**Study on Water Saving Potential and Net Profit of** *Zea mays* **L.: The Role of Surface Mulching with Micro-Spray Irrigation**

### **Zhaoquan He 1,2,\*, Xue Shang <sup>1</sup> and Tonghui Zhang 2,3**


Received: 27 November 2019; Accepted: 3 January 2020; Published: 5 January 2020

**Abstract:** Water shortage threatens agricultural sustainability in Horqin Sandy Land, northeast China. To explore the effects of various surface mulching patterns with micro-spray irrigation on the yield, water consumption (ETc), and water-saving potential of maize (*Zea mays* L.), we used three treatments: straw mulching (JG), organic fertilizer mulching (NF), and no mulching (WG; control). In each treatment, plants were supplied with 500 mm of total water (irrigation plus precipitation) during the entire growing season and were irrigated with the amount of total water supply minus precipitation. Yield and water use efficiency (WUE) showed a significant negative correlation with water saving potential per unit yield (Py) and water saving potential per unit area (Sp), which were also consistent with their relationships in the function model. Meanwhile, a remarkably positive correlation occurred between yield, WUE, and net economic profit, respectively. The JG treatment, which was mainly affected by light and temperature production potential (Yc), grain yield, and ETc, showed the lowest Py (0.16 m3 kg−1) and Sp (2572.31 m<sup>3</sup> hm−2), and the maximum increase in yield, WUE, and net economic profit, extending to 16,178.40 kg hm−2, 3.25 kg m−3, 17,610.09 yuan hm−2, respectively, which were significantly higher than those in NF and WG, (*p* <0.05). Thus, straw mulching with micro-spray irrigation was the best treatment for maximizing yield and WUE. Organic manure mulching and no mulching need further investigation, as these showed high Py and Sp, which were together responsible for lower WUE.

**Keywords:** water resources; surface mulching; water saving potential; micro-spray irrigation; economic profit

### **1. Introduction**

Maize (*Zea mays* L.) is the second most important grain crop in China. Inner Mongolia is one of the major maize production regions of China that supports the livelihood of farmers in the region [1]. Maize requires a substantial amount of water during the growth phase but 80% of Inner Mongolia is arid and semi-arid, and water shortage in these areas continues to worsen with global climate change [2]. In addition, increase in the cultivable area is remarkably slow; therefore, the crop production area is unlikely to expand in the region so increasing yield per unit area is the only solution to continually support maize-production dependent households [3]. The improvement of maize yield per unit area primarily depends on the extension and reform of agricultural water saving planting technologies [4]. On the one hand, surface mulching is commonly used because of its notable positive impact on water conservation and yield. There are reports that mulching conserves

soil moisture by reducing evaporation and saving 12–84% of the irrigation water [5]. Common surface mulching methods include plastic mulching, straw mulching, organic manure mulching, and biochar mulching [6–8]. Effective use of crop straw and animal manure is conducive to the intensive management of agricultural resources, which is of practical significance for improving crop cropping systems, developing sustainable biodiversity of agro-ecosystems, and implementing national poverty reduction policies [9]. In addition, crop straw mulching and fully rotten organic manure mulching can improve the topsoil physics and chemistry nature of soil remarkably, without causing environmental pollution [10]. On the other hand, efficient irrigation contributes to increases in yields of crops and in income for the local farmers, providing evidence of the significance of irrigation in the past and for future poverty alleviation in China [11]. Therefore, combining mulching with advance irrigation method (drip/micro irrigation) increase more significantly crop productivity and water use efficiency (WUE), reduce water consumption (ETc) in a region where water shortage is the major factor limiting agricultural sustainability [12–14].

The potential and configuration of climate resources (light, heat, and water) affect their utilization and ultimately limit the sustainable development of agricultural production [15]. Therefore, investigation of the characteristics of crop climate production potential is essential. The production potential of light and temperature is the main focus of studies in China. Water saving potential refers to the amount of irrigation water that can be saved per unit scale of crops, involving four scales, that is, crop, field, irrigation area, and regional/basin, respectively, and is a major factor affecting agricultural structure [16,17]. The definition of agricultural water saving potential is mainly based on water efficient methods, various water saving techniques, and management [16,18]. Previously, many approaches have been adopted to measure water saving potential, including efficient irrigation technology and method, irrigation scheduling improvement [19]. For example, some studies used the water balance method and remote sensing technology to calculate water consumption for obtaining the theoretical water saving potential [20,21], while many studies utilized crop models, such as DASSAT and WOFOST [22], to simulate the water consumption of crops and water saving ability, generating favorable results. Additionally, economic profit evaluation of agricultural technology has been largely explored in maize, confirming that the water saving irrigation method is significantly better than traditional irrigation technology [23], and no-till and permanently fixed ridge is better than conventional tillage [24].

Although responses of yield and water consumption of maize to surface mulching have been explored extensively, as mentioned above, research on the water saving potential per unit yield (Py) and water saving potential per unit area (Sp) of maize (two angles of water-saving potential calculation, that is, crop scale and area scale, respectively) with surface mulching and micro-spray irrigation technology, and on the primary factors affecting the water saving potential of maize, is lacking. In this study, we used the meteorological observation data of the Naiman desertification station to achieve the following two main objectives: (1) to adjust the crop planting structure and farmland irrigation system through the improvement of the efficiency of irrigated agriculture by determining the surface mulching patterns with the greatest water-saving ability; and (2) to establish the optimal crop planting pattern with high net economic benefits and prominent water saving ability to provide theoretical support for the evaluation measures of economic benefits of water-saving planting patterns of farmland crops, thus expanding the appropriate ecological planting scale in this region.

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

### *2.1. Experimental Site*

The experiment was carried out at the Naiman Desertification Research Station of the Chinese Academy of Sciences (42◦58 N, 120◦43 E; 360 m a.s.l.) in April–September 2017, located in the eastern part of Inner Mongolia Autonomous Region, China (Figure 1). Naiman, located in the southwest of Horqin Sandy Land [25], has a semiarid continental monsoon climate. The experimental site, with sandy soil texture sensitive to wind erosion, has a mean annual precipitation of approximately 360 mm (265 mm in this growing season), an annual mean evaporation of around 1950 mm, and an annual mean temperature of 6.40 ◦C, of which the minimum monthly average temperature of −13.50 ◦C occurred in January and the maximum of 23.80 ◦C in July. In the soil depth of 0–60 cm, soil organic carbon content, pH (1:2.5 water), and electrical conductivity (1:5 water) at 0–30 cm depth before planting are 2.48 g kg<sup>−</sup>1, 9.23, 62.73 μS cm<sup>−</sup>1, respectively. Field capacity was 12.77%, wilting point was 5.40%, water saturation was 30.24%, saturated hydraulic conductivity was 0.93 mm min<sup>−</sup>1, as well as bulk density being 1.55 g cm<sup>−</sup>3.

**Figure 1.** Location of Naiman desertification research station in China. Notes: the blue region represents Horqin Sandy Land; the yellow region represents the Naiman banner; the red flag represents Naiman desertification research station, CAS.

### *2.2. Test Design*

The trial was laid out in a completely randomized plot design. A total of 51 plots (2 m × 2 m), representing 3 treatments and 17 replicates, were created. The *Zea mays.* L. cv. Jingke 958 variety was chosen as the tested cultivar for all the treatments and was planted by a tube-shaped seeder on 27 April at a depth of 5 cm. The hole of the seeds was 2.5 cm in the direction of the row and was 3.0 cm in the direction of the column, which was reserved for the space of maize growth. The planting density of maize was 60,000 plants hm−2. Each plot contained 4 rows, with 6 maize plants per row, that is, 4 rows of 6 columns, at a plant-to-plant spacing of 20 cm and row-to-row spacing of 36 cm (Figure 2). The spacing of 0.5 m between each plot was provided for maintaining independence among treatments, and a buffer channel of 1 m width was provided in the neighborhood of experimental fields to avoid edge effects. The experimental field was oriented west to east. The field was rectangular in shape, with 9 m in the north-south direction and 44 m in the east-west direction. Three treatments were established: straw mulching (JG), organic manure mulching (NF) and no mulching (WG; control).

