Irrigation Scheduling for Green Bell Peppers Using Capacitance Soil Moisture Sensors

Plots were harvested on 60, 74, and 77 days after transplanting (DAT) in 2005; on 58, 70, and 74 DAT in 2006 and on 69 and 83 DAT in 2007. The harvested area consisted of a central 10.5-m long region within each plot. Pepper fruits were graded into culls, U.S. Number 2 (medium), U.S. Number 1 (large), and Fancy (extralarge) according to USDA (2005) grading standards for fresh market sweet peppers. Marketable weight was calculated as total harvested weight minus the weight of culls. The number and weight of fruits per grading class were recorded for individual plots. IWUE expressed in kg of fruits ${\text{m}}^{-3}$ was calculated by taking the quotient of the marketable yields $\left(\text{kg}{\text{ha}}^{-1}\right)$ and the total applied seasonal irrigation depth $\left({\text{m}}^3{\text{ha}}^{-1}\right)$ .

Maximum biomass accumulation was evaluated by harvesting two representative plants per treatment replicate at 74, 70, and 78 DAT for 2005, 2006, and 2007 trials, respectively. Vegetative and reproductive plant parts were separated. Shoot and fruit tissues were dried at $65°\text{C}$ for subsequent dry weight determination.

Shoot and fruit tissue samples were ground in a Wiley mill to pass through a 2-mm screen and a thoroughly mixed 5-g portion of each sample was stored. Tissue material was digested using an aluminum block digestion procedure of Gallaher et al. (1975) and analyzed for total Kjeldahl N at the Analytical Research Laboratory (University of Florida, Gainesville) using U.S. EPA method 351.2 (Jones and Case 1991). Nitrogen accumulation by the plant was calculated by multiplying weights of stems plus leaves, and fruit tissue by the corresponding N concentrations. NUE was defined as N uptake by the plants divided by the total amount of N supplied from weekly fertigation plus initial soil nitrate. In 2005, leaf samples were also taken at 47 DAT. The 12 most recently matured leaves were collected between 9 and 11 a.m. from each treatment plot. The petiole base was cut at 15 mm from the stem attachment and crushed in a stainless steel garlic crusher. Sap ${\text{NO}}_3–\text{N}$ concentration in the sap was determined by an ion electrode method using C-141 Cardy Nitrate Meter (Horiba, Spectrum Technologies, Inc., Plainfield, Ill.)

The VWC of the top soil of the production beds was monitored by coupling time domain reflectometry (TDR) probes (CS-616, Campbell Scientific, Inc. Logan, Utah) with a data logger (CR-10X, Campbell Scientific, Inc., Logan, Utah). Soil moisture probes were placed in the beds at two subsequent soil layers which recorded soil moisture values. The upper probe was inserted at an angle in order to capture soil moisture in the top 0.25 m of the profile and the lower probe was inserted vertically below the upper probe recording soil moisture between 0.25 and 0.55 m.

Zero tension drainage lysimeters were located 0.75 m below the surface of the bed (Zotarelli et al. 2007). The drainage lysimeters were constructed out of 208-L capacity drums that were cut in half lengthwise with a length of 0.85 m, a diameter of 0.55 m, and a height of 0.27 m. A total of 24 lysimeters were used to evaluate percolated volume and nitrate leaching in all irrigation treatments under N-rate treatments of 220 and $330\text{kg}{\text{ha}}^{-1}$ (four replicates for each treatment). A vacuum pump was used to extract the leachate accumulated at the bottom of the lysimeter. The leachate was removed weekly one day prior to the next fertigation event by applying a partial vacuum (35–40 kPa) using 20-L vacuum bottles for each drainage lysimeter. The use of weekly samplings combined with a partial vacuum allowed for an effective extraction of leachate at the bottom of the drainage lysimeter and the absence of anaerobic conditions. After sampling, soil water in the bottom of the barrel dropped to 15 to 20% VWC and the soil system remained oxygenated between samplings, thereby minimizing denitrification potential. Total leachate volume was determined gravimetrically and subsamples collected from each bottle were analyzed for ${\text{NO}}_3–\text{N}$ and thus total N loading rates could be calculated. Nitrate samples were analyzed using an air-segmented automated spectrophotometer (Flow Solution IV, OI Analytical, College Station, Tex.) coupled with a Cd reduction approach [modified U.S. EPA Method 353.2 (Jones and Case 1991)].

