Assessing micro-irrigation clogging risk through water quality classification systems

Clogged irrigation systems affect irrigation uniformity (Capra & Scicolone 1998), the emitter-discharge rate (Ravina et al. 1992), crop yields (Sadeh & Ravina 2000), and the useful life of irrigation equipment (Gilbert et al. 1981), which leads to crop losses and increased labor and production costs (Raudales et al. 2017). Clogging is mainly caused by the accumulation of suspended particles such as chemical precipitates (e.g., calcium, magnesium, iron, and manganese), minerals (e.g., silt), organic matter, or the accumulation of biofilm and algae (Bucks et al. 1979; Adin & Sacks 1991; Capra & Scicolone 1998). Therefore, monitoring water quality and implementing an adequate clogging risk assessment is critical to establish management strategies that prevent crop loss and mitigate the economic risks associated with clogged irrigation systems.

Two water quality classification systems assess the overall risk of clogging and classify water sources by risk level in drip irrigation systems. The system reported by Bucks et al. (1979) – from now on referred to as the Bucks system – works by rating water sources based on the concentration of chemical, physical, and biological parameters and then estimating a combined score (Supplemental Material 1). This system, or some of its components, has been adopted by the Food and Agriculture Organization (FAO) (Ayers & Westcot 1985), the United States Department of Agriculture – Natural Resources Conservation Service (USDA-NRCS 2013), and several university extension publications (Storlie 1995; Hanson et al. 1997; Rogers et al. 2003; Grabow et al. 2005; Shortridge & Benham 2018) to evaluate clogging risks in micro-irrigation systems. Although some publications referenced Hanson et al. (1997) and Hassan (1998) as the source for the guidelines, their guidelines can be traced back to the classification system proposed by Bucks et al. (1979).

The system reported by Capra & Scicolone (1998) – from now on referred to as Capra & Scicolone system – works by assigning a clogging hazard rating of ‘Minor, Moderate, or Severe’ based on the electrical conductivity (EC) and the concentration of total suspended solids (TSS), iron, manganese, calcium, and magnesium (Supplemental Material 2). Details on how to use the rating systems are outlined in the Methods section of this article. Both classification systems were developed based on a combination of observational data – samples collected on farms, and the researchers' experience – instead of data from controlled experiments. Moreover, there is limited evidence that classification systems may be applied as a tool to identify and predict the causes of clogging. Identifying the source of clogging may help develop robust water quality thresholds and assist operators and system designers in making decisions to effectively manage water to minimize clogging risks.

Our objective was to evaluate if the existing water quality classification systems could be used to identify the cause of clogging in micro-irrigation systems in greenhouses. In this project, we sampled water at commercial greenhouse operations that reported frequent clogging of irrigation systems. Where applicable, we rated the water parameter values using the Bucks and Capra & Scicolone water classification systems.

Randomized and equal probability sampling was not possible because we were looking for greenhouse operations that had persistent and complete clogging of irrigation emitters in Connecticut, USA. We selected eight commercial operations producing ornamental crops in containers using different water sources and reporting different clogging points in their operations. Sampling occurred once annually, between late spring and summer, 2015 and 2017. Four operations were sampled in April to May 2015, one in June 2016 and eight in July to Aug. 2017. The description of each location is listed in Table 1.

To understand the nature of clogging and the design of irrigation water distribution systems, we conducted informal interviews and conversations with the growers. We inquired about their water sources, the points and frequency of clogging, and the availability of any water treatment at their farms. Table 1 summarizes the information gathered through these interviews, conversations, and walkarounds.

Three 1-L water samples were collected directly from either the source or closest outlet to the source and from the farthest emitter in each location. Location B had two water sources, the well water and the recirculated nutrient solution that were sampled after water treatment, which included multistage filtration (membrane filter, fiber media, and granular-activated carbon filters), peracetic acid injection, and iron-chelates injection. For all samples, water ran for at least 1 min before sample collection. We kept the samples at 4 °C and processed them within 24 h of collection.

We shipped the samples to a commercial analytical laboratory to measure pH, EC, concentration of calcium (Ca²⁺), magnesium (Mg²⁺), bicarbonate (⁠⁠), calcium carbonate (CaCO₃), iron (Fe²⁺), and manganese (Mn²⁺), among the standard nutrients measured for plant growth. We estimated the total dissolved solids (TDS) 640 scale by multiplying the EC in mS·cm⁻¹ by a constant of 640 (Corwin & Yemoto 2020). In 2017, we measured total and soluble iron using the Orion™ AQUAfast™ AC4078 kits and the Orion™ AQUAfast™ AQ4000 colorimeter (Thermo Fisher Scientific Inc., Beverly, MA). The difference between total and soluble iron indicated the concentration of insoluble iron. We measured TSS using the American Public Health Association (APHA) standard method 2540 D (APHA 2012). We used the APHA standard method 9215 C heterotrophic plate counts (HPC) to enumerate the total amount of culturable aerobic bacteria in the water samples.

We rated the risk of emitter clogging of each sample according to the Bucks and the Capra & Scicolone systems. The Bucks system (Supplemental Material 1) ratings were conducted as follows: (1) physical, chemical, and biological parameters were rated for their clogging risk from 0 (low risk) to 10 (high risk) depending on the concentration in the samples; (2) for the chemical contributors to clogging (dissolved solids, iron, and manganese), the parameter with the highest rating was selected to represent the chemical category rating; (3) if the water pH was above 7.5, then the iron/manganese rating was multiplied by two; (4) a combined score was created by adding the ‘Physical–Chemical–Biological’ independent scores. A combined score below 10 (e.g., ‘0–0–0’) represented a low risk of emitter clogging. A combined score between 10 and 20 (e.g., ‘0–10–5’) was considered a moderate risk of emitter clogging, and a combined score between 20 and 30 (e.g., ‘10–10–10’) was considered a high risk of emitter clogging.

The Capra & Scicolone system (Supplemental Material 2) rates clogging hazards as minor, moderate, or severe according to specific levels of chemical and physical water quality parameters. We compared the ratings of the samples from the source and emitters to observe how water quality changed as it flowed through the irrigation system in each location.

In 2017, we extracted DNA from microorganisms in the water from locations A, B, E, F, G, and H. Briefly, up to 1 L of the water passed through a 0.22 μm pore size hydrophilic polyethersulfone membrane filter (Supor^® PES, Pall Corporation, Port Washington, NY), then DNA was extracted from microorganisms on the filter using the DNeasy PowerWater kit (QIAGEN GmbH, Germany). The University of Connecticut Microbial Analysis, Resources, and Services (MARS) laboratory conducted library preparation and sequencing. The DNA was quantified using the Quant-iT PicoGreen kit (Thermo Fisher Scientific Inc.). Bacterial 16S rRNA (V4) and fungal ITS2 genes were amplified using 30 ng of DNA. The V4 region was amplified using 515F (5′-GTGCCAGCMGCCGCGGTAA-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) primers with Illumina adapters and dual indices – 8 basepair Golay on 3′ (Caporaso et al. 2012) and 8 basepair on the 5′ (Kozich et al. 2013). The ITS2 region was amplified with ITS3 (5′-GCATCGATGAAGAACGCAGC-3′) and ITS4 (5′-TCCTCCGCTTATTGATATGC-3′) primers (White et al. 1990) using the same dual indexing design as the V4 region. Three subsamples were amplified using Accuprime PFX PCR master mix (Thermo Fisher Scientific Inc.) with the addition of 10 μg bovine serum albumin (BSA). (New England BioLabs, Ipswich, MA). The polymerase chain reaction (PCR) conditions consisted of 95 °C for 2 min, then 30 cycles of 15 s at 95.0 °C, 1 min at 55 °C, 1 min at 68 °C, and a final extension at 68 °C for 5 min. PCR products were pooled for quantification and visualization using the QIAxcel DNA Fast Analysis (QIAGEN GmbH, Germany). PCR products were normalized based on the concentration of DNA from 250 to 400 bp and then pooled using the QIAgility liquid handling robot. Combined PCR products were cleaned with Mag-Bind RXNPure Plus (Omega bio-tek, Norcross, GA) and then sequenced on the MiSeq using a v2 2 × 250 base pair kit (Illumina, Inc., San Diego, CA).

Sequences were demultiplexed using onboard bcl2fastq and then processed in Mothur v. 1.39.4 following the MiSeq SOP (Kozich et al. 2013). Specific commands can be found here https://github.com/krmaas/bioinformatics/blob/master/mothur.batch. Merged sequences that had any ambiguities or did not meet length expectations were removed. Sequences were aligned to the Silva nr_v119 alignment (Quast et al. 2013). Taxonomic identification of OTUs was done using the RDP Bayesian classifier (Wang et al. 2007) against the Silva nr_v119 taxonomy database. Alpha (inverse Simpson index) and beta diversity (Bray–Curtis dissimilarity) statistics were calculated by taking the average of 1,000 random subsampling to 10,000 reads per sample in Mothur. Nonmetric multidimensional scaling (NMDS) and permutational multivariate analysis (PERMANOVA) were conducted using the vegan package (Oksanen et al. 2015) in R 3.3.2. A subsampled species matrix was used for indicator species analysis (de Cáceres & Legendre 2009). Figures were drawn in R 3.3.2 using ggplot2 2.2.1 (Wickham 2011) and RcolorBrewer 1.1-2.