In the JG treatment, when the maize seedlings reached a height of 20 cm, crushed maize straw was evenly applied to 100% of the soil surface with 4000 kg hm−<sup>2</sup> (1.6 kg per plot). In the NF treatment, cattle and poultry manure were the sources of organic manure. They were collected from the cattle and poultry farms located in Naiman banner and were uniformly mixed with soil 15 days before applying, making it fully rotten, preventing environmental pollution. Because organic manure that is not fully rotten contains various bacteria, this would cause a high incidence of crop diseases and insect pests, and affect the ecological environment when applied directly to farmland. Physical and chemical properties of the organic manure were, 145.77 g kg−<sup>1</sup> of soil organic carbon, 260.49 g kg−<sup>1</sup> of soil organic matter, 10.25 g kg−<sup>1</sup> of total nitrogen, 8.64 g kg−<sup>1</sup> of total phosphorous, and 11.57 g kg−<sup>1</sup> of total potassium. When the maize seedlings reached a height of 20 cm, organic manure was evenly distributed on 100% of the soil surface with 30,000 kg hm−<sup>2</sup> (12 kg per plot). In the WG (control) treatment, no surface mulching was used during the entire growth period of the maize.

**Figure 2.** Sketch of experimental site. (**a**) represents the front view; (**b**) represents the lateral view.

### *2.3. Irrigation Scheme*

In each treatment, zonal micro-spray irrigation was used. Maize plants were supplied with 500 mm of water, including irrigation and precipitation, during the entire growth season. The irrigation amount was the total water supply minus precipitation. The maize growth season was divided into five stages: seeding, jointing, heading, filling, and ripening. According to the water requirements of maize at each growth stage, 15%, 35%, 22%, and 28% of the total water supply was applied to the seedling-jointing, jointing-heading, heading-filling, and filling-ripening stage, respectively [26]. The irrigation level was the same across all treatments. In each growth phase, if the precipitation exceeded the upper limit of the designed water supply, the water increment needed to be subtracted from the next irrigation, ensuring the same total water supply for all treatments throughout the entire growing season. Maize was irrigated on days with no or low wind (<1.5 m s−1) to achieve uniform irrigation. Irrigation regimes are summarized in Table 1. Spray lines (42 m) were installed in the middle of the plot along the east–west direction, with a nozzle spacing of 50 cm and discharge rate of 1Lh<sup>−</sup>1. Irrigation groundwater was measured continuously using flowmeters.


**Table 1.** The irrigation regimes across the growing season (mm, April–September 2017).

Note: N represents no irrigation; + represents irrigation increment and needs to be subtracted from the next irrigation; − represents irrigation loss, and needs to be added at the next irrigation. Date between precipitation and irrigation is continuous because the precipitation during the days of irrigation is automatically counted as the amount of precipitation in the interval between that irrigation and the next irrigation.

### *2.4. Field Management*

The field was tilled approximately 1 week before sowing. At the time of tilling, a basal dose of fertilizer was evenly and equably distributed in the topsoil at a rate of 375 kg ha−<sup>1</sup> (1.35 kg per plot) of diammonium phosphate (N-P2O5-K2O, 18-46-0) based on the N and P requirements; the fertilizer was applied in spade slits to avoid loss over the soil surface and sprinkled near the maize roots to ensure full absorption by the crops. No pesticides and insecticides were used during the whole growth period of maize to prevent the test results from being affected.

### *2.5. Climate Data*

Slight fluctuations in temperature occurred among the different growth stages. The mean temperature was 24.60 ◦C (Figure 3). Relative humidity was significantly higher in July and August, reaching a maximum of 100%, compared with other months (Figure 3). It was found that, from the calculation of the food and agriculture organization (FAO) Penman–Monteith Equation (2), reference crop evapotranspiration (ETo) and precipitation were 501 and 261 mm, respectively (Figure 4). The average ETo was 3.61 mm d<sup>−</sup>1, with remarkable seasonal variation; ETo increased from April to July and then declined significantly after August, along with the decrease in solar radiation intensity and temperature. In August, lower ETo was positively related to lower temperature and higher relative humidity.

**Figure 3.** Variation of the temperature and relative humidity of the experimental site during the growth period of maize. Note: Tmax represents the maximum of temperature; Tmin represents the minimum of temperature; RHmax represents the maximum of relative humidity; RHmin represents the minimum of relative humidity.

**Figure 4.** Variation of the reference crop evapotranspiration and precipitation during the growth period of maize. Note: ETo represents reference crop evapotranspiration, mm. ETo was calculated by means of the FAO Penman–Monteith Equation (2) for the entire growth season of maize.

### *2.6. Measurement of Indicators*

Soil water content of 0–20 cm, 20–40 cm, 40–60 cm, 60–80 cm, and 80–100 cm in each growth stage was determined by the gravimetric method.

Soil temperature in the soil layer of 0–20 cm was measured by a thermometer during each growth period of maize.

Maize of each plot was harvested at the ripening stage, and 6 ears that were growing well were selected randomly in each plot. Drying the grain to constant weight at 85 ◦C, weighing by an electric balance for grain yield, the grain yields were then converted to a standard grain water content of 15.50% wet basis [27].

*ETc* was calculated daily during the growing season by the soil water balance Equation (1) [28]:

$$ET\_c = I + P + C\_r - D\_w - R\_f \pm \Delta s \dots \tag{1}$$

where *ETc* was the total amount of actual evapotranspiration for the entire season (mm), *I* was the amount of irrigation water applied (mm), *P* was the precipitation (mm), *Cr* was the capillary rise (mm), *Dw* was the amount of drainage water (mm), *Rf* was the amount of runoff (mm), and *s* was the change in the soil moisture content (mm). The soil moisture content measurement was used by the conventional oven-dry method in soil layers (0–20 cm, 20–40 cm, 40–60 cm, 60–80 cm, and 80–100 cm). No runoff was observed during the trials. Capillary rise was considered as negligible due to the deep water table level. Drainage water included precipitation under the effective rooting depth, according to the soil water content measurements in the soil layer at the effective rooting depth, was determined.

*ET*<sup>0</sup> was calculated per day during the growing season by using the FAO Penman–Monteith equation. The FAO Penman–Monteith equation is given by (2):

$$ET\_0 = \frac{0.408\Delta (R\_\text{fl} - G) + \gamma \frac{900}{T + 273} \mu\_2 (\varepsilon\_\text{s} - \varepsilon\_\text{t})}{\Delta + \gamma (1 + 0.34\mu\_2)} \tag{2}$$

where *ET*<sup>0</sup> was the reference evapotranspiration (mm day<sup>−</sup>1), *Rn* was net radiation at the crop surface (MJ m−<sup>2</sup> day<sup>−</sup>1), *G* was soil heat flux density (MJ m−<sup>2</sup> day<sup>−</sup>1), *T* was mean daily air temperature at 2 m height (◦C), μ<sup>2</sup> was wind speed at 2 m height (m s<sup>−</sup>1), *es* was the saturation vapor pressure (kPa), *ea* was the actual vapor pressure (kPa), *es - ea* was the saturation vapor pressure deficit (kPa), γ was the slope of the saturation vapor pressure curve (kPa ◦C<sup>−</sup>1), and Δ was the psychrometric constant (kPa ◦C<sup>−</sup>1). Meteorological parameters needed to calculate *ET*<sup>0</sup> were derived from a local meteorological station.

Water use efficiency (kg m<sup>−</sup>3) was calculated as (3) [29]:

$$\mathcal{W}UE = \Upsilon / ET\_{\varepsilon} \dots \tag{3}$$

where *WUE* was the water use efficiency (kg m<sup>−</sup>3), *Y* was the grain yield (kg hm<sup>−</sup>2), *ETc* was the total amount of actual evapotranspiration for the entire season (mm).

Net economic profit (yuan hm<sup>−</sup>2) was calculated as (4):

$$\text{Net profit} = \text{total revenue} - \text{total cost} \tag{4}$$

where, total revenue = grain yields × average local price. The average local price for maize was 1.70 yuan kg<sup>−</sup>1. The total cost included the cost of seed, fertilizers, sows, micro-spray irrigation belts, water pipes, maize straw, and organic fertilizers during the trial.

Light and temperature production potential (kg hm−2) was calculated by the photosynthetic production potential multiplied by the revised function of temperature effect, and the expression is given by (5):

$$Y\_{\mathcal{C}} = Y\_p \cdot f(T) \dots \tag{5}$$

*Yc* was the light and temperature production potential, kg hm<sup>−</sup>2;

*Yp* was the photosynthetic production potential, kg hm<sup>−</sup>2;

*f*(*T*) was the revised function of the temperature effect.

