1. 1 Iowa Conservation Practices: Historical Investments, Water Quality, and Gaps (Final progress report) February, 2007 (revised version) October, 2007

2. 2 Background—CARD/DNR Project DNR goals: 1. Provide estimate of Iowa’s “needs” for non-point source pollution water quality control 2. Inform the debate in Iowa concerning water quality Study: 1. Summarize baseline water quality and land use 2. Identify set of conservation practices in each watershed 3. Predict water quality (sediment and nutrients) under this set of practices 4. Compute cost of those practices (provides “needs” answer)

3. 3 Project Overview Goals 1. What conservation practices are currently in place in Iowa, what is their coverage, and what is the cost of these practices? 2. What are (and have been) the effects of this investment on water quality? 3. What would it take to improve water quality to obtain specific standards?

4. 4 Conservation Programs Used EQIPIFIPCRP Description Federal Conservation Program State Conservation Program Federal Conservation Program Program Function Cost Sharing and Incentive Payments Cost Sharing, Maintenance Incentive Payments, and Rental Payments Coverage Contour Farming Filter Strips Grassed Waterways Nutrient Management Terraces Tillage Practices Contour Farming Contour Stripcropping Filter Strips Grassed Waterways Terraces Tillage Practices Land retirement Years 1997 to 2005 2004 Positives 1. The entire population of contracts that received assistance is included. 2. During the practices contract period the practice is required to be maintained in working condition. 1. The entire population of contracts that received assistance is included. 2. During the practices contract period the practice is required to be maintained in working condition. The entire population of CRP land is included. Negatives 1. Only includes records where the practice was cost-shared on or offered incentive. 2. Available records are more highly aggregated than IFIP records. 1. Only includes records where the practice was cost-shared on or offered incentive. Aggregated for all practices

5. 5 Surveys Used NRICTIC Description USDA Survey Reported findings from the USDA’s Crop Residue Management Survey Coverage Contour Farming Contour Stripcropping Filter Strips Grassed Waterways Terraces Erodibility Measures CRP Tillage Practices Years19972004 Nature of Survey statistically based survey. Data were collected using a variety of imagery, field office records, historical records and data, ancillary materials, and a limited number of on- site visits “best estimates” of district conservationist are combined with Cropland Transect Surveys Positives 1. Records the total coverage of the practices (both those that received financial assistance and those that were installed voluntarily without funding). Negatives 1. Conservation practice usage is calculated from a sample and so assumptions are made about the population. 1. Aggregate data (at the county level) that cannot be directly used in water quality modeling.

6. 6 Statewide average cost PracticeStatewide Average Cost Grass Waterway$2,127/acre Terrace$3.57/ft Water & Sediment Control Basin$3,989/structure Grade Stabilization Structure$15,018/structure Filter Strip$116.83/acre Contour Buffer Strip$78/acre Riparian Forest Buffer$486/acre Wetland Restoration$245/acre Nutrient Management$4.09/acre Contour Farming$6/acre No Till$17.94/acre Continuous CRP$142/acre General CRP$97/acre

7. 7 Statewide Coverage PracticeStatewide Coverage Grass Waterway (NRI)2,225,900 acres Terrace (NRI)1,997,900 acres Contour Stripcropping (NRI)236,800 acres Contour Farming (NRI)5,148,200 acres Mulch Till (CTIC)8,290,000 acres No Till (CTIC)5,220,000 acres CRP (CRP program)1,894,488.2 acres

8. 8 Total Costs of the Practices PracticeCost Terraces$692,147,676 Grassed Waterways $95,090,424 Contour Farming $30,889,200 Contour Stripcropping $3,552,000 No-Till$104,308,740 Much Till $82,861,900 CRP$175,878,365 $37,194,949 $397,490,205  The first two practices are structural practices.  Divide the installation costs over the lifespan of the practices (terrace: 25yrs, GW: 10 yrs), then the sum of annual payment is: $37,194,949.  The cost numbers for the rest of the practices are annual payments.  Then the total annual costs would be: $37.2million + $397.5million = $434.7

