The role of rivers and streams in the global carbon (C) cycle remains unconstrained, especially in headwater streams where CO2 evasion (FCO2) to the atmosphere is high. Stream C cycling is understudied in the tropics compared to temperate streams, and tropical streams may have among the highest FCO2 due to higher temperatures, continuous organic matter inputs, and high respiration rates both in-stream and in surrounding soils. In this paper, we present paired in-stream O2 and CO2 sensor data from a headwater stream in a lowland rainforest in Costa Rica to explore temporal variability in ecosystem processes. Further, we estimate groundwater CO2 inputs (GWCO2) from riparian well CO2 measurements and assess all fluxes to examine the relative contributions of sinks and sources of dissolved inorganic C (DIC) to a headwater stream. Paired O2 - CO2 data reveal stream CO2 supersaturation driven by groundwater CO2 inputs and large in-stream production of CO2. Areal fluxes in our study reach show FCO2 is supported by both GWCO2 inputs and in-stream metabolism and the seasonality in GWCO2 reflects the hydrology of the site. Using a mass balance approach, we show FCO2 is the dominant loss of DIC from the stream, greater than dissolved exports, and is sustained by both internal production of DIC and terrestrial inputs of DIC. Our results underscore the importance of tropical headwater streams as large contributors of greenhouse gases to the atmosphere among inland waters and show of this C derives from both in-stream and terrestrial sources.
Research Article
Partitioning Inorganic Carbon Fluxes from Paired O2 - CO2 Gas Measurements in a Neotropical Headwater Stream, Costa Rica
https://doi.org/10.21203/rs.3.rs-832711/v1
This work is licensed under a CC BY 4.0 License
You are reading this latest preprint version
tropical streams
carbon cycling
inorganic carbon
CO2 evasion
net ecosystem production
Inland waters play an important role in the global carbon (C) cycle and estimates of C fluxes between inland waters, the atmosphere, and terrestrial ecosystems are being revised at a rapid rate (Cole et al. 2007; Raymond et al. 2013; Drake et al. 2018; Tank et al. 2018). Streams, wetlands, and lakes transform, export, and store terrestrial C prior to delivery downstream to the ocean (Cole et al. 2007) and contribute CO2 to the atmosphere (Battin et al. 2009). These processes are particularly important in headwaters streams, which comprise 79% of stream length (Colvin et al. 2019) and disproportionately contribute to evasion of CO2 to the atmosphere on an areal basis (Raymond et al. 2013; Borges et al. 2019). However, these findings represent the synthesis from predominantly temperate systems, whereas tropical streams have received less study (Aufdenkampe et al. 2011; Oviedo-Vargas et al. 2015). As a result of higher year-round temperatures, high and continuous organic matter inputs, and drain soils with high respiration rates, tropical streams are likely to be hotspots of greenhouse gas emissions and fluxes (Borges et al. 2019). In-stream production of CO2 from processing of organic matter can further contribute to high CO2 fluxes (Richey et al. 2002; Mayorga et al. 2005; Hotchkiss et al. 2015).
Streams are generally supersaturated with respect to atmospheric CO2 (Wetzel and Likens 2000), reflecting large fluxes of CO2 into headwater streams of both terrestrially derived and internally produced CO2. Terrestrially derived CO2 originates from soil respiration and CO2-rich geologic formations and is transported to headwaters via overland, subsurface, and groundwater flow. In contrast, CO2 from internal processes is the result of respiration by in-stream heterotrophs. In headwater streams, the influence of terrestrially derived CO2 is typically greater than internal production and decreases in importance in larger streams and rivers (Hotchkiss et al. 2015). Losses of CO2 from streams include evasion to the atmosphere, uptake through photosynthesis, and hydrologic export downstream. Evasion of CO2 from freshwaters is a large flux of C sometimes unaccounted for in terrestrial budgets (Genereux et al. 2013) and is important at global scales (Raymond et al. 2013). Hydrologic export of dissolved inorganic carbon (DIC) includes all forms of C in equilibrium with the carbonate system, as aqueous CO2, H2CO3, HCO3-, and CO32--, which speciate according to pH (Stumm and Morgan 1996). High concentrations of CO2 may reduce pH by an amount that depends on the buffering capacity of the water (Wetzel and Likens 2000; Small et al. 2012) and accelerate dissolution of carbonate minerals (Stoddard et al. 1999).
Advances in sensor technology allow measurement of CO2 and O2 at high frequency in freshwaters (Johnson et al. 2010), permitting estimation of fluxes of these gases at different hydrologic conditions and as drivers of stream physicochemistry. Results from discrete sampling of evasion and lateral inputs of CO2 reveal important aspects of underlying processes and estimates of upscaled fluxes (Oviedo-Vargas et al. 2015). Aquatic sensors allow partitioning into sources and sinks at higher frequency and under conditions hard to capture using discrete sampling (e.g. floods). Further, coupling net ecosystem productivity (NEP) data with CO2 sensor data allows for finer parsing of CO2, variability across the hydrologic conditions, and evaluate processes that affect concentrations of both gases together (e.g. gas exchange, respiration) and separately (e.g. anaerobic respiration) (Vachon et al. 2020). Previous studies spanning various stream types across the continental US identified temporal patterns in headwater stream CO2, from diel to annual and influenced by biology, physics, and hydrology (Crawford 2017). While combining CO2 and O2 sensor data in headwater streams to partition and account for various sources and sinks has been applied in Arctic and temperate streams (Lupon et al. 2019; Rocher-Ros et al. 2019), these methods have not been applied in tropical streams.
