Gravel bars are sites of increased CO2 outgassing in stream corridors

Streams are significant sources of CO2 to the atmosphere. Estimates of CO2 evasion fluxes (f CO2) from streams typically relate to the free flowing water but exclude geomorphological structures within the stream corridor. We found that gravel bars (GBs) are important sources of CO2 to the atmosphere, with on average more than twice as high f CO2 as those from the streamwater, affecting f CO2 at the level of entire headwater networks. Vertical temperature gradients resulting from the interplay between advective heat transfer and mixing with groundwater within GBs explained the observed variation in f CO2 from the GBs reasonably well. We propose that increased temperatures and their gradients within GBs exposed to solar radiation stimulate heterotrophic metabolism therein and facilitate the venting of CO2 from external sources (e.g. downwelling streamwater, groundwater) within GBs. Our study shows that GB f CO2 increased f CO2 from stream corridors by [median, (95% confidence interval)] 16.69%, (15.85–18.49%); 30.44%, (30.40–34.68%) and 2.92%, (2.90–3.0%), for 3rd, 4th and 5th order streams, respectively. These findings shed new light on regional estimates of f CO2 from streams, and are relevant given that streamwater thermal regimes change owing to global warming and human alteration of stream corridors.

2 calcareous rock (hydraulic conductivity: 8 × 10 -5 m s -1 ), with a median sediment size of 23.1 mm and an 9 average porosity of 29%; reach slope is 0.41%. The typical hydrology, physiochemistry 57 , and stream and 10 hyporheic fCO2 9 have been described in detail by other authors.

11
The catchment study was conducted across a range of streams of varying size, discharge, bed-gradient, 12 stream order and degree of anthropogenic influence. The course of the Grosse Erlauf and to a lesser 13 extent, lower section of the Ybbs River catchments ( Figure S1) are interrupted by hydraulic constructions. 14 The Bodingbach (Bod) and Ybbs-Steinbach (Ysteinb) locations (Table S2)  chambers, over each chamber, and placing a thick bag over the ring and CO2 chamber within -filling the 27 plastic covered space between the chamber and ring with water which was acclimatized to the air 28 temperature. This also had the benefit of reducing heating of the chambers beyond ambient temperature 29 when exposed to the sun, potentially altering fCO2 measurements. We chose free-flowing streamwater 30 sampling locations with low levels of turbulence in order to avoid possible formation of eddy currents 31 below the chamber which could lead to over-estimation of streamwater fCO2 75 . 32 33 Physiochemical sampling and groundwater hydraulic flux: In OSB, we monitored stream and 34 groundwater levels (HT Type 255-Hydrotechnik GmbH, Limburgrange: 1 -300m, accuracy: < 0.05% 35 measured value) at 10 and 30 minute intervals, respectively. Hydraulic heads were monitored (every 10 36 min) in 23 wells distributed over the GB using Trutrack ® WT-HR 1500 capacitive water level sensors 37 (Tru Track Ltd, Christchurch, NZrange: 0 -1500mm, accuracy: ± 1mm). Sediment hydraulic 38 conductivity across the OSBGB was determined via the Hvorslev slug method 76 .

39
We conducted diurnal sampling of porewater from piezometers at 2 depths (0.75 and 1.25 m below GB 40 surfacereference: GB crest) next to vertical temperature monitoring locations. Porewater was slowly 41 pumped into 250 mL pre-combusted Schott © bottles via a through-flow system for DOC (Sievers TOC   42 Analyzer GE) and electrical conductivity (WTW Cond 3310, Weilheim, Germany) determination.

43
The hydraulic conductivity of GB sediments in OSB was measured using the Hvorslev slug method 76 .

44
Average downwelling stream flux travel times, estimated by the lag times between electrical conductivity 45 time series in the streamwater and the hyporheic zone 77 at 0.75m and 1.25m below the GB surface, 46 approximated 10.2 h, 1.5 h and 1.3 h to the head, crest and tail, respectively (Table S1). The percent 47 groundwater contribution to the gravel bar sub-surface was calculated according to a two-component 48 mixing equation 78 : X1 = (Cmix -C2) / (C1 -C2)  100, where: X1 is the proportion (%) of groundwater; 49 Cmix is the specific electrical conductivity of the sampled porewater within the gravel bar; C1 is the 50 specific electrical conductivity of the sampled groundwater and C2 is the is the specific electrical 51 conductivity of the streamwater.

53
Upscaling of CO2 sources and fluxes: We conducted in-stream roaming surveys across 8 mid-order 54 streams (2 nd to 5 th order) within the Ybbs catchment (~ 0.25 -0.60 km for 2 nd order streams and ~1.0 km 55 for 3 -5 th order streams) to determine the prevalence and contribution of GBs to stream corridor CO2 56 outgassing within the stream network ( Figure S1). Gravel bar area was estimated from GB maximum 57 length and width, as the area of an ellipse (A = π ab), where "A" is the area of an ellipse, "a" is the 58 distance from the center to a vertex and "b" is the distance from the center to a co-vertex. The area within 59 the wetted stream boundary was calculated from the mean stream width (6 equally spaced cross-sections 60 were taken during each survey) and the total surveyed length of each stream within the specified stream 61 order class. The percentage GB cover within the stream corridor was determined according to equation

62
(1). We assumed that the average GB fCO2 : SW fCO2 ratio for each stream order, determined during our     Tables  229  230  231   232   Table S1 : Seasonal OSB gravel bar and stream physiochemical properties along 2 vertical depth planes (0.75m and 233 1.25m below the gravel bar surfacereference: gravel bar middle). Shown is the mean, standard deviation (s.d.) and