9. 9 USDA payments for farms in Iowa ($million, From EWG) Year Conservation Payments Disaster Payments Commodity Payments Total USDA Payments 2000 $169.25$15.63$2,118.23$2,303.10 2001 $208.16$20.26$1,742.67$1,971.09 2002 $223.19$18.00$498.03$739.22 2003 $219.12$65.30$761.06$1,045.48 2004 $221.21$1.30$1,030.59$1,253.10 2005 $219.22$85.00$1,936.61$2,240.83

10. 10 13 Iowa Watersheds

11. 11 % Improvement due to Existing Practices FlowNitrateOrg NMin POrg PTotal NTotal P Boyer-8235645503848 Des Moines-8143929371533 Floyd-4194241442542 Iowa-510410401325 Little Sioux-7205240502447 Maquoketa84237421739 Monona-3154954612658 Nishnabotna-3215245473346 Nodaway-2285649513750 Skunk-6144640442142 Turkey-553830371834 Upper Iowa-374838471845 Wapsipinicon-864634451140

12. 12 Various Nutrient Criteria and Measured Levels (for basis of comparison) Nitrate (mg/L) Total N (mg/L) Total P ug/L EPA Ecoregional Nutrient Criteria (Lakes) 0.7837.5 EPA Ecoregional Nutrient Criteria (Rivers) 2.1876.25 Drinking Water Limit 10 5-year average for Raccoon River in 1978 4.5 5-year average for Raccoon River in 2004 7.7 Projected average, Cedar River in 1940s 2 Recent concentrations, Cedar River 6

13. 13 What would it take to improve water quality? Numerous conservation practices could be implemented on any field, each with different levels of effectiveness and costs Numerous conservation practices could be implemented on any field, each with different levels of effectiveness and costs Solving for the least-cost solution requires comparison among a very large number of possible land use scenarios Solving for the least-cost solution requires comparison among a very large number of possible land use scenarios

14. 14 Search Space If there are N conservation practices possible for adoption on each field and there are F fields, this implies a total of possible N F configurations to compare. If there are N conservation practices possible for adoption on each field and there are F fields, this implies a total of possible N F configurations to compare. In a watershed with 20 fields, if we consider only two possible options: implement a BMP or leave it alone, we would have to evaluate 2 20 possibilities (over 1 million) In a watershed with 20 fields, if we consider only two possible options: implement a BMP or leave it alone, we would have to evaluate 2 20 possibilities (over 1 million) With 30 fields would need to evaluate over 1 billion possibilities With 30 fields would need to evaluate over 1 billion possibilities

15. 15 Approach Evolutionary algorithms provide one search strategy Evolutionary algorithms provide one search strategy mimic the power of evolution, which, in effect, is a method of searching for solutions among an enormous amount of possibilities (Mitchell, 1999) mimic the power of evolution, which, in effect, is a method of searching for solutions among an enormous amount of possibilities (Mitchell, 1999) Outline basic idea only Outline basic idea only

16. 16 Pareto Optimality Point A is not clearly preferred to point B (since B has lower costs, but higher pollution) Point A is not clearly preferred to point B (since B has lower costs, but higher pollution) Point C ought to be preferred to point D (since D exhibits both lower costs and lower pollution levels) Point C ought to be preferred to point D (since D exhibits both lower costs and lower pollution levels) We try to identify a frontier like ABC and then push it toward the origin We try to identify a frontier like ABC and then push it toward the origin

17. 17 Why a Pareto frontier? While a Pareto-optimal frontier by itself does not indicate the solution that the public should choose, it allows for two important questions to be answered: While a Pareto-optimal frontier by itself does not indicate the solution that the public should choose, it allows for two important questions to be answered: 1. What is the minimum cost of achieving a certain level of pollution, and 2. What is the minimum level of pollution that can be achieved with a given conservation budget?

