Study area
The study area for this analysis contains the Western Lake Erie Basin, or WLEB, plus a 5-mile buffer extent outward from the basin boundary. Inclusion of the buffer area allowed for modeling of manure phosphorus that may be produced outside the basin and applied within, and vice versa. The WLEB represents 11,870 square miles and the buffer area adds an additional 2,473 square miles, for a total study area of 14,343 square miles. Ohio makes up the majority of land area at 68 percent, followed by Michigan and Indiana at 19 percent and 13 percent, respectively. A total of 45 counties intersect the study area among the three states, with 24 counties having over 95 percent of their land area within the study area extent.
AFO identification
EWG received a list of permitted concentrated animal feeding facilities, or CAFFs, from the Ohio Department of Agriculture, or ODA, in October 2020, which included the location, dominant animal type and maximum number of animals housed at each operation. A dataset of permitted confined feeding operations, or CFOs, in Indiana was obtained from the Indiana map website, last updated in May 2020. This dataset included the location, distribution of animal types and maximum number of animals housed at each operation. The MIWATERs website was used to locate and attribute permitted CAFO operations (November 2020) in Michigan. The most recent inspection report provided on the MIWATERs website was used to determine the location, distribution of animal types and maximum animal counts for each operation.
To identify unpermitted operations, EWG scanned aerial photography (2020 NAIP imagery in Michigan and Indiana; 2019 NAIP imagery in Ohio) to find additional AFOs. Locations were attributed with “animal type” (cattle, dairy, beef, poultry or swine), “size” and “access to pasture” designations. Size was defined using a small (50 to 199 cows), medium (200 to 499) or large (more than 500) designation for cattle operations and a barn count for poultry and swine. We could not confidently assign dairy or beef to small cattle operations, so they were assigned a “cattle” animal type. We gave a “dairy” or “beef” designation to each medium or large cattle operation.
Multiple geographic information system, or GIS, analysts (at least three and up to five) independently attributed each operation, and we accepted the majority animal type, size and pasture designation. Where a consensus did not exist, the operation was revisited by at least two GIS analysts together to agree on final attribution.
Table 1. Permitted status of operations identified
State |
Permitted operations |
Unpermitted AFOs |
Total |
Indiana |
|||
Swine |
77 |
60 |
137 |
Poultry |
15 |
74 |
89 |
Cattle (dairy + beef) |
24 |
277 |
301 |
Total |
116 |
411 |
527 |
Michigan |
|||
Swine |
13 |
5 |
18 |
Poultry |
0 |
4 |
4 |
Cattle (dairy + beef) |
12 |
160 |
172 |
Total |
25 |
169 |
194 |
Ohio |
|||
Swine |
53 |
546 |
599 |
Poultry |
20 |
164 |
184 |
Cattle (dairy + beef) |
26 |
989 |
1015 |
Total |
99 |
1699 |
1798 |
Overall total |
240 |
2279 |
2519 |
Source: EWG via NAIP aerial photography, ODA, EGLE and IDEM
Animal counts
Cattle
The 2017 Department of Agriculture’s Census of Agriculture guided the number of animals allocated to each unpermitted cattle operation, based on its attributed size designation. The same number of cows was assigned to all small, medium and large categories in each state. To calculate the number of animals assigned to small operations, we divided the number of cattle and calves housed in farms between 50 and 199 cows, as listed in the census, by the number of farms in this same size category. We performed this same calculation for medium operations (cattle and calves in farms with 200 to 499 cows) and large operations (more than 500 cows). All counties that intersected the study area were used to determine rates from the census for each state.
When determining the rate for large operations, permitted facilities (and permitted animal counts) in the counties analyzed were first subtracted from the census numbers to avoid artificially inflating the large rate. No unpermitted large cattle operations were identified in Indiana, a finding supported by the state’s lower permitting requirement for all cattle operations housing more than 300 cows.
