Trends in Pesticide Concentrations in Corn- Belt Streams, 1996-2006 Skip Vecchia, Dan Sullivan, Bob Gilliom, Jeff Martin, Dave Lorenz National Water Quality Monitoring Conference April 28, 2010 NAWQA Topics ¾Study background • Trend assessment technique • Summary of atrazine and metolachlor concentration trends • Relation between concentration trends and pesticide use 2 USGS Report By D.J. Sullivan A.V. Vecchia D.L. Lorenz J.D. Martin R.J. Gilliom http://pubs.usgs.gov/sir /2009/5132/ 3 Study Objectives 1) Evaluate the performance and applicability of statistical trend-assessment methods for pesticide data. 2) Assess trends in concentrations of commonly occurring pesticides in selected Corn Belt streams. 3) Assess the relative influences of pesticide use patterns and other potential factors, such as changes in management practices, on pesticide trends 4 31 USGS pesticide sampling sites in Cornbelt (culled from a national network of 201 NAWQA sites) 5 Topics • Study background ¾Trend assessment technique • Summary of atrazine and metolachlor concentration trends • Relation between concentration trends and pesticide use 6 Acetochlor data for Mad River and Platte River 3 uncensored concentration MAD o censored concentration 2 Trend-Method 1 Requirements 0 -1 • Repeatable -2 o o oo oo o ooo oo o oooo ooo o oo oo oo o ooo o o oo oo ooooo oo oo o o o ooo -3 • Unbiased -4 -5 • Efficient 2000 2001 2002 2003 2004 2005 2006 4 uncensored concentration PLATTE o censored concentration • Robust: 3 2 •Censoring 1 •Sparse and 0 changing sampling -1 •High variability o o -2 oo o oo oo oooooo o ooooo oo -3 •Changing flow CONCENTRATION, AS BASE-10 LOGARITHM OF MICROGRAMS PER LITER -4 2000 2001 2002 2003 2004 2005 2006 7 Acetochlor data for Mad River and Platte River 3 uncensored concentration SEAWAVE-Q MAD o censored concentration 2 fitted concentration (SEAWAVE-Q) trend “seasonal wave 1 w/streamflow 0 -1 adjustment” -2 o o oo oo o ooo oo ooooo ooo o oo oo oo o ooo o o oo oo ooooo oo oo o o oooo -3 -4 Statistical model -5 2000 2001 2002 2003 2004 2005 2006 for concentration 4 uncensored concentration with: PLATTE o censored concentration 3 fitted concentration (SEAWAVE-Q) trend • Seasonal 2 1 variability 0 • Flow-related -1 variability -2 o o oo o oo oo oooooo o ooooo oo -3 • Trend CONCENTRATION, AS BASE-10 LOGARITHM OF MICROGRAMS PER LITER -4 2000 2001 2002 2003 2004 2005 2006 • Error 8 SEAWAVE-Q Model “Seasonal wave with streamflow adjustment” Log(Conc) = SWAVE + FLRV + TND + ERR ¾ SWAVE = γ0 + γ1 W(t) W(t) is a periodic function (period=1 yr) (d/dt) W(t) = λ(t +s) - φ W(t) λ(.) is a pulse input with λ(.)>0 during the “application season(s)” and λ(.)