18. 18 Flow diagram

19. 19 Land use options considered Option #Option descriptionAssumed per acre cost, $/acre 1CRP (land retirement)110 2Corn-Soybeans, Conventional Tillage0 3Corn-Soybeans, Conventional Tillage, 20% Fertilizer ReductionRF (varies) 4Corn-Soybeans, No-Till20 5Corn-Soybeans, No-Till, 20% Fertilizer Reduction20+RF 6Corn-Soybeans, Conventional Tillage, Terracing18 7Corn-Soybeans, Conventional Tillage, Terracing, 20% Fertilizer Reduction 18+RF 8Corn-Soybeans, No-Till, Terracing38 9Corn-Soybeans, No-Till, Terracing, 20% Fertilizer Reduction38+RF 10Corn-Soybeans, Conventional Tillage, Contouring15 11Corn-Soybeans, Conventional Tillage, Contouring, 20% Fertilizer Reduction 15+RF 12Corn-Soybeans, No-Till, Contouring35 13Corn-Soybeans, No-Till, Contouring, 20% Fertilizer Reduction35+RF

20. 20 Land use options considered 1.Land Retirement 2.Corn-Soybeans, Conventional Tillage 3.Corn-Soybeans, No-Till 4.Corn-Soybeans, 20% Fertilizer Reduction 5.Contouring 6.Terracing 7.Combinations (as sensible) of above practices

21. 21 Costs of scenarios and baseline Costs for land use options drawn from Kling et al. (2005) Costs for land use options drawn from Kling et al. (2005) Cost of 20% N fertilizer reduction is derived: Cost of 20% N fertilizer reduction is derived: Using corn yield curves (J. Sawyer) Using corn yield curves (J. Sawyer) Using baseline N application data (C. Wolter) Using baseline N application data (C. Wolter) Cost=Yield reduction*Corn price Cost=Yield reduction*Corn price At rates where yield is predicted to be 150 bu/acre, cost=$4.43/acre At rates where yield is predicted to be 150 bu/acre, cost=$4.43/acre Cost varies with the N application data by 8-digit HUC Cost varies with the N application data by 8-digit HUC Baseline options are costed at the same rates Baseline options are costed at the same rates Alternative crop rotations are costed based on ISU Extension production budgets Alternative crop rotations are costed based on ISU Extension production budgets At assumed corn prices of $2.4 and soybean prices of $6.0 all rotations carry a cost relative to the Corn-Soybean rotation At assumed corn prices of $2.4 and soybean prices of $6.0 all rotations carry a cost relative to the Corn-Soybean rotation Cost of baseline=Cost of practices+Losses from other rotations Cost of baseline=Cost of practices+Losses from other rotations

22. 22 Baseline land use Watershed Area (km 2 ) PastureCrop # of NRI Points % of watershed Floyd2,3871284 646 Monona2,3381480 396 Little Sioux8,3561082 1,929 Boyer2,8302268 602 Nishnabotna7,6752271 1,500 Nodaway2,0683252 432 Des Moines36,1251170 8,154 Skunk10,8412061 2,332 Iowa31,528776 8,072 Wapsipinicon6,4341871 1,515 Maquoketa4,7812855 1,210 Turkey4,3532255 913 Upper Iowa2,4441953 658

23. 23 State-wide results Apply the evolutionary algorithm to the 13 watersheds Apply the evolutionary algorithm to the 13 watersheds A challenging task, given the number of HRUs and current computing capacity A challenging task, given the number of HRUs and current computing capacity Results presented are based on over 91 days of CPU time Results presented are based on over 91 days of CPU time Over 116,000 SWAT model runs Over 116,000 SWAT model runs Overall, the algorithm is able to identify scenarios which result in significant reductions in loadings of nitrates and phosphorus relative to baseline Overall, the algorithm is able to identify scenarios which result in significant reductions in loadings of nitrates and phosphorus relative to baseline Results must be interpreted with caution: Results must be interpreted with caution: it is possible that better solutions could be identified, given enough CPU time it is possible that better solutions could be identified, given enough CPU time