Water saving potential per unit yield (m3 kg<sup>−</sup>1) was calculated as (6):

$$P\_{\mathcal{Y}} = \frac{1}{\mathcal{W}UE\_{\mathfrak{a}}} - \frac{1}{\mathcal{W}UE\_{\mathfrak{t}}} \quad \dots \tag{6}$$

*Py* was water saving potential per unit yield, m<sup>3</sup> kg<sup>−</sup>1;

*WUEa* was actual crop water productivity, kg m<sup>−</sup>3;

*WUEt* was the theoretical crop water productivity, kg m<sup>−</sup>3.

Water saving potential per unit area (m3 hm<sup>−</sup>2), was calculated as (7):

$$S\_p = P\_p \times \mathcal{W} \dots \tag{7}$$

*Sp* was the water saving potential per unit area, m<sup>3</sup> hm<sup>−</sup>2;

*Pp* was the proportion of crop water saving potential, %;

*W* was the irrigation quota, m<sup>3</sup> hm<sup>−</sup>2.

### *2.7. Data Collations and Statistical Methods*

Effects of various surface mulching patterns on soil water content, soil temperature, yield, water consumption, water saving potential, and net profit of maize were plotted using Origin 8.0. Variance, correlation and stepwise regression analyses were performed using SPSS 20.0, while significant differences were detected using the least significant difference (LSD) test. The cost and net profit of various surface mulching patterns were compared using the quota comparison method. Tables were created in Excel 2010.

### **3. Results**

### *3.1. Changes in the Soil Temperature and the Soil Water Content*

Under micro-spray irrigation, variation in soil temperature and soil water content during the entire growth period of maize was not affected by the different surface mulching patterns. Soil temperature increased from the seedling stage to the heading stage, which remained relatively stable until the filling stage, and then declined. In each treatment, the soil water content was significantly higher at heading and maturity than at other growth stages (Figure 5).

**Figure 5.** Soil temperature (soil layer of 0–20 cm) and soil water content (soil layer of 0–100 cm) of maize at different growth stages. Note: JG represents straw mulching; NF represents organic fertilizer mulching; WG represents no mulching; data represents mean ± standard error (*n* = 17). Bars labeled with different letters (lowercase) differed significantly among the treatments (*p* <0.05).

The temperature of the 0–20 cm soil layer was the highest from the seedling stage to the heading stage, and the rate of increase of soil temperature reached 42.99%, 36.71%, and 37.55% in the JG, NF, and WG treatments, respectively (Figure 5). Then, the soil temperature decreased from the heading to the ripening stage by 28.29%, 26.51%, and 28.24% in the JG, NF, and WG treatments, respectively. Overall, during the entire growth period of maize, the mean soil temperature (0–20 cm layer) in different treatments was in the order: NF > WG > JG.

Soil water content at each growth stage was lower in JG and WG treatments than in the NF treatment (Figure 5); thus, the soil water content was directly related to the variation in soil temperature in different mulching treatments. However, there was no significant difference between JG, and WG, respectively, for the mean soil water content. In all three treatments, the soil water content was the highest at the heading stage, which was directly responsible for the higher irrigation proportion (35% of the total water supply). Compared with the seedling stage, the heading stage showed an increase in soil water content of 18.70%, 91.05%, and 8.33% in the JG, NF, and WG treatments, respectively.

### *3.2. Water Saving Potential (Per Unit Yield and Per Unit Area)*

Values of Py and Sp of maize in the JG treatment were significantly lower than those in the NF and WG treatments. Compared with the JG treatment, values of Py and Sp were significantly higher by 31.73% and 34.51%, respectively, in the NF treatment and by 20.21% and 20.08%, respectively in the WG treatment (*p* <0.05); however, no significant differences were detected in Py and Sp between the

NF and WG treatments. Thus, under micro-spray irrigation, Py and Sp of maize in various treatments were in the order: NF and WG >JG (Figure 6).

**Figure 6.** Water saving potential per unit yield and per unit area of maize under various surface mulching patterns. Note: JG represents straw mulching; NF represents organic fertilizer mulching; WG represents no mulching; Py represents water saving potential per unit yield, m3 kg<sup>−</sup>1; Sp represents water saving potential per unit area, m<sup>3</sup> hm<sup>−</sup>2. Bars labeled with different letters (lowercase) differed significantly among the treatments (*p* <0.05).

The *P* regression model affecting the water-saving potential per unit yield and per unit area was 0.00, R2 was up to 0.92, 1.00, respectively, with high fitting degree (Table 2). This elucidated that under various surface mulching patterns, the index that affected Py was yield (R2 = 0.92), and indexes that affected Sp were light and temperature production potential (Yc), yield and water consumption (ETc).


**Table 2.** Stepwise regression coefficient of indexes affecting water saving potential of maize.

Notes: Py represents water saving potential per unit yield, m3 kg<sup>−</sup>1; Sp represents water saving potential per unit area, m<sup>3</sup> hm<sup>−</sup>2; Yc represents light and temperature production potential of crop, kg hm<sup>−</sup>2; Y represents grain yield, kg hm−2; ETc represents water consumption, mm. t represents significance test values of regression parameters. *P* represents significant value.

### *3.3. Light and Temperature Production Potential, Yield and Water Consumption*

The JG treatment showed the lowest ETc and the highest yield and WUE (Figure 7); however, no significant differences were detected in these parameters between the NF and WG treatments. The value of ETc was essentially the same in JG, NF, and WG treatments. Values of Yc in the NF and WG treatments were 16.18% and 13.32% higher than those in the JG treatment, respectively (*p* <0.05). However, maize yield in the JG treatment was 9.53% and 12.10% higher than that in the NF and WG treatments (*p* <0.05). In the current study, ETc was approximately 500 mm in the JG, NF, and WG treatments during the entire growth season; no significant differences were detected among them, probably because of the thickness of maize straw in the JG treatment and organic fertilizer in the NF treatment, which requires further investigation. Moreover, this difference was closely related to differences in the region, maize variety, and irrigation approach.

**Figure 7.** Light and temperature production potential, yield, water use efficiency and water consumption of maize under various mulching patterns. Note: JG represents straw mulching; NF represents organic fertilizer mulching; WG represents no mulching. Yc represents light and temperature production potential; ETc represents water consumption of maize; WUE represents water use efficiency. Bars labeled with different letters (lowercase) differed significantly among the treatments (*p* <0.05).

In the JG, NF, and WG treatments, correlation coefficients between water saving potential (Py and Sp) of maize and indexes affecting water saving potential were greater than 0.80 (Table 3), indicating a high correlation. Values of Py and Sp changed significantly with yield (R >0.90).


**Table 3.** Correlations between indexes affecting water saving potential.

Notes: \* represents the significant difference at the level of 0.05 (bilateral); \*\* represents the significant difference at the level of 0.01 (bilateral).

### *3.4. Economic Profits*

Analysis of the economic profits of various treatments showed that the WG treatment with micro-spray irrigation was the least expensive (9575.33 yuan hm<sup>−</sup>2; Figure 8) because this treatment had no associated cost of surface mulching material. However, the WG treatment showed no significant difference compared with the JG and NF treatments because of the higher labor cost associated with no mulching in the WG treatment. Net profit was the highest in the JG treatment, which was significantly higher by 18.26% and 17.71% than those in the NF and WG treatments, respectively (*p* <0.05). The profit:cost ratio was 1.78 in the JG treatment, which was 22.14% and 13.93% higher than that in the NF and WG treatments, respectively. According to the regression fitting analysis, the net profit of maize showed a significantly negative linear correlation with Py and Sp, with the coefficient of determination (R2) of 0.898 and 0.989, respectively (Figure 9). The higher the water saving potential, the smaller the net economic profit of maize; this explained why the net economic profit of maize in the JG treatment was higher than that in the NF and WG treatments.

**Figure 8.** Cost and profit of maize under various mulching planting patterns. Note: JG represents straw mulching; NF represents organic fertilizer mulching; WG represents no mulching. The average local price for maize was 1.70 yuan kg<sup>−</sup>1. The total cost included the cost of seed, fertilizers, sows, micro-spray irrigation belts, water pipes, maize straw, organic fertilizers during the trial. Labor costs included the layout of the sample plot, weeding, sampling, irrigation, spreading fertilizer, and determination of sample. Bars labeled with different letters (lowercase) differed remarkably among different indexes (*p* <0.05).

**Figure 9.** Relationships between water saving potential and net economic profit of maize. Note: N represents net economic profit, yuan hm<sup>−</sup>2; Py represents water saving potential per yield, m3 kg<sup>−</sup>1; Sp represents water saving potential per unit area, m3 hm<sup>−</sup>2. \* represents regression effect was significant; \*\* represents regression effect was remarkably significant.

