Phosphorus Indicator

Authors: Keith Reid, Kimberley Schneider and Antoun El Khoury

The issue

Phosphorus (P) is an essential nutrient for all plants and animals. It is added to agricultural soils as mineral fertilizer where soil reserves are inadequate to sustain crop yields, or as livestock manure or biosolids. Since the early 1950s, intensified crop and livestock production has resulted in P applications in some regions greater than the removal of P in the harvested portion of crops. Over time, cumulative P surpluses have enriched the soil and increased the risk that soil P will be released and transported from agricultural fields in runoff. The risk of P loss is dependent on the co-occurrence of two factors, P source and P transport, which may vary with local and regional conditions and with weather. Therefore, periods of high runoff such as spring snowmelt and heavy rain events result in a greater risk of P being released and transported into freshwater bodies.

In natural freshwater systems, P occurs in very low concentrations but may vary significantly as a function of the stream size and ecosystem characteristics. Relatively small increases in P concentration in surface fresh water can contribute to eutrophication of rivers and lakes, and to formation of harmful or nuisance algal blooms. These result in decreased water quality, and limitations on bathing, drinking and recreational activities, subsequently leading to economic and social impacts on the local community (Watson et al. 2016). This has been most evident in relatively shallow lakes with a large proportion of the watershed under agricultural or urban land uses, like Lake Winnipeg, Manitoba, Lake Erie, and Missisquoi Bay in Lake Champlain. While the source of the phosphorus is not only agricultural —municipal and industrial wastewater as well as residential contributions from waste water or leaky septic systems also contribute — significant effort has been expended on reducing or mitigating agricultural contributions. Agricultural sources of P can be difficult to manage compared to urban due to the transport mechanisms being a diffuse source of pollution with no single point to ‘cut-off’. Agricultural Beneficial Management Practices (BMPs) have focused on using P more efficiently for plant uptake resulting in requiring less input, as well as mitigating P movement within the landscape to reduce contamination into freshwater bodies.

The indicator

The IROWC-P (Risk of Water Contamination by Phosphorus) was developed to assess the status and trends over time for the risk of surface water contamination by P from Canadian agricultural land. It is calculated for 3,487 Soil Landscapes of Canada (SLC) polygons with greater than 5% agricultural land. Previous reports had consolidated the data for these polygons into 280 watersheds, but this report has maintained the greater detail. The structure of IROWC-P follows that of a component P Index (Reid, Schneider, & McConkey, 2018), considering the following four components:

• Dissolved P from the soil, where the P source (estimated from the cumulative P balance and the resulting degree of P saturation) is multiplied by the modeled volume of surface plus tile runoff.
• Dissolved P incidental to the application of mineral P fertilizers or livestock manure. The P available for runoff is estimated from application rates, solubility of P source, application timing and degree of incorporation. A P distribution factor is used to calculate how much P is retained in the field as water infiltrates into the soil, and how much is carried in runoff.
• Dissolved P from over-wintering vegetation, estimated from the quantity of P in the vegetation that is released as soluble P modified by the proportion of this P which infiltrates into the soil. Losses are greatest when there is no loss until spring thaw, and the runoff occurs over frozen soils that cannot absorb the P.
• The bioavailable portion of particulate P from soil erosion. The amount of water erosion is calculated in the soil erosion indicator (WatERI) and multiplied by the soil test P concentration to estimate the portion of particulate P that could contribute to eutrophication.

For each of the components, transport off the field is estimated using both the surface runoff, and the portion of tile drainage that has reached the tiles through quickflow/preferential flow through earthworm burrows, root channels, and desiccation cracks in the soil. Delivery from the field to surface water is estimated considering the stream density, dominant surface landforms, and connectivity through tile drains.

