Discussion
For the sampling period, the riparian forest soils examined in this study were a net sink of CH4, and a net source of CO2 and N2O. The reference sites had significantly higher N2O emissions and significantly higher CH4 uptake than the sites with no buffer. CH4 uptake was significantly higher in the ND areas compared to the DIS areas, reflecting the difference in soil moisture and resulting aerobic conditions required for methanotrophy and anaerobic conditions required for methanogenesis (Oertel and others 2016). Soil temperature, soil moisture, and depth to the groundwater table were significant predictors of CO2 emissions. Soil moisture and depth to the groundwater table were significant predictors of CH4 fluxes.
Effects of forest harvest
Although riparian zones are widespread across the global landscape, we found very few studies examining GHG fluxes from the riparian zones of small streams (Goodrick and others 2016; Soosaar and others 2011), and even fewer for headwater streams (Chang and others 2014; De Carlo and others 2019a). Due to the distinct environmental conditions of riparian zones (Vidon and others, 2018), the effects of forest harvest on GHG fluxes are likely to differ here from upland soils. Riparian zones of headwater streams in particular are unique due to their steep geomorphology, high groundwater influence, and susceptibility to low flows (Richardson and Danehy 2007). Therefore, this study fills the knowledge gap of the effects of forest harvest on GHG fluxes from riparian zones of headwater streams.
Clear-cutting has inconsistent effects on CO2 emission rates (Striegl and Wickland 1998; Ullah and others 2009), with studies reporting a decline (Striegl and Wickland 1998), an increase (Lavoie and others 2013; Paul-Limoges and others 2015), or no change (Kähkönen and others 2002) in CO2 emissions following harvest. In this study, there was no significant difference in CO2emission rates between treatments, although CO2emissions were highest at the buffer sites. This lack of a significant treatment difference may have been because of the distinct conditions in riparian forests compared to upland forests. Riparian zone soils have high moisture levels due to shallow water tables and a strong groundwater influence (Goodrick and others 2016). Therefore, the rise in temperature following forest harvest that can explain the rise in CO2 following harvest in upland forests (Lavoie and others 2013), may be buffered by the generally moister and cooler soils of riparian forests (Clinton and others 2010), resulting in a less distinct rise in CO2 emissions following harvest in riparian forests compared to upland forests. Moreover, the mild mean annual temperature and narrow range in annual temperature in our study region may have meant limited temperature differences, resulting in a small effect size of the treatments.
Soil temperature, soil moisture, and depth to the groundwater table were significant predictors of CO2 fluxes. This is consistent with the large body of literature on the drivers of soil respiration (Luo and Zhou 2006). Soil respiration usually increases exponentially with temperature, reaches a maximum, and then declines (Luo and Zhou 2006). Temperature controls many aspects of soil respiration from the activity of cellular enzymes, to root growth and microbial activity (Luo and Zhou 2006). Soil moisture is another well-established driver of soil CO2 emissions, with the common conceptual relationship where soil respiration is low under dry conditions, reaches a maximum at intermediate soil levels, and decreases at high soil moisture content where anaerobic conditions depress aerobic microbial activity (Luo and Zhou 2006). The interactive effects of these two variables on soil respiration is a key knowledge gap (Meyer and others 2018). Our research provides valuable information to help understand the effects of these factors on soil respiration in the unique riparian environment.
The results of our study were in line with other studies, which have also found that clear-cutting increased CH4 efflux from the soil (Kähkönen and others 2002; Wu and others 2011). This rise can be attributed to higher average summer soil temperatures, greater soil moisture, and higher dissolved organic carbon concentrations in clear-cuts (Ullah and others 2009; Wu and others 2011). Forest harvest can result in soil compaction, reduced transpiration, and a rise in the groundwater table, all of which promote waterlogging and anaerobic conditions (Christiansen and others 2017; Gundersen and others 2010). In our study, soil moisture and depth to the groundwater table were significant predictors of CH4 fluxes, while temperature was not. In agreement with our results, CH4 uptake was three times lower following clear-cutting in a temperate spruce forest in southern Germany (Wu and others 2011). In another study, clear-cutting turned a spruce forest soil in Finland from a sink to a source of CH4, with a 40% decrease in CH4 consumption rates (Kähkönen and others 2002). The lower CH4 uptake following forest harvest may be explained by the harmful impacts of soil disturbance on methanotrophic bacteria, resulting in the inhibition of CH4-oxidation (Le Mer and Roger 2001; Wu and others 2011). Alternatively, the anaerobic conditions created by higher soil moisture concentrations following forest harvest can promote the production of CH4 (Wu and others 2011). A clear-cut wetland in Québec, Canada produced 131 times more CH4 than the undisturbed wetland soil, likely due to higher soil temperature and soil moisture in the clear-cut (Ullah and others 2009). In our study, the no buffer sites still were a net sink, albeit a weak sink, of CH4 over the growing season. This is likely because unlike many wetlands, most riparian soils do not have consistently anoxic soils, which promote methanogenesis (Dalal and Allen 2008). Given that there was no significant difference in CH4 flux rates between the buffer and reference sites, it appears that riparian buffers may be effective in preserving soil ecosystem conditions contributing to CH4 fluxes. Consequently, riparian buffer zones may be an effective strategy for forest managers interested in maintaining CH4 balance in riparian zone soils.