### **4. Discussion**

### *4.1. Water Saving Potential*

Currently, climate change is one of the major concerns of the environmental problem facing mankind, and agriculture is highly sensitive to climate change [30]. Temporal and spatial distribution patterns of climate resources, such as water and heat, are directly affected by meteorological factors, including solar radiation, temperature, and precipitation. Fluctuations in temperature and precipitation during the growing season affect crop productivity, ultimately affecting regional agricultural production [31]. Therefore, spatiotemporal distribution characteristics of the crop climate production potential represent the basis of comprehensive food production potential and provide a crucial theoretical basis for agricultural productivity planning and agricultural structural adjustment [32]. Water saving potential is a vital evaluation index for adjusting agricultural structure. In this study, soil water content at each growth stage was lower in JG and WG treatments than in the NF treatment (Figure 5). However, there was no significant difference between JG, and WG, respectively, for the mean soil water content. This resulted mainly from the type and rate of the irrigation and the mulches. Because the soil moisture strongly depended on the precipitation and irrigation pattern (see irrigation scheme), the precipitation, irrigation pattern, and irrigation amount applying to all three treatments were the same. In addition, Zhao et al. [33] found that, compared with deep tillage with no mulch, mean soil water content of sunflowers was only higher by 5.75%, 2.50%, respectively, in 2011 and 2013, when straw mulch was used at a rate of 12,000 kg hm−2, which was significantly higher than that of our study (4000 kg hm−2). Teame et al. [34] indicated that, by exploring the efficacy of organic mulching, sorghum straw mulching and rice straw mulching with a rate of 10,000 kg hm−<sup>2</sup> increased mean water soil content of sesame by 33.29%, 42.05%, respectively, compared to no mulch. The efficiency of increasing soil water was more significant than that in Zhao et al. [33], which would be due to the difference in the crops and the mulches. Therefore, it was suggested that yield, Py, and Sp were not affected significantly by the water soil content, based on the results of the stepwise regression analysis (Table 2) and the insignificant difference among the three treatments for water soil content (Figure 5). The water saving potential was mainly affected by Yc, grain yield and ETc, and showed a positive correlation with Yc and negative correlation with grain yield and ETc. Therefore, both Py and Sp were the lowest in the JG treatment; however, the WUE and water saving capacity of the JG treatment were significantly higher than that of the NF and WG treatments.

Photosynthetic production potential represents the maximum crop yield achieved only under light conditions. Yc was the maximum yield of crop subjected to both light and temperature constraints [35]. Regions that experienced a rapid decrease in the climate production potential of maize over the last 30 years show a dramatic increase in Yc, resulting from a dry climate [32]. Radiation and temperature are the most critical factors affecting Yc; these factors decreased by −12.70% and −6.10%, respectively, when solar radiation of maize decreased by 10% or temperature increased by 1 ◦C during the growing season [35]. Under micro-spray irrigation, maize yield was the highest in the JG treatment, in which Yc was considerably lower than that in the NF and WG treatments, although no significant differences were detected in ETc among the three treatments. The photosynthetic production potential of the yield has been the main focus of research in northeast China [36]. The decrease in solar radiation is primarily responsible for the decrease in the photosynthetic production potential of maize. Spatial variation characteristics of the photosynthetic production potential of maize were similar to those of the surface solar radiation, both of which showed a decreasing trend from the southwest to the northeast [37]. Licker et al. systematically analyzed global maize yields and concluded that the maize yield would increase by 50%, if 95% of the maize cultivation areas worldwide met the climate potential. However, the difference between climate production potential and actual production was approximately 20% because of improvements in traditional farming patterns, economic costs, and technological measures, and the yield potential of maize was mainly affected by unreasonable farmland management and low technical levels. Therefore, conclusions explored in our study on the water saving potential of maize using different planting patterns and micro-spray irrigation are extremely conducive to the rational planning of the layout of planting areas, effectively improving maize yield and ensuring food security.

### *4.2. Yield and Water Consumption*

Mulching has been adopted in numerous parts of the world as an approach to increase crop productivity [38]. Maize yield in the JG treatment was 9.53% and 12.10% higher than that in the NF and WG treatments (*p* <0.05), and was significantly higher than that reported by Sharma et al. [39] and similar to that reported by many scholars. Li and colleagues [40] showed that plastic mulching dramatically increases crop yield. Yin et al. [41] found that compared to conventional tillage without straw residue, integrating no tillage with two-year plastic and straw mulching improved grain yields by 13.8%, reduced soil evaporation by 9.0%, and reduced soil evaporation by 9.0%. However, in another study, straw mulching did not significantly affect yield under limiting soil water content [42]. In addition, ETc was approximately 500 mm in the JG, NF, and WG treatments during the entire growth season; no significant differences were detected among them, probably because of the thickness of maize straw in the JG treatment and organic fertilizer in the NF treatment, which requires further investigation. Our results of ETc were in contrast to those of previous studies. For example, Yin et al. [43] reported that plastic film together with straw mulching decreased total evapotranspiration by an average of 4.60% (*p* <0.05) compared with no mulching; Brar et al. [44] suggested that straw mulching resulted in 19.00% higher yield compared with no mulching, resulting in 36.20 mm higher transpiration and 44.20 mm lower soil evaporation; Sun et al. [45] confirmed that crop water consumption was reduced by 32 mm under straw mulching compared with no mulching, with no significant differences in WUE. Zhou et al. [46] indicated that compared with no mulching, straw mulching increased maize yield by 10.6% and 12.5% under a drip irrigation system in 2016 and 2017, respectively, and achieved 6.1% lower water consumption. The contradiction was mostly responsible for the enhanced maize straw mulching amount in their study, which was significantly higher than that of our study. Moreover, this difference was closely related to differences in the region, maize variety, and irrigation approach.

### *4.3. Economic Profits*

Economic profits include the rate of input application and the rate of consumptive use in irrigation and fertilizer [47]. Net profit was the highest in the JG treatment, which was significantly higher by 18.26% and 17.71% than those in the NF and WG treatments, respectively (*p* <0.05). Similar net profit has been reported in rice using straw mulching in water saving production systems [48]. In this study, the profit: cost ratio was 1.78 in the JG treatment, which was 22.14% and 13.93% higher than that in the NF and WG treatments, respectively. This was consistent with the results of Sharma et al. [39]; the authors showed that the profit: cost ratio was the highest (0.62) with straw mulching, although this ratio was markedly lower than that obtained in our study. In other studies, drip irrigation resulted in net economic profits of 4359.58–6240.19 yuan hm−2, which were higher than those obtained by furrow irrigation [44]. Small amounts of maize cob biochar would also attain higher net profit through increased yields [49]. Sweet maize and green beans grown in rotation resulted in a greater increase in net profits compared with potato monoculture [50]. Therefore, straw mulching with micro-spray irrigation elevated furthest economic profits of maize, compared to organic manure mulching and no mulching.

### **5. Conclusions**

Under micro-spray irrigation, maize yield and WUE were the highest in the JG treatment with 16,178.40 kg hm−2, 3.25 kg m−3, respectively, in which Yc was significantly lower than that in the NF and WG treatments. The three treatments showed no significant differences in ETc. The water saving potential (including Py and Sp) of maize was positively affected by Yc and negatively affected by grain yield and WUE. Therefore, values of Py and Sp were the lowest in the JG treatment, were just 0.16 m<sup>3</sup> kg−<sup>1</sup> and 2572.31 m3 hm<sup>−</sup>2, respectively, but no significant differences were found, compared to NF and WG treatments. The net economic profit of maize was negatively correlated with the water saving potential in all treatments, which was primarily responsible for the negative relationship between water saving potential and yield, WUE, respectively. So, the maximum of net economic profit appeared in the JG treatment, was up to 17,610.09 yuan hm<sup>−</sup>2, and was higher than that in the NF and WG treatments (*p* <0.05).

The yield, WUE, and water saving capacity of the JG treatment were significantly higher than that of the NF and WG treatments. This suggests that straw mulching with micro-spray irrigation should be applied in local appropriate farmland. Given the lower WUE and higher water saving potential of the NF and WG treatments, it is important to explore these planting patterns further.

**Author Contributions:** Z.H. designed, performed the experiments, and wrote the original draft; X.S. analyzed the data; T.Z. reviewed and edited the draft. All authors read the final manuscript and approved the submission. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the National Natural Science Foundation of China (Grant No. 41371053, 30972422, 51669034, 51809224), National key research and development project of China (Grant No. 2017YFC0506706, 2017YFC0504704) and Science research launch project of PhD (205040305).