For small cattle operations for which a dairy-versus-beef distinction could not be assigned, we assumed the ratio of beef to milk cows listed in the census for each small operation in that county. For example, a small cattle operation in Lenawee County, Mich., would be assigned 87 dairy cows and 9 beef cows, totaling 96, using the census’ 91-to-9 dairy to beef-cow ratio.
Table 2. Census-derived rates for unpermitted cattle operations
Number of cows allocated to each unpermitted operation |
|||
State |
Small |
Medium |
Large |
Indiana |
89 |
274 |
NA |
Michigan |
96 |
294 |
817 |
Ohio |
101 |
303 |
714 |
Source: EWG via the 2017 USDA Census of Agriculture
Swine and poultry
Building footprints were manually digitized for each unpermitted poultry or swine operation using NAIP imagery (2020 NAIP imagery in Michigan and Indiana; 2019 NAIP imagery in Ohio). In total, 1,421 barn footprints were digitized for 242 poultry operations and 611 swine operations. The mean number of barns per operation was 1.92 for poultry and 1.56 for swine. The mean square footage was 17,655 ft2 for poultry barns and 11,089 ft2 f or swine barns.
Leading government and/or industry guidelines were used to estimate a stocking density for each animal type (Table 3). Animal counts for swine operations, for example, were estimated by dividing the mapped square footage of each barn by the square footage allotted per animal (9 ft2 per hog).
Table 3. Assumed stocking density for swine and poultry operations
Animal type |
Stocking density (square foot per animal) |
Note |
Source |
Swine |
9 |
2.9 ft2 (50 lb pig) – 9.6 ft2 (300 lb pig) |
|
Layers |
1 |
.465 ft2 (caged) – 1.5 ft2 (cage free) |
|
Broilers |
1 |
.5 ft2 – 1.15 ft2 (based on live weight) |
|
Pullets |
1 |
Assumed same as layers |
|
Turkeys |
3 |
Commercial tom maximum floor space |
Source: EWG via sources listed in Table 3
The 2017 USDA Census of Agriculture lists poultry inventories for the four dominant poultry production systems in each county – layers, broilers, pullets and turkeys. Undisclosed values, listed as (D) values, are designed to protect producer anonymity but also present significant challenges for estimating poultry inventories in a given county. EWG backfilled undisclosed values to estimate county-level inventories for layers, broilers, pullets and turkeys.
The census lists the distribution of layer farms by size, even for counties with an undisclosed value for total layers. Permitted layer operations in each county were first matched to those listed in the census according to size. For example, in Auglaize County, Ohio, one permitted layer operation exists that houses 317,000 layers. This operation was matched to the single operation in the census larger than 100,000 birds, the largest size category listed for layers. Permitted operations were matched only to operations of equal or lesser size, and permitted layer counts were used when matched to a census operation. Remaining operations listed in the census, representing those not matched to a permitted operation, were allocated the minimum inventory value of its corresponding size range. For example, 50,000 birds were assumed for each remaining operation in the 50,000-to-99,999 size category.
Once this backfilling process was completed, a new layer inventory value was generated for each county. This inventory value was substituted for all counties with an undisclosed value for total layers and for those counties where it exceeded the total number of layers listed in the census. This latter scenario occurred in five of the 45 counties, where the number of permitted layers exceeded the total layer inventory listed in the census.
Unlike layers, pullet, broiler and turkey inventories are given no size distribution in the census. Only a total inventory of birds for each animal type is listed for each county. Permitted inventories were substituted for pullet, broiler and turkey inventories when these values exceeded census inventories or when an undisclosed value was listed. When an undisclosed value was listed and no birds were permitted, the 2012 USDA Census of Agriculture inventory value was used, if available. Where both 2012 and 2017 census values were undisclosed and no birds were permitted, zero birds for that production system were estimated for that county.
