Monthly sampling and processing.

Samples were collected from water sources in MD, MS, and VA. The 10 water sources were labeled first by the state abbreviation, second by a sequential numbering of the nurseries sampled within a state (1 or 2), and third by sequential numbering of RCBs (1 to 3) connected in series by overflow culverts, as follows: MD11, MD21, MS11, VA10, VA11, VA12, VA13, VA21, VA22, and VA23. MD11 and MD21 receive irrigation runoff directly from the production area and are direct sources for irrigation water. MS11, VA11, and VA21 receive irrigation runoff directly from the production area and indirectly serve as a source for irrigation water. VA12 and VA23 are direct sources for irrigation water, and receive overflow second or third in series, respectively, from RCBs that directly receive runoff. VA13 and VA22 neither receive runoff directly nor serve as direct sources of irrigation water, but are connected to other RCBs that serve those functions. The stream (VA10) is a local natural water source that does not receive runoff from the nursery, thus used as a comparison with VA11, VA12, and VA13. Descriptions of MD and VA RCBs and stream have been previously published using the same label designation (Zhang et al., 2015a, 2015b, 2016). A description of MS11 also was previously described (Copes et al., 2016).

All nine RCBs and one stream (VA10) in MD, MS, and VA were sampled monthly from Feb. to July in 2013; however, water samples were not collected in every month from all 10 water sources, resulting in an incomplete design. MS11 was sampled for three additional months from Aug. to Oct. in 2013. A single set of water samples were collected monthly between 11 am and 1 pm at ≈0.3 m below the RCB surface. Multiple sets of water samples were obtained from five RCBs and one stream (MD11, MD21, MS11, VA12, VA13, and VA10) in either June or July in 2013 and from MS11 in September by collecting every 2 h from 8:00 am to 6:00 pm. Every sampling set consisted of three 1-L subsamples taken about 1 m apart from each other in a triangle pattern around a Sonde buoy located about in the center of the RCB. Water samples were collected in 1-L polycarbonate bottles in VA and MD and in 1-L polypropylene (PP) bottles in MS. All water samples were placed into an ice cooler for transport to the laboratory. Samples were divided into aliquots to nearly fill 20- and 100-mL high-density polyethylene (HDPE) bottles in MD and MS; and 45-mL PP and 100-mL HDPE bottles in VA. Water in the 20-mL HDPE and 45-mL PP bottles were acidified with 50 or 100 μL, respectively, of 32 n sulfuric acid to preserve NO₃⁻-N and NH₄⁺-N. All aliquots were stored at −20 °C until delivery to the analytical laboratory. All collection containers were previously treated with a 1.21 n hydrochloric acid wash to remove contaminants adhered to surfaces. Water analysis of monthly samples consisted of the determination of NO₃⁻-N, NH₄⁺-N, PO₄-P, T-P, potassium (K), calcium (Ca), magnesium (Mg), sulfur (S), aluminum (Al), B, Cu, Fe, manganese (Mn), Zn, pH, T-Alk, EC, and sodium (Na), and were performed by the University of Delaware Soils and Plant Analysis Laboratory, Newark, DE.

Gulf coast regional assessments.

Fifteen RCBs were sampled from 10 ornamental plant container production nurseries to assess water quality variation across the Gulf Coast region, including six RCBs from four nurseries in AL; four RCBs from three nurseries in MS; and five RCBs from three nurseries in LA. One liter of water was collected from about 0.2 m below the surface from about 2 m from the shore of each RCB. Water samples were collected in 1-L PP bottles between 9:30 am and 3:00 pm in Aug. 2014 and between 11 am and 1 pm in May 2016. Water samples were stored in an ice chest with cold packs for transportation back to the laboratory and divided into aliquots in 125 mL (sulfuric acid preserved), 125 mL (nitric acid preserved), and 250- and 500-mL HDPE bottles. Samples were stored at −20 °C in Aug. 2014 and at 4 °C in May 2016, because samples were collected, respectively, over 3 weeks in 2014 and within 7 d in 2016 before shipping for analysis. Water analysis of water samples consisted of NO₃⁻-N, NH₄⁺-N, total-nitrogen (T-N), PO₄-P, T-P, K, Fe, Mn, Zn, pH, T-Alk, EC, salinity, and biochemical oxygen demand performed by Brookside Laboratories, Inc., New Bremen, OH.