24. 24 Baseline Results WatershedGenerationsHRUs Base Cost, $Base NO3, kg Base Phosphorus, kg Baseline NO3 concentration, mg/L Baseline P concentra- tion, mg/L Boyer64642510,241,7002,478,9561,086,5443.781.22 Des Moines782122344,099,40070,250,4002,538,7045.830.24 Floyd11615245,277,7502,420,400566,1826.550.92 Iowa625205583,576,10080,908,0003,623,2607.960.28 Little Sioux596187921,192,70011,512,8001,392,4805.020.73 Maquoketa500204117,999,100490,02036,6585.770.19 Monona20743799,245,0801,974,400331,7266.210.77 Nishnabotna501299729,903,1009,996,1203,075,1803.860.98 Nodaway6125934,770,1701,967,580574,3623.090.71 Skunk526328421,182,90013,882,4002,896,4603.700.80 Turkey460164015,350,4007,277,0001,376,3804.300.54 Upper Iowa6926647,297,9702,760,780215,8022.320.16 Wapsipinicon588314120,918,30016,150,800610,9605.690.20

25. 25 Maximum reductions observed WatershedGenerationsHRUsMax NO3 reduction,%Max P reduction,% Boyer64642535.048.7 Des Moines782122326.951.4 Floyd116152455.270.4 Iowa625205525.032.9 Little Sioux596187941.043.2 Maquoketa500204150.066.5 Monona207437956.160.0 Nishnabotna501299735.532.7 Nodaway61259330.343.2 Skunk526328425.936.4 Turkey460164026.928.7 Upper Iowa69266429.041.9 Wapsipinicon588314132.032.4

26. 26 Nutrient Targeting strategies To be consistent across the watersheds, we focus on 25% reduction of nitrate loading and on 40% reduction in total phosphorus loading To be consistent across the watersheds, we focus on 25% reduction of nitrate loading and on 40% reduction in total phosphorus loading Potentially, have 3 distinct scenarios in each watershed: Potentially, have 3 distinct scenarios in each watershed: Focus on NO 3 without regard to P Focus on NO 3 without regard to P Focus on P without regard to NO 3 Focus on P without regard to NO 3 Focus on both P and NO 3 simultaneously Focus on both P and NO 3 simultaneously

27. 27 Targeting strategies (cont’d) However, in our analysis, we find that However, in our analysis, we find that If we focus on P, NO 3 is achieved automatically If we focus on P, NO 3 is achieved automatically We believe this is due to having 20% N reduction interacted with practices which should help reduce P We believe this is due to having 20% N reduction interacted with practices which should help reduce P In the following presentation, focus on 2 scenarios for each watershed: In the following presentation, focus on 2 scenarios for each watershed: N-targeting: aim to reduce nitrate loadings by 25% (whatever the consequences for P may be) N-targeting: aim to reduce nitrate loadings by 25% (whatever the consequences for P may be) P-N-targeting: aim to reduce total P loadings by 40% AND reduce nitrate loadings by 25% P-N-targeting: aim to reduce total P loadings by 40% AND reduce nitrate loadings by 25%