Permitted birds in each county were then subtracted from the backfilled county inventory estimates for each production system. As both permitted and census inventories were provided at the county-level, unpermitted poultry inventory estimates reflect the entire spatial extent of all 45 counties analyzed. These unpermitted inventory estimates were summed by state and are reported by poultry production system in Table 4.
The distribution listed in Table 4 was then assumed for each unpermitted poultry operation identified in each state. In other words, each unpermitted operation was allocated a proportion of each poultry production system that mirrored the distribution of unpermitted poultry production systems in that state. For example, 68 percent of the square footage of an unpermitted poultry operation in Ohio was allocated to turkeys, 14 percent to broilers, 11 percent to layers and 7 percent to pullets.
Table 4. Estimated distribution of unpermitted poultry production systems in the study area
Estimated unpermitted poultry inventory by production system (% of total) |
||||
State |
Layers |
Pullets |
Broilers |
Turkeys |
Ohio |
1,290,612 (11%) |
737,558 (7%) |
1,653,281 (14%) |
2,618,186 (68%) |
Michigan |
34,857 (62%) |
5,887 (10%) |
9,589 (17%) |
1,969 (11%) |
Indiana |
1,286,662 (33%) |
179,003 (5%) |
893,869 (23%) |
514,141 (39%) |
Source: EWG via ODA, EGLE, IDEM and 2017 USDA Census of Agriculture
Validation of animal counts with 2017 Census of Agriculture
The 2017 USDA Census of Agriculture represents the most reliable source of data on county-level animal inventories in the U.S. Collected every five years, the census is very useful for understanding trends in the animal agricultural industry. To validate animal counts assigned to unpermitted AFOs, we compared the animal count with census inventory estimates. Due to unique reporting methods for each animal type, approaches for interpreting census inventories varied and are described in detail below.
Cattle and calves
The total number of cattle and calves housed in operations with more than 50 cows, as listed in the census under “Farms by inventory,” was calculated for each county intersecting the study area. For farm size categories that listed an undisclosed value for inventory, the number of farms in that size category was multiplied by the minimum value in its corresponding size range. Only cows housed on farms with more than 50 cows were included, assuming that smaller hobby farms could not be confidently identified from aerial photography. Census cattle and calf inventories in each county were then multiplied by the proportion of the county land area falling within the study area. Census inventory estimates were then summed by state and are compared to cattle inventories used in this study (Table 5).
Table 5. Cattle animal counts compared with 2017 USDA Census of Agriculture
Beef |
Dairy |
Total cattle and calves |
||||||
|
EWG |
Permitted |
EWG |
Permitted |
EWG |
Permitted |
Total |
Census |
Indiana |
11,838 |
3,139 |
25,152 |
25,149 |
36,990 |
28,288 |
65,278 |
69,994 |
Michigan |
5,209 |
2,950 |
30,259 |
24,334 |
35,468 |
27,284 |
62,752 |
67,778 |
Ohio |
47,317 |
8,565 |
149,740 |
61,527 |
197,057 |
70,092 |
267,149 |
274,524 |
TOTAL |
64,364 |
14,654 |
205,151 |
111,010 |
269,515 |
125,664 |
395,179 |
412,296 |
Source: EWG via ODA, EGLE, IDEM and 2017 USDA Census of Agriculture
Swine
The census “total hogs and pigs” was used to represent swine inventories. When more hogs were permitted in a county than what the census listed, permitted hogs were substituted for the total in that county. This occurred in 12 of the 45 counties – five in Ohio and seven in Indiana – and these higher values are reflected in the census column in Table 5. Census hog inventories in each county were then multiplied by the proportion of the county land area falling within the study area. Census inventory estimates were then summed by state and are compared to swine inventories used in this study (Table 6).
