Landscape-scale carbon storage associated with beaver dams

1 Introduction

[2] Beaver (Castor canadensis in North America and C. fiber in Europe) are ecosystem engineers [Rosell et al., 2005] because of the diverse effects created by their construction of dams and canals along small- to medium-sized rivers [Pollock et al., 2003]. Beaver dams obstruct flow, creating backwater areas that store sediment [Butler and Malanson, 1995; Pollock et al., 2007] and organic matter [Naiman et al., 1986, 1994]. Dams also enhance magnitude, duration, and frequency of overbank flow [Westbrook et al., 2006]. Overbank flow forms stable, multithread channels [John and Klein, 2004; Green and Westbrook, 2009; Polvi and Wohl, 2012], floodplain deposition of fine sediment and organic matter [Wohl et al., 2012], and high riparian water tables and floodplain wetlands [Hood and Bayley, 2008]. The resulting wet valley bottoms are known as beaver meadows [Ruedemann and Schoonmaker, 1938; Ives, 1942; Westbrook et al., 2011]. Greater diversity of instream and floodplain habitats associated with beaver promotes greater biomass and biodiversity than valley segments not inhabited by beaver [McDowell and Naiman, 1986; Wright, 2009], as well as greater standing stocks and longer retention of nutrients [Naiman et al., 1986, 1994; Correll et al., 2000]. By reducing longitudinal river connectivity [Burchsted et al., 2010] and enhancing lateral [Westbrook et al., 2006] and vertical [Briggs et al., 2012] river connectivity, beaver fundamentally alter river networks [Naiman et al., 1988] and create alternative stable states [Wolf et al., 2007; Wohl, 2013]. Areas with ideal habitat can support beaver populations for thousands of years [Kramer et al., 2012; Polvi and Wohl, 2012].

[3] Most studies documenting physical and ecological effects of beaver meadows focus only on a few beaver colonies (Johnston and Naiman [1990] is an exception). Very few studies address the cumulative effects of widespread dam removal across North America and Europe over the past few centuries [Pollock et al., 2003], despite the likelihood of substantial hydrologic, geomorphic, and biological cumulative effects. Sixty to 400 million beaver inhabited ~15 million km² of North America [Naiman et al., 1988], for example, and substantial or complete reductions in beaver populations occurred following European settlement of the continent.

[4] I present landscape-scale estimates of cumulative sediment and organic carbon storage associated with beaver dams in mountainous headwater catchments within Rocky Mountain National Park (RMNP) in Colorado, USA, as an example of the regional assessments necessary to understand the cumulative effects of reduced beaver populations. I use detailed characterizations of two headwater catchments and reconnaissance-level surveys of the extent of beaver dams in an additional 25 catchments to infer regional cumulative effects of dam removal associated with 19th and 20th century declines in beaver populations. In headwater catchments of the U.S. Rocky Mountains, the greatest beaver-enhanced aggradation occurs in low-gradient valleys and depressions of glacial origin [Persico and Meyer, 2009]. As much as half of the postglacial sediment in such areas is associated with beaver dams [Kramer et al., 2012; Polvi and Wohl, 2012], and high percentages of organic carbon, along with greater sediment volume per unit length of valley, result in beaver meadows disproportionately serving as carbon sinks within mountainous river networks [Wohl et al., 2012].

2 Study Area

[5] The study area is the eastern side of Rocky Mountain National Park (RMNP). Beaver were historically present in RMNP from the highest portions of the subalpine conifer forest zone (3400–2740 m ) down to and beyond the lowest elevations within the park [Packard, 1947]. Valley and channel geometries in the region exhibit substantial longitudinal variability, with predominantly steep, narrow valley segments and limited broad, low-gradient valley segments (here referred to as unconfined). Most channel segments have a single-thread planform, but a stable multithread planform can occur within unconfined valleys where logjams or beaver dams obstruct flow and create backwaters that facilitate sedimentation, overbank flow, and formation of secondary channels [Wohl, 2011, 2013].

[6] The most extensive and well-developed multithread channel segments occur in unconfined valleys immediately upstream from glacial terminal moraines (Figure 1). I refer to these as extensive beaver meadows. RMNP was glaciated multiple times during the Pleistocene, with glaciers extending down to ~2300–2400 m elevation. Terminal moraines raised local base levels and facilitated accumulation of outwash sediment and formation of relatively broad, low-gradient valley segments during glacial retreat. Beaver subsequently colonized these valley segments, in which the broad valley bottom maximizes ponded and overbank deposition associated with beaver dams. Ground penetrating radar (GPR) surveys of Upper Beaver Meadows, a broad valley segment immediately upstream from the moraine that affects Beaver Brook, indicate maximum postglacial sediment thicknesses of 6 m, with an average of 1.3 m. As much as half of this sediment is associated with beaver dams [Kramer et al., 2012].

[7] Beaver-related sedimentation depends on riverine scale. Smaller streams along which dams persist longer have more pond deposition, whereas beaver dams primarily enhance overbank deposition along larger streams, which have more frequent dam failure [Levine and Meyer, 2013]. The streams examined in RMNP are relatively small streams (<10 m wide) in wide valleys along which beaver dams promote both ponded and floodplain deposition. I refer to all beaver-induced deposition as beaver sediment.

[8] Beaver populations partly recovered from early 19th century trapping, reaching a maximum within RMNP circa 1940, before declining again, particularly after circa 1980 [Kramer, 2011]. Although there may be multiple reasons for the decline, intensive grazing by elk (Cervus canadensis) of woody riparian vegetation may be partially responsible by limiting food and dam-building materials for beaver. Elk numbers within RMNP have climbed since wolves were hunted to extinction in the region during the 1920s, although park managers have tried to manage the elk population since the 1930s [Hess, 1993]. Evidence for much more widespread beaver activity in the past remains in the form of relict, breached dams and ponds filled with fine-grained, organic-rich sediments. Data collection focused on Beaver Brook and the upper Poudre River (Figure 1 and Table S1 in the supporting information). A 1947 survey of beaver populations in RMNP recorded numerous beaver at these sites [Packard, 1947]. No beaver were present at either site by the time of the next survey in 1999 [Mitchell et al., 1999].

3 Methods

[9] Field methods included longitudinally continuous channel surveys to map relict beaver dams and sediment. Although this study emphasizes Beaver Brook and the upper Poudre River, I mapped the location of relict beaver dams and beaver on an additional 13 drainages on the eastern side of RMNP during 2010 and 2011 (Figure 1). Features were mapped using a handheld GPS with horizontal resolution of ±3 m. Channel surveys began close to treeline and continued downstream to the park boundary. Some large beaver meadows continue farther downstream, but this boundary provides a consistent, readily identifiable stopping point. I supplemented this data set with 12 additional drainages on the eastern side of the park for which I could identify the presence of extensive beaver meadows using topographic maps and aerial photographs. I delineated the length and width of each beaver meadow, assumed a depth of 1.3 m of fine-grained sediment based on GPR data from Beaver Brook meadow [Kramer et al., 2012], and calculated the total volume of stored sediment (see supporting information). Based on the GPR surveys and interpretations from the Beaver Brook meadow [Kramer et al., 2012], I conservatively assumed that half of the total volume of sediment was beaver related. The complete data set thus includes 27 streams.

[10] Detailed surveys of Beaver Brook and the upper Poudre River form the intensive data set. Relict beaver dams in steep, narrow channel segments were identified based on geometry and vegetation (supporting information), and the sediment volume was estimated for each dam. The data set included 148 dams along Beaver Brook and 100 dams along the upper Poudre River, as well as an extensive beaver meadow along each channel.

[11] I sampled and analyzed beaver sediment for total organic carbon (TOC) analysis (supporting information). Fifteen samples of beaver sediment from Beaver Brook and 14 samples from the Upper Poudre River were collected and analyzed for total organic carbon (TOC) (supporting information). These 29 samples were in addition to 10 TOC samples of the Beaver Brook meadow from an earlier study, as well as 19 samples from active beaver meadows in RMNP [Wohl et al., 2012]. I estimated the magnitude of stored organic carbon using the average TOC content of all samples, an average bulk density of 1.8 g/cm³ for beaver sediment [Wohl et al., 2012], and the volume of beaver sediment.

4 Results

[12] The largest estimated volumes of beaver sediment occur in extensive beaver meadows along rivers influenced by terminal glacial moraines (Table 1 and Figure S3 in the supporting information). Although individual relict beaver dams extend most of the channel length on Beaver Brook, 95% of the total beaver sediment occurs within the large beaver meadow near the moraine (Table S1 and Figure S4 in the supporting information). These proportions provide a rationale for estimating the landscape-scale storage of beaver sediment using only beaver meadows. Extensive beaver meadows occupy approximately a quarter of the total length of primary valleys on the eastern side of RMNP (Table 1).

[13] Using an average value of 3.3% TOC for beaver sediment in relict beaver meadows (Table S2 in the supporting information), ~735,800 Mg of organic carbon is stored in beaver meadows. In relict beaver meadows, riparian water tables have dropped, floodplain deposition of organic matter has declined, and carbon has been lost through drying and oxidation [Naiman et al., 1994; Trumbore and Czimczik, 2008]). Upland forest soils in the region average 7.3% TOC [Rueth and Baron, 2002]. Based on samples collected from active beaver meadows, I used an average TOC value of 12% [Wohl et al., 2012] to estimate the TOC storage when beaver meadows were active. Assuming that all beaver meadows were active simultaneously results in ~2,675,645 Mg of organic carbon storage.

[14] To compare these values to the total estimated organic carbon stored in the data set of 27 watersheds, I used the total drainage area of 717 km² with published values of 260 Mg C/ha ecosystem carbon stock in forested uplands [Bradford et al., 2008] and 105 Mg/ha in tundra uplands [Hartman et al., 2009] in the region. Proportions of land cover within the Fall River, Big Thompson, and North St. Vrain Creek within RMNP average 0.33 unvegetated, 0.23 tundra, 0.39 forest, and 0.05 subalpine meadow, with relatively little variation between drainages [Clow and Sueker, 2000]). I extrapolated these land cover values to the data set of 27 watersheds and estimated 9,012,500 Mg of C in the uplands. The estimated contemporary carbon storage (average 3% TOC) in beaver meadows thus represents ~8% of the total carbon storage in the landscape. The former carbon storage when these beaver meadows were active (average 12% TOC), prior to abandonment and drying, represents ~23% of the landscape total.

5 Discussion

[15] The estimates of TOC stored in beaver meadows on the eastern side of RMNP represent a first-order approximation for several reasons, as detailed in the supporting information, including use of single average values for thickness of beaver sediment, soil bulk density, and percent soil TOC. Similarly, the estimated change in total carbon storage by a factor of more than three from the last period of active beaver meadows to the present is a first-order approximation. Each set of TOC samples includes substantial variation (Figure S5 in the supporting information), as do published values of upland carbon. The estimated mean values of TOC for the two populations of active and abandoned beaver meadows are consistent across multiple sampling sites, however, and differ significantly from one another.

[16] Although values of carbon storage can be refined, the trends indicated in this analysis are robust: beaver meadows store the great majority of carbon along rivers in these mountainous headwater catchments; this source of carbon storage cumulatively represents ~8% (relict) to 23% (active) of estimated total landscape carbon storage; and riverine carbon storage has declined substantially with the replacement of beaver meadows with drier “elk grasslands” [Wolf et al., 2007]. The estimated 23% of total landscape carbon storage within active beaver meadows also agrees well with an earlier estimate of 25% of total carbon storage in unconfined valley bottoms that included beaver meadows and old-growth forest with substantial downed wood [Wohl et al., 2012].

[17] Estimated total carbon storage during periods of active beaver meadows does not imply that all of the extensive meadows were continuously occupied during the Holocene. Although these sites represent optimal beