of contaminated soil (above 6500 mg/kg) were much lower than those of the control.

5

**Figure 4.** Effect of soil contamination with crude oil on K and Na concentrations in maize leaf. (*p* < 0.05). Different letters on top of the bar indicate they are significantly different at *p* < 0.05.

#### *3.5. Phenanthrene Concentration in Maize Leaf*

As shown in Figure 5, PHE concentrations of maize leaf in contaminated treatments were lower than that in the control group, and there were significant differences when TPH levels reached 3700 mg/kg. Furthermore, in the soil treatments with TPH levels ranging from 3700 to 48,800 mg/kg, PHE concentrations in maize did not exhibit significant changes.

Concentration of total peroleum hydrocarbon in soil (mg/kg) **Figure 5.** Phenanthrene concentration of maize leaf in different treatments. (*p* < 0.05). Different letters on top of the bar indicate they are significantly different at *p* < 0.05.

#### *3.6. Relationship*

The relationships between different parameters were presented in Table 1. TPH level in soil was negatively correlated to soil water content, leaf water content, stomatal length and density, K, Na and PHE contents in maize leaf, but positively correlated to soil electrical conductivity and transpiration rate of maize leaf. Especially, significant negative correlations were observed between soil TPH level and soil water content (r = −0.82, *p* < 0.05), as well as stomatal density (r = −0.91, *p* < 0.05). In this study, correlation analysis also covered the stomata structure, ion contents, and PHE concentration in maize leaf. Stomatal length was significantly positively correlated to leaf K content (r = 0.92, *p* < 0.01). Besides, PHE concentration had a significantly positive correlation with stomatal length (r = 0.98, *p* < 0.01).


**Table 1.** Simple correlation coefficient (r) between parameters.

TPH: total petroleum hydrocarbon in soil; SW: soil water content; EC: soil electrical conductivity; LW: leaf water content; TR: transpiration rate; SL: stomatal length; SD: stomatal density; K: leaf K content; Na: leaf Na content; PHE: phenanthrene concentration in leaf. \* indicate the r values are significant at *p* < 0.05. \*\* indicate the r values are significant at *p* < 0.01.

#### **4. Discussion**

#### *4.1. Soil Properties in Crude Oil Contaminated Soil*

Soil properties play an important role in soil microorganism activity and plant growth. According to previous work, plants exposed to soils contaminated with petroleum hydrocarbons were subjected to growth limitations, due to low water uptake and reduced nutrient availability [8]. Mineral nutrient availability can be reflected by soil electrical conductivity. Therefore, soil water content and soil electrical conductivity need to be well understood in phytoremediation of soil contaminated with petroleum hydrocarbons. In this study, water contents increased in soil contaminated with certain TPH levels (3700–17,800 mg/kg) but decreased significantly when TPH levels reached 48,800 mg/kg. Water-stable aggregates in soil are related to the content of soil organic matter [32]. The addition of crude oil to soil increased the soil organic matter content, possibly resulting in enhanced water holding capacity at a certain limited range, as soil organic matter was an important determinant of the available water capacity [33]. However, the high concentration of crude oil in soil might prevent water from entering the pores of soil, which decreases water holding capacity.

Soil electrical conductivity increased in crude oil contaminated soils compared to the control soil in this study. This result was in agreement with previous work, which showed that the value of electrical conductivity in contaminated soil was higher than that of the control site [5]. Soil electrical conductivity represents soil salinity, which is mainly composed of cation ions such as Na+, K+, and Ca2+. The result of this study illustrated that availability of these ions in soil was not the limiting factors for plant uptake. The addition of crude oil in soil leads to higher soil electrical conductivity, possibly resulting from the production of metabolites from crude oil biodegradation [34].

#### *4.2. Stomata and Other Leaf Parameters*

Stomata are the pores in the epidermis of botany leaf controlling gas exchange, mainly CO<sup>2</sup> and water vapor, between the atmosphere and plants [15]. In the present work, stomatal length and density of maize leaf in contaminated soil treatments decreased compared with those in the control, indicating that crude oil contaminated soil harmed stomatal structures. And the downtrend showed stomatal length to be more sensitive than stomatal density to contaminated soil. Previous work showed stomata in plants were not only influenced by air pollution of automobile emissions [35], but also by soil or water contamination of environmental stress [19,36]. It is interesting to observe the changes in maize stomata induced by the contamination of air or soil in the present work.