Table 6. Swine animal counts compared with 2017 USDA Census of Agriculture
Swine |
||||
EWG |
Permitted |
Total |
Census |
|
Indiana |
70,987 |
314,360 |
385,347 |
421,745 |
Michigan |
8,926 |
42,820 |
51,746 |
51,073 |
Ohio |
1,099,273 |
292,239 |
1,389,615 |
1,332,079 |
Total |
1,179,186 |
649,419 |
1,828,605 |
1,804,897 |
Source: EWG via ODA, EGLE, IDEM and 2017 USDA Census of Agriculture
Poultry
Census poultry inventories were calculated for each of the four production systems – layers, broilers, pullets and turkeys – as described in detail in the “Animal counts” section above. An important note is that backfilled inventory values were substituted for census inventory values when they either exceeded that listed in the census or when an undisclosed value was listed. These backfilled estimates are reflected in the census column in Table 7. Census poultry inventories in each county were then multiplied by the proportion of the county land area falling within the study area. Census inventory estimates were then summed by state and are compared to poultry inventories used in this study (Table 7).
This approach assumes animals are evenly distributed throughout a county, which is not always the case and is particularly evident in the poultry industry. An example is Mercer County, Ohio, which houses 35 permitted poultry operations with a capacity for over 12 million birds. Although 70 percent of Mercer County lies within the study area, only nine of the 35 operations, with a capacity for just over 2 million birds, fall within the study area. Using this example, adjusting county-level census inventories by the proportion of the county land area falling within our study area results in a higher than anticipated number of poultry in Mercer County.
Table 7. Poultry animal counts compared with 2017 USDA Census of Agriculture
Poultry (layers + broilers + pullets + turkeys) |
||||
EWG |
Permitted |
Total |
Census |
|
Indiana |
1,273,579 |
3,619,496 |
4,893,075 |
3,191,809 |
Michigan |
196,812 |
0 |
196,812 |
80,684 |
Ohio |
3,421,209 |
15,401,217 |
18,822,426 |
27,935,505 |
Total |
4,891,600 |
19,020,713 |
23,912,313 |
31,207,998 |
Source: EWG via ODA, EGLE, IDEM and 2017 USDA Census of Agriculture
Manure nutrients
Manure nutrients were estimated using literature values from the Midwest Plan Service, which is the same resource found in the Ohio Administrative Code (Table 8).
Although some permitted operations provided information about the distribution of animal types, such as a grow-finish pig versus nursery pig, this data was not available for AFOs identified from aerial photography. A single animal type was assumed for all unpermitted beef AFOs, equating to a 1,100-pound finishing beef cow. Similarly, a grow-finish pig (150 pounds) was assumed for all unpermitted swine AFOs. Each unpermitted poultry AFO was assumed to have a distribution of the four different poultry types, as described in the above sections. Given the dominance of the dairy industry relative to total phosphorus generation, animal counts for unpermitted dairy AFOs were divided into mature dairy cows, heifers and calves using a 75/12.5/12.5 percent distribution, respectively, which was approximated using the state permit data provided.
Animals were assumed to be present 365 days a year. To estimate manure volumes, all poultry and beef manure was assumed to be solid and all swine and dairy manure liquid. Mature dairy cows were assumed to be lactating 305 days of the year and dry 60 days of the year, or a .85 to .15 ratio.
Table 8. Literature values for manure and nutrient production
Animal type |
Size (pounds) |
Manure excreted (pounds per animal per day) |
||
Pounds |
Gallons |
P205 |
||
Mature dairy |
1400 |
– |
17.19 |
0.4645 |
Dairy heifer |
750 |
– |
5.21 |
0.08 |
Dairy calf |
150 |
– |
1.38 |
0.01 |
Beef |
1100 |
54 |
– |
0.12 |
Beef calf |
450 |
48 |
– |
0.09 |
Swine nursery |
25 |
– |
0.23 |
0.01 |
Grow-finish pig |
150 |
– |
0.89 |
0.03 |
Sows |
375 |
– |
2.08 |
0.11 |
Boars |
350 |
– |
0.865 |
0.04 |
Layers |
3 |
0.15 |
– |
0.0008 |
Pullets |
1.5 |
0.075 |
– |
0.0004 |
Broilers |
2 |
0.19 |
– |
0.0014 |
Turkeys |
20 |
0.74 |
– |
0.0074 |
Source: EWG via Midwest Plan Service, Ohio Administrative Code
Modeling scenarios
A primary objective of this study was to minimize uncertainty at each stage of the analysis. Efforts to do this entailed the participation of multiple GIS analysts attributing animal feeding operations and rigorous comparison of animal counts to the 2017 USDA Census of Agriculture.
