Effects of nettle slurry (Urtica dioica L.) used as foliar fertilizer on potato (Solanum tuberosum L.) yield and plant growth

Experimental design

In this experiment, six treatments were planned; three with different Urtica doses (A–C) as so: one with Urtica and Equisetum slurry (D); one with conventional foliar manure (E); one control treatment (F) treated only with water (Fig. 1). The experiment was run with a randomized complete block design with six replicates per treatment and 36 plots distributed in six blocks. Each plot was 5 m long and 3 m wide (three ridges) and each plot covered 15 m² (Fig. 1). All the plants in each plot were treated. However, measurements were only taken in the three central meters of the central ridge to avoid the edge effect (Fig. S1). Plots were coded for unbiased management.

Experimental site

The experiment was conducted in the town of Godella in the province of Valencia, Spain (39°31′12.9″N, 0°24′34.9″W) in a calcareous silty clay loam alluvial soil from February to June 2016. The general site climate is Mediterranean oceanic, with a long-term average annual rainfall of 468 mm and an average annual air temperature of 17 °C (w.s. 39°29′N, 0°23′W 13 m a.s.l.). In 2016, the annual rainfall was 382 mm and it rained 58 mm during the experimental period. The plot was fertilized once at the beginning of the experiment before sowing. For this, horse manure (25 t/ha) was used. Subsequently, no additional soil fertilization was carried out; therefore, the use of nettle slurry was the essential fertilizer in our fertilization program. The plot was prepared and managed and ridges were built in collaboration with a local organic farmer. Eighty kilograms of the Dutch potato variety “agria” was sown for the experiment. The total cycle lasted 107 days, with five irrigation sessions (Table 1).

Soil, water, and slurry chemical analyses

The site soil was analyzed at the start of the experiment. The experimental site was divided into four equal plots for soil sampling (A, B, C, D; Table 2). Each sample resulted from mixing five uniformly distributed soil subsamples (detail of subsamples location in Fig. S1). Soil analyses were carried out in the laboratory following standard protocols (AOAC, 1995). Electrical conductivity, pH, organic matter, calcium carbonate, total nitrogen, and available potassium, sodium, phosphorus, calcium, and magnesium, were determined per soil sample. For each soil parameter, high (H), medium (M) and low (L) levels were assigned according to previous studies in the same geographical area (Jackson, 1976; Villalbi & Vidal, 1988; Guigou et al., 1989; Legaz et al., 1995). The water analysis results were provided by Ambitec laboratorios (laboratory reference: 7938A) using a standard methodology (Table 3). Electrical conductivity, pH, organic matter, ash (%), K₂O (%), P₂O₅ (%), total N (%) and protein were determined for Urtica and Equisetum slurry (Table 4). Slurry analyses were carried out in the laboratory following standard protocols (Horwitz, 1989). Slurry chemical analyses were redone several times and with different batches to confirm the obtained values.

Treatments

Commercial foliar fertilizers were used to prepare the treatment solutions. The U. dioica slurry (fermenting extract) and the Equisetum hyemale slurry came from the “Ortiga Amiga” trading house (Spanish company tax code (NIF) 46339215b). The base substance was produced according to regulation (EC) No. 1107/2009. The recommended Urtica and Equisetum slurry dose (RD) was 10% v/v. The conventional foliar manure was “Isabión” from the Syngenta trading house. The Isabión label indicates 10.9% of total nitrogen. The recommended Isabión dose was 200–300 cc/100 l of water.

Following the company recommendations, in a 10-l final volume solution the following volumes of foliar fertilizers were added to water: for treatment A (RD) 1 l; for treatment B (½ RD) 0.5 l; for treatment C (2 × RD) 2 l; for treatment D 1 l of the Urtica and 1 l of Equisetum slurry; for treatment E 25 ml of the conventional foliar manure. Control treatment F was composed of only water. Slurry was applied using a 15 l knapsack sprayer, to which we added a conical nozzle (PULMIC) to ensure a constant pressure of 2.5 bar. Treatments were applied three times throughout the crop cycle (Table 1). In the first application, plants displayed the first four true leaves (15th April); in the second, plants flowered (May 4); and in the third, plants had fully developed and came closer to the end of the cycle (May 23).

Measurements

Crop yields, expressed as kilogram of potatoes, were determined by harvesting the central 3 m of the central ridge (Fig. S1). Three samplers worked at the same time by harvesting 1 m each to obtain three measurements per plot. The sampler effect was analyzed with no significant differences found among samplers (Fig. S2 Samplers). A total of five parameters related to plant growth and size were measured. Height, number of leaves and leaf length were measured three times throughout the plant’s growth cycle (T1 is April 14, T2 is May 3, and T3 is June 1), while the number of flowers (May 3) and the final biomass (June 11) were measured once (Table 1). Each parameter was measured on five individuals from all 36 plots (n = 180).