**Acknowledgments:** We are grateful to all the members of Naiman Desertification Research Station, Chinese Academy of Sciences, for their help in field work. We are also grateful to other anonymous reviewers for their valuable comments on the manuscript.

**Conflicts of Interest:** The authors declare no competing interests.

### **Abbreviations**


### **References**


© 2020 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* **E**ff**ect of Salinity on the Gut Microbiome of Pike Fry (***Esox lucius***)**

### **Tomasz Dulski 1,\*, Roman Kujawa 2, Martyna Godzieba <sup>1</sup> and Slawomir Ciesielski <sup>1</sup>**


Received: 15 February 2020; Accepted: 1 April 2020; Published: 5 April 2020

**Abstract:** The increasing popularity of pike in angling and fish farming has created a need to increase pike production. However, intensive pike farming is subject to limitations due to diseases and pathogens. Sodium chloride (NaCl) could be a good alternative to chemotherapeutics, especially for protecting the fish against pathogens and parasites at early life stages. However, the impact of high salinity on the symbiotic bacteria inhabiting freshwater fish is still unclear. Therefore, our objective was to analyze the gut microbiome to find possible changes caused by salinity. In this study, the influence of 3‰ and 7‰ salinity on pike fry was investigated. High-throughput 16S rRNA gene amplicon sequencing was used to profile the gut microbiome of the fish. It was found that salinity had a statistically significant influence on pike fry mortality. Mortality was highest in the 7‰ salinity group and lowest in the 3‰ group. Microbiological analysis indicated that *Proteobacteria* and *Actinobacteria* predominated in the pike gut microbiome in all examined groups, followed by lower percentages of *Bacteroidetes* and *Firmicutes*. There were no statistically significant differences in the percent abundance of bacterial taxa between the control group and groups with a higher salinity. Our results suggest that salinity influences the gut microbiome structure in pike fry, and that 3‰ salinity may be a good solution for culturing pike at this stage in their development.

**Keywords:** 16S rRNA; gastrointestinal tract; gut microbiota; sequencing; salinity; *Esox lucius*; microbiome; metagenomics

### **1. Introduction**

The pike (*Esox lucius*) is a large, iteroparous, long-lived, top-predator fish species that occupies a broad range of aquatic environments, such as eutrophic and oligotrophic lakes, rivers, and brackish waters [1]. It is a keystone predator that can exert a top-down influence on fish communities [2]. The pike has become an important species for recreational and commercial fishing because of its size, wide distribution, presence in waters in urban areas, and locally high abundance [3–5]. It is also economically important for inland fisheries [6]. The increasing popularity of pike in angling and fish farming has created a need to increase pike fry production [7]. However, intensive pike farming is subject to limitations due to diseases and pathogens.

Although antimicrobial treatments in aquaculture effectively reduce or prevent mortalities caused by primary pathogens, they may have harmful side-effects that affect the overall health of the fish. Chemotherapeutics like antibiotics, vaccines, and immunostimulants can control and prevent disease outbreaks in fish, but these methods can harm both the fish and consumers, by accumulating in fish tissues [8] and causing immunosuppression of the fish and development of microbial resistance to antibiotics.

In some situations, sodium chloride (NaCl) can be added to tanks to protect fish against pathogens and parasites. NaCl may be a good alternative for controlling fungal outbreaks and external parasites [9]. In a number of studies, NaCl has been found to be an effective prophylactic treatment against important protozoans, helminthes, and fungal pathogens [10–12]. Increased salinity helped to suppress trichodiniasis (one of the major diseases in fish aquaculture worldwide, which causes massive fish mortality) in farmed freshwater tilapia (*Oreochromis niloticus*) [13]. However, changes in salinity can exert stress on fish, and thus affect the growth condition of the animals [14,15]. Furthermore, the salinity influences the osmotic gradients between the environment and the animal, which can impair basic physiological processes and even cause death [16].

Pike appear to be able to tolerate some changes in salinity [17–19]. In the Baltic Sea, this predator is found in both estuaries and coastal areas where salinity is at the average level of 3‰ and 7‰, respectively [20,21]. Therefore, pike might be a good candidate for experiments on the temporary prophylactic addition of NaCl to tanks, especially at the beginning of the juvenile stage, when the morbidity and mortality are highest. However, although there have been some studies on the effect of salinity on freshwater and marine fish [22–24], including pike [17–19], the studies on pike have focused in physiological changes (body weight and length, immune indicators, and cortisol level). It would be interesting to study how the microbiome can change under the influence of salinity levels that freshwater fish can tolerate. So far, information on the effect of salinity on the pike gut microbiota is lacking.

Information on fish gut microbiota is generally useful for understanding the factors that affect fish health, as research has shown that the gut microbiota play a key role in the health and nutrition of the host [25–27]. Fish gut microbiota contribute to digestion and can affect growth, reproduction, overall population dynamics, and the vulnerability of the host fish to disease [28]. Many studies have reported that the structure of the fish gut microbiota is influenced by factors such as the environment, diet, temperature, and pH [29–31]. These factors can affect the microbiome by changing the relative abundance of individual groups of microorganisms. Such changes can have repercussions for physiological, hormonal, or cellular functions, which can result in the development of diseases. To date, only a few studies have investigated the effect of salinity on fish gut microbiota [15,32,33]. However, not only were these studies conducted on other species, but even basic information on the pike gut microbiota is lacking.

It seems that this gap in our knowledge should be filled, because information on fish gut microbiota is generally useful in fish domestication [34]. Identification of the gastrointestinal microbiota contributes to our understanding on the functional interactions between microbes and the host [35]. Furthermore, examining the effects of various factors on the structure of the gut microbiome can help to maintain the fish in a good condition by adjusting those factors. For example, feed can be properly composed and enriched with appropriate probiotics and other supplements, and the length and intensity of prophylactic adjustments to salinity can be optimized.

Therefore, the objective of this study was to study the influence of salinity on the composition of the pike gut microbiome. Furthermore, characterization of the gastrointestinal microbiome of pike (*E. lucius*) using a mass sequencing approach based on genes coding for 16S rRNA was conducted. The differences between the microbiomes of the gastrointestinal tracts of pike from the control tank and tanks with 3‰ and 7‰ salinity levels were investigated. Additionally, we tried to determine if the mortality of juvenile fish in salinity tanks was correlated to changes in the microbiome. Information concerning the influence of salinity on the fish gut microbiome may help to develop our understanding on the functions of the fish gut microbiota, provide insight into growth condition differences, and explain the influence of salinity on host nutrition and/or physiology.

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

### *2.1. Conditions of Fish Rearing*

The fish were cultured at the Department of Lake and River Fisheries, University of Warmia and Mazury in Olsztyn, Poland. The experiment was conducted in rectangular glass aquarium tanks with a capacity of 25 dm3. Experimental groups of fish were kept in salinity levels of 3‰ and 7‰, obtained by adding NaCl to tap freshwater in proper amounts. The control group of fish was kept in a freshwater tank. The experiment was conducted by producing two replicates per group of treatment. Each tank had a filter and its own water pump with a separated water cycle. The fish density in the aquarium was 10 fish/dm3. During the experiment, the water temperature was 15 ◦C. The fish were fed with commercial Artemia Premium Cysts three times per day, with the amount depending on the number of fish in the tank. The experiment lasted for 10 days, which is a typical period of time for fry pike rearing. During the experiment, the mortality of fish in each tank was counted daily. Each day of the experiment, fish in all tanks were observed. At the end of each day, dead fish were noted and removed from tanks.

An ethics statement is not required for this type of research. No specific permissions were required for the described studies. The studies did not involve endangered or protected species. All experimental procedures were conducted in accordance with Polish law.

### *2.2. Sampling*

At the end of the experiment, pike were randomly chosen from the control group (group TS\_0‰) and two studied groups (TS\_3‰ and TS\_7‰). To obtain the morphological parameters, 20 random fish from each group were weighted and measured. Length and weight measurements were also used to calculate the body condition index (BCI; defined as weight/length3) as an indicator of the overall physiological state [36] and compared among salinity treatments over time. For microbiological analyses, 30 fish were analyzed (10 fish from each salinity group). To sacrifice each fish, MS-222 (150 mg/dm3) was used. To remove excess mucus from the ventral body surface, it was wiped with a paper towel. Then, to sanitize all instruments and surfaces, and the exterior of each fish, they were treated with 70% ethanol, after which, the instruments were flame-sterilized for dissection. For the removal of any remaining ethanol, the ventral body surface was dried with a paper towel. Next, the body cavity was opened and the entire gastrointestinal tract and its contents were aseptically removed from each fish. After this, the guts and their contents were stored at −20 ◦C until analysis.