Despite these efforts, literature values for the estimation of manure nutrients can vary +/- 30 percent, even with precise knowledge on animal counts and manure management. The fraction of manure considered recoverable is also highly uncertain, and will vary over time, region of the country and by farm size. In a study of all confined U.S. AFOs with potentially recoverable manure, 76 percent of manure phosphorus was assumed recoverable from the total amount of as-excreted phosphorus in 2012, a statistic that is showing an upward trend. The use of literature values also does not reflect temporal changes in the phosphorus concentration of excreted manure, such as the addition of phytase to swine feed.
In addition to the uncertainty noted above, it is likely that not all animal feeding operations operate at maximum capacity in any particular year. Market fluctuations and disease outbreaks, for example, may cause certain industries to reduce inventories.
A sensitivity analysis was performed to quantify the impact of these variables on manure P205 production and the land area required for manure disposal. Seven manure allocation scenarios were analyzed, as described in Table 9. Results from Scenario 5 were used in the primary text and map of this EWG report.
Table 9. Manure allocation scenarios
Scenario |
Recoverability factor |
Recoverability factor by operation size |
Operation capacity |
Phytase reduction |
1 |
100% |
None |
100% |
No |
2 |
75% |
Small, medium |
100% |
No |
3 |
50% |
Small, medium |
100% |
No |
4 |
50% |
All |
100% |
No |
5 |
50% |
All |
95% dairy, beef, swine |
No |
80% Poultry |
||||
6 |
50% |
All |
95% dairy, beef, swine |
Yes |
80% Poultry |
||||
7 |
50% |
All |
80% dairy, beef, swine |
Yes |
70% Poultry |
||||
Recoverability factor: fraction of manure considered recoverable from cattle, dairy or beef operations with a pasture flag of “YES” |
||||
Recoverability factor by operation size: size of operations to which the recoverability factor is applied – small (< 199 cows), medium (200-499) or large (> 500 cows) |
||||
Operation capacity: fraction of maximum animal capacity assumed present |
||||
Phytase reduction: 25% reduction in P205 excretion from swine operations |
Source: EWG via discussion with stakeholder advisory group
Crop phosphorus removal
The Agricultural Conservation Planning Framework , or ACPF, database provided crop rotation history for each field from 2015 to 2020. Every crop field that intersected the study area was included in the analysis. Six-year average county crop yields (NASS Surveys, 2015-2020) for five major crop types were used to estimate the amount of phosphorus removed annually from each field and for each year in the rotation. Phosphorus removal was estimated for each of the major crops listed in Table 10. Average county yields were assigned to fields based on the county in which the centroid of the field is located.
Table 10. Phosphorus removal rates for major crop types in Ohio, Michigan and Indiana
Phosphorus removal rates by major crop type |
||
Crop |
P205 removal |
Unit |
(pounds per unit yield) |
||
Corn for grain |
0.35 |
Bushel |
Soybeans |
0.8 |
Bushel |
Wheat |
0.5 |
Bushel |
Small grains |
0.36 |
Bushel |
Alfalfa |
12 |
Ton |
Source: EWG via Tri-State Fertilizer Guidelines and Ohio State University Extension
Modeling manure application
Manure application was simulated to represent the land area required for manure application over a six-year crop rotation period from 2015 to 2020. Manure was spatially applied from feedlots to fields using the methods described in Porter and James, 2020. Briefly, manure is spatially applied from feedlot to proximal fields, moving outward from each feedlot until no manure nutrients remain.
