1. Midwestern rivers: Are they hotspots, buffers, or sentinels of agricultural land use? J. L. Tank 1, M. Dee 1, A. Marzadri 2,3, D. Tonina 2, A. Bellin 3 1 University of Notre Dame, 2 University of Idaho, 3 University of Trento, Italy F UNDING SOURCES

2. Goal: reduce “dead zone” in Gulf of Mexico caused by agricultural land use in Midwestern corn belt Excess Nutrients Algal Blooms Algal Death & Decomposition Water Column Hypoxia Gulf of Mexico (NO 3 - ) Summer 2015 = 16,770 km 2 (> avg) = “Connecticut + Rhode Island” “Dead Zone”: 5 yr avg (2009-2013) = 14,000 km 2. Call by EPA Mississippi River/Gulf of Mexico Hypoxia Task Force: reduce size to 5000 km 2 State Action Plans = Nutrient Loss Reduction Strategies to achieve 45% reduction in NO 3 - export (by 2025).

3. Mechanism: runoff entering streams/ditches draining row crops Example: Indiana land cover >80% ag, 90% of >48,000 km of streams are located within 500m of a field. Channelization, tile drainage, and excess fertilizer increase N losses. Peak run-off often occurs during spring snowmelt and storms in hydrologically “flashy” systems.

4. Wabash River Basin (WRB) = “hotspot” for N export from MRB. WRB drains over 103,500 km 2 in IN, OH, IL (>65% ag land use) and currently contributes a disproportionate N load to the Gulf of Mexico relative to its area in the MRB. “big water year” = WRB exports >200K metric tons (200M kg) N

5. Rivers can also remove and transform nutrients Results: riverine demand not always saturated & V f comparable to streams. Network Model (Ye et al. in prep) shows % riverine nutrient removal > than predicted by flowpath length  bioreactive hotspots in fluvial networks. Prediction: high inorganic nutrient loads saturate nutrient removal capacity.

6. Synoptic sampling: characterize role of agricultural land use in two contrasting watersheds Goal: identify riverine nutrient signature in context of headwater streams. Approach: use seasonal, synoptic sampling in two river networks with contrasting land use. Across each watershed, chose 80 stream/river sites: ~30 % mainstem and ~60% tributaries. For each surge, sampled over 24hrs  “snapshot” of the stream network for water chemistry, dissolved gasses, and physical characteristics. TIP Site A7, Winter2014 MAN Site A4, Summer2015 TIP Site C7, Winter2014 MAN Site C14, Winter2014

7. INDIANA Tippecanoe River TRIBUTARY MAINSTEM 82% AGRICULTURAL (AG) MICHIGAN TRIBUTARY MAINSTEM Manistee River 83% FORESTED (FOR) Synoptic sampling site Synoptic sampling site Study sites: Mainistee R (MI) and Tippecanoe R (IN)

8. AGRICULTURAL (AG) FORESTED (FOR) MAINSTEM SITES TRIBUTARY SITES Su Wi Sp Fa Su Wi Sp Fa SEASON [NO 3 -N] 10X higher in AG watershed, and highest in winter due to runoff. No seasonal differences in FOR watershed. Tribs more variable than river mainstem. [N 2 O] higher in AG, also higher/more variable in tribs, but few seasonal differences,. In FOR, seasonally, winter/spring higher in both mainstem and tribs. MAINSTEM SITES TRIBUTARY SITES Nitrate Concentration (μg/L) AGRICULTURAL (AG) FORESTED (FOR) Su Wi Sp Fa Su Wi Sp Fa SEASON a b a a [NO 3 -N] a b a a N2ON2O Results: NO 3 - and N 2 O show distinct seasonal, spatial patterns N 2 O Concentration (μmol/L)

9. Predicted Measured Model the spatial distribution of dissolved nutrients and gasses Combine seasonal empirical observations with network-based geostatistical models to visualize how patterns occur across scale. Modeling Approach: STARS (Spatial Tools for the Analysis of River Systems) in ArcGIS Geostatistical analysis using SSN (Spatial Stream Network) model in R to fit spatial models based on network structure. Using framework from McGuire et al. (2014) to quantify spatial patterns using empirical semivariograms that incorporate network topology.

10. AGRICULTURAL [NO 3 -N] FORESTED N2ON2O Spring 2015: Spatial variability at the stream-network scale Both Ag and For watersheds, saw broad-scale gradients in [NO 3 -N]. Contrasts with small-scale “patchiness” for [N 2 O]. DN- weak relationship- also influenced by temp & hydrology. Small-scale “patchiness” Broad-scale gradient

11. River transformations: relative role of mainstem vs. tribs in linkage between NO 3 - and N 2 O Density plots comparing tribs vs. mainstem rivers. Tribs = greater range in conc, tightly linked to surrounding land use, acting as “sentinels” of ag impact. Rivers= lower conc, act as “integrators”, buffering upstream impacts.

12. Marzadri et al. unpublished Next steps: mechanistic model to predict N 2 O production in stream network. Exploring inverse relationship between N 2 O and stream size. For rivers, suggests reduced role of hyporheic zone in N 2 O production  increasing role of water column production.

13. What can we do? Restoration of river floodplains (reactive) From 1990s-2012: TNC & NRCS restored ~29K floodplain acres Increased connectivity btwn river and floodplains improves N retention via increased WRT and biological processing. Literature review + N export data for WRB (via USGS): floodplains potentially reduce N export by 7%. To reach 45% reduction = ~180,000 acres …costly!

14. Goal: maintain ag production, but prevent nutrient runoff to streams and rivers. Since 2013: watershed- scale cover crops. Cumulative NO 3 - export: cover crops = drought yr. Annual load reductions = 29%-43%. Pre (320 acres) vs. Post (~1600 acres) Cover Crops What can we do? Watershed-scale cover crops (preventative)

15. Source: USGS Data: Shiklomanov 1993 in “Water in Crisis”, P. Gleick (ed) 2120 km 3 0.006% of FW Signature of land use impact clear: [NO 3 - ] strongly influenced by agriculture, tribs are interfaces and can be hotspots. Rivers are bioreactive and integrate signature of land use; they remove and transform nutrients, altering patterns of solute and gas transport. Ag impacts on rivers can be mitigated via floodplain restoration. Nutrient impacts can be prevented via watershed-scale conservation (e.g., cover crops)  decrease runoff. Bold strategies will be needed to solve environmental challenges at ag/fw interface. Take Homes: Global Riverine Freshwater:

16. Midwestern rivers: Are they hotspots, buffers, or sentinels of agricultural land use? Jennifer Tank, Martha Dee, Alessandra Marzadri, Daniele Tonina, Alberto Bellin The Midwest has undergone extensive land use change as forest, wetlands, and prairies have been converted to agroecosystems, and excess nutrients are a key stressor in agricultural river networks. Given their nutrient status, we expected that Midwestern rivers would have compromised nutrient removal capacity, yet recent research has shown that riverine demand is not necessarily saturated. Nutrient uptake per unit distance was higher in rivers compared to well-studied headwater streams, suggesting they act as hotspots of nutrient processing. In order to understand the role of riverine nutrient dynamics in light of their headwaters, we studied two river networks with contrasting land use using a seasonal synoptic sampling regime. As expected, we found that nutrient concentrations were strongly influenced by agriculture, but river mainstems also acted as land use “buffers” compared to impacted tributaries. Finally, while rivers can act as sentinels of land use change in their catchments, agricultural impacts on rivers can be mitigated. We estimate that recent restoration of ~30,000 acres of floodplain in the Wabash River Basin may equate to measurable regional reductions in nutrient and sediment export.