=0 otherwise s is a seasonal shift φ is a “decay rate” 9 Seasonal wave for acetochlor – Illinois River at Ottawa, IL (2000-2006) 2-Month Application Season (Apr-June) Decay rate too fast Decay rate “just right” 2 2 ILLI-OTT uncensored concentration ILLI-OTT uncensored concentration o censored concentration o censored concentration 1 seasonal wave 1 seasonal wave 0 0 -1 -1 o o o o -2 o -2 o o oooooo ooooooo o oooooo ooooooo OF MICROGRAMS PER LITER -3 -3 JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC CONCENTRATION, AS BASE-10 LOGARITHM 10 Seasonal wave for simazine - Ohio River at Cannelton Dam (2000-2006) Two, 2-month application 6-month application seasons (Oct-Dec and season (Dec-June) Apr-June) 1 1 OHIO-CA uncensored concentration OHIO-CA uncensored concentration o censored concentration o censored concentration seasonal wave seasonal wave 0 0 -1 -1 o o o o o o -2 -2 ooooo o ooooo o OF MICROGRAMS PER LITER -3 -3 JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC JAN FEB MAR APR MAY JUNE JULY AUG SEPT OCT NOV DEC CONCENTRATION, AS BASE-10 LOGARITHM Poor fit – model rejected Optimum fit – model accepted 11 SEAWAVE-Q Model Log(Conc) = SWAVE + FLRV + TND + ERR ¾ FLRV = γ2 LTFA(t) + γ3 MTFA(t) + γ4 STFA(t) “Flow-related variability” Flow anomalies computed using log- transformed daily flow for USGS streamgage LTFA(t) “long-term flow anamoly” ¾ Low-frequency (annual) flow variability MTFA(t) “mid-term flow anomaly” ¾ Mid-frequency (seasonal) flow variability STFA(t) “short-term flow anomaly” ¾ High-frequency (day-to-day) flow variability 12 Acetochlor for Platte River at Louisville, NE Without FLRV With FLRV 3 3 PLATTE uncensored SEAWAVE PLATTE uncensored SEAWAVE-Q o o 2 censored trend 2 censored flow-adjusted trend 1 1 0 0 -1 -1 -2 o o -2 o o ooo oo oo oooooo oooooo oo ooo oo oo oooooo o ooooo oo -3 -3 OF MICROGRAMS PER LITER -4 -4 2000 2001 2002 2003 2004 2005 2006 2000 2001 2002 2003 2004 2005 2006 CONCENTRATION, AS BASE-10 LOGARITHM Significant (p=0.006) No significant flow- upward trend (28 percent adjusted trend per year) 13 SEAWAVE-Q Model Log(Conc) = SWAVE + FLRV + TND + ERR ¾ TND = γ5 t Log-linear trend “Flow-adjusted trend” (adjusted for FLRV) ¾ ERR Random error Assumed normal, uncorrelated Model parameters estimated using maximum likelihood with censored data (“Survival” or “Tobit” regression) 14 Topics • Study background • Trend assessment technique ¾Summary of atrazine and metolachlor concentration trends • Relation between concentration trends and pesticide use 15 31 Pesticide Sampling Sites on Corn-Belt Streams 16 Average daily Average annual Flow-adjusted trends in concentration flow, 1996-2006 flux, 1996-2006 Estimated value (m^3/s) (kg/yr) ATRAZINE <80 <4,000 90-percent confidence limits TRENDS FROM 800 40,000 nonsignificant significant highly significant 8,000 400,000 (p>0.1) (p<0.1) (p<0.01) SEAWAVE-Q 1996-2002 2000-2006 1. MAD 2. OHIO-CA Ohio Basin 3. LBUCK 4. SUG-NP 5. WHITE 6. WABASH 7. OHIO-GRCH 8. DUCK Great Lakes 9. MILW na 10. CLINT na 11. STJOS Basins 12. AUGL 13. MAUM 14. LCOBB 15. MSSP-HA 16. MSSP-CL Upper 17. WAPS-TR 18. IOWA-ROW 19. SFIOWA Mississippi 20. IOWA-WAP 21. SUG-MIL Basin 22. DPLAI 23. ILLI-OTT 24. SANG 25. ILLI-VC 26. MSSP-GR 27. MIZZ-OM Missouri 28. MAPLE 29. PLATTE Basin 30. MIZZ-HE 31. MSSP-TH -60 0-+60 600 +60 TREND, IN PERCENT PER YEAR 17 Average daily Average annual Flow-adjusted trends in concentration flow, 1996-2006 flux, 1996-2006 Estimated value (m^3/s) (kg/yr) METOLACHLOR <80 <1,600 90-percent confidence limits TRENDS