28. 28 State-wide Results for N-targeting Watershed Nitrate loadings, kg NO3, % of baseline NO3 concentration, mg/LGross Cost, $ Cost relative to baseline, $ P loadings, in % of baseline P concentration, mg/L Boyer1,865,22275.22.3437,847,016-2,394,684112.21.511 Des Moines52,568,80074.84.775182,563,053138,463,65368.30.151 Floyd1,186,756494.0465,705,151427,401850.878 Iowa60,808,00075.26.156164,070,15580,494,055840.234 Little Sioux8,254,22071.74.13314,264,234-6,928,46688.50.768 Maquoketa300,00056.2143.717569,241-17,429,859162.7580.348 Monona1,168,60059.24.3105,837,135-3,407,94583.30.796 Nishnabotna7,526,26075.32.68323,730,976-6,172,124127.21.228 Nodaway1,490,60075.82.1634,012,947-757,224121.20.844 Skunk10,421,80075.12.67737,496,07516,313,175100.10.756 Turkey5,461,000752.96910,206,733-5,143,668120.20.658 Upper Iowa2,051,76074.31.6147,226,572-71,398129.50.220 Wapsipinicon10,000,00075.49844.2738,320,590-12,597,710167.9460.341

29. 29 State-wide summary of N-targeting Focusing on nitrates exclusively, without any regard for phosphorus levels leads to an increase in total phosphorus loadings in 8 out of 13 watersheds Focusing on nitrates exclusively, without any regard for phosphorus levels leads to an increase in total phosphorus loadings in 8 out of 13 watersheds State-wide gross cost of reducing nitrates: about $472 million annually (on cropland HRUs) State-wide gross cost of reducing nitrates: about $472 million annually (on cropland HRUs) We do not consider non-cropland HRUs (like CRP) in the search process We do not consider non-cropland HRUs (like CRP) in the search process Existing CRP adds to the cost Existing CRP adds to the cost Cost relative to baseline: $181 million annually Cost relative to baseline: $181 million annually The total loading of nitrates from the watersheds in the state would be reduced by 27% The total loading of nitrates from the watersheds in the state would be reduced by 27% while the loadings of phosphorus would increase by about 12% while the loadings of phosphorus would increase by about 12%

30. 30 State-wide results: P-N targeting Watershed P loading, kg Actual P loading, % of baseline P concentration, mg/LGross Cost, $Net cost, $ N loadings, in % of baseline N concentration, mg/L Boyer648,39859.70.74214,950,5854,708,88566.22.232 Des Moines1,482,27058.3870.129190,010,000145,910,60075.90484.514 Floyd295,67652.20.5608,056,5822,778,83264.24.494 Iowa2,432,16067.10.206193,575,632109,999,53275.26.189 Little Sioux800,92457.50.49133,142,35011,949,65064.23.617 Maquoketa14,58939.80.3232,218,699-15,780,40160.63.682 Monona200,11660.30.5467,753,342-1,491,73855.14.158 Nishnabotna2,068,92067.30.62543,579,66213,676,56265.12.408 Nodaway354,08861.60.4018,140,2473,370,07772.92.049 Skunk1,842,72063.60.48250,379,56129,196,66175.22.595 Turkey981,90071.30.38117,057,4901,707,090742.999 Upper Iowa130,64060.50.10011,038,4593,740,48975.81.635 Wapsipinicon413,10067.60.13533,030,35912,112,059703.861

31. 31 State-wide summary of P-N targeting The total gross cost of implementing the management scenarios is almost $613 million a year The total gross cost of implementing the management scenarios is almost $613 million a year The net cost is estimated to run just under $322 million a year The net cost is estimated to run just under $322 million a year Following the prescriptions of the algorithm for each of the watersheds results in the state-wide reduction in phosphorus loadings of over 36% Following the prescriptions of the algorithm for each of the watersheds results in the state-wide reduction in phosphorus loadings of over 36% would simultaneously result in the state-wide reduction in nitrate loadings of over 31% would simultaneously result in the state-wide reduction in nitrate loadings of over 31%