Statistical analyses were performed using PROC GLM procedure of SAS (SAS 2002) to determine treatment effects for total plant biomass and N accumulation, yield, IWUE, and NUE. When the $F$ value was significant, a multiple means comparison was performed using Duncan’s multiple range test at a $P$ value of 0.05. For repeated measurements such as percolated volume and nitrate leaching, the PROC MIXED procedure of SAS with residual maximum likelihood estimation approach and least-squares means of fixed effects were pairwise compared.

During the crop season, programmed irrigation events were skipped, which significantly reduced the amount of water applied to SMS based treatments. The overall volume of irrigation increased in order ${\text{SS}}_{10}<{\text{SS}}_{12}<\text{TIME}$ , except in 2006 when ${\text{SS}}_{10}$ received similar volume of irrigation water as ${\text{SS}}_{12}$ (Fig. 1). It was observed that both SMS treatments failed to bypass irrigation events in the beginning of the season. The locations of both treatment sensors were assessed 35 DAT and reburied in a different location in the bed. The location of the sensor relative to drip line and plant plays an important role in the sensing irrigation systems. Regardless that the treatments were set at different thresholds, if the ${\text{SS}}_{10}$ soil moisture probe was placed in a drier spot, it could result in higher irrigation volume applied than the ${\text{SS}}_{12}$ . However, after 35 DAT, both treatments were bypassing and irrigating the same scheduled events and applying same irrigation amounts. The problem was attributed to cross communication between the capacitance sensors, causing each of the irrigation controllers to receive signals from only one of the two wired sensors. Several adjustments were made, but the problem was not solved until each controller was wired to a separate individual irrigation timer, which occurred at 58 DAT, when the SMS treatments began to irrigate independently of each other (Fig. 1).

The contribution of rainfall to pepper water requirements was not considered in the calculations, due to the presence of plastic mulch and the absence of a perched water table, while coarse sandy soils also typically demonstrate very limited lateral flow. Although, it was observed that high intensity precipitation events $\left(>8\text{mm}{\text{h}}^{-1}\right)$ increased the soil water content as measured by TDR. For both SMS treatments, there was no increase in number of skipped events after rainfall. For example, precipitation events of 44, 48, and 45 mm occurring in 2005, 2006, and 2007, respectively [Figs. 2(d-f)], showed a slight increase (around 1%) in VWC in the 0 to 0.25-m depth layer (data not shown). The ${\text{SS}}_{10}$ treatment received an average of 1.0, 4.7, and $2.6\text{mm}{\text{day}}^{-1}$ of irrigation water in 2005, 2006, and 2007, respectively. For the same respective year order, the corresponding average irrigation rates for the ${\text{SS}}_{12}$ treatment were 2.2, 4.5, and $3.7\text{mm}{\text{day}}^{-1}$ and for the TIME treatment were 4.3, 5.1, and $4.4\text{mm}{\text{day}}^{-1}$ .

The soil moisture had a noticeable increase after each irrigation event for SMS treatments and TIME (Fig. 3). However, due to the reduced number of irrigation events for the ${\text{SS}}_{10}$ treatment, variations in soil moisture at 0–0.25 m soil depth layer were not as distinct as the ${\text{SS}}_{12}$ and TIME treatments. SMS-based irrigation treatments irrigated for short periods of time, in this case, 24 min, which resulted in a relatively small increase in soil moisture in the upper soil layer, consequently decreasing the volume of percolate [Figs. 3(b-d)]. In fact, slight oscillations in soil moisture were during the irrigation events observed in the deep monitored layer for ${\text{SS}}_{10}$ and ${\text{SS}}_{12}$ treatment. On the other hand, the TIME treatment was irrigated for a longer time period (2 h), which resulted in very pronounced soil moisture fluctuations [Figs. 3(e, f)]. For TIME treatment, after each irrigation event, the soil moisture content from 0.10 to $0.15{\text{m}}^3{\text{m}}^{-3}$ . Soil moisture content returned to field capacity within 12 h. The spikes also indicate that the soil water content as measured by the TDR probes rapidly reaches a point above the soil water holding capacity in the soil upper layer, inducing percolation to deeper soil layers and explaining the higher percolate values for the TIME treatment compared to the SMS treatments. In fact, similar spikes in soil water content were observed at 0.25–0.55 m showing appreciable soil water percolation though the soil profile throughout the entire production cycle. In terms of soil water availability to plants, the TIME treatment initially may provide more favorable growth conditions since the soil remains wetter, thus reducing potential water stress. However, excessive water percolation also may reduce N retention and crop N supply and thereby reduce yield for green bell pepper (Table 1).