The lower N2O emissions in the no buffer sites compared to undisturbed riparian zones was a surprising result because many other studies have reported an increase in N2O emissions following forest harvest. Typically, forest harvest can increase soil moisture and mobilize soil nitrogen, promoting N2O emissions from logged forest sites (Kreutzweiser and others 2008). Higher N2O emissions were seen following forest harvest in the taiga region of eastern Siberia, Russia (Takakai and others 2008). Additionally, N2O emissions were 2.7 times higher in clear-cut than in mature black spruce forest soil in Québec, Canada (Ullah and others 2009). However, these studies were not conducted in riparian forests, which have unique conditions such as shallow water tables, high soil organic matter quality and quantity, and high soil nitrogen availability, unlike most upland forests (Knoepp and Clinton 2009; Vidon and others 2018). Given that forests are typically sources of N2O (Dalal and Allen 2008), the lower N2O fluxes in the no buffer compared to the reference sites is a departure from undisturbed ecosystem function. The unexpectedly low N2O fluxes at the no buffer sites could be explained by the mechanical soil disturbance in the riparian zone caused by forest harvest. The disruption of the structure and function of microbial communities responsible for nitrification and denitrification could contribute to the comparatively low N2O fluxes at the no buffer sites (Tan and others 2005). Meanwhile, N2O fluxes at the buffer sites were not significantly different from the reference sites, thus, riparian buffers may be effective in preserving soil ecosystem conditions contributing to N2O fluxes. A decline in soil CO2emissions was attributed to the disruption of the soil surface and death of tree roots following the clear-cutting of a jack pine stand in Saskatchewan (Striegl and Wickland 1998). Moreover, soil compaction as a result of forest harvest has been found to reduce net nitrification rates in the forest floor and mineral soil as well as reduce the soil microbial biomass nitrogen (Tan and others 2005).
None of our measured environmental variables (i.e. soil temperature, soil moisture, and depth to the groundwater table) were significant drivers of N2O fluxes. Thus, perhaps some environmental variables not measured in this study could explain some of the unexplained treatment differences. For instance, soil nitrogen concentrations are an important driver of N2O fluxes (Christiansen and others 2012). A large proportion of nitrogen in upland soils is transferred to riparian ecosystems, where it can be retained via biological assimilation or removed via denitrification (Pinay and others 2018). Denitrification rates were positively correlated with soil nitrate content in a riparian forest along the Louge River in south-west France (Pinay and others 1993). Moreover, streamwater nitrate concentrations have been found to increase in the short term following forest harvest in western North America due to enhanced soil nitrification rates (Feller 2005). This increased flow of nitrate from the soil into streams in the short term, subsequently returns to pre-harvest levels due to uptake by rapidly growing biomass (Feller 2005). This phenomenon could explain the lower N2O emissions we observed at the sites with rapidly re-growing vegetation, harvested over three years earlier, compared to the reference sites.
In addition to the effects of external nutrient inputs, the potential for nitrogen retention and removal can vary between climatic regions, based on differences in soil moisture conditions and nitrogen supply. In temperate regions with high humidity, such as our study region, high soil moisture promotes nitrification and denitrification (Christiansen and others 2012). In contrast, Mediterranean and arid regions have lower denitrification rates due to low water availability and short residence time of water and solutes (Pinay and others 2018).