Results are reported for the accumulation of P in soils across Canada (soil P source), the aggregate risk of P losses at the edge-of-field scale (EOF), and the risk of P losses to surface water (IROWC-P). EOF and IROWC-P values were grouped into five risk classes (very low, low, moderate, high and very high). The risk classes are relative rankings wherein 50% of SLC polygons are classified in the very low risk class while the highest 5% fall into the high and very high-risk classes.

Updates from Report 4

Several changes were made to better reflect the risk of P movement from agricultural land in the following ways:

• Initial soil test P values were adjusted for Ontario to reflect the high applications of fertilizer and manure during the 1960s and 70s. Previous versions of the indicator had adjusted the initial values on a provincial basis, but for this indicator, data on the P balance was collected at the county scale, providing improved spatial discretization.
• Allocation of fertilizer P to SLC polygons was based on the relative value of fertilizer purchases by different farm types within each polygon, rather than assuming manure P displaced fertilizer P.
• Components were added to account for P losses incidental to the application of fertilizer or manure, and for losses from over-wintering vegetation.
• Calculation of potential P losses through tile drains was updated to reflect better estimates of the proportion of quickflow to tiles.
• Improved alignment with other indicators, including the use of a common hydrology model to predict runoff (the hydrology module of the DNDC model); using the allocation of manure to agricultural land calculated for the IROWC-N indicator; and using the water erosion outputs from the soil erosion indicator to calculate particulate P losses.
• Runoff calculations were based on a synthetic weather file that represents the normal temperature and precipitation patterns for each SLC polygon, so changes in indicator status are due to changes in crop management rather than extremes in weather.
• Results have been reported by SLC polygon, rather than consolidated into watersheds, so there is more detail in the spatial patterns of risk of P loss.

Limitations

IROWC-P assesses the risk originating from agricultural P; non-agricultural P is not considered. Calculations of P inputs and cumulative P balance follow Census of Agriculture data from 1981 to 2016. Except for Ontario, there is insufficient data to allow accounting for P enrichment prior to 1976.

Data for nutrient application timing and application method follow the Farm Environmental Management Survey. This does not differentiate among different fertilizer types, so there is some uncertainty regarding the incorporation and timing of P fertilizer specifically.

The spatial extent of tile drainage is poorly enumerated, particularly in provinces where it is a relatively new practice.

The calculation of IROWC-P accounts for most BMPs that lower P losses at the source, but only accounts for a few BMPs that mitigate the movement of P in the landscape. This is due to a lack of comprehensive national BMP adoption data for practices such as buffer strips or cover crops.

Data is aggregated at the SLC polygon level, so considerable variation is expected within these polygons in soil P accumulation, application rate, method and timing, and water erosion. Depending on the location of “hot spots” for P loss, the actual risk of P loss to surface water can vary from the aggregated values.

Results and interpretation

Areas of high or very high soil P source (defined as >4 or >5 mg water extractable P (WEP) kg-1 of soil, respectively) are present in every province except Prince Edward Island, with the highest proportion of agricultural land in these categories in Newfoundland and Labrador, Manitoba, and Ontario (Table 2 and Figure 1. Phosphorus accumulation in the soil occurs when P additions as fertilizer, manure, or a combination of both, is greater than the amount removed in the harvested portion of the crop (Figure 2). The largest accumulations at the start of the study period were in Southwestern Ontario, but soil P levels in this region have been stable or declined slightly in the intervening 35 years. There have been increases in soil P in pockets across the country where there has been intensive livestock production or concentrations of high-value horticultural crops (Reid and Schneider 2019). Provincial trends in P inputs over time are shown in Figure 3 (manure P) and Figure 4 (fertilizer P). Soil WEP is estimated as a function of both the accumulation of P in the soil, and the ease with which that P can desorb into runoff water. This desorption is greater for the predominantly alkaline soils of the prairie provinces than for the acidic soils in Quebec and Atlantic Canada. It represents the long-term potential source of P export from agricultural land (that is, legacy P), but the risk of water contamination from this P source also depends on the amount of annual runoff, and the connectivity of the landscape to surface water.