There were no interactions between irrigation and N-rate treatments for total and marketable yield, maximum plant biomass accumulation, or IWUE (Table 1). The overall total yield for green bell pepper ranged between 26.7 to $32.0\text{Mg}{\text{ha}}^{-1}$ in 2005; 18.2 to $19.4\text{Mg}{\text{ha}}^{-1}$ in 2006; and 24.7 and $28.2\text{Mg}{\text{ha}}^{-1}$ in 2007. Except in 2006, when unfavorable environmental conditions occurred, bell pepper yield obtained in these experiments were in the range of those reported in the literature for sandy soils in Florida (Dukes et al. 2003; Maynard and Santos 2007; Simonne et al. 2006). The lower yield in 2006 compared to 2005 and 2007 was attributed to the effect air temperature on plant development and flowering. Low night temperatures were shown to have a considerable effect on flower morphology and functioning, larger flowers, with swollen ovaries and shorter styles in comparison with flowers grown under higher temperature conditions. This effect of low temperatures has a direct effect on pepper production by decreasing the total number of pollen grains formed and by reducing their viability and germination capacity (Aloni et al. 1999; Polowick and Sawhney 1985; Pressman et al. 2006; Shaked et al. 2004). According to Polowick and Sawhney (1985), nonextreme conditions such as $18°\text{C}$ day and $15°\text{C}$ night caused stamen deformation with abnormal, nonviable pollen, which resulted in functionally male-sterile flowers. A detailed analysis of measured air temperature during the entire crop cycle revealed that in 2006, pepper plants were exposed to temperatures below $14°\text{C}$ during 311 h, while in 2005 and 2007, the cumulative hours with low temperatures $\left(<14°\text{C}\right)$ were 181 and 185 h, respectively. In addition, temperatures below $14°\text{C}$ occurred during the entire plant development and reproduction stages in 2006 (Fig. 2), while in 2005, the low temperatures occurred exclusively during the beginning of the season (until 35 DAT). In 2007, low temperatures occurred throughout the season for short periods of time, however, between 49 and 63 DAT (peak of flowering stage) there was no occurrence of low temperatures. Air temperature may have affected the relation of marketable and total yield. In 2006, the weight of fruits classified as culls (total yield minus marketable yield) reached $3.4\text{Mg}{\text{ha}}^{-1}$ , which corresponded to 18% of total yield. Reduced proportion of culls was reported in 2005 and 2007, 10 and 3%, respectively (Table 1).

Differences in pepper yield between seasons also can be attributed to substantial differences in solar radiation. The cumulative solar radiation measured was 1,535, 1,482, and $1,849\text{MJ}{\text{m}}^{-2}$ for 2005, 2006, and 2007, respectively [Fig. 2(g-i)]. During the period between the end of May and beginning of June in 2005 and 2006, there was a reduction in solar radiation availability due to elevated number of cloudy/rainy days. Conversely, in 2007 the reduced number of rain events resulted in higher solar radiation availability compared to 2005 and 2006. The higher cumulative solar radiation in 2007 may have benefited pepper yield in response to the low temperatures occurred during reproductive phase compared to 2005, since both years yielded similarly.

There was no significant effect $\left(P\le 0.05\right)$ of N-rates on pepper yield. Only for 2007 conditions, a linear response of application of N-rates was observed (Table 1). The increase of N-rate from 176 to $220\text{kg}{\text{ha}}^{-1}$ and from 220 to $330\text{kg}{\text{ha}}^{-1}$ resulted in an increase of 6% (nonsignificant) of marketable yield. Complementally, the only statistical difference was observed between N-rates of 176 and $330\text{kg}{\text{ha}}^{-1}$ for that particular year. The lack of yield response to N-rates above $220\text{kg}{\text{ha}}^{-1}$ has been reported for other vegetable crops for drip irrigated production systems in Florida (Hochmuth and Cordasco 2000; Zotarelli et al. 2008; Zotarelli et al. 2009).

The use of SMS to control irrigation significantly affected the IWUE in 2005 and 2007 (Table 1). The treatment ranking for IWUE was as follows: ${\text{SS}}_{10}>{\text{SS}}_{12}>\text{TIME}$ . The TIME treatment had a lowest IWUE values $\left(<8\text{kg}{\text{m}}^{-3}\right)$ due to the high irrigation rates applied. In 2006, reduced yields associated to the high volume of irrigation applied for all treatments were responsible for the lower IWUE values ( $<5.7\text{kg}{\text{m}}^{-3}$ , Table 1). It is important to point out that high irrigation rates as applied for TIME did not increased yield, conversely, the use of scheduling irrigation by using SMS allowed application of less water, divided in five possible irrigation events per day (low volume, high frequency), which resulted in higher IWUE values. While TIME treatment had a single irrigation event (high volume, low frequency), which promotes excessive water percolation (Fig. 4).