FROM 800 16,000 nonsignificant significant highly significant 8,000 160,000 (p>0.1) (p<0.1) (p<0.01) SEAWAVE-Q 1996-2002 2000-2006 1. MAD 2. OHIO-CA Ohio Basin 3. LBUCK 4. SUG-NP 5. WHITE 6. WABASH 7. OHIO-GRCH 8. DUCK Great Lakes 9. MILW na 10. CLINT na 11. STJOS Basins 12. AUGL 13. MAUM 14. LCOBB 15. MSSP-HA 16. MSSP-CL Upper 17. WAPS-TR 18. IOWA-ROW 19. SFIOWA Mississippi 20. IOWA-WAP 21. SUG-MIL Basin 22. DPLAI 23. ILLI-OTT 24. SANG 25. ILLI-VC 26. MSSP-GR 27. MIZZ-OM Missouri 28. MAPLE 29. PLATTE Basin 30. MIZZ-HE 31. MSSP-TH -60 0-+60 600 +60 TREND, IN PERCENT PER YEAR 18 Topics • Study background • Trend assessment technique • Summary of atrazine and metolachlor concentration trends ¾Relation between concentration trends and pesticide use 19 Questions ¾How do pesticide trends in corn-belt rivers compare to changes in agricultural use? ¾Are there other factors (land use changes, management practices, non-ag use, etc.) that may be contributing? 20 Metolachlor and Atrazine Use 12 States, >80% Corn and Soybean Acreage 140,000,000 120,000,000 100,000,000 2000-2006 80,000,000 1996-2002 ATRAZINE 60,000,000 40,000,000 Pounds of active ingredient 20,000,000 METOLACHLOR 0 1992 1994 1996 1998 2000 2002 2004 2006 2008 S-metolachlor introduced 21 Graphic by USGS based on data provided by Dmrkynetec, Inc. Focus on 11 sites on major rivers Annual pesticide use estimates for 4 sites in Upper Mississippi each basin Basin obtained from GIS 3 sites in coverage of crop Missouri data and pesticide Basin application rates 3 sites in Ohio Basin Large basins ~ more accurate use estimates 1 site on Mississippi downstream of Missouri (but upstream of Ohio) 22 Metolachlor Concentration and Agricultural Use Trends Wabash River: 1996-2002 3 SUT = -11.9 (-14.3 to -9.4) pct/yr SCT = -16.8 (-21.6 to -11.6) pct/yr 2 1 0 -1 Log Concentration or Use -2 1996 1997 1998 1999 2000 2001 2002 EXPLANATION Uncensored concentration Use intensity Censored concentration Use intensity trend SEAWAVE-Q SCT Slope of concentration trend (with 90-pct confidence limits) Concentration trend SUT Slope of use intensity trend (with 90-pct confidence limits) 23 Concentration and Use Trends for Metolachlor 1996-2002 2000-2006 OHIO-CA Ohio WABASH Basin OHIO-GRCH MSSP-CL Upper IOWA-WAP Miss. Basin ILLI-VC MSSP-GR MIZZ-OM Misso. PLATTE Basin MIZZ-HE MSSP-TH -60 0-+60 600 +60 TREND, IN PERCENT PER YEAR Estimated concentration or agricultural use trend 90-pct confidence bounds for agricultural use trend 90-pct confidence bounds for concentration trend 24 Concentration and Use Trends for Atrazine 1996-2002 2000-2006 OHIO-CA Ohio WABASH Basin OHIO-GRCH MSSP-CL Upper IOWA-WAP Miss. Basin ILLI-VC MSSP-GR MIZZ-OM Misso. PLATTE Basin MIZZ-HE MSSP-TH -60 0-+60 600 +60 TREND, IN PERCENT PER YEAR Estimated concentration or agricultural use trend 90-pct confidence bounds for agricultural use trend 90-pct confidence bounds for concentration trend 25 Implications for major agricultural herbicides in large Cornbelt rivers ¾ A multi-year trend in watershed use intensity generally produces a similar trend in river concentrations.