### *2.3. DNA Extraction*

For DNA isolation, the entire fish gut and its contents were used. Plastic spatulas were used to homogenize samples. DNA was then extracted using a QIAmp DNA Stool Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. To quantify DNA, a Qubit 2.0 Fluorometer (Invitrogen, Carlsbad, CA, USA) was used. For verification of the DNA integrity, 1% agarose gel electrophoresis was performed. Purified DNA was suspended in 60 μL of elution buffer and then stored at −20 ◦C.

### *2.4. 16S rRNA Gene Amplicon Library Preparation and Sequencing*

To verify the quality and bacterial origin of DNA, PCR was performed with two universal 16S rRNA primers: 8F and 534R [37]. Based on these results, the DNA yield, and quality, 20 samples were chosen (seven from both TS\_0‰ and TS\_7‰, and six from TS\_3‰) and sent to Genomed S.A (Poland) for the mass sequencing of 16S rRNA amplicons. To prepare the 16S rRNA gene amplicons, the Illumina protocol "16S Metagenomic Sequencing Library Preparation" was used. In the PCR reaction, the variable V3 and V4 regions of the 16S rRNA gene were amplified using the following primers: forward (5 CCTACGGGNGGCWGCAG 3 ) and reverse (5 GACTACHVGGTATCTAATCC 3 ) [38]. The reaction was performed according to the Illumina protocol. To index the amplicons, a Nextera®XTIndex

Kit (Illumina, San Diego, CA, USA) was used according to the producer's instructions. For DNA sequencing, an Illumina MiSeq instrument was used, employing a 2 × 250 paired-end protocol, along with a Miseq reagent kit v3 (Illumina, USA).

### *2.5. Microbiome Analysis*

To process raw paired-end sequences (1,809,106 reads from 20 samples), the QIIME 2 [39] software package (https://qiime2.org; version: 2018.8) was employed. Paired ends were merged, reducing the 1,809,106 reads to 1,533,836 reads, and reads that could not be merged were excluded from further analyses. For quality control of the sequences, the Deblur plugin in Qiime 2 [40] was used to associate erroneous sequence reads with the true biological sequence from which they were derived, thus producing high-quality sequence variant data. First, 213,390 reads (13.91% of the data) were removed by applying an initial filtering process quality score (q = 20). Second, using Deblur, all reads were trimmed to 285 bp, based on the median quality score. In addition, chimeric sequences were detected and excluded from analyses. 16S rRNA Operational Taxonomic Units (OTUs) were picked from the Illumina reads using a closed-reference OTU picking protocol against the Greengenes database (https://docs.qiime2.org/2018.8/data-resources; data files: 13\_8) clustered at 99% identity and trimmed to span only the 16S rRNA V4 region flanked by sequencing primers 515F-806R. In the next step, taxonomy assignments were associated with OTUs based on the taxonomy associated with the Greengenes reference sequence defining each OTU. This step discarded most of reads which were not found in the Greengenes database and found only 228,667 reads which were assigned taxonomy. Out of the 228,667 Illumina reads from the V4 region of the bacterial 16S rRNA genes that passed the QIIME quality filters, 90.6% (207,197 reads) matched a reference sequence at a 99% nucleotide sequence identity. Next, OTU counts were binned into genus-level taxonomic groups for plot preparation.

Sequencing data were exported as individual fastq files and have been deposited in the Sequence Read Archive (SRA) NCBI (https://www.ncbi.nlm.nih.gov/) under the accession no. SRP226690. Samples can be found under accession numbers from SRX7043689 to SRX7043708.

### *2.6. Statistical Analysis*

Alpha and beta diversity statistics were recorded using the QIIME 2 scripts diversity plugin, which supports metrics for calculating and exploring the community alpha and beta diversity through statistics and visualizations in the context of sample metadata. In the calculation of alpha diversity metrics, normalization was performed using the "rarefaction" QIIME 2 process with standard parameters, setting the max\_rare\_depth (upper limit of rarefaction depths) to the mean sample size. Alpha diversity metrics were calculated using 'observed species', 'Chao1 index' (species richness estimator), 'Shannon's diversity index', and 'Good's coverage'. An alpha-rarefaction plot was created for each metric. The alpha diversity values at the same rarefaction level were calculated.

The normality and homogeneity of variance of all weight, length, and alpha diversity indices obtained from fish gut microbiome analysis were checked by Shapiro–Wilk's and Levene's test. Next, one-way ANOVA followed by Tukey–Kramer's post-hoc test (*p* = 0.05) was used to check the statistical differences between fish of different tanks (STATISTICA v.13.1 (StatSoft, Inc, Tulsa, OK, USA)).

The number of reads across samples was normalized by the sample size and the relative abundance (%) of each taxon was calculated. All taxa found in the gut microbiome were considered for statistical analysis. Statistical analysis of intestinal microbial profiles was performed using the Statistical Analysis of Metagenomics Profiles (STAMP) program (http://kiwi.cs.dal.ca/Software/STAMP), retaining unclassified reads [41]. To conduct a reliable statistical analysis, one sample (from TS\_0‰) was rejected from the analysis due to the low number of reads assigned to taxon levels. All *p*-values were calculated by ANOVA followed by Tukey–Kramer's post-hoc test and corrected for multiple comparisons using the Benjamini–Hochberg method for a False Discovery Rate (FDR) of 5% [42].

The beta diversity metric is an estimation of the between-sample diversity of the microbial profile and it was calculated by the QIIME 2 "diversity beta-group-significance" script. Both weighted (abundance matrix) and unweighted (presence/absence matrix) UniFrac distances [43,44] were used. The distance matrices were graphically visualized by three-dimensional principal coordinates analysis (PCoA) representations.

Differences in the beta diversity of bacterial communities were verified using a nonparametric Permutational Multivariate Analysis of Variance (PERMANOVA) test with 999 permutations (*p* = 0.05). A pairwise significance test was also performed by comparing groups from tanks with different salinity levels using the same distance matrix metrics (weighted and unweighted UniFrac distances). These tests were available in QIIME 2.

### **3. Results**

### *3.1. Fish Culturing*

Fish mortality was observed in all tanks during the experiment. The number of dead fish was noted and used to prepare the cumulative mortality of fish plot (Figure 1). The lowest mortality was observed in tanks with 0‰ and 3‰ salinity, representing less than 6% of the cumulative mortality at the end of the experiment. In the TS\_7‰ group, mortality increased more intensively and reached 37% at the end of the experiment. At the end of the experiment, the weight and length of fish from each tank (n = 20) were measured (Figure 2A). The length and weight were used to calculate the BCI and were presented on a plot (Figure 2B). Statistical analysis showed significant differences in both the length and weight between different salinity groups.

**Figure 1.** Cumulative mortality of pike fry during 10 days of the experiment. The experiment was conducted in rectangular glass aquarium tanks with a capacity of 25 dm3 (n = 2 per each group). Each tank had a filter and its own water pump with a separated water cycle. The density of fish in the aquarium was 10 fish/dm3 and the water temperature was 15 ◦C. The fish were fed with commercial Artemia Premium Cysts three times per day.

**Figure 2.** (**A**) Average weight and length of fish at the end of the experiment (n = 20/group) and (**B**) the <sup>e</sup>ffects of salinity on the pike condition index <sup>=</sup> weight (g)/fork length (mm)3; the mean <sup>±</sup> SEM for each salinity level (n = 20).

### *3.2. Qiime Analysis of Sequencing Data*

The fastq files with data on the gut microbiome of 20 chosen fish obtained from the Illumina MiSeq were analyzed using QIIME 2 software. After filtering for quality, trimming length, and assigned taxonomies, the number of reads taxonomically classified according to the Greengenes database was 207,197 (Table 1). This value corresponded to an average number of 10,360 per sample (range 4260–17,585). A total of 937 OTUs at a 99% nucleotide sequence identity in pike gut content samples were identified. After rarefaction, which normalized the sample to the max rare depth of sequences, the observed species number per sample was between 93 and 235, corresponding to an average number of counts per group between 140 and 196 (Table 1). Good's coverage values for all groups were ≥0.99, indicating that sequencing coverage was attained and that the OTUs found in the samples were representative of the sampled population (Table 1). All the rarefaction curves tended to plateau (Figure S1). By statistically analyzing the OTU index of the examined groups, it was found that salinity was a factor which affected the species richness. As for the rest of the indices, salinity was not a factor with a significant affect.