Yield measurements showed that reduced irrigation volume application and proper irrigation scheduling showed potential to produced more pepper fruits with less irrigation water. In 2006, even with different irrigation scheduling (one single application for TIME versus five irrigation windows for ${\text{SS}}_{10}$ and ${\text{SS}}_{12}$ ), the volume of irrigation was very similar between treatments, therefore the pepper yield was identical. Differences between ${\text{SS}}_{10}$ and ${\text{SS}}_{12}$ in 2005 and 2007 can be also explained based on the estimated cumulative crop evapotranspiration and the cumulative irrigation water application for each treatment. When comparing pepper yield of treatments ${\text{SS}}_{10}$ and ${\text{SS}}_{12}$ in 2005, it is important to point out that after the establishment period (14 DAT), the cumulative irrigation volume for ${\text{SS}}_{10}$ was below the cumulative ${\text{ET}}_c$ (Fig. 1), while, the cumulative irrigation volume for ${\text{SS}}_{12}$ was always on or slightly above the ${\text{ET}}_c$ curve during the entire plant cycle including flowering (Fig. 1). Thus, yields for ${\text{SS}}_{12}$ were higher than ${\text{SS}}_{10}$ due to reduced irrigation water application for ${\text{SS}}_{10}$ , but no statistical differences in biomass accumulation were observed. However, in 2007, the cumulative irrigation for ${\text{SS}}_{10}$ was above the ${\text{ET}}_c$ curve until 50 DAT and slightly below after 50 DAT. On the other hand, ${\text{SS}}_{12}$ cumulative irrigation volume was always above the ${\text{ET}}_c$ curve. Thus, ${\text{SS}}_{10}$ had adequate water most of the season which optimized yields, whereas, yield on ${\text{SS}}_{12}$ was reduced due to excessive watering (and N leaching) which was observed on TIME in all years and ${\text{SS}}_{10}$ and ${\text{SS}}_{12}$ in 2006.

Shoot biomass accumulation ranged between 1.30 to $1.58\text{Mg}{\text{ha}}^{-1}$ in 2005, 0.81 to $0.84\text{Mg}{\text{ha}}^{-1}$ in 2006 and 1.08 to $1.30\text{Mg}{\text{ha}}^{-1}$ in 2007. There was no significant $\left(P\le 0.05\right)$ difference between SMS-based treatments and TIME treatments, neither between N-rates. However, in 2007 the fruit biomass accumulation was superior under ${\text{SS}}_{10}$ treatment (Table 1), following the same pattern for marketable yield. In the same year, the N-rates of 220 and $330\text{kg}{\text{ha}}^{-1}$ resulted in higher fruit and shoot biomass accumulation, however they did not differed between each other.

Pepper plant N nutrition was slightly enhanced by SMS-based irrigation systems and N accumulation in fruits and shoots were significant $\left(P\le 0.05\right)$ greater in 2005 and 2007 compared to TIME treatment (Table 1). The corresponding increase in shoot N uptake for SMS-based systems occurred for ${\text{SS}}_{12}$ in 2005 and ${\text{SS}}_{10}$ in 2007 in the order of 20 and 14%, respectively. Moreover, the overall N accumulation increased in the order of $2005>2007>2006$ , thus mirroring the overall plant dry weight accumulation and yield trends. Complementary measurements of plant N levels using petiole sap ${\text{NO}}_3–\text{N}$ status were taken at 46 DAT. Petiole sap ${\text{NO}}_3–\text{N}$ concentration of ${\text{SS}}_{10}$ $\left(1,648\text{mg}{\text{L}}^{-1}\right)$ and ${\text{SS}}_{12}$ $\left(1,353\text{mg}{\text{L}}^{-1}\right)$ were significantly $\left(P\le 0.05\right)$ higher than TIME $\left(1,061\text{mg}{\text{L}}^{-1}\right)$ treatment. The petiole sap ${\text{NO}}_3–\text{N}$ concentration found was within the range of optimum concentration reported Hochmuth (1994), except for TIME treatment, which showed a ${\text{NO}}_3–\text{N}$ sap concentration 20% below the lower limit. These differences ${\text{NO}}_3–\text{N}$ sap concentration among irrigation treatments corroborated with nitrate leaching patterns and NUE, in general, higher nitrate leaching resulted in lower nitrate sap concentration in the plant.