Manure was applied to crop fields at the average annual phosphorus removal capacity of each field. Average annual phosphorus removal for each field was estimated by summing the phosphorus removal for each year in a six-year rotation, from 2015 to 2020, and dividing by six. No losses were assumed for manure phosphorus – the estimated amount of phosphorus recovered was available for field application.
The use of an average annual phosphorus removal rate does not limit which crops are available for manure application in the model. For example, in a six-year rotation of alternating corn and soybean, manure application may occur in corn years only, with soybean years representing drawdown years, or every year in the rotation. The year of manure application within the six-year crop rotation is irrelevant, if the total crop phosphorus removal over the rotation is not exceeded.
Adjustment for manure moving out of the WLEB
Before we modeled manure application, we adjusted manure P205 for all operations within the 5-mile buffer of the WLEB. This was done to account for the proportion of manure that may be applied to crop fields outside the study area. A unique circle of 5-mile radius was delineated around each operation within the buffer region, and manure P205 was reduced proportionally by the percentage of the circle that fell within the study area. For example, the amount of manure P205 available for crop application was reduced to 70 percent for an operation if 70 percent of its 5-mile radius circle fell within the study area.
County-level commercial fertilizer sales
Commercial fertilizer estimates of phosphorus as P were obtained from a 2021 U.S. Geological Survey report, which provided estimates of kilograms of P for farm use for each county in the conterminous U.S from 1950 to 2017. A multiplication factor of 2.29 was used to convert from P to P205 and a factor of .0011023 was used to convert from kilograms to tons.
To adjust county-level sales to the study area, 2017 values were adjusted by the proportion of cropland within each county (excluding pasture) falling within the study area. Cropland area for each county was determined using the 2020 NASS CDL.
Manure allocation scenario results
Table 11. Excreted and recoverable tons of manure P205 across scenarios
Scenario |
P205 excreted |
P205 recoverable |
Percent crop P205 removal |
1 |
38,311 |
38,311 |
28% |
2 |
38,311 |
36,481 |
27% |
3 |
38,311 |
34,650 |
26% |
4 |
38,311 |
33,706 |
25% |
5 |
35,553 |
31,178 |
23% |
6 |
33,245 |
28,871 |
22% |
7 |
28,144 |
24,460 |
18% |
133,872 tons of P205 are removed on average by all crop fields in the study area each year |
|||
Indiana: 16,660 tons; Michigan: 16,519 tons; Ohio: 100,693 tons |
Source: EWG via Midwest Plan Service and Ohio Administrative Code
Table 12. Acres required for manure application across scenarios
Scenario |
Indiana |
Michigan |
Ohio |
Total |
1 |
239,288 |
185,131 |
1,165,090 |
1,589,509 |
2 |
224,195 |
172,226 |
1,114,765 |
1,511,186 |
3 |
208,952 |
159,634 |
1,065,077 |
1,433,663 |
4 |
206,688 |
147,645 |
1,042,314 |
1,396,647 |
5 |
191,851 |
140,162 |
959,442 |
1,291,455 |
6 |
174,480 |
137,650 |
884,368 |
1,196,498 |
7 |
147,772 |
115,820 |
749,788 |
1,013,380 |
5,904,037 acres were available for potential manure spreading across the three states |
||||
Indiana: 731,481 acres; Michigan: 782,173 acres; Ohio: 4,390,383 acres |
Source: EWG via Tri-State Fertilizer Guidelines, Ohio State University Extension, and USDA/ARS ACPF Database
References
American Public Health Association (2019). Precautionary Moratorium on New and Expanding Concentrated Animal Feeding Operations. Available at: https://www.apha.org/policies-and-advocacy/public-health-policy-statements/policy-database/2020/01/13/precautionary-moratorium-on-new-and-expanding-concentrated-animal-feeding-operations.