**Table 1.** Mean number of reads per sample assigned to OTUs, and alpha diversity metric values (without one sample with less than 737 reads) of the gut microbial community of pike from different environments.

### *3.3. Characterization of the Pike Gut Microbiome*

The gut microbiome of 20 fish in three groups that were kept at different levels of salinity was examined to characterize its structure and to reveal the differences between groups and individual fish. The microbiome structures of each investigated group of fish at the phylum, class, order, and family level were successfully described. In our study, 15 phyla, 32 classes, 53 orders, and 88 families were classified. The gut microbial community in the groups as a whole and in individual fish are presented at the phylum (Figure 3), class (Figure 4), and order (Figure 5) levels. The figures present taxa with a relative abundance ≥0.5% (up to the class level) and ≥2.5% at the order level. All taxa are shown in Table S1.

**Figure 3.** Relative abundance (%) of the most prevalent bacterial phyla in three groups kept at different levels of salinity (**A**) and in individual pike (**B**). Figures show all phyla with an abundance >0.5%. Phyla with an abundance ≤0.5% are pooled and labeled as "Others".

**Figure 4.** Relative abundance (%) of the most prevalent bacterial classes in three groups kept at different levels of salinity (**A**) and in individual pike (**B**). Figures show all classes with an abundance >0.5%. Classes with an abundance ≤0.5% are pooled and labeled as "Others".

**Figure 5.** Relative abundance (%) of the most prevalent bacterial orders in three groups kept at different levels of salinity (**A**) and in individual pike (**B**). Figures show all orders with an abundance >2.5%. Orders with an abundance ≤2.5% are pooled and labeled as "Others".

At the phylum level, *Proteobacteria* predominated in each group (mean abundance of 40%–53%), followed by *Actinobacteria*, *Firmicutes*, and *Bacteroidetes* (Figure 3A, Table S1). In individual fish, the composition of the bacterial community was generally similar to that in the groups as a whole, with four exceptions. In fish S3-3, *Firmicutes* predominated (42.7%) (Figure 3B), whereas in fish S3-0, S6-0, and S9-7, *Actinobacteria* predominated (43%–72%). Interestingly, in the 7‰ salinity group, *Planctomycetes* were more abundant in three fish than in the other fish (5.3%–30.2% vs. ≤0.5%)

At the class level, in almost all fish in the investigated groups, the gut microbiome was dominated by members of *Gammaproteobacteria* (17.7%–37.5%) and *Actionobacteria* (15.6%–22.2%) (Figure 4A, Table S1). *Alphaproteobacteria* were third in terms of the percent abundance in all groups. Some of the gut microbiome profiles of individual fish were different from those of other individuals. For example, the gut microbiome of fish S6-0 was dominated by *Acidimicrobiia* (48.64%), whereas in

other fish, its abundance was less or even absent (Figure 4B). The gut microbiome of fish S3-0 was completely dominated by *Gammaproteobacteria* (93.8%), whereas in fish S9-7, the gut was dominated by *Actionobacteria* (71.4%). *Bacili* dominated the gut microbiome of fish S3-3 (40.6%), whereas *Alphaproteobacteria* dominated that of fish S2-7.

At the order level, a higher percent abundance of unclassified bacteria (23.8–33.7%) was observed than at higher taxonomic levels. In all groups, there was a similar predominance of *Actinomycetales* (21.8%–22.8%) (Figure 5A). In TS\_7‰, the mean abundance of *Rhodobacterales* was greater (10.84%) than in the rest of the groups (0.97%–2.44%). *Burkholderiales* were present in all groups, but more abundant in TS\_3‰ (12.66%) (Figure 5A). The bacterial community composition of the gut of an individual fish tended to be more similar to that of other fish in the same group than to that of fish in other groups. *Rhodobacterales* were present in all fish microbiomes, but in TS\_7‰, they were more abundant in five fish (5.8%–25.2%) (Figure 5B). Interestingly, in the 3‰ salinity group, *Aeromonadales* were more abundant in four fish than in the other fish (5.97%–79.4% vs. ≤1.57%).

One-way ANOVA was calculated and differences were considered significant at *p* < 0.05 after Benjamini–Hochberg FDR correction for multiple comparisons. In addition, the corresponding effect size (ETA-Squared) was calculated. This analysis showed no statistically significant differences between groups in terms of the abundance of each taxa of bacteria. Statistical analyses of all taxa and their relative abundance (%) are reported in Table S1. Although any significant differences in relative abundance (%) of all taxa between groups were not found, it is interesting to note that the abundance of some bacteria was higher for specific salinity concentrations in most of the fish gut microbiome group. At 7‰ salinity, *Planctomycetes*, *Rhodobacterales*, and *Alphaproteobacteria* were more abundant than at other levels of salinity; at 3‰, in contrast, *Betaproteobacteria* and *Burholderiales* were more abundant; and in the control tanks, *Aeromonadales* were more abundant (Table S1).

Permutation multivariate analysis (PERMANOVA) indicated that, overall, there was a significant divergence between groups only in terms of unweighted UniFrac distance matrices (*p* = 0.001; Pseudo-F = 1.63) (Table 2). The beta-diversity pairwise test on the unweighted UniFrac data showed that the gut microbiomes of fish in 7‰ salinity were significantly different to those of the 3‰ salinity group. Weighted Unifrac did not show a significant overall difference in the abundance of bacteria.


**Table 2.** Permutation multivariate analysis (PERMANOVA) of weighted and unweighted Unifrac data of intestinal microbiomes of pike living in different environments.

QIIME 2 was used to compute microbial beta diversity metrics. Analyses were performed using weighted and unweighted UniFrac distances. Data from UniFrac metrics were used to prepare three-dimensional plots using principal coordinates analysis (PCoA) (Figure 6). PCoA reflected the PERMANOVA results. Weighted Unifrac showed that the 0‰ and 3‰ salinity groups clustered together. Fish from 7‰ salinity were separated from the rest of the samples. PCoA of the unweighted Unifrac distance matrix showed that samples from the 3‰ salinity group clustered together and samples from the other groups were scattered.

**Figure 6.** Beta diversity metrics. Principal coordinates analysis (PCoA) of unweighted (**A**) and weighted (**B**) Unifrac distances of gut microbial communities associated with different salinity levels. The figures show plots of individual fish according to their microbial profile.

### **4. Discussion**

Recent studies on the effect of salinity on fish have focused on examining how salinity affects the growth, mortality, health condition, and osmotic stress of fish farm [9,13,14,22,45,46] and hatching eggs [47]. This type of research is necessary because it might help to properly choose salt concentrations and times of exposition of fish to keep them healthy during farming. Salt is commonly used as a disinfectant for the prophylactic prevention of disease development in fish farms, and thus might help to keep fish in good health, especially at the beginning of fish fry life. It is a cheaper, and probably healthier, replacement for other chemotherapeutics. However, the effect of salinity on the fish gut microbiome has been poorly investigated. Although there have been studies on the effect of salinity on the gastrointestinal microbiome of a few fish species [15,32,33,48], to the best of our knowledge, there is a lack of studies on the gut microbiome of pike fry *(E. lucius*). Therefore, it was hypothesized that the gut microbiota of pike living in freshwater differ from those of pike living in different salinity concentrations. What is more, a comparison of the morphological parameters and mortality of fish under osmotic stress was conducted. Our results provide information on the gut microbiota of pike and highlight associations between environmental factors (salinity) and gut microbiota. An understanding of these associations provides information that may be useful for addressing problems during the domestication of these valuable freshwater fish.

In our study, the size and weight of pike fry were influenced by long-term salinity tolerance. All examined fish reached a developmental stage where the organs involved in osmoregulation, e.g., the gills, were expected to be fully functional. Although the average length of fish showed little difference between each tank, the mean weight differed to a greater extent (Figure 2A). Generally, along with increasing the salinity, the environment should be cleaner and healthier, which could prevent the excessive development of pathogens, leading to better welfare of animals. However, in our study, the highest growth of fish from the 7‰ group was related to the lowest density of fish caused by the highest mortality. In these conditions, high-salinity fish had better access to food, and thus showed the highest weight values.