Data analysis.

Replications consisted of monthly means calculated from all subsets from a single or multiple sampling sets collected during a single day, with five or six replications per RCB and the stream. Monthly means were compared using analysis of variance with the MIXED procedure (version 9.4; SAS Institute Inc., Cary, NC). Fixed effect was waterways (nine RCBs and one stream) and random effect was replications. Analysis was performed on log-transformed mean water quality variables plus 0.001 to reduce heterogeneous variances that generally increased with increasing means. Goodness of fit was determined based on the smallest Akaike’s, corrected Akaike’s, and Sawa Bayesian information criterion values. The null hypothesis of all treatments being equal was tested using Tukey’s multiple comparison adjustment of least squares means (P = 0.05). The CORR procedure was used to compute Pearson correlation coefficients to measure the strength and direction of a linear relationship between all possible pairings of the 18 water quality variables by RCB. A P = 0.01 was used to select significant correlation coefficients.

Summary statistics for water samples taken in three states within the Gulf Coast region for Aug. 2014 and May 2016 consisted of the minimum, mean, and maximum of 19 variables by state.

Water quality assessments.

The means of all 18 water quality variables measured in the nine RCBs and one stream were within or below water quality ranges given in a University of Arkansas extension bulletin (Robbins, 2010). The guidelines given by Robbins (2010) stated for irrigating greenhouse and nursery crops were very similar to ranges stated by several water testing laboratories (not shown). The upper ranges of a few of the water quality variables did extend above the preferred level and would be considered high, but only marginally posed potential risk. Water quality variables from the nine RCBs and one stream were similar to those found in stormwater basins, but generally lower than those reported in a composite summary of several large natural rivers, which included the MS and Amazon rivers, although this later comparison is not directly applicable to RCBs (Chiandet and Xenopoulos, 2016; Hem, 1985). Without the availability of more detailed reference sources, the conclusion at this stage is water quality in the nine RCBs and one stream, based on the 18 water quality variables measured, should be suitable for crop production without detrimental impact.

Macronutrients comprise the highest percent of ingredients in fertilizers. These nutrients also tend to be the most labile. Based on the classification by Robbins (2010), NO₃⁻-N and T-P levels were low to moderate in all RCBs except for VA11 and VA12 where upper values extended above the preferred range (Table 1). Putting this in perspective, the upper levels recorded were still considerably less than rates recommended for promoting plant growth. Chen et al. (2003) also measured low levels of NO₃⁻-N (0.04 mg·L⁻¹) and T-P (0.87 mg·L⁻¹) in an RCB. An initial conclusion is RCBs have the chemical and biological capacity to convert and use N, P, and other macronutrient forms (Chiandet and Xenopoulos, 2016; Song et al., 2015). However, we lack a measure of how much stormwater flows dilute and flush nutrients from each RCB (Chiandet and Xenopoulos, 2016).