Figure 1: Soil P saturation (water extractable P) in Canada in 2016

Total P inputs to agricultural land have stayed within a narrow range over the period from 1981 to 2016, with more year-to-year variation in mineral P (Figure 4) than manure P (Figure 3). When this is broken down to the provincial scale, however, it is evident that there has been a gradual decline in manure P in the eastern provinces, accompanied by a gradual increase in AB, MB and BC (Figure 3).

Mineral P applications are more variable, as farmers respond to the relative prices of crops and fertilizer. There is a trend of higher mineral P application in the prairie provinces (Figure 4), which may be related to the transition from a summer-fallow system to one of continuous cropping. The very high mineral P inputs to the Atlantic provinces in 2016 appear to be an anomaly.

The annual risk of P loss to the edge of the field (Figure 5) is the summation of the four components of P loss, but it does not account for P delivery from EOF to surface water. Areas of high or very high risk have a combination of high soil P source, high runoff and erosion potential due to a relatively moist climate, and high annual P application rates as either fertilizer or manure. The model used to predict runoff accounts for changes in transport/hydrology due to changes in land use (for example forages to row crops) and tillage systems (conventional tillage to no-till). Southwestern Ontario, southern Quebec and parts of Manitoba show up with the highest risk, but it is notable that the proportion of Manitoba in high risk category has shrunk significantly relative to the soil P source. This pattern in Manitoba is due to the relatively low annual runoff in a prairie environment, and also to low risk from the other components of P loss (low soil erosion in a flat landscape, and high adoption of subsurface fertilizer placement so low incidental losses). The EOF risk is modified by the proportion of runoff from the field likely to reach surface water to calculate the risk of water contamination. This is a function of the stream density, landform, and proportion of tile-drained land.

Figure 5: Risk of phosphorus loss from edge-of-field in Canada in 2016

The overall risk of water contamination is calculated by multiplying the edge-of-field risk by a transport modifier that accounts for runoff amount and the connectivity of the landscape to surface water (distance to streams, or prevalence of tile drains). The high and very high-risk categories of risk of water contamination are concentrated in Ontario and Quebec, followed by Manitoba and Prince Edward Island (Table 1 and Figure 6). There are also small areas of high risk in British Columbia, Alberta, and New Brunswick. In most of the country, this is associated with high concentrations of livestock, while in Prince Edward Island it is a result of the high erosion and high annual P fertilizer rates associated with potato production.

Figure 6: Risk of water contamination by phosphorus in Canada in 2016

On a national scale, the risk of water contamination by P has been stable since 1981, although there have been significant geographic shifts (Table 1 and Figure 7). Southern Ontario has shown improvement in most SLCs, although this is primarily a movement from very high or high risk to moderate risk so there is still much work to be done. Several pockets in the Atlantic provinces show declines. Large areas in Western Canada show worsening trends (Figure 7), but these are from very low risk to low or moderate risk.

Figure 7: Change in risk of water contamination from agricultural land in Canada from 1981 to 2016

Full datasets for the outputs from the IROWC-P algorithm for each year from 1981 to 2016 can be found on the Open Data website.

Response options

The data generated for each of the P loss components for this assessment indicates that the magnitude of P loss risk is similar for soil erosion and for P desorption from the soil, while losses incidental to application of fertilizer or manure are roughly one-tenth as large on average. P loss from overwintering vegetation does not appear to be a significant contributor, although the authors acknowledge that the input data necessary to properly assess this source is not available.

The relatively small magnitude of the incidental losses to P application shows that many farmers have already adopted the principles of 4R Nutrient Stewardship by applying most P fertilizers in subsurface bands at planting time (Bruulsema, Peterson, & Prochnow, 2019). This is not the case, however, in every area, and encouraging all farmers to adopt or continue best practices for P application will have significant local effects. There is also a risk that as farming operations expand, the time and labour savings of broadcast P over banding could move some farmers to adopt less appropriate P management practices.