Mortality is a common problem in fish fry farms due to diseases mainly caused by pathogens [49]. Salt used in farming freshwater fish which tolerate some salinity is quite effective, cheaper, and much healthier (for fish and consumers). Sometimes, in aquaculture, fish are bathed prophylactically in brine to avoid pathogen development [50]. Our results showed that salinity had a statistically significant effect on cumulative mortality in fish. The greatest mortality was observed in the 7‰ salinity tank, increasing drastically on day 8 of the experiment to about 37% (Figure 1). Here, the length of the fish

was the greatest. It seems that larger pike fry cannot acclimate to these long-term osmotic conditions. This is in good agreement with other studies on pike, where the mortality of larger fish was greater than that of smaller fish [18]. The cumulative mortality of fish from tanks with a 3‰ salinity was surprisingly similar to that of the control group. What is more, the BCI was greater than in the control group. These findings suggest that this level of salinity is well-tolerated by pike, which is consistent with reports of this species in coastal areas all around the Baltic Sea, where salinities vary from 4‰ to 7‰ [18]. The fact that around 97% of fish survived this salinity treatment and had a condition factor similar to that of the controls indicates that the fish in 3‰ salinity did not experience a loss of water due to salinity stress. A loss of water is believed to be one of the reasons why fish have died in salinity treatments [18].

Our study indicated that, in the pike gut microbiome, the phyla *Proteobacteria* and *Actinobacteria* predominate, followed by lower percentages of *Bacteroidetes* and *Firmicutes* (Figure 3), in both control fish and fish in different salinity levels. This observation suggests that these phyla could play important roles in pike gut functioning. Although it is difficult to estimate the contribution of specific bacteria to the function of the whole gut ecosystem, it is reasonable to expect that the overall gut microbiome will be strongly influenced by the predominant microorganisms [28]. The presence of similar bacterial taxa in the gut microbiota of multiple fish species suggests that these bacteria are valuable for the host and could play important roles in digestion, nutrient absorption, and immune responses [51]. These phyla have been found in the intestines of many marine and freshwater fish species [52–56]. However, the proportions of these phyla present in our study were different to those in other studies [33,57,58]. In studies on eight freshwater fish species [55], marine Atlantic cod [59], and Fine flounder [60], *Proteobacteria* has been described as the predominant phylum. On the other hand, *Firmicutes* predominate in the intestinal content of aquacultured Siberian sturgeon [61], grass carp [62], and Atlantic salmon [63]. Our study showed that the mean abundance of *Proteobacteria* of a group might be a predominant phylum (Figure 3A). However, looking at individual fish, the proportions of the percent abundance of main phyla were not similar. Sometimes, *Actinobacteria* predominate over *Proteobacteria*. This observation was noticed in control and salinity tanks, so salinity does not have a significant influence on the gut structure at the phylum level. This may be related to fish development. It appears that the diversity of bacteria increases as fish develop [64]. Ringø and Birkbeck [65], in their review, summarized 24 studies on the microbiome of fish larvae and fry and showed changes during fish development. Furthermore, relatively stable gut microbiota are established within the first 50 days of life for many species [66]. Therefore, gut microbiota of pike might just have been forming during the experiment. Despite these facts, our results suggest that *Proteobacteria, Firmicutes, Bacteroidetes*, and *Actinobacteria*, which dominate in the fish gut, may play important roles in the gut microbiome of pike at an early life stage.

Various salinity concentrations can lead to osmotic stress in fish. Physiological changes in the host that occur during osmotic stress force its gut microbiome to adapt to the new conditions. Therefore, new species may develop in the gut microbiome. Although significant differences in the abundance of taxa between control and salinity groups (Table S1) were not found, there were small changes in the percent abundance of some groups. In some fish reared in 7‰ salinity, a higher percent abundance of *Planctomycetes* were observed, whereas in the other groups, these bacteria were absent. *Planctomycetes* were found in yellow grouper, which is a marine fish [67]. Another study reported that *Planctomycetes* were found in macroalgae biofilm [68]. This suggests that *Planctomycetes* like salty environments and could multiply under the influence of this stress factor.

In most fish, at all salinity concentrations, *Proteobacteria* usually predominate over other phyla. In contrast to our results, a study on *Oreochromis niloticus* (Nile tilapia) showed that the percent abundance of *Proteobacteria* changed and was higher in fish from a marine environment than in those from a freshwater environment [15]. Furthermore, the study found that *Actinobacteria* and *Bacteroidetes* were more abundant in fish from freshwater than in those from marine water, whereas our study found similar levels of these phyla at all salinity levels. The reason for these results is not yet completely

understood, but the results might depend on many conditions, such as the salinity concentration, fish species, length of exposition, diet, age of the fish, and others. Although significant changes in phyla were not observed, differences in the percent abundance of *Proteobacteria* classes between salinity levels were observed. *Alphaproteobacteria* predominated in 7‰ salinity, whereas in 3‰ and control tanks, *Betaproteobacteria* were more abundant. An effect of salinity on the microbiome structure was also found in black molly (*Poecilia sphenops*), where an increase in salinity induced changes in the dominant bacterial taxa in the microbiomes [32]. In the more saline treatments in their study, unknown *Enterobacteriaceae* (18‰ and 30‰) replaced *Aeromonas* and *Cetobacterium* OTUs that were present in the freshwater treatments (0‰ and 5‰). Although their results differ slightly from our results, they confirm that salinity can influence the gut microbiome. Furthermore, it is interesting to note that *Aeromonadales*, which was more abundant in control fish in our study, was less abundant in fish with salinity treatment. This supports previous findings in other studies [33,34], where microbes differ on broader scales between freshwater and saltwater fish, with the bacteria *Aeromonadales* being enriched in freshwater specimens and anadromous fish collected from freshwater habitats. Those researchers also found that *Vibrio* exhibited a greater prevalence in marine species. Although an increased percent abundance of *Vibrio* in fish with 3‰ and 7‰ salinity was not observed in our study, *Burholderiales* was more abundant in 3‰ salinity, and *Rhodobacterales* was more abundant in 7‰ salinity. However, it is necessary to note that, in the cited studies [33,34], different fish from freshwater or marine water were studied, so it is difficult to evaluate the exact role of salinity in shaping the intestinal microbiota. In contrast, our study focused on one species in different conditions and suggests that the gut microbiota composition is related to salinity.

### **5. Conclusions**

In conclusion, our research provides the first detailed description of the structure of the gut microbiome of juvenile pike (*E. lucius*). The dominant phyla were *Proteobacteria* and *Actinobacteria*, followed by lower percentages of *Bacteroidetes* and *Firmicutes.* These phyla are the main components of the gut microbiome in many other fish species [56,65]. Despite the fact that there were no significant differences in the percent abundance of gut bacteria species, some taxa were more abundant at certain levels of salinity. At 7‰ salinity, *Planctomycetes, Rhodobacterales*, and *Alphaproteobacteria* were more abundant than at other levels of salinity; at 3‰, in contrast, *Betaproteobacteria* and *Burholderiales* were more abundant; and in the control tanks, *Aeromonadales* were more abundant. Furthermore, the alpha diversity based on the OTU index showed that salinity affected the number of species. Taken together, these findings indicate that salinity influences the bacterial biodiversity. Our results also suggest that salinity influences the composition of the pike gut microbiome in terms of the percent abundance of bacteria. At 3‰ salinity, mortality was low (similar to that in the control tanks), suggesting that this concentration is tolerated by the pike, beneficial for their development, and protects them from pathogens at an early life stage. Our findings suggest that salinity adjustment can improve fish welfare and aquaculture practices. To test this hypothesis, future interspecific studies should test a wider range of salinity levels and include long-term exposition.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2076-3417/10/7/2506/s1: Figure S1: Alpha diversity metrics. Rarefraction curves of the gut microbiome structure from pike kept at different levels of salinity. (A) Observed OTU's, (B) species richness (Chao1), and (C) Shannon's diversity index; Table S1: Statistical analysis of the mean relative abundance (%) ± SD of phyla, classes, orders, and families between groups.

**Author Contributions:** Conceptualization, S.C. and T.D.; methodology, T.D. and S.C.; software, T.D.; validation, T.D.; formal analysis, T.D. and M.G.; investigation, T.D.; resources, R.K.; data curation, T.D.; writing—original draft preparation, T.D.; writing—review and editing, T.D., S.C., R.K., and M.G.; visualization, T.D.; supervision, S.C.; project administration, S.C. and T.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** The project was financed as part of a grant statutory project (No. 29.610.024-110) of the University of Warmia and Mazury in Olsztyn and financially co-supported by the Minister of Science and Higher Education in the scope of the program entitled "Regional lnitiative of Excellence" for the years 2019–2022, Project No. 010/RID/2018/19, amount of funding 12.000.000 PLN.

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

### **References**


© 2020 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/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Applied Sciences* Editorial Office E-mail: applsci@mdpi.com www.mdpi.com/journal/applsci

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18