Electrical conductivity as a proxy measurement for nitrogen

In the regulatory context, California dairies must account for the nitrogen content of the manure generated on, applied to, and removed from their operations. Nitrogen is an essential macronutrient for plant growth, yet it is difficult to manage for environmental outcomes because its form, availability, and susceptibility to loss depend on manure composition, weather, soil properties, and management practices.

Laboratory analyses provide accurate measurements of manure N, but they are periodic and do not offer real-time feedback. Electrical conductivity (EC), by contrast, can be measured continuously in the field using commercially available sensors. This article evaluates whether EC can serve as a practical proxy for nitrogen concentration in dairy manure. Using data from three California dairies, we assess how well EC predicts total Kjeldahl nitrogen (TKN), ammonium (NH₄⁺), and plant available nitrogen (PAN) under real-world conditions.

Background

Total Kjeldahl nitrogen (TKN) 

TKN is an important metric because it represents the organic and ammoniacal N fractions, which make up the majority of N in organic sources such as manure. The organic fraction is not immediately available to the plant but will become available via mineralization over timescales of days to years (Pettygrove et al., 2009). Ammonium is plant-available and will convert to nitrate (NO₃^-) via nitrification, typically in days to weeks depending on field conditions. Once it has been converted to nitrate, N is water-soluble and vulnerable to leaching to groundwater. Nitrate quick tests are available, but currently there is not a feasible in-situ measurement for dynamic N levels in soil water or manure. 

Electrical conductivity (EC) 

EC has been shown to correlate with nutrient content in liquid materials such as manure or soil water (Stevens et al., 1995; Peralta & Costa, 2013). It is hypothesized that the relationship between EC and manure changes with specific operations’ attributes such as diet, breed, and manure handling (Rieck-Hinz et al., 1996). There are existing sensors that continuously measure EC of soil or manure water to monitor salts.   

Manure Subsurface Drip Irrigation System (MSDI) 

The manure subsurface drip irrigation system (MSDI) uses raw dairy manure or digestate, blends it with fresh water, and flushes the mixture through sand media filters. This dynamic blend is then pumped through subsurface drip tape to irrigate dairy forage crops. The ratio of effluent to fresh water is controlled by a butterfly valve based on an EC sensor set by the user, depending on their salt tolerance. 

If EC proves to be an adequate proxy measurement for N, this sensor and the MSDI system could provide an estimate and point of control for N applied per irrigation event, promoting more precise nutrient application and expanding the beneficial use of dairy manure. Our objective is to present the correlation of EC and various N forms on three California dairies. 

Data sources and methods 

Manure samples were collected from three dairies biweekly to monthly during the 2017–2019 irrigation seasons. These farms are in Merced (Farm 1), Madera (Farm 2), and Kern (Farm 3) counties. All three farms have an anaerobic lagoon, and the Kern farm installed a digester in 2018. This effort was part of a Conservation Innovation Grant (69-3A75-17-53) to determine the viability and environmental impact of MSDI on dairy forage. 

There were 16 total blend samples taken and 73 total manure samples (Figures 1 and 2).  Each manure sample was tested for EC, NO₃^-, NH₄⁺, organic N, total N, P, K, Cl, SO₄, Ca, Mg, Na, Total Dissolved Solids (TDS), pH, and Total Suspended Solids (TSS), collected and processed by Dellavalle Laboratories.

Manure composition 

As Rieck-Hinz et al. (1996) and Davis et al. (2002) have documented, manure was highly heterogenous within and across dairies over time (Figure 1). Over the entire period, the organic fraction represented 40%, 50.3%, and 28.4% of anaerobic lagoon N for Farm 1, Farm 2, and Farm 3, respectively. Ammonium made up 59.9%, 49%, and 71.3% of N for the same farms. The total N from digestate (Farm 3) was composed of 75.7% NH₄⁺ and 24.2% organic N. Nitrate–N represented less than 1% of total manure N on all farms. 

Blend composition 

The EC of the blend (measured in dS/m, sample collected by hand and analyzed in a lab) ranged from 1.4–2.91 at Farm 1, 0.18–2.08 at Farm 2, and 0.29–1.4 at Farm 3. Averaged over all farms, the organic N fraction of the blend was 39.4% and NH₄⁺ was 59.7%, with less variability than the lagoon water and digestate (Figure 2). 

Plant available N (PAN) of blend 

PAN can be estimated by summing the inorganic fractions (NO₃^- and NH₄⁺) and the fraction of organic N that will become available during the crop’s growing season (20%–50% mineralization rate for lagoon water/slurry in California (Pettygrove et al., 2009)). PAN ranged from 1–38 pounds N per acre-inch of blend (assuming a mineralization rate of 35%). 

EC and N relationships 

We fit a linear mixed-effects model with farm as a random effect (Miller et al., 2022). For TKN, there was a total explanatory power (conditional R²) of 83% and fixed effects (marginal R²) explain 81% of the total variance. In other words, the farm only accounted for 2% of the variance. For ammonium, there was a conditional R² of 78% where fixed effects (EC) explain 71% of total variance. For PAN (assuming 35% mineralization rate), conditional R² was 83% and marginal R² was 78% over all samples (Figures 3 and 4). EC was not meaningfully correlated with the ratio of inorganic N to total N (Figure 5, conditional R² = 0.21 and marginal R² = 0.03). 

Measured variables and their correlation with TKN 

We analyzed the correlation (r) of all measured parameters for all types of manure (lagoon, digestate, and blend). A correlation matrix (Table 1) shows the relationship between all combinations of variables, and the scale ranges from -1 to 1, representing perfectly negatively or positively correlated. EC was positively correlated with TKN (Table 1, r = 0.91). This indicates a relatively strong relationship between TKN and EC, especially considering ammonium and TKN have an r of 0.93. Others have hypothesized that the correlation would get stronger if we analyzed each farm individually, but that was not true in this case. 

Takeaways and conclusions 

Electrical Conductivity and Nitrogen were strongly correlated, especially compared to other nutrients measured at the same time. We did not find that the relationship between EC and N differed by farm. Our correlation is stronger than what Miller et al. (2022) reported, but we sampled biweekly (three farms, 90 samples) as opposed to quarterly self-reported values (91 lagoons, 1,274 samples). Additionally, our overall conclusions differed, as they reported an increase of 55 mg/L TKN with a 1 unit (dS/m) increase in EC, where we estimate 83 mg/L TKN increase. Their samples represent a larger population which likely more accurately capture the variation of manure across California dairies. 

TKN (and EC as a proxy), while valuable for total N balance, does not give insight into the relative amounts of inorganic vs. organic N available. Additionally, even a relatively well-fit model can still result in substantial practical deviations at scale. However, considering our current logistical and systematic constraints, existing EC sensors could be an acceptable solution to improve understanding of N applications of dairy manure. This process would need to be continually validated with local data while considering its limitations. Additionally, it will require buy-in from the user to incorporate the information available to improve the N use efficiency and overall N balance on their operation.