Even though all locations reported a high incidence of clogging, water source samples were classified as a low to moderate risk of clogging based on both classification systems (Table 2). Well water sources were classified as a low risk of clogging, mainly associated with chemical factors, except in location C where manganese levels led to a moderate risk rating. The municipal treatment facility water source was classified as a low risk of clogging. Pond water sources were classified as a moderate risk of clogging, mainly attributed to bacteria in both locations G and H plus TSS in location G. Based on the classification systems, it was not possible to identify the causes of clogging in locations A, B (well water sources), D, and E.

Recirculated irrigation water in location B was rated as low quality with a high risk of clogging mostly attributed to high bacterial counts and high mineral levels from dissolved fertilizers (Table 2). Recirculating water is a strategy to reduce nutrient runoff into the environment, save water, and lower costs (DeVicentis et al. 2015; Raudales et al. 2017), with the added risk of clogging (Table 2 and Supplemental Material 3). In general, recirculated water has more microbial load, nutrients from fertilizers, and organic matter. It remains to be determined whether the increases in microbial load are beneficial or detrimental for plant health or irrigation efficiency.

The grower in location B reported that screen filters and emitters were clogged with recirculated nutrient solutions despite using a multiple-stage water treatment system. Capra & Scicolone (2007) indicated that clogging prevention depends on the type of filter, water quality, and emitter. They observed that vertical gravel media filters (1.5 mm crushed granite with a 20 m³·h⁻¹ flow rate) resulted in high emitter performance and lower clogging incidence and severity, while screen filters were clogged and did not prevent emitter clogging. Liu & Huang (2009), Oliver et al. (2014), and Zhou et al. (2017) found that emitters with discharges above 2 L·h⁻¹ and pathways wider than 1.5 mm were less sensitive to clogging when using urban wastewater. The evidence from the literature suggests that the risk of clogging depends on the design of the irrigation system, not water quality alone. Therefore, understanding the risks of clogging must involve modeling water quality compatibility with the design of the distribution and treatment systems. In other words, prevention of micro-irrigation emitters requires designing water treatment and distribution systems that match the characteristics of the water source.

The risk of clogging ratings increased from the source to the emitters in locations A, B, C, E, and F because of a high number of bacteria and mineral content from dissolved fertilizers (Table 2 and Supplemental Material 3). The risk of clogging ratings for the treated recirculated solution did not change from the source to the emitters (Table 2). Our observations are similar to those of Meador et al. (2012) who reported a higher number of bacteria at the farthest outlet compared with the source, thus increasing the risk of biological clogging based on the ratings. The organic matter, nutrients, biofilms inside pipes, and irrigation scheduling may influence planktonic bacteria in water. Ravina et al. (1992) and Zhou et al. (2013) observed that emitter clogging is a gradual process for which the biofilm and microbial growth inside the pipes is the leading cause, with emitters in the middle or end part of an irrigation line being more prone to clogging than those at the head of the irrigation line. These observations suggest that the environment inside the irrigation system is favorable for microbial growth and water treatments with residual effects are required to prevent an increase in biological clogging risk as water reaches the farthest emitters.

Bucks et al. (1979), Capra & Scicolone (1998), and Shinde et al. (2012) argue that the risk of chemical clogging increases when fertilizers are added to the nutrient solution. Bozkurt & Bulent (2006) reported that drip emitters are susceptible to clogging when using fertilizers that supply calcium and sulfates. However, fertigation is a standard practice in greenhouses, and not all systems experience clogged emitters, suggesting the existing classification systems may overestimate the risk of chemical clogging when using fertigation. In addition to water quality parameters, meta-analysis studies have shown that emitter design features (such as flow path length, width, and depth) and operating time can influence the risk of clogging (Liu et al. 2022; Lv et al. 2024). Although the meta-analyses did not account for biological water quality indexes, Liu et al. (2022) highlighted the potential interactions of microorganisms with the studied water quality parameters. Their results indicate that Fe and Mn promote extracellular polymers (EPS) and biofilm formation in clinical settings, and highlighted that TSS particles can serve as sites for microbial aggregation, growth, and EPS production, which further increase TSS concentrations. Lv et al. (2024) stated that further research should evaluate the impact of interactions between microorganisms and water quality on clogging. Therefore, it is suggested that classification systems should include indicators that consider specific interactions between water quality parameters that lead to chemical precipitation and biological clogging rather than looking at the concentration of individual elements.