The buildup of P in agricultural soils represents a long-term risk to water quality, as these fields will continue to release elevated concentrations of both dissolved and particulate P for many decades until soil P levels are drawn down by crop removal. This is a particular challenge for livestock operations, where the manure generated while producing meat or milk cannot be easily or economically transported to distant fields that would benefit from the nutrients in the manure. Reducing or eliminating P inputs to fields that have more P than is needed for optimum crop production is the first step. Soil testing is an important tool to identify those fields that need additional P as opposed to those that do not. Enabling the transfer of manure from livestock to crop farms can alleviate local surpluses. Depending on the local circumstances, this may be whole manure, or some component of manure extracted during a treatment process (for example, solid-liquid separation to concentrate the P in the solid fraction). Precision feeding practices can reduce the P concentration in the manure by providing only the P that is required by a particular livestock species at a specific age. In the long run, such crop and livestock management can progressively reduce the quantity of soil P available for transport to surface waters and return agro-ecosystems to lower risk of P contamination.

Reducing soil erosion and surface runoff through reduced tillage and crop rotations including forages will help to prevent P losses to surface water, although current evidence shows this may have more impact on particulate than dissolved P losses (Duncan et al., 2019). Implementation of BMPs to impede the movement of P into the drainage network, such as the establishment of buffer strips around surface water bodies will reduce the risk of P contamination of surface waters. However, buffer strips are not equally effective in all regions, and they can impede agricultural activities. To render this BMP more economically acceptable for producers, plant species that offer potential economic returns should be prioritized for buffer strips.

IROWC-P enables the identification of areas with a high risk of water contamination by P from agricultural sources. A more detailed examination of the agricultural practices in these regions could reveal which regional characteristics contribute to the risk of water contamination by P. This could facilitate targeted mitigation practices or research efforts, as opposed to a “one-size-fits-all” approach (Macrae et al., 2021).

IROWC-P could be further developed by incorporating information about new or existing BMPs that have significantly impacted P source and transport. Currently, there is a lack of national data regarding the extent and location of such BMPs, and so they are not adequately considered in the current indicator algorithm.

References

Bruulsema, T. W., Peterson, H. M., & Prochnow, L. I. (2019). The Science of 4R Nutrient Stewardship for Phosphorus Management across Latitudes. Journal of Environment Quality, 48(5), 1295-1299. doi:10.2134/jeq2019.02.0065

Duncan, E. W., Osmond, D. L., Shober, A. L., Starr, L., Tomlinson, P., Kovar, J. L., . . . Reid, K. (2019). Phosphorus and Soil Health Management Practices. Agricultural & Environmental Letters, 4(1), 1-5. doi:10.2134/ael2019.04.0014

Macrae, M., Jarvie, H., Brouwer, R., Gunn, G., Reid, K., Joosse, P., . . . Zwonitzer, M. (2021). ..One size does not fit all: towards regional conservation practice guidance to reduce phosphorus loss risk in the lake erie watershed. J Environ Qual, 50(3), 529-546. doi:https://doi.org/10.1002/jeq2.20218

Reid, K. and Schneider, K. (2019). Phosphorus accumulation in Canadian agricultural soils over 30 years. Canadian Journal of Soil Science, 99(4), 520-532. doi: 10.1139/cjss-2019-0023

Reid, K., Schneider, K., & McConkey, B. (2018). Components of Phosphorus Loss From Agricultural Landscapes, and How to Incorporate Them Into Risk Assessment Tools. Frontiers in Earth Science, 6(135). doi:10.3389/feart.2018.00135

Watson, S. B., Miller, C., Arhonditsis, G., Boyer, G. L., Carmichael, W., Charlton, M. N., ... & Wilhelm, S. W. (2016). The re-eutrophication of Lake Erie: Harmful algal blooms and hypoxia. Harmful algae, 56, 44-66.