Becot, F., Inwood, S., Jackson-Smith, D., & A. Katchova, A. (2020). The Status and Changing Face of Ohio Agriculture: Summary of Ohio Farm Trends 1997–2017. College of Food, Agricultural, and Environmental Sciences, SENR Technical Report. Columbus, OH: The Ohio State University.
Culman, S., Fulford, A., Camberato, J., & Steinke, K. (2020). Tri-State Fertilizer Recommendations. Bulletin 974. College of Food, Agricultural, and Environmental Sciences. Columbus, OH: The Ohio State University.
Falcone, J.A., 2021, Estimates of county-level nitrogen and phosphorus from fertilizer and manure from 1950 through 2017 in the conterminous United States: U.S. Geological Survey Open-File Report 2020–1153, 20 p., https://doi.org/10.3133/ofr20201153.
Indiana Business Research Center, Kelley School of Business, Indiana University, March 2017. The Economic Impact of Animal Agriculture in Indiana’s Regions. Available at: https://www.ibrc.indiana.edu/studies/Livestock-Report-2017.pdf.
Indiana Department of Environmental Management. Confined Feeding Operations Facilities in Indiana. Downloaded November 2020 from IndianaMap website: https://maps.indiana.edu/LayerGallery.html.
International Joint Commission (2018). Fertilizer Application Patterns and Trends and Their Implications for Water Quality in the Western Lake Erie Basin.
Kast, J.B., Long, C.M., Muenich, R.L., Martin, J.F., Kalcic, M.M., 2019. Manure management at Ohio confined animal feeding facilities in the Maumee River watershed. J. Great Lake. Res. 45 (6), 1162-1170.
Kast, J., Apostel, A.M., Kalcic, M.M., Muenich, R.L., Dagnew, A., Long, C.M., Evenson, G., Martin, J., 2020. Source contribution to phosphorus loads from the Maumee River watershed to Lake Erie. J. Environ. Manag.
Long, C. M., Muenich, R. L., Kalcic, M. M. & Scavia, D. Use of manure nutrients from concentrated animal feeding operations. J. Great Lakes Res. 44, 245–252 (2018).
Lorimor, J.; Powers, W.; Sutton, A. Manure Characteristics; MWPS-18, Section 1; Midwest Plan Service: Ames, IA, USA, 2004.
Michigan Department of Environment, Great Lakes, and Energy. MIWater Wesbsite. Available at: https://miwaters.deq.state.mi.us/ncore/external/home.
Ohio Department of Agriculture, Division of Livestock Environmental Permitting. Public Data Request for Permitted CAFO locations in Ohio, received October 2020.
Ohio Lake Erie Commission (OLEC), 2018. State of Ohio’s domestic action plan. https://lakeerie.ohio.gov/Portals/0/Ohio%20DAP/DAP%201-1%20FINAL%202018-0%208-27.pdf.
Porter, S.A.; James, D.E. Using a Spatially Explicit Distribution Model to Assess the Contribution of Animal Agriculture to Minnesota’s Agricultural Nitrogen Budget. Agronomy. 2020, 10(4), 1–15.
Tomer, M.D., Porter, S.A., James, D.E., Boomer, K.M.B., Kostel, J.A. & McLellan, E. Combining precision conservation technologies into a flexible framework to facilitate agricultural watershed planning. Journal of Soil and Water Conservation. 2013. 68, 113A-120A.
USDA-NASS. County Agricultural Production. United States Department of Agriculture-National Agriculture Statistics Service: Washington, D.C., 2020. Available at: https://www.nass.usda.gov/Surveys/Guide_to_NASS_Surveys/County_Agricultural_Production/index.php.
USDA NASS, 2017 USDA Census of Agriculture, Available at: https://www.nass.usda.gov/Publications/AgCensus/2017/.