In the context of other studies, the soils in our study area have a moderate nutrient content and the GHG flux rates in this study were comparable to those reported in other temperate riparian zones. The soils in undisturbed stands at Malcolm Knapp Research Forest have previously been categorized as medium-nutrient (Collins and others 2001), with reported mineral soil nitrate content of 2.5 kg ha-1 and a C:N ratio of 23 (Turk and others 2008). Mean gas flux rates at the reference sites (June to September) were 65.7 ± 29.9 mg CO2 m-2h-1, -26.3 ± 17.7 µg CH4-C m-2 h-1, and 3.6 ± 2.6 µg N2O-N m-2 h-1. Since we measured gas fluxes from June to September, our average flux rates should be slightly higher than annual flux rates reported from temperate studies that include winter months when microbial activity is lower, and hence flux rates are lower. The average annual CO2 flux rate in flood-prone riparian forest in an agricultural landscape along the White River in Indiana, USA measured 48.8 mg CO2m-2 h-1 and was significantly related to the C:N ratio (16.4) of soil organic matter (Jacinthe and others 2015). At a temperate riparian forest in an agriculturally dominated landscape in Ontario, Canada, the mean annual CO2 emissions were 18 mg CO2m-2 h-1, with a soil organic carbon content of 84.1 g C kg-1 and a total nitrogen content of 12.5 g N kg-1 (De Carlo and others 2019a). At the same location, mean annual N2O emissions were 5.93 µg m-2 h-1, and were significantly correlated with soil total nitrogen (12.5 g N kg-1) (De Carlo and others 2019b). A riparian forest in New Jersey had a C:N ratio of 21 and a N2O emission rate of 0.9 µg N m-2 h-1 (Ullah and Zinati 2006). Nevertheless, there is a lack of studies about GHG emissions from riparian zones of headwaters streams, moreover most of the literature is in agriculture-dominated landscapes and not in forested riparian areas impacted by forest harvest practices.
Effects of local groundwater conditions
Landscape features that dictate soil characteristics, such as local microtopography, can be important for predicting riparian GHG emissions as they may affect the spatial distribution of soil moisture, nutrients, and organic matter, thus consequently affecting the intensity of GHG emissions (Jacinthe and Vidon 2017; Soosaar and others 2011). Local groundwater discharge conditions may create particularly important microsite variation in the riparian zones of streams. Soil conditions in DIS areas have been found to have higher base cations, soil moisture, pH levels, and nitrogen concentrations when compared to surrounding soils (Giesler and others 1998). These soil conditions may influence the processes controlling soil GHG fluxes. However, it is unknown how forest harvest in the riparian zone might influence conditions in DIS areas, and how GHG fluxes may subsequently be influenced. We hypothesized that due to the higher soil moisture at DIS areas, the emission of anaerobically produced CH4 and N2O would be higher than at ND areas. There were no significant differences between the DIS and ND areas for CO2 and N2O emissions, although ND areas generally had higher CO2 emissions on average and DIS areas generally had higher N2O emissions. However, CH4uptake was significantly lower in the DIS sites compared to the ND sites. This means that DIS sites were more likely to be CH4 sources, while ND sites were more likely to be CH4 sinks. Similar results were found in riparian zones in central Indiana, where a topographic depression in the riparian forest accounted for 78% of annual CH4 emissions, despite only covering <8% of the total land area (Jacinthe and others 2015). Additionally, GHG fluxes from flowing stream waters have been found to peak downstream of DIS areas, due to their lateral gas inputs from riparian soils (Lupon and others 2019). Given that CH4 fluxes were highest in DIS areas at no buffer sites, the results of our study provide additional support for the use of hydrologically adapted buffers, which provide more protection for wet areas, such as DIS areas, in the riparian zone (Tiwari and others 2016). The variable buffer width adapted to site-specific hydrological conditions can protect biogeochemical and ecological functions as well as provide economic savings when compared to fixed-width buffers (Tiwari and others 2016).
Conclusions
In conclusion, our work shows that forest harvest and local groundwater conditions influence GHG emissions from riparian forest soils alongside headwater streams. Our results demonstrate that riparian buffers may be effective in protecting soil ecosystem functions contributing to CH4 and N2O fluxes. Nevertheless, our findings should not diminish the importance of maintaining intact riparian forests for other benefits, such as the unique biodiversity riparian areas sustain. Given our finding that local groundwater conditions play an important role in driving CH4 fluxes, forest managers may choose to consider hydrologically adapted buffers to reduce CH4 emissions. This new information about the effects of forest harvest in the riparian zone on GHG flux rates should be considered in local policy discussions regarding how forest harvest in the riparian zone can contribute to climate change, and what can be done to mitigate or diminish the impacts. Considering that many small streams are left unprotected in British Columbia (Kuglerová and others 2020) and elsewhere, perhaps the leverage of climate change mitigation may influence policy changes leading to increased retention of riparian buffer zones in the region. In sum, this research may be useful to forest managers interested in managing riparian buffer zones for GHG balance and climate change mitigation.