Chemical and biological conversions of N and P have been characterized in other waterways than RCBs, such as in stormwater basins. NO₃⁻-N is the most soluble and mobile form of N and denitrification is the main process converting it to other N-forms, such as nitrogen gas (N₂) that is released into the atmosphere (Burgin and Hamilton, 2007). Denitrification likely would be occurring in the anoxic zones of the RCB during summer stratification periods when fertilizer is most heavily applied. Phosphorus also undergoes multiple reactions. Song et al. (2015) measured low levels of P with a mean T-P of 0.056 mg·L⁻¹ in nine urban stormwater basins. Despite the low levels, Song et al. (2015) found P in multiple forms. The predominate form of dissolved organic P shifted from phosphomonoesters at the inflow to phosphodiesters at the outflow, whereas T-P was higher in the middle of the basin, which corresponds to the location of samples taken for this study. Particulate-P increased toward the outflow, likely associated with the assimilation of dissolved-P by algae and bacteria that converted it to particulate P forms. Phosphorus is considered a limiting nutrient for phytoplankton production in freshwater systems (Rabalais, 2001). Algae and phytoplankton communities have not been characterized in RCBs, although algal blooms are evident in many RCBs during summer months. Algae and phytoplankton colonizing soil at the bottom of RCBs (benthic) and vertical zones of the water body (sestonic) would use these nutrients. Morphometric pond variables (pond depth, length:width ratio, and area:perimeter ratio) can serve as predictors of water quality that affect N and T-P levels (Chiandet and Xenopoulos, 2016). Stormwater ponds with simple circular shapes and without forebays have higher levels of T-P than deeper ponds with sinuous shorelines. Most of the RCBs in ornamental plant nurseries have an intermediate irregularity to their oval shape, thus few of them have all the morphometric features deemed beneficial. Low DO levels in bottom water layers can also promote release of P from sediments (Chiandet and Xenopoulos, 2016). Understanding the chemical and biological cycling of where nutrients accumulate, under what conditions and in what form, along with hydrological cycling involving rain, evaporation and stormwater flow, and capacity of the system is still evolving.

Micronutrient levels followed the same trend as macronutrients of being below or at the lower part of the preferred range listed by Robbins (2010). Chen et al. (2003) also measured micronutrients in the lower preferred range. In our study, samples were only collected from the upper horizontal water strata (epilimnion) in the center of RCBs. Water quality levels in other zones of the RCB, such as the lower horizontal water strata (hypolimnion) were not measured. A larger national sampling of RCBs in more regions might find levels of Fe and Mn to be high enough to negatively affect water quality. Although Cu levels were low in the epilimnion of the RCBs sampled, Cu levels could differ in other zones, such as the hypolimnion. Cu is relatively immobile if assimilated by benthic biota and if bound to soil in RCB bottoms. Some of the di- and trivalent micronutrients, such as Al, Cu, Mn, and Zn, have potential to be toxic to Phytophthora zoospores. For example, if Cu levels attained the upper scale of the preferred range (0.15 mg Cu/L) (Table 3) zoospores of many Phytophthora sp. might be inactivated (Erwin and Ribeiro, 1996).

The nurseries in VA provided an interesting comparison due to differences in fertilizer delivery and water overflowing into successive RCBs connected in series. Fertilizer at VA1 is delivered in a soluble form as liquid fertigation on a regular schedule. Water quality variables in VA11 ranged from the upper part of the preferred range to a little above that level (Tables 1–5). A concern exists whether this can create an imbalance with applied fertilizer rates. P toxicity has been reported for some plants, such as Ilex crenata, when a higher than optimum level of P resulted from a combination of fertilizer in the plant media and from irrigation water when K levels were low (Wright and Niemiera, 1987). However, fairly dramatic reductions in many of the water quality variables occurred from VA11 and VA12 to VA13. More of the water quality levels were within the preferred range in VA12 from which water is pumped for irrigation (Tables 1–5). In contrast, fertilizer at VA2 is delivered to plants from control-release fertilizer. Water quality variable levels in VA21 ranged from low to within the preferred range. Few differences in water quality variable levels existed between RCBs VA21, VA22, and VA23 (Tables 1–5). These limited comparisons seem to demonstrate that a single RCB has more capacity to use nutrients when they are low, but the capacity of that system may be challenged when nutrient levels are higher than the preferred range. The limited comparison is extended to show the possibility that multiple RCBs in series provide added capacity to lower nutrient levels in successive RCBs.

The fact that nutrients were low in many of the RCBs shows plant toxicity was not an issue; furthermore, some RCBs do not provide a measurable contribution to plant fertility. An issue that needs further consideration is determining whether elevated levels of nutrients that are still below fertility recommendations contribute an additive effect that could cause nutritional imbalances in plant fertility.

Mean pH measurements varied between RCBs and the range varied within RCBs (Table 5). Mean pH measurements were within the preferred pH range (5.2 to 6.8 pH units) for two RCBs (VA11 and VA12); higher than preferred but below an upper acceptable level (8.0 pH units) for four RCBs (MD21, VA13, VA21, and VA23) and one stream (VA10); and above 8.0 pH units for three RCBs (MD11, MD21, and MS11). Table 5 gives the impression that RCBs in MS and MD might have higher pH values, yet Table 6 shows a larger range of pH ranges occurred when a greater number of RCBs were sampled in AL, LA, and MS, similar to the range in MD, MS, and VA. Most pH values in this study were taken about near noon to minimize diurnal influences (Zhang et al., 2016). The diurnal fluctuation in pH of surface water, as well as DO, is related to respiration and photosynthetic activities of algal and phytoplankton populations (Hong et al., 2009). Annual pH ranges would still be influenced by other seasonal and chemical aspects of individual RCBs, such as alkalinity.

RCBs in this study were characterized predominately by very low to low T-Alk means, low to moderate EC means, and low Na or salinity means (Tables 5 and 6). Alkalinity is an acid-neutralizing capacity of carbonates in water to buffer pH changes, whereby more rapid and wider pH fluctuations will occur with lower T-Alk levels. EC is the overall EC generated by all ions. The lower EC values indicate RCBs have the capacity to absorb the EC levels delivered in runoff, although as stated the influence from stormwater is unknown. Chen et al. (2003) recorded similar measurements for pH, T-Alk, and EC in a research-constructed RCB. Na levels were not a problem in most of the RCBs. Na was at the upper part of the range for water samples collected from VA22 and VA23, thus an issue that needs monitoring.

Fifteen RCBs were sampled in southern AL, southern MS, and southeastern LA to measure diversity of water quality across the Gulf Coast region (Table 6). Ranges for the 14 water qualities measured in AL, LA, and MS were similar to ranges from MD, MS, and VA, which provide further evidence that water quality variables measured may be acceptable as an irrigation water source from many RCBs. Yet, the spotty detection of correlations in Table 7 shows no prominent repeatable effect in RCBs and indicates a complexity to these systems might exist that makes it difficult to generalize about potential interactions.

Management considerations.

In the studies being reported, levels of water quality variables ranged from low to preferable levels, thus indicating good irrigation water quality for the variables tested. Even though Robbins (2010) listed the ranges as preferred, the basis for these values have not been thoroughly researched. Chen et al. (2003) showed commercial marketable plants were produced using RCBs as a source for irrigation water. The RCBs tested in AL, LA, MD, MS, and VA in this study demonstrate that same result, as these RCBs are actively used as a source of irrigation without any noticeable reduction in plant quality. Chen et al. (2004) produced plant quality equal to or better than plants watered from a well. In contrast, Meador et al. (2012) identified a lower water quality from 24 RCBs in nine states than from municipal and well water sources based on total dissolved solids (TDS). Concerns about TDS include clogging of emitters and interference with water sanitation treatments, such as ultraviolet light. These problems do exist, however, they are not derived solely from TDS levels, but also from mineral salts accretion and biofilm growth in irrigation lines.

Accumulation and movement of solutes in RCBs have not yet been incorporated as a fertility management tool. It may not always be clear whether water quality from an RCB is negatively or positively affecting crop quality. Grower concern exists whereby water from some RCBs is applied only on hardier plants. The problem is RCBs are complex dynamic systems that will require understanding interactions between water chemistry processes, sestonic and benthic biological activity, and weather influences. Because of the importance of water, the RCB capacity and water quality management need to be better understood (Wilson et al., 2010).

Several RCBs had pH values at 8.0 or above and could require acidification treatment before applying certain disinfestant treatments, such as chlorine gas, peroxygen compounds, and ozone. Fortunately, T-Alk levels were low enough that only low acid rates would be required in these specific cases. As an alternative approach, water could be drawn during morning hours when water is more acidic, which occurs due to a diurnal pH change in response to the metabolic activity of algae (Hong et al., 2009). Additional opportunity exists by drawing water from deeper depths where it is more acidic than surface water during stratification periods from April to October (Zhang et al., 2016). Zhang et al. (2016) demonstrated the occurrence of stratification in some of the same RCBs (MD11, MD21, VA11, VA12, VA13, VA21, VA22, and VA23) sampled in 2013. High pH levels could potentially cause an interaction with plant nutrient solubility, particularly micronutrients such as Fe and Zn. However, low water alkalinity values allow for rapid pH change, as a result buffered organic growing media would readily compensate and shift water to the desired pH to allow nutrient availability and uptake by plants (Chen et al., 2004). In such cases, plants should be monitored for micronutrient deficiencies and fertilized according to extension recommendations.

Influence on plant pathogens.

Dispersal of plant pathogenic Phytophthora sp. in irrigation water is well documented and a primary concern of plant industries (Ghimire et al., 2011; Loyd et al., 2014; Parke et al., 2014). Water quality has been shown to reduce survival of several plant pathogenic Phytophthora sp. to as short as 1 d (Kong and Hong, 2014; Kong et al., 2012a, 2012b; Kong and Lea-Cox, 2014). Information about how Phytophthora sp. survive and traverse the aquatic environment of RCBs is in its infancy. Of the RCBs in this study, even the higher mean EC values of 0.23 to 0.36 mmho·cm⁻¹ (0.23 to 0.36 dS·m⁻¹) would be unfavorable for the long-term survival of Phytophthora alni, Phytophthora kernoviae, and Phytophthora ramorum zoospores that favored an EC >1.89 dS·m⁻¹ (Kong et al., 2012a). Chen et al. (2003) also measured ECs lower than 0.6 mmho·cm⁻¹. These results are encouraging, yet too few Phytophthora species have been evaluated to understand the larger ecological implications of EC on zoospore longevity and how variable the response might be by Phytophthora species. The influence of pH on zoospore longevity varies by Phytophthora species. Of Phytophthora species that have been evaluated, preferred pH for zoospore longevity over 7 d was pH 5 to 11 for P. alni and P. ramorum; pH 9 for Phytophthora citricola; pH 7 for Phytophthora citrophthora, Phytophthora insolita, Phytophthora irrigata, Phytophthora megasperma, and Phytophthora nicotianae; pH 5 for Phytophthora tropicalis; and pH 3 to 9 for P. kernoviae (Kong et al., 2009, 2012b). Since RCBs vary in pH means and standard errors (Table 5), this brings into question whether survivability of Phytophthora sp. will be more favored in some RCBs than in others. Higher preferred levels of some micronutrients (Tables 3 and 4) could further reduce the survivability of zoospores without being detrimental to plant growth. It is known that some Phytophthora species are commonly recovered in RCBs, such as Phytophthora gonapodyides, Phytophthora hydropathica, P. irrigata, and Phytophthora chlamydospora (Copes et al., 2016; Ghimire et al., 2011; Loyd et al., 2014; Parke et al., 2014), these species may be more tolerant of low ECs and a wider pH range than the more virulent plant pathogenic Phytophthora species reported by Kong et al. (2012a). A picture is developing that shows various water quality variables (EC, DO, and pH) could affect the population of Phytophthora species in RCBs. Additional research is needed to better understand the potential of zoospore survival within the main water body and niche habitats of RCBs, to characterize the risk of pathogen spread in irrigation water and to better understand the potential for developing management techniques that limit pathogen dispersal.