32. 32 Patterns across watersheds Some patterns can be noted, which appear to hold across the entire state Some patterns can be noted, which appear to hold across the entire state The CS/Conventional Till area in each watershed is higher in all of the watersheds for the N-targeting scenarios both relative to the P-targeting scenario and relative to the baseline The CS/Conventional Till area in each watershed is higher in all of the watersheds for the N-targeting scenarios both relative to the P-targeting scenario and relative to the baseline The area in which no-till is used is significantly smaller for the N-targeting scenarios across all of the watersheds relative to the P-N-targeting scenarios The area in which no-till is used is significantly smaller for the N-targeting scenarios across all of the watersheds relative to the P-N-targeting scenarios The same is true relative to the baseline in all of the watersheds except Floyd The same is true relative to the baseline in all of the watersheds except Floyd The area in which terracing is used is smaller in the N-targeting scenarios than in the P-N-targeting scenarios The area in which terracing is used is smaller in the N-targeting scenarios than in the P-N-targeting scenarios No strong pattern relative to baseline No strong pattern relative to baseline The area where contour cropping is used is smaller in the N-targeting scenarios than in the P-N-targeting scenarios (except in Des Moines) The area where contour cropping is used is smaller in the N-targeting scenarios than in the P-N-targeting scenarios (except in Des Moines) no consistent pattern relative to baseline data no consistent pattern relative to baseline data

33. 33 Comparison of selected watersheds: Des Moines

34. 34Nodaway

35. 35 Extracting two scenarios

36. 36 Distributions of options (alleles) and practices

37. 37 N Map Scenarios

38. 38 Scenario Maps Scenario Maps

39. 39 Nodaway

40. 40 Distributions of options (alleles) and practices

41. 41 Conclusions and Recommendations Significant reductions in both phosphorus and nitrogen loadings can be achieved in all of the watersheds in the state Significant reductions in both phosphorus and nitrogen loadings can be achieved in all of the watersheds in the state The costs are significant The costs are significant But, much smaller than they would be if didn’t carefully select which practices go where (targeting) But, much smaller than they would be if didn’t carefully select which practices go where (targeting) Focus on N alone produces a markedly different watershed configuration than when the focus is both on P and N Focus on N alone produces a markedly different watershed configuration than when the focus is both on P and N Practices which are efficient for phosphorus and erosion control are, in general, not efficient in controlling nitrogen Practices which are efficient for phosphorus and erosion control are, in general, not efficient in controlling nitrogen If we wish to focus to both nutrients, combining traditional structural practice with measures directly targeting nitrogen will be necessary If we wish to focus to both nutrients, combining traditional structural practice with measures directly targeting nitrogen will be necessary

42. 42 Caveats and Research Needs Inclusion of other options will very likely alter results Inclusion of other options will very likely alter results Results needed to be interpreted with caution Results needed to be interpreted with caution E.g., we do not recommend actual removal of conservation practices or abandonment of conservation tillage E.g., we do not recommend actual removal of conservation practices or abandonment of conservation tillage Lack of detailed spatial data only allows identification of desired conservation option mix at the subbasin level Lack of detailed spatial data only allows identification of desired conservation option mix at the subbasin level Better hydrologic, water quality, land use, and farm- level data would help to add realism to the analysis Better hydrologic, water quality, land use, and farm- level data would help to add realism to the analysis Studying sensitivity of the results to changes in costs, efficiencies of conservation practices and weather conditions would be very helpful in providing Studying sensitivity of the results to changes in costs, efficiencies of conservation practices and weather conditions would be very helpful in providing

43. 43 Continuing Work Sergey Rabotyagov’s PhD thesis, Iowa updates Sergey Rabotyagov’s PhD thesis, Iowa updates 2007 land and corn prices 2007 land and corn prices Added grassed waterways as an option Added grassed waterways as an option Continued improvements to SWAT Continued improvements to SWAT 30% N and P reductions ~ $300 million/yr cost 30% N and P reductions ~ $300 million/yr cost UMRB work (also Sergey) UMRB work (also Sergey) Smaller watershed work (field scale) Smaller watershed work (field scale) Walnut Creek (CEAP project – USDA) Walnut Creek (CEAP project – USDA) Boone (USDA-CSREES) Boone (USDA-CSREES) Application of GA for biofuels sustainability Application of GA for biofuels sustainability