In this paper, we present the results of continuous deployment of O2 and CO2 sensors in a lowland Neotropical rainforest headwater stream in La Selva Biological Station, Costa Rica. We supplemented in stream sensors with a riparian well CO2 sensor to measure groundwater CO2 concentration and from that, calculate groundwater CO2 inputs (GWCO2). We combined sensor data to estimate CO2 fluxes into and out of a 75-m study reach, including hydrologic export, NEP using O2 data, evasion, and groundwater inputs. The hourly measurement frequency for six months allowed capturing short (e.g. rainfall) and long-term (e.g. seasonal) dynamics of C fluxes. We hypothesized that: 1) CO2 would be supersaturated and O2 undersaturated with respect to atmospheric equilibrium and the variation in paired gas concentration over time would reflect inputs from groundwater; 2) all C fluxes would increase during the wet season; 3) CO2 evasion would be a greater than hydrologic export of C via stream flow; 4) discharge would drive increases in areal CO2 fluxes; and 5) higher fluxes of GWCO2 would decrease stream pH.
2.1. Site description
La Selva Biological Station, Costa Rica (LSBS), is a 1600 ha tropical wet forest reserve, with elevation ranging from 22 to 146 meters above sea level (Fig 1). LSBS is located at the transition from the upland foothills of the Cordillera Central of Costa Rica and the Caribbean lowlands, and is the lowland terminus of the altitudinal transect in the Braulio Carrillo National Park (Fig 1). Mean annual rainfall from is 4300 mm, with dry season January to April and wet season July to December (Sanford et al., 1994).
We focused on the lower 75 m reach of the Taconazo stream (10.432, -84.013) to partition and quantify CO2 input (upstream input, NEP, and GWCO2) and output (FCO2, downstream hydrologic export) fluxes. The Taconazo watershed area is 0.28 km2, all of which is forested with a closed canopy (Table 1). The Taconazo is a low solute stream at LSBS, with long-term (1997 - 2015) mean water temperature of 25.1 ˚C, pH of 5.5, discharge of 0.06 m3 s-1, and NO3- and SRP of 192.4 µg L-1 and 4.9 µg L-1, respectively (Ganong et al. 2015). The Taconazo has received significant study as a model headwater lowland tropical stream. Numerous studies from the Taconazo have contributed to better understand of the hydrology (Genereux et al. 2005; Genereux and Jordan 2006), biogeochemistry (Small et al. 2012; Oviedo-Vargas et al. 2015; Osburn et al. 2018), and ecology (Ardón et al., 2006; Ramírez et al., 2003; Rosemond et al., 2002) of tropical lowland headwater streams.
2.2. Data collection
We collected stream data from April 1, 2013 - September 30, 2013. Discharge (m3 s-1) was continuously measured from the V-notch weir constructed near the outflow of the Taconazo; further descriptions of the weir in Zanon et al. (2014). Rainfall (mm) was collected from the LSBS weather station, located ~900 m from the Taconazo. Meteorological data are available at: https://archive.tropicalstudies.org/meteoro/default.php?pestacion=201.
Groundwater discharge into the 75 m study reach was measured using conservative tracer injections to the stream. In the dry season (April), groundwater flux into the reach was 20% of total discharge at the end of the reach (Oviedo-Vargas et al. 2015), compared to the wet season (May-September) flux of 9% (Oviedo-Vargas et al. 2015, Ardón et al. 2013). For the dry and wet season, we multiplied discharge by the respective percentages as an estimate of groundwater discharge (QGW). To calculate discharge into the upstream end of the study reach (Qu), the groundwater discharge in the reach was subtracted from stream discharge measured at the end of the study reach (Qd).
A YSI 600xlm multiprobe sonde was secured 5 cm above the stream bottom in a PVC tube with holes drilled to allow for exchange of stream water. The multiprobe continuously measured temperature (˚C), pH, and DO (mg L-1 and %-saturation) at hourly intervals. The sonde was retrieved every two weeks for recalibration and to clean fouling on the sensor heads. Partial pressure of CO2 (ppmv) was measured using a Vaisala GMT221 infrared gas analyzer (IRGA) in the stream and in a riparian well. The sensors were prepared for submerged deployment as described in Johnson et al. (2010) and pCO2 was logged hourly using a Campbell CR1000 datalogger. pCO2 data were processed according to Johnson et al. (2010) for depth and temperature. We removed data from all sensors when the water level fell below the height of the sonde (e.g. conductivity < 0.005, DO near air saturation). Sensor data underwent further QAQC following the guidance of Taylor and Loescher (2013) and using the datacleanr R package (Hurley 2021).
2.3. Evaluating aqueous concentrations of CO2 and O2
We evaluated the departure of stream CO2 and O2 concentrations relative to concentrations expected at atmospheric equilibrium, and the magnitude of gas departure over time. Departure of O2 and CO2 from equilibrium with the atmosphere in freshwaters is mediated by biochemical reactions that directly link O2 and CO2 (e.g. aerobic metabolism), water temperature, bicarbonate chemistry, and gas exchange (Vachon et al. 2020). We calculated CO2 departure as the difference of measured aqueous concentration ([CO2]aq) minus the concentration at atmospheric equilibrium at 400 ppmv ([CO2]sat) using a temperature-corrected Henry’s Law constant. O2 departure was calculated as the difference between in-stream concentration minus the concentration at saturation which is calculated as a function of temperature and barometric pressure, following the outline in Hall and Hotchkiss (2017) and detailed below.
Paired O2 and CO2 departure were plotted as a data cloud, O2 departure vs. CO2 departure. To evaluate movement of the cloud over time, we estimated 95% confidence interval ellipses for each month of data collection using the ellipse v0.4.2 R package (Murdoch and Chow 2020). In each ellipse, we calculated the: 1) centroid, which indicates the location in Cartesian space for both O2 and CO2; 2) 1/[slope] through each ellipse, which informs the efficiency of metabolism between O2 and CO2, interpreted as the moles of CO2 produced per moles of O2 consumed through ecosystem respiration. See Vachon et al. (2020) for more details on these metrics.
To examine the role of individual rainfall events (2067 mm over 180 days) on pCO2, we selected three storm events: one event from the dry season (April), early wet season (June), and late wet season (August). For the three events, we examined the hysteresis response of pCO2 to discharge. For each event, we selected data 1 hour prior to the rising limb, through the peak in discharge, and for 5 hours following return to pre-storm discharge and before the next rising limb in the hydrograph (Evans and Davies 1998; Dinsmore and Billett 2008). We evaluated each hysteresis response for diagnostics describing the solute-discharge rotation (clockwise vs. counterclockwise vs. ‘figure 8’) and trend (positive vs. negative) (Evans and Davies 1998).