The risk of clogging ratings for pond water sources decreased from the source to the emitters because TSS and bacteria were lower at the emitters than at the source (Table 2). This reduction could be caused by filtration (0.15 mm screen filters in location G) and settling of particles in the storage tank (location H). Bacteria may attach to suspended particles (Jeng et al. 2005), which emphasizes the importance of removing suspended particles as part of an integrated treatment approach to reduce risks associated with microbial loads such as biofilm, algae, and pathogens. The interaction between TSS and the EPS secreted by microorganisms has been linked as a possible cause of drip emitter clogging (Zhou et al. 2013; Oliver et al. 2014) and should be explored to improve the reliability of the classification systems.

The classification systems failed to identify the sources of clogging. The following case studies are provided to discuss why the systems failed and to propose ways to improve the systems to accurately identify the sources of clogging.

Locations B and E were classified as having a low risk of clogging (Table 2) despite the visible filamentous growth blocking screen filters, drip emitters, and tubing. Well water in location B contained cyanobacteria (Obscuribacteriales and Elstera), iron-oxidizing bacteria (Gallionella), biofilm-forming bacteria (Hymenobacter and Phenylobacterium), and floc-forming bacteria (Zooglea) that may have contributed to biological clogging (Figure 4). The water in location E had low numbers of aerobic bacteria (Table 2 and Supplemental Material 3), but the samples had bacteria taxa with phenotypic adaptations that may have contributed to clogging. These adaptations include photosynthetic cyanobacteria (Chromatiales and Obscuribacteriales), iron oxidation (TRA3-20 unclassified and Acidiferrobacter), biofilm formation (Rhizomicrobium), and floc formation (Crenothrix, Singulisphaera, and Zooglea) (Figure 4). These observations suggest that the presence and abundance of specific bacteria in the irrigation water may be a more relevant indicator of clogging than the number of aerobic bacteria, which has been the sole indicator to assess the biological risk of clogging. The HPC method quantifies only culturable organisms and gives no indication of bacterial phenotypic characteristics. Therefore, the number alone is not a reliable indicator to assess the risk of clogging. Finding a single molecular or biochemical marker that screens for biological clogging–related traits is challenging due to the redundancy and multiplicity of genes and pathways involved in iron oxidation (Bonnefoy & Holmes 2012) and biofilm formation (Beloin & Ghigo 2005; Karatan & Watnick 2009) in bacteria. Therefore, additional research is needed to determine common microbial taxa among a wide array of clogged systems, which could lead to the identification of potential genetic and phenotypic markers. Furthermore, this research can aid in the development of methods that can be used to quantify biological clogging risks in irrigation systems.

Management practices change the quality of irrigation water in the distribution system, which, in consequence, may affect the risk of emitter clogging. The risk of clogging in location F rated low and moderate at the source and emitter, respectively (Table 2). The rating changed at the emitter because the number of bacteria, TDS, and magnesium levels increased relative to the source water (Table 2 and Supplemental Material 3). This location had a constant injection of fertilizers, which explains the increase in the nutrient levels and consequential higher chemical risk of clogging rating at the emitter (Table 2 and Supplemental Material 3). The nutrient solution was prepared in an open container that was placed in a ground-level trench, pumped to storage tanks, and passed through a 60-mesh screen filter with a transparent housing. A green filamentous growth was visible inside the screen filter. The nutrient solution at the emitter contained photosynthetic, biofilm-forming (Rhodobacter and Rhizomicrobium), floc-forming (Gordonia), and surface-attaching (Asticcacaulis and Prosthecobacter) bacteria (Figure 4). The biological risk of clogging rating increased because of possible contamination in the open mixing tank and the transparent screen filter housing that allowed light transmission, resulting in the growth of photosynthetic organisms. These results show that inadequate management practices can decrease water quality and increase the risk of clogging.