A total of three leaf samples from different individuals in each plot (108 samples) were taken for the chlorophyll analyses. Sampling was performed when plants had fully developed on June 1 (Table 1). Chlorophylls were extracted with acetone protocol (Val, Heras & Monge, 1985) and their content was determined by absorbance spectrophotometry at wavelengths 645, 652, and 663 nm. The presence of pests and diseases was observed three times (T1, T2, and T3). Each time, five plants were noted per repetition and treatment (n = 180). Only Colorado potato beetle (Leptinotarsa decemlineata Say) and potato blight (Phytophthora infestans (Mont.) de Bary) were found.

Statistical analysis

The average, standard error, skew, kurtosis, frequency distribution and density curve of the yields were assessed for each treatment. ANOVAs were used to compare the mean values between treatments and blocks. Shapiro–Wilk tests were calculated to check normality requirements. In some cases due to lack of normality, nonparametric methods were selected to compare the means among treatments and blocks by a Kruskal–Wallis rank sum test. When significant differences were found, Levene’s test and eta-squared statistics were calculated to assess the homogeneity of variances and the effect size in the ANOVA, respectively. Tukey honest significant difference (HSD) was selected as the post hoc test.

To exclude the block effect (watering) from the main effect (yield), the residuals from the block ANOVAs were compared. Linear models were selected to analyze the effect of numerical variables (plant size and chlorophyll variables) on yield, while Pearson correlations were calculated to assess the relationships between independent variables. A principal component analysis was used to analyze how the chlorophyll parameters were related. All the statistical analyses were done using R (R Core Team 2017) with some extra packages: car (Fox & Weisberg, 2011); plotrix (Lemon, 2006); ggpubr (Kassambara, 2017); agricolae (de Mendiburu, 2017); vcd (Meyer, Zeileis & Hornik, 2006); writexl (Ooms, 2017); ggplot2 (Wickham, 2009); readxl (Wickham & Bryan, 2017); plyr (Wickham, 2011); tidyr (Wickham & Henry, 2017) and knitr (Xie, 2017).

Soil, irrigation water, and slurry chemical composition

Soil nutrient status was homogeneous throughout the site (Table 2). Although natural organic matter contents were medium and high, the presence of high levels of carbonates and C/N ratio indicated a slow release of nitrogen to soil. Therefore, the total nitrogen value in soil was low. We obtained homogeneous pH values, which indicates strongly alkaline soil (average pH of 8.6; Table 2).

The irrigation water analyses indicated a low-medium nitrate content (Table 3). According to Riverside classification C3-S1, water salinity was medium, and therefore, appropriate to irrigate well-drained soil (Aragues et al., 1979). The analyses showed a high content of carbonates and soluble potassium, and a low sodium content (Table 3).

The slurry chemical analyses indicated neutral pH values (7.6–7.9) and medium electrical conductivity (1.2–1.3 mS) (Table 4). Very low levels of macronutrients (N-P-K) were obtained for both slurries. Nettle slurries from other trading houses were also analyzed with similar low levels of macronutrients (Table S1).

Effect of nettle slurry on potato yields

Including all the treatments, 36 plots were treated and 108 yield weight measures were recorded (Fig. 1 and Fig. S3). The ANOVA for the effect of treatments on yields showed no significant differences among treatments (p = 0.7697, R² = 0.0243). With the control treatment, a mean of 2.35 kg/m of potatoes were collected, while the mean was 2.16 kg/m with the recommended nettle treatment dose (A), and no significant differences were found (Table 5). None of the evaluated Urtica doses (recommended dose, half the recommended dose, double the recommended dose and Urtica in combination with Equisetum), meaning 24 independent plots, showed any significant increase in potato tuber yields compared with the control treatment (Table 5 and Fig. 2).

In order to check whether there was any overlapping factor that masked the possible effect of nettle slurry on potato yields, we analyzed whether there were any significant differences between blocks (Table S2 and Fig. S4). The ANOVA for the effect of blocks on yield showed significant differences (p = 0.0004, R² = 0.3743). The mean yield values increased from block 1–2 (closer to the water entry point) to block 5–6 (far away from the water entry point). Accordingly, the watering effect was analyzed. Due to the plot slope, a similar amount of water was applied to blocks 1 and 2, blocks 3 and 4, and blocks 5 and 6 (Fig. 1). Therefore, pairs of blocks were used to analyze the effect of watering (Table 6 and Fig. 3). Significant differences among watering blocks were observed (p = 4.1507⁻⁴, R² = 0.1378), with a yield reduction as blocks approached the water entry point (e.g., 1.97 ± 0.08 kg/m for B1-2, and 2.57 ± 0.12 kg/m for B5-6) (Table 6).