2.4. CO2 flux calculations and reach scale DIC mass balance
Our data collection allowed for independent calculation of CO2 input and output fluxes in the study reach. We calculated aqueous CO2 from the in-stream (see section above) and riparian well ([CO2]GW) stations from pCO2 data using the temperature dependent Henry’s Law constant from Plummer and Busenberg (1982). Stream CO2 at saturation ([CO2]sat) was determined using the same method, taking 400 ppm as the atmospheric CO2 concentration (Rocher‐Ros et al. 2019).
We compared sensor estimates of DIC to DIC estimates from samples of stream water. Samples were collected in 10 mL syringes and 4 mL of water was injected into an inverted sealed serum vial with a 22-gauge needle and 0.45 µm pore filter. Samples were equilibrated on a shaker table, and 250 µL headspace was removed and pCO2 analyzed on an SRI Instruments gas chromatograph (Las Vegas, Nevada, USA). Conversions from pCO2 to DIC were calculated as described in the supplementary file for sensor data. We matched the dates of syringe samples with sensor derived estimates and compared the two methods using the Student’s t-test.
We defined CO2 inputs to the study reach as the groundwater CO2 flux (GWCO2) and in-stream net ecosystem production (NEP). Groundwater CO2 flux was calculated as
Where SA is the benthic area of the stream and the flux units are mol C m-2 h-1
CO2 inputs from in-stream aerobic metabolism was estimated as hourly NEP using the instantaneous direct metabolism method outlined in Hall and Hotchkiss (2017). Briefly, NEP was estimated from the volumetric O2 mass balance equation:
Where NEPi is hourly instantaneous metabolism (mol O2 m-2 h-1), Oi is the O2 concentration at each timepoint (mol m-3), Osat,i is the concentration of O2 at saturation for the same timepoint (mol m-3), z is stream depth (m), and KO is the gas exchange coefficient for O2 (h-1). We converted KO from KCO2 (see equation 3) via Schmidt scaling (Raymond et al. 2012). NEP was converted to mol C m-2 h-1 assuming a 1:1 respiratory quotient between O2 and CO2 (Rocher-Ros et al. 2019). We selected the direct metabolism method for determining NEP because the lack of a diel O2 signal (Fig S1) and relatively high reaeration make the Taconazo ill-suited to alternative stream metabolism methods (Appling et al. 2018). Further, the direct method allows for estimation of NEP at finer temporal compared to alternative approaches, which are resolved at the daily scale.
CO2 output as areal CO2 evasion (FCO2, mol C per stream surface area per hour, m-2 h-1) was calculated as
Where KCO2 is the dry (1.13 h-1) or wet season (0.43 h-1) gas exchange coefficient reported in Oviedo-Vargas et al. (2015), and 
is depth in meters measured at the weir.
We conducted a volumetric mass balance in terms of DIC at the reach scale to compare reach scale estimates with watershed scale hydrologic inputs (upstream inputs and downstream export). The mass balance scales the fluxes calculated in equations 1-3 from mol C m-2 h-1 to reach as mol DIC h-1; conversion of CO2 to DIC are detailed in the supplementary file. Upstream DIC inputs to the study reach were calculated as the product of upstream discharge and [DIC] and upstream discharge was as the difference of downstream discharge and groundwater flux in the reach, which is an assumption supported in previous work (Ardón et al. 2013, Oviedo-Vargas et al. 2015). Thus, the full mass balance is:
Where C is DIC (mol m-3), Q is discharge (m3 h-1), L is the reach length (75 m), w is wetted width (1.5 m in the dry season, 2.8 m in the wet season), z is depth (m) measured at the weir, NEP is volumetric net ecosystem production (mol DIC m-3 h-1) and subscripts d, u, and GW refer to downstream, upstream, and groundwater, respectively. We evaluated the mass balance through the dDIC dt-1 term. If all relevant fluxes are accounted for, the change in should average zero, representing DIC inputs and outputs in balance with no net change in C storage in the reach.
We aggregated fluxes from the DIC mass balance to daily rates to evaluate the correlation with daily rainfall and the extent groundwater flux could predict stream pH. Hourly fluxes were aggregated to daily rates as the sum of hourly fluxes, daily rainfall the sum of hourly rainfall, and pH as mean daily pH. To evaluate univariate control of CO2 fluxes by discharge, we used the Kendall rank correlation, τ, on daily aggregated data. We evaluated the relationship of groundwater inputs of CO2 as a driver of stream acidification events in the Taconazo. All analysis, statistics, and figures were made in R v4.0.3 (R Core Team 2020).
3.1. Stream data
Total rainfall during the 180-day monitoring period was 2067 mm, with median daily rain of 2.29 mm (range: 0 – 107 mm) and 49 days with no rainfall recorded (Fig 2, a). April had the lowest rainfall (104.1 mm) and June had the most rainfall (498.6 mm). Median hourly stream discharge was 0.004 m3 s-1 (0 – 1.378 m3 s-1 range). Median discharge was lowest in April (0.0002 m3 s-1) and greatest in July (0.0160 m3 s-1). Median estimated groundwater flux into the reach was 0.0003 m3 s-1 (0 – 0.0346 m3 s-1 range) (Fig 2, b). Median pCO2 in the stream was 6343.6 ppmv (range 773 – 11994 ppmv), lower than that measured in the riparian well (median pCO2 46924 ppmv, range 28663 – 48683 ppmv). Over the monitoring period, 14% of hourly pCO2 measurements were missed in the stream, the largest missing section corresponding with the highest flows in late July and early August (Fig 2, d). In the riparian well, 27% of hourly measurements were missing, including the same period missing from the stream time-series and a period in May, during the transition from dry to wet season (Fig 2, e). Estimates of DIC from both discrete sampling (mean 0.23 ± 0.06 mmol DIC L-1) and from the sensor (mean = 0.25 ± 0.04 mmol DIC L-1) were similar (p = 0.34).
3.2 Gas dynamics and CO2 hysteresis
In departure space, all data points show strong CO2 super-saturation and O2 undersaturation (Fig 3). CO2 departure was ~6.6-times greater than O2 departure, on average. Ellipse centroid location varied little over time. The value of 1/[slope] was greatest in September (467.9) and lowest in August (0.7). The general position of the cloud away from the 1:-1 line indicates aerobic metabolic processes are not responsible for controlling CO2 or O2, and the higher concentration of CO2 than O2 suggests anaerobic production of CO2 or external inputs of CO2.
There is little evidence of hysteresis between pCO2 and discharge in the three storm events we examined. In the dry season, pCO2 varies little across the rise in discharge, though during the falling limb, there is slight ‘figure 8’ rotation (Fig 4, a). Similarly, pCO2 varied little preceding and during the wet season events, while decreasing during the falling limb before returning to pre-storm concentration (Fig 4, b). In the late wet season, pCO2 stays relatively constant during the rising limb, before oscillating during the falling limb (Fig 4, c). Of note, during the early wet season event, overall pCO2 was greater than during the dry and late wet season events, and peak event discharge during the dry season (0.01 m3 s-1) was less than the early (0.08 m3 s-1) and late (0.08 m3 s-1) wet season events.
3.3. Areal CO2 fluxes and DIC mass balance
The relationship between mean hourly CO2 inputs and evasion to the atmosphere varied temporally. GWCO2 (mean 1.39 mmol C m-2 h-1) showed strong monthly variation, with lower magnitudes in April and May compared to June and July (Fig 5, a). The other input, NEP (3.16 mmol C m-2 h-1), showed little variation over time (Fig 5, b). Evasion showed little temporal variation (mean 20.5 mmol C m-2 h-1).
Reach scale mass balance inputs reveal shifts in dominant contributions over time. The input with least temporal variation was NEP (mean 0.75 mol DIC h-1), and was greater, on average, than upstream longitudinal inputs (mean 0.40 mol DIC h-1) and groundwater (0.31 mol DIC h-1) (Fig 6, a). Upstream DIC and groundwater fluxes each reflect the seasonal hydrology of the stream, with lower fluxes in April, and greater values during the remaining wet season. DIC outputs at the reach scale were predominately to the atmosphere rather than hydrologic export (Fig 6, b). Mean evasion (4.8 mol DIC h-1) was ~11 times greater, on average, than hydrologic export (mean 0.44 mol DIC h-1), and did not reflect changes in seasonal hydrology. The mass balance reveals that the Taconazo study reach is a net source of DIC (Fig 6, c). The central tendency of the mass balance near -4 suggests we accounted for major fluxes in the reach, but not centered at 0 suggests a discrepancy in this approach. Using the Kendall rank correlation statistic, τ, daily upstream DIC inputs and downstream outputs of DIC were significantly correlated with daily rainfall. In contrast, none of the upscaled areal fluxes were statistically correlated with rainfall.
Mean daily pH was decreased as GWCO2 increased (Fig 7) (F1,134 = 8.84, p = 0.06, R2 < 0.01). There were differences in pH response between months and seasons. In the dry season (April), mean daily pH fell to 4.50 at GWCO2 of 0.01 mol DIC d-1. During the peak of the wet season (July), GWCO2 was >0.1 mol DIC d-1 and corresponded to a mean daily pH of 5.0.
We showed that an in-stream stream and well sensor approach can be used to estimate inputs of terrestrial C into streams via groundwater and to evaluate the relative contribution of the different inputs and outputs in a stream DIC budget. Further, we showed that CO2 fluxes responded seasonally to changes in rainfall. We found evasion of CO2 to be a major loss of C from the Taconazo, underscoring the importance of considering fluxes from headwater streams in complete C assessments and budgets.
4.1. pCO2 and O2 variability
pCO2 values measured in the Taconazo were high relative to similarly sized streams. Compared to a monitoring study across headwater streams in North America, Taconazo pCO2 was higher than 6 of the 7 streams monitored, only less than a site in a southern hardwood forest and greater than a highland tropical stream in Puerto Rico (Crawford et al. 2017). We observed little diel variability in pCO2 or O2 (Fig S1), indicating gross primary productivity (GPP) is low in the Taconazo, and supported by the NEP estimates > 0 mol C m-2 h-1 (Fig 3). In addition, the paired O2 - CO2 cloud (Fig 3) is located far from the 1:-1 line for all time points, suggesting that aerobic metabolism is not a dominant process in the Taconazo and is not responsible for driving O2 and pCO2. In tundra headwater streams, the location of the CO2-O2 departure cloud was closer to the 1:-1 line and reflected the greater influence of in-stream NEP on FCO2 (Rocher-Ros et al. 2019).
The monitoring period included one month of the dry season with the remaining five months during the wet season, following the seasonality reported at LSBS (Sanford et al. 1994). The transition from dry to wet season is reflected by the increased rainfall in mid-May. The dry to wet transition coincides with an increase in stream pCO2. During the wet season, there was greater variability in both stream pCO2 and well CO2. The higher pCO2 in the wet season reflects higher GWCO2 flux corresponding to higher discharge and groundwater flux into the Taconazo and lower fluxes in the late wet season, following the peak of the wet season in July and August. Throughout the study period, some storm events resulted in decreases in pCO2 (Fig 4, b), but was not a consistent pattern. The lack of consistent hysteresis between pCO2 and discharge is perhaps limited by the events we examined, but and clock-wise hysteresis is present during larger storms with greater turbulence that increases gas exchange rates.
4.2 Variability of areal and reach-scale fluxes
Among the areal fluxes, FCO2 was greater than the input fluxes NEP and GWCO2 (Fig 5). GWCO2 showed strong seasonal variability, reflecting shifts in rainfall and discharge in the Taconazo (Fig 5, a), whereas NEP was similar in magnitude across the six-month period (Fig 6, b). Surprisingly, GWCO2 and NEP values are similar in magnitude, contrasting with the gas departure data (Fig 3). We expected, as is characteristic in headwater streams, for external C contributions (GWCO2) to exceed in-stream production (NEP) (Hotchkiss et al. 2015). While the gas departure data suggest that aerobic processes are not responsible for mediating O2 and CO2 concentrations in the stream, gas flux data (Fig 5) suggests that NEP and GWCO2 are similar C inputs in the lower Taconazo. This discrepancy could be explained by the independent calculation of each flux. First, the hydrology of groundwater discharge varies spatially within the study reach, and may exhibit a temporal lag between rainfall and discharging into the stream, and thus underestimates of GWCO2. Second, NEP is difficult to estimate with low diel O2 variation (Appling et al. 2018). Third, estimates of evasion are likely underestimated due to the limited gas exchange data measured during low flows and missing high discharge events.
CO2 evasion showed little temporal variability (Fig 6, b). We attribute the lack of variability to the sustained high pCO2 relative to the atmosphere. Theory predicts gas exchange to increase with turbulence and geomorphological parameters, such as depth and velocity, and therefore with discharge (Raymond et al. 2012; Ulseth et al. 2019). Since our results are based on only two estimates of gas exchange estimates, the potential influence of discharge-driven temporal variation of gas exchange on evasion may be substantially reduced in our estimates. Streams like the Taconazo, with relatively high gas exchange rates and little diel O2 variation are ill-suited to use inverse modeling outputs (Appling et al. 2018) or night-time regression to estimate gas exchange, using direct tracer injections of gases is an important method (Raymond et al. 2012). Our gas exchange rates were measured during low flows, missing high discharge events, and there may be finer spatial scale areas within the study reach that have higher evasion but are masked in our one sensor station approach (Schneider et al. 2020). Our mean daily FCO2 estimate (4.6 g C m-2 d-1) was within the range of discrete estimates from the same reach during the same year (dry 10.8 ± 4.8, wet 7.2 ± 3.6 g C m-2 d-1, Oviedo-Vargas et al. 2015) and was greater than the average of six Arctic streams (1.6 g C m-2 d-1, Rocher-Ros et al. 2019).
At the reach scale, the mass balance reveals that FCO2 was the dominant loss of DIC (Fig 6); this highlights the need for estimates of greenhouse gases from tropical inland waters, particularly from headwater streams (Cole et al. 2007, Raymond et al. 2013). Elevated FCO2 compared to the dominant DIC inputs (Fig 5) also contributes to the reach being a source, as determined by the negatively skewed d[DIC]/dt distribution (Fig 6, c), and the reach is perhaps a greater DIC source given the potential underestimation of FCO2. With little change in stream pCO2 during storms (Fig 4), hydrologic exports of DIC follow short-term patterns and the seasonality in discharge. The mass balance deserves further scrutiny given the magnitude of the fluxes regulating DIC within the 75 m reach and the assumptions about [DIC] at the upstream and downstream ends of the reach.
4.3. Drivers of CO2 fluxes
The location of the O2 - CO2 cloud represents CO2 supersaturation and slight O2 undersaturation, on average (Fig 3) and based on the predictions in Vachon et al. (2020), we conclude that groundwater CO2 inputs are largely responsible between internal and external sources of CO2. However, the flux data suggest that NEP is a similarly sized contributor to DIC in the reach. The 1/[slope] for each monthly cloud (Table 2) suggests the ecosystem respiratory quotient, as mol CO2 produced for mol of O2 consumed in aerobic processes, are >1 (except August), further contributing to the supersaturation of CO2. We conclude that pCO2 and the fluxes that affect pCO2 in the Taconazo are mediated primarily by groundwater inputs (Figure S2), at broad timescales and is true for similar headwater streams in lowland tropical wet forests.
Rainfall explained variation in hydrologic inputs and exports in the mass balance, though not for GWCO2 which has a hydrologic component in its calculation (Table 3). The lack of correlations for the other fluxes with rainfall is informative for headwater tropical streams. For FCO2, as previously discussed, the sustained high pCO2 drives this flux and the influence of rainfall or storm events have little effect on pCO2. The sustained high pCO2 measured in the riparian well explains why GWCO2 is unaffected by rainfall. The lack of temporal variability in DO, and therefore NEP, indicate storm events are less important. Storm events do influence dissolved organic C (DOC) in the Taconazo (Osburn et al. 2018), which may contribute to increased respiration. Changes in the delivery of DOC that fuels in-stream respiration may be reflected in the differences in time of the 1/[slope] (Table 2) which reflects the ecosystem respiratory quotient, or conversion of O2 into CO2 during respiration at the ecosystem scale. In temperate streams with organic soils, rainfall events increased [DOC] and stimulated in-stream ecosystem respiration (Demars 2019). It is unlikely primary productivity affects NEP in a measurable way. The dense canopy cover attenuates light at the stream bed and the food webs are fueled through detrital pathways (Rosemond et al. 2002), and there is little diel variation in DO (Fig S1), all suggesting photosynthetic uptake of C is small.
4.4. Influence on pH
Small et al. (2012) suggested that CO2 from terrestrial sources enter streams with low solutes and reduce pH through carbonic acid creation, primarily during the early wet season in streams at La Selva. Our results provide some support for this hypothesis. We show GWCO2 has a weak negative correlation with stream pH (Fig 7). In the late dry and early wet season (April – May, Fig 2), small inputs of GWCO2 (<10-3 mol DIC d-1) can reduce mean daily pH to as low as 4.5 and these single day acidification events detract from the correlation (R2 = 0.06), but are ecologically important. During the wet season (June onward), greater GWCO2 reduce pH to between ~5. The Taconazo has low alkalinity (<5 mg L-1 as CaCO3, unpub. data), therefore the acid neutralizing capacity is overcome by small inputs of CO2. CO2 is supplemented by increased DOC during storm events (mean stormflow DOC = 2.25 mg L-1, Ganong et al. 2015; Osburn et al. 2018) and redox reactions (e.g. iron oxidation) which can collectively reduce pH. During the wet season, DOC inputs to the Taconazo are disproportionately organic acids (Osburn et al. 2018) which have a pKa that contributes to lower pH.