Plasticity of stomatal development may be determined by many exogenous and environmental cues, of which abscisic acid (ABA) is considered as a vital regulator of environmentally determined stomatal development [37]. According to previous work, ABA can increase progressively in the root with responses to abiotic environmental stress [38,39]. In particular, the level of endogenous ABA significantly increased in pea (*Pisum sativum* L.) plant with increasing fluoranthene concentrations [40]. Therefore, crude oil contaminated soil might stimulate the synthesis of ABA in maize root and then increase the amount of ABA, thus decreased stomatal length and density.

In addition, the changed stomata might affect other leaf parameters. Stomatal length has an extremely positive correlation with leaf K content (r = 0.92, *p* < 0.01), and positive correlation with leaf Na content (r = 0.75, *p* < 0.01). It seems that K and Na content in leaf may be influenced by the stomata structure. But on the other hand, K<sup>+</sup> is considered to involve in controlling stomatal movements, in which guard cell K<sup>+</sup> uptake from the apoplast is mediated by a proton-extruding adenosine triphosphatase on the plasmalemma [41]. Moreover, tonoplast-localized NHX proteins as Na+, K+/H<sup>+</sup> antiporters are essential for active K<sup>+</sup> uptake at the tonoplast for stomatal function [42], so that K<sup>+</sup> and Na<sup>+</sup> in plant are considered as twins [43]. The contents of K and Na in maize leaf might also affect the length and density of stomata. The relation between those ions and stomata structure still needs further study. Additionally, leaf water content and transpiration rate did not significantly change in different treatments, indicating that the response of water balance in maize plant was different from nutrient ions in contaminated soil.

#### *4.3. PHE Uptake and Stomata*

Accumulation of PAHs in aerial plant tissues may be from root through transpiration stream and from diffusion via leaf stomata [14]. In the present work, transpiration rates in all plants were similar (Figure 3B). Besides, PHE concentrations of maize leaf in contaminated treatments were lower than that of the control. PHE could volatilize from contaminated soil. These results indicated PHE accumulation in maize aerial tissues might be from foliar uptake pathway which was controlled by stomata. This suggestion was also confirmed by the fact that PHE concentration in maize leaf was significantly positively correlated with stomatal length (r = 0.98, *p* < 0.01). Besides, previous studies also confirmed that foliar uptake was the dominant pathway of PHE accumulation by plant [44–46].

Since stomata play an important role in uptaking pollutants by plants grown in contaminated soils, measurements influencing stomatal movement can be applied in phytoremediation of contaminated soil for different purposes. For example, ABA can be used on maize husk for inhibition of PAH accumulation by grain due to being able to induce stomatal closure and inhibit a light-induced stomatal opening [47], when maize plant is considered for phytoremediation of PAHs-contaminated soil. Thus, safe food will be produced. In contrast, since fusicoccin can prevent dark-induced stomatal closure [48], it can be used on hyperaccumulators for extracting more pollutants in remediation of soil contaminated with heavy metals, which are taken up by plant root and transferred to aboveground tissues with a transpiration stream that closely relates to a stomatal opening. Therefore, phytoremediation of soils contaminated with organic pollutants or heavy metals can benefit from the controlling of stomatal movement. However, the data for the effect of phytohormone on maize plant was not provided in this study, which needs to be investigated in future work.

#### **5. Conclusions**

Stomatal response and the change of related parameters of the maize plant (*Zea mays* L.) to crude oil contaminated soil were investigated in this study. Soil water content and electrical conductivity increased to a certain extent in contaminated soil, whereas the TPH level exhibited a negative relationship with soil water content (r = −0.82, *p* < 0.05). Stomatal length and density, leaf K, and Na contents decreased in contaminated soil compared with that of the control group. Stomatal length is positively correlated to leaf K content (r = 0.92, *p* < 0.01), while stomatal density is negatively correlated to soil TPH level (r = −0.91, *p* < 0.05). Moreover, it is found that the accumulation of PAHs in maize mainly occurred through the foliar uptake pathway. And PHE concentration exhibits a significantly positive relationship with stomatal length (r = 0.98, *p* < 0.01). Based on this study, measurements should be applied to control stoma closure or opening for different purposes in phytoremediation of contaminated soils.

**Author Contributions:** C.L. and C.Z. conceived and designed the experiments; C.Z. and H.H. performed the experiments; H.H., Y.Z. and T.X. analyzed the data; C.L. and C.Z. written original draft; H.L. and H.H. reviewed and edited the draft.

**Funding:** The research was financially supported by the Program for the National Key R&D Program of China (2018YFD0800700, 2017YFD0801300).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Technical Note*