The ANOVA for the effect of treatments on yield, excluding the watering effect, showed no significant differences among treatments (p = 0.7065, R² = 0.0282; Fig. 4). F value, p value and HSD changed slightly by eliminating the effect of block pair, but not enough to be significant. Consequently, none of the nettle treatments resulted in significantly increased tuber production after excluding the watering effect. When considering only tendencies (as differences were nonsignificant), the plants treated with the half dose and the control treatment yielded more tuber kilograms than those with the recommended nettle dose (Table 5).

The interaction between treatments and block pairs (Fig. 5 and Table S3) showed that a higher yield was obtained in block pair 5–6 for all the treatments. Inside the block pair, differences among treatments (Table S3) were not significant in any case.

Influence of nettle slurry on potato growth and plant size parameters

The results of measuring the growth and size parameters (height, number of leaves, leaf length measured at T1, T2, and T3, number of flowers, and the final biomass measured once) are provided in Table 7. The ANOVA showed that the effect of treatments on these parameters was weak, with only some significant differences (Table 8). The analyses of plant height and number of flowers, both at T2, were not meaningful because the standard deviations were not homogenous among treatments (Levene’s test in Table 8). The ANOVA for plant height at T3 showed significant differences (p = 0.0051), which indicates that the plants treated with a double RD (mean of 67.27 cm) were slightly higher at the end of the cycle than the control treatment (mean of 60.13 cm), with significant differences even though the effect size was small (η² = 0.09) (Fig. 6).

The ANOVA for leaf length at T1 also showed significant differences (p = 0.00001), which indicates that initially the leaves treated with slurry (RD and ½ RD) were larger than those from the control (e.g. 21.9 ± 0.6 cm for ½ RD, and 16.4 ± 1 cm for the control), although this effect disappeared as plants grew (T2 and T3) (Fig. 6). The differences in the number of leaves among treatments were significant at all times, but effect sizes were small (η² = 0.09 for T1 and η² = 0.07 for T3). A variation at T2 of only two leaves between the lowest (13.7) and highest (15.7) mean values was found. Nevertheless, the results of number of leaves were not consistent as the plants treated with the slurry RD had two leaves less than those of the control treatment, while the plants treated with the ½ RD had the same number of leaves as the control treatment (15.7). Aerial biomass was also evaluated at the end of the cycle, and no significant differences among treatments were found (Table 8).

Influence of treatments on chlorophyll content

Ten days before harvest, three leaves per plot were collected for chlorophyll determination purposes (n = 108). A chlorophyll, B chlorophyll, total chlorophyll 1 and total chlorophyll 2 were calculated with the absorbance data. No significant differences in the leaf chlorophyll content among treatments were observed (total chlorophyll 2 p value = 0.3186) (Table 9). The chlorophyll parameters correlated highly (Fig. S5), therefore total chlorophyll 2, which was nearly parallel to Principal Component 1, was selected to analyze the effects of chlorophylls (Fig. S6). The plants treated with double RD had the highest total chlorophyll 2 content (mean value of 0.82 mg/g), but no significant differences compared with other treatments were found (Table 9).

Effect on pests and presence of diseases

Pests and presence of diseases were recorded at three different times and in five plants from each plot. The Colorado beetle L. decemlineata appeared at T2 and slightly decreased at T3 (Fig. 7), whereas P. infestans appeared only at T3 for all treatments and water regimes, except for treatment E (Fig. 8). The Chi-square tests for pest occurrence were done to compare treatments. Nevertheless, there were not enough expected values for the pests present analyses, which indicates that it was not possible to correctly assess significant differences among treatments. Similar small amounts of pests were observed in all the treatments, and low levels of pests present indicated that no treatment significantly increased the appearance of pests.

Correlations among the evaluated variables

Plant size and the chlorophyll parameters did not affect yield, while some of these parameters were affected by treatments or watering. Chlorophyll content was not affected by watering blocks (p > 0.05), but correlated positively with biomass (R = 0.54), height (R = 0.62) and number of leaves (R = 0.47) at T3, and to a lesser extent at T2. These findings indicate that large plants had a higher chlorophyll content.

Pests and diseases did not correlate with one another, nor with yield (R = 0.06 for beetle; R = 0.18 for blight) or watering blocks. P. infestans correlated negatively with plant height (R = −0.52), number of leaves (R = −0.57) and number of flowers (R = −0.44), which indicates that this disease appeared more frequently in small or weak plants. The plant size parameters correlated highly with one another (i.e., R = 0.70 for height T3—number of leaves T3), but surprisingly did not relate with yield or watering blocks (R² between 0.01–0.001 and p > 0.05).