The lowest hourly pH measured during our study was 4.0 and is similar to lowest pH measured in the long-term record from the Taconazo, which occurred following the global El Niño Southern Oscillation event in 1997/98 (Small et al. 2012). During this event at LSBS, tree growth was decreased (Clark et al. 2003, 2010) and root mortality increased (Espeleta and Clark 2007). Small et al. (2012) hypothesize the stock of labile C in the soil increased and soil respiration causes high soil CO2 (Schwendenmann and Veldkamp 2006). When soil moisture increases during the early wet season, soil CO2 is quickly flushed from groundwater into streams. While acidification events are common in low solute headwater streams at La Selva, where pH ranges from 4 – 6 in the span of days, the negative effects on organisms and ecosystem processes are limited (Ganong 2015), though there is evidence of short term behavior modification in macroinvertebrates and fish (Ardón et al. 2013).
We deployed a trio of sensors, in a stream and in a riparian well, to independently estimate external and internal fluxes of CO2 in a lowland headwater tropical stream. We found that values for terrestrial flux of CO2 are similar to internal production as NEP. High FCO2 was sustained and greater than hydrologic export of C. In a global synthesis of efflux from inland waters, both Cole et al. (2007) and Raymond et al. (2013) stress the importance of evasion estimates from headwater tropical streams, which exhibit higher reaeration velocities and higher pCO2. We document sustained high pCO2 leads to higher FCO2 compared to similar-sized headwater streams across the globe (Raymond et al. 2013). Our study provides estimates of globally relevant C fluxes from a stream type with disproportionate influence among inland waters and provides a sensor-based approach to independently estimate reach scale C fluxes.
Funding
Funding was provided by National Science Foundation (NSF) Long-Term Research in Environmental Biology (LTREB) program (award #1655869).
Conflicts of interest
The authors have no relevant financial or non-financial interests to disclose.
Availability of data and material
Data, code, and supplementary material from this study can be accessed at https://github.com/nmarzolf91/Taconazo_CO2. Until acceptance of the paper, this repository is not publically available but all files in the repository will be uploaded during manuscript submission.
Code availability
Data, code, and supplementary material from this study can be accessed at https://github.com/nmarzolf91/Taconazo_CO2.
Author contributions
Conceptualization: N. Marzolf, C. Small, D. Oviedo-Vargas, C. Ganong, J. Duff, D. Genereux, M. Ardón; Methodology: N. Marzolf, C. Small, C. Ganong, D. Oviedo-Vargas, D. Genereux, M. Ardón; Formal analysis and investigation: N. Marzolf, C. Small, D. Oviedo-Vargas, D. Genereux, M. Ardón; Writing - original draft preparation: N. Marzolf, C. Small, M. Ardón; Writing - review and editing: all authors; Funding acquisition: M. Ardón, A. Ramírez, C. Pringle; Resources: C. Small, D. Oviedo-Vargas, D. Genereux, C. Ganong, M. Ardon; Supervision: M. Ardón, A. Ramírez, C. Pringle.
Ethics approval
Not applicable.
Consent to participate
Not applicable.
Consent for publication
Not applicable.
Acknowledgements
We are grateful to Minor Hidalgo for help in the field and sensor calibration and maintenance. This study was aided by Laura Willson and the Organization for Tropical Studies Research Experience for Undergraduates program. Dr. Mark Johnson aided in setting up and interpreting CO2 sensor data. Funding was provided by National Science Foundation DEB award #1655869. We are grateful to members of the Ardón Lab and Dr. Diego Riveros-Iregui for their comments on early drafts of the manuscript.
- Appling AP, Hall RO, Yackulic CB, Arroita M (2018) Overcoming Equifinality: Leveraging Long Time Series for Stream Metabolism Estimation. J Geophys Res Biogeosciences 123:624–645. https://doi.org/10.1002/2017jg004140
- Ardón M, Duff JH, Ramírez A, et al (2013) Experimental acidification of two biogeochemically-distinct neotropical streams: buffering mechanisms and macroinvertebrate drift. Sci Total Environ 443:267–277. https://doi.org/10.1016/j.scitotenv.2012.10.068
- Ardón M, Stallcup LA, Pringle CM (2006) Does leaf quality mediate the stimulation of leaf breakdown by phosphorus in Neotropical streams? Freshw Biol 51:618–633. https://doi.org/10.1111/j.1365-2427.2006.01515.x
- Aufdenkampe AK, Mayorga E, Raymond PA, et al (2011) Riverine coupling of biogeochemical cycles between land, oceans, and atmosphere. Front Ecol Environ 9:53–60. https://doi.org/10.1890/100014
- Battin TJ, Luyssaert S, Kaplan LA, et al (2009) The boundless carbon cycle. Nat Geosci 2:598–600. https://doi.org/10.1038/ngeo618
- Borges A V, Darchambeau F, Lambert T, et al (2019) Variations in dissolved greenhouse gases (CO2, CH4, N2O) in the Congo River network overwhelmingly driven by fluvial-wetland connectivity. Biogeosciences 16:3801–3834
- Clark DA, Piper SC, Keeling CD, Clark DB (2003) Tropical rain forest tree growth and atmospheric carbon dynamics linked to interannual temperature variation during 1984-2000. Proc Natl Acadademy Sci United States Am 13:5852–5857