We suspected that iron build-up was the cause of clogging in locations A and C because we observed rust stains on emitters and inside the housing of disk filters. However, the samples in location A rated low for chemical clogging, while samples in location C rated moderate for manganese clogging (Table 2). The results from the commercial laboratory test showed no iron present in the irrigation water in 2015 (Supplemental Material 3). In 2017, we measured total soluble (iron II) and insoluble (iron III) iron along with the commercial laboratory tests for all locations. The measured and laboratory-reported levels of iron were 0.8 and 0.3 mg·L⁻¹ for location A, and 2.5 and 0.2 mg·L⁻¹ for location B. With these concentrations of insoluble iron, the samples were classified as having a moderate to high risk of chemical clogging. We also detected iron-oxidizing bacteria (Acidiferrobacter, Ferriphaselus, Gallionella, Sideroxydans, and TRA3-20 unclassified) in the water samples from location A (Figure 4) at near-neutral pH (Supplemental Material 3). This is in accordance with observations made by de Vet et al. (2011) in which Gallionella spp. grew and accumulated inside oxidation filters at pH from 6.5 to 7.73 and in water with oxygen at near saturation. This suggests that iron-oxidizing bacteria play an important role in the precipitation of iron that results in clogging, highlighting the importance of more specific qualitative and quantitative assessments of water quality.

Iron precipitation depends on interactions between chemical and biological processes. The standard water suitability tests for irrigation water in horticultural operations only measure the available iron for the plants, not the total iron or biological oxidation of iron. Therefore, models capable of estimating the biological and chemical oxidation of iron should be tested to determine the clogging risk of iron in water.

In location D, a white mineral precipitate clogged the emitter in an irrigation boom. However, the irrigation water in location D was classified as having a low risk of clogging (Table 2). The water was classified as moderately hard based on the levels of CaCO₃ and according to the US Geological Survey 2016 (Supplemental Material 3). This suggests that the degree of chemical precipitation depends on the interaction between multiple parameters rather than the concentration of individual elements. A single indicator that explains these interactions is needed to estimate the chemical clogging risks. For example, the Langelier saturation index (LSI) estimates the degree of saturation of water with respect to calcium carbonate by considering the temperature, alkalinity, ionic strength, TDS, and pH of a sample (Bower et al. 1965). LSI values between − 2 and 0 indicate that CaCO₃ is likely to remain in solution, and values between 0 and 2 indicate that CaCO₃ precipitation may occur (Rafferty 1999). The LSI for this location was below 0, suggesting no CaCO₃ precipitation. However, the sample at the emitter had no dissolved fertilizers, and we do not know the temperature of the water inside the pipes at the moment of sample collection. The classification systems can include the LSI as an indicator of scale formation to assess the chemical clogging risk of irrigation water. It is important to determine the levels of LSI at which emitter performance is affected.

The classification systems recommended by major international organizations were a useful tool to compare the relative risks between samples in the same location (i.e., from source to emitter and between water sources). However, they were not effective as a standalone predictor of clogging or as a tool for identifying the specific sources of clogging. The systems did not provide enough insight into the specific cause of the clogging or hazard level. The major limitations of the classification systems were that their assumptions do not consider interactions among chemical and microbial factors or the qualitative characteristics of microbial communities and their assumptions overemphasize the value of chemical clogging – even when levels within the severe range included levels typically applied in fertigation in greenhouses. A reliable classification system serves as a valuable decision-making tool for practitioners to assess water source suitability and develop water quality management strategies aimed at mitigating the risk of emitter clogging. This supports in minimizing the economic impact of clogged irrigation systems. The classification system can be effectively used in tandem with real-time on-field clogging detection methods, such as the evaluation of emitter performance data and operating pressure.

Our observations indicate that assessing clogging risk based solely on the concentration of physical, chemical, and biological indexes can lead to unreliable results. To improve classification systems, it is essential for researchers to develop new indicators that account for interactions between water quality parameters and environmental conditions that lead to clogging. It is crucial to establish the relationships between these new water quality indexes and performance indicators for various categories of emitters.

Furthermore, relying solely on HPC to assess biological clogging risk is not advisable. Our results highlight the need to identify methods that screen for specific phenotypes that contribute to clogging, such as microbial surface attachment, biological iron oxidation, filamentous growth, and production of flocs, EPS, and biofilm. Research should also determine the levels at which these indexes may affect irrigation performance and may impact the economic and technical viability of farms. Identifying the specific cause of clogging is important for growers who aim to prevent issues and minimize economic risks with targeted treatment options.

This work was supported by Critical Agricultural Research and Extension grant no. 11947449 from the USDA National Institute of Food and Agriculture and from the USDA-Agricultural Research Services Floriculture Research Initiative. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the view of the US Department of Agriculture. A partial modified version of this article was published in 2019 as part of a Ph.D. dissertation by the primary author in fulfilling a degree requirement (Cabrera Garcia 2019).

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.