Except in 2006 when shoot and fruit N accumulation were not affected by the N-rate (Table 1), the N-rate of $176\text{kg}{\text{ha}}^{-1}$ showed reduced N shoot and fruit accumulation. In 2007, with more favorable climatic conditions, a linear response of biomass and N accumulation to the N-rate treatments was detected (Table 1). In other words, more favorable growth conditions resulted in higher N uptake capacity, increasing the N-rate from 176 to $220\text{kg}{\text{ha}}^{-1}$ and from 220 to $330\text{kg}{\text{ha}}^{-1}$ resulted in an overall increase of 10 to 20% in total N accumulation. The effect of N-rate treatments on petiole sap ${\text{NO}}_3–\text{N}$ concentration evaluated at flowering stage showed values of 1,141, 1,419, and $1,342\text{mg}{\text{L}}^{-1}$ for N-rates of 176, 220, and $330\text{kg}{\text{ha}}^{-1}$ , respectively, with no significant differences between N-rates of 220 and $330\text{kg}{\text{ha}}^{-1}$ . It is also important to note that there were no differences in fresh marketable yield between those three N-rates in 2005 and 2006 seasons and between N-rates of 220 and $330\text{kg}{\text{ha}}^{-1}$ in 2007. Results of petiole sap ${\text{NO}}_3–\text{N}$ concentration indicated that the N-rate of $330\text{kg}{\text{ha}}^{-1}$ may be exceeding the N of a pepper plant needs and possibly inducing N luxury consumption. Though, the extra consumption did not result in higher yields, in fact, for TIME treatment this N-rate showed higher ${\text{NO}}_3–\text{N}$ leaching.

NUE was significantly $\left(P\le 0.05\right)$ higher when lower N-rates were applied (Tables 2). The use of SMS resulted in higher NUE in 2005 for ${\text{SS}}_{10}$ and ${\text{SS}}_{12}$ , and ${\text{SS}}_{10}$ in 2007. Again, there was no interaction between N-rate and irrigation treatment on NUE. The NUE increased with the increase in yield level. The NUE values ranged between 33 and 42% in 2005, 14 and 20% in 2006, and 25% and 30% in 2007. Linear NUE values decreased with increasing N-rates (Table 1) may be related to limitation in uptake and sink capacities resulting in a saturation response (e.g., “law of the diminishing returns”). In other words, as N supply continues to increase, factors other than N limit growth and yield. In 2005 and 2007, the TIME treatment showed consistently lower values of NUE, which was related to the higher irrigation volumes being applied resulting in increased N dilution and displacement thus reducing N uptake efficiency (Scholberg et al. 2002). In addition, a significant increase in nitrate leaching was observed with the increase of N-rate.

Water percolation during crop establishment was identical for all treatments; 8, 10, and 8 mm for 2005, 2006, and 2007, respectively. According to the statistical analysis for the postestablishment period, an overall decrease $\left(P\le 0.05\right)$ in soil water percolation was obtained when SMS-based irrigation controlled the water application. The TIME treatment resulted in the highest volume of water percolated below the effective root zone and captured in the lysimeters (Fig. 4). The volume percolated ranged between 32 and 52 mm (Fig. 4), which corresponded to 10 to 16% of the applied irrigation water. Excluding 2006 season, the volume percolated under ${\text{SS}}_{10}$ treatment was 6 and 10 mm, in 2005 and 2007, respectively, which corresponded to only 5% of the total irrigation water applied. Similar comparison showed that ${\text{SS}}_{12}$ treatment percolated 21 and 24 mm in 2005 and 2007, respectively, which was translated to 9 and 11% of the total irrigation water applied, respectively.

There was no interaction between irrigation and N-rate treatments for cumulative nitrate loads below root zone. The TIME treatment resulted in the most ${\text{NO}}_3–\text{N}$ leaching. Cumulative ${\text{NO}}_3–\text{N}$ leaching values were ranged between 40 and $47\text{kg}{\text{ha}}^{-1}$ for TIME treatment (Table 2). The single high volume daily application of the TIME treatment is likely the cause of the appreciable drainage and ${\text{NO}}_3–\text{N}$ leaching below the rootzone compared to the irrigation scheduling. A consistent reduction in ${\text{NO}}_3–\text{N}$ leaching was observed when scheduling irrigation associated to the use of SMS was adopted. These reductions were on the order of 25 to 74%, which can be translated to range of 10 to $35\text{kg}{\text{ha}}^{-1}$ of N. Independently of the irrigation treatment, the increase in N-rate from 220 to $330\text{kg}{\text{ha}}^{-1}$ , significantly $\left(P\le 0.05\right)$ increased the ${\text{NO}}_3–\text{N}$ leaching in 2005 and 2006 (Table 2).