- Clark DB, Clark DA, Oberbauer SF (2010) Annual wood production in a tropical rain forest in NE Costa Rica linked to climatic variation but not to increasing CO2. Glob Chang Biol 16:747–759. https://doi.org/10.1111/j.1365-2486.2009.02004.x
- Cole JJ, Prairie YT, Caraco NF, et al (2007) Plumbing the Global Carbon Cycle: Integrating Inland Waters into the Terrestrial Carbon Budget. Ecosystems 10:172–185. https://doi.org/10.1007/s10021-006-9013-8
- Colvin SAR, Sullivan SMP, Shirey PD, et al (2019) Headwater Streams and Wetlands are Critical for Sustaining Fish, Fisheries, and Ecosystem Services. Fisheries 44:73–91. https://doi.org/10.1002/fsh.10229
- Crawford JT, Stanley EH, Dornblaser MM, Striegl RG (2017) CO2 time series patterns in contrasting headwater streams of North America. Aquat Sci 79:473–486. https://doi.org/10.1007/s00027-016-0511-2
- Demars BOL (2019) Hydrological pulses and burning of dissolved organic carbon by stream respiration. Limnol Oceanogr 64:406–421. https://doi.org/10.1002/lno.11048
- Dinsmore KJ, Billett MF (2008) Continuous measurement and modeling of CO2 losses from a peatland stream during stormflow events. Water Resour Res 44:1–11. https://doi.org/10.1029/2008WR007284
- Drake TW, Raymond PA, Spencer RGM (2018) Terrestrial carbon inputs to inland waters: A current synthesis of estimates and uncertainty. Limnol Oceanogr Lett 3:132–142. https://doi.org/10.1002/lol2.10055
- Espeleta JF, Clark DA (2007) Multi-scale variation in fine-root biomass in a tropical rain forest: a seven-year study. Ecol Monogr 77:377–404. https://doi.org/10.1890/06-1257.1
- Evans C, Davies TD (1998) Causes of concentration/discharge hysteresis and its potential as a tool for analysis of episode hydrochemistry. Water Resour Res 34:129–137. https://doi.org/10.1029/97WR01881
- Ganong CN (2015) Effects of Precipitation Regime on Neotropical Streams: Biogeochemical Drivers of Stream Physicochemistry and Macroinvertebrate Tolerance to pH Shifts. University of Georgia
- Ganong CN, Small GE, Ardón M, et al (2015) Interbasin flow of geothermally modified ground water stabilizes stream exports of biologically important solutes against variation in precipitation. Freshw Sci 34:276–286. https://doi.org/10.1086/679739
- Genereux DP, Jordan M (2006) Interbasin groundwater flow and groundwater interaction with surface water in a lowland rainforest, Costa Rica: A review. J Hydrol 320:385–399. https://doi.org/10.1016/j.jhydrol.2005.07.023
- Genereux DP, Jordan MT, Carbonell D (2005) A paired-watershed budget study to quantify interbasin groundwater flow in a lowland rain forest, Costa Rica. Water Resour Res 41:. https://doi.org/10.1029/2004wr003635
- Genereux DP, Nagy LA, Osburn CL, Oberbauer SF (2013) A connection to deep groundwater alters ecosystem carbon fluxes and budgets: Example from a Costa Rican rainforest. Geophys Res Lett 40:2066–2070. https://doi.org/10.1002/grl.50423
- Hall RO, Hotchkiss ER (2017) Stream Metabolism. In: Lamberti GA, Hauer FR (eds) Methods in Stream Ecology, Vol 2: Ecosystem Function, 3rd Edition. Academic Press Ltd-Elsevier Science Ltd, pp 219–233
- Hotchkiss ER, Hall RO, Sponseller RA, et al (2015) Sources of and processes controlling CO2 emissions change with the size of streams and rivers. Nat Geosci 8:696–699. https://doi.org/10.1038/ngeo2507
- Hurley A (2021) datacleanr: Interactive and Reproducible Data Cleaning
- Johnson MS, Billett MF, Dinsmore KJ, et al (2010) Direct and continuous measurement of dissolved carbon dioxide in freshwater aquatic systems-method and applications. Ecohydrology 68–78. https://doi.org/10.1002/eco.95
- Lupon A, Denfeld BA, Laudon H, et al (2019) Groundwater inflows control patterns and sources of greenhouse gas emissions from streams. Limnol Oceanogr. https://doi.org/10.1002/lno.11134
- Mayorga E, Aufdenkampe AK, Masiello CA, et al (2005) Young organic matter as a source of carbon dioxide outgassing from Amazonian rivers. Nature 436:538–541. https://doi.org/10.1038/nature03880
- Murdoch D, Chow ED (2020) ellipse: Functions for Drawing Ellipses and Ellipse-Like Confidence Regions
- Osburn CL, Oviedo-Vargas D, Barnett E, et al (2018) Regional groundwater and storms are hydrologic controls on the quality and export of dissolved organic matter in two tropical rainforest streams, Costa Rica. J Geophys Res Biogeosciences. https://doi.org/10.1002/2017JG003960
- Oviedo-Vargas D, Genereux DP, Dierick D, Oberbauer SF (2015) The effect of regional groundwater on carbon dioxide and methane emissions from a lowland rainforest stream in Costa Rica. J Geophys Res Biogeosciences 120:2579–2595. https://doi.org/10.1002/
- R Core Team (2020) R: A language and environment for statistical computing
- Ramírez A, Pringle CM, Molina L (2003) Effects of stream phosphorus levels on microbial respiration. Freshw Biol 48:88–97
- Raymond PA, Hartmann J, Lauerwald R, et al (2013) Global carbon dioxide emissions from inland waters. Nature 503:355–359. https://doi.org/10.1038/nature12760
- Raymond PA, Zappa CJ, Butman D, et al (2012) Scaling the gas transfer velocity and hydraulic geometry in streams and small rivers. Limnol Oceanogr Fluids Environ 2:41–53. https://doi.org/10.1215/21573689-1597669
- Richey JE, Melack JM, Aufdenkampe AK, et al (2002) Outgassing from Amazonian rivers and wetlands as a large tropical source of atmospheric CO2. Nature 416:617–620
- Rocher-Ros G, Sponseller RA, Bergstr A, et al (2019) Stream metabolism controls diel patterns and evasion of CO2 in Arctic streams. Glob Chang Biol 0–3. https://doi.org/10.1111/gcb.14895
- Rocher‐Ros G, Sponseller RA, Lidberg W, et al (2019) Landscape process domains drive patterns of CO2 evasion from river networks. Limnol Oceanogr Lett. https://doi.org/10.1002/lol2.10108
- Rosemond AD, Pringle CM, Ramírez A, et al (2002) Landscape variation in phosphorus concentration and effects on detritus‐based tropical streams. Limnol Oceanogr 47:278–289
- Sanford RL, Paaby P, Luvall JC, Phillips E (1994) Climate, geomorphology, and aquatic systems. In: McDade LA, Bawa KS, Hespenheide HA, Hartshorn GS (eds) La Selva: Ecology and natural history of a Neotropical rain forest. University of Chicago Press, pp 19–33
- Schneider CL, Herrera M, Raisle ML, et al (2020) Carbon Dioxide (CO2) Fluxes From Terrestrial and Aquatic Environments in a High-Altitude Tropical Catchment. J Geophys Res Biogeosciences 125:e2020JG005844. https://doi.org/10.1029/2020JG005844
- Schwendenmann L, Veldkamp E (2006) Long-term CO2 production from deeply weathered soils of a tropical rain forest: evidence for a potential positive feedback to climate warming. Glob Chang Biol 12:1878–1893. https://doi.org/10.1111/j.1365-2486.2006.01235.x
- Small GE, Ardón M, Jackman AP, et al (2012) Rainfall-driven amplification of seasonal acidification in poorly buffered tropical streams. Ecosystems 15:974–985
- Stoddard JL, Jeffries DS, Lükewille A, et al (1999) Regional trends in aquatic recovery from acidification in North America and Europe. Nature 401:575–578
- Stumm W, Morgan JJ (1996) Dissolved Carbon Dioxide. In: Aquatic chemistry: chemical equilibria and rates in natural waters, 3rd editio. Wiley, New York
- Tank SE, Fellman JB, Hood E, Kritzberg ES (2018) Beyond respiration: Controls on lateral carbon fluxes across the terrestrial-aquatic interface. Limnol Oceanogr Lett 3:76–88. https://doi.org/10.1002/lol2.10065
- Taylor JR, Loescher HL (2013) Automated quality control methods for sensor data: a novel observatory approach. Biogeosciences 10:4957–4971. https://doi.org/10.5194/bg-10-4957-2013
- Ulseth AJ, Hall RO, Boix Canadell M, et al (2019) Distinct air–water gas exchange regimes in low- and high-energy streams. Nat Geosci. https://doi.org/10.1038/s41561-019-0324-8
- Vachon D, Sadro S, Bogard MJ, et al (2020) Paired O2–CO2 measurements provide emergent insights into aquatic ecosystem function. Limnol Oceanogr Lett 5:287–294. https://doi.org/10.1002/lol2.10135
- Wetzel RG, Likens GE (2000) The Inorganic Carbon Complex: Alkalinity, Acidity, CO 2, pH, Total Inorganic Carbon, Hardness, Aluminum. In: Limnological analyses. Springer, pp 113–135
- Zanon C, Genereux DP, Oberbauer SF (2014) Use of a watershed hydrologic model to estimate interbasin groundwater flow in a Costa Rican rainforest. Hydrol Process 28:3670–3680. https://doi.org/10.1002/hyp.9917
Table 1) Reach characteristics from the Taconazo and summary statistics for physicochemical data during the study period. High frequency data (temperature, DO, pH, pCO2, and stream discharge) are presented as the median (range in parentheses) during the monitoring period.
|
Measurement (unit) |
Value |
|
Reach length (m) |
75 |
|
1Mean reach width (m) |
1.3 |
|
1Mean reach velocity (m s-1) |
0.014 |
|
1Travel time (min) |
82.5 |
|
Gradient (m m-1) |
0.0024 |
|
Temperature (˚C) |
24.5 (21.2 – 29.6) |
|
DO (mg L-1) |
5.94 (0.26 – 8.32) |
|
pH |
5.32 (4.02 – 6.91) |
|
pCO2 (ppmv) |
6443.6 (773.8 – 11994) |
|
2Stream Discharge (m3 s-1) |
0.017 (0 – 1.378) |
1 Measured in March 2013
2 excludes backflooding events when the downstream Rio Puerto Viejo floods its tributaries, including the Taconazo (Zanon et al., 2014).
Table 2) Paired O2 - CO2 metrics for ellipses for each month of the monitoring period. Mean CO2-Dep and O2-Dep are mean monthly concentrations followed by standard deviation in parentheses. The 1/[slope] is the slope of the linear model fit through each months’ data.
|
Month |
Mean CO2-Dep (SD) |
Mean O2-Dep (SD) |
1/[Slope] (µM µM-1) |
|
April |
175.5 (46.6) |
-48.6 (18.4) |
20.5 |
|
May |
204.2 (30.0) |
-35.3 (14.2) |
7.0 |
|
June |
243.5 (33.9) |
-32.0 (15.2) |
12.1 |
|
July |
207.7 (15.4) |
-11.1 (5.2) |
13.4 |
|
August |
183.9 (14.9) |
-30.0 (45.3) |
0.7 |
|
September |
221.2 (25.2) |
-30.6 (12.8) |
467.9 |
|
All Dates |
204.3 (37.9) |
-31.8 (24.8) |
9.9 |
Table 3) Correlation of daily DIC fluxes with daily rainfall. Units for mass balance terms are mol C d-1. Bold text indicates statistically significant correlations (p < 0.05).
|
|
Mass Balance term (mol DIC d-1) |
τ |
p value |
|
Inputs |
GWCO2 |
0.08 |
0.15 |
|
NEP |
-0.01 |
0.78 |
|
|
DICu |
0.16 |
<0.01 |
|
|
Outputs |
DICd |
0.16 |
<0.01 |
|
FCO2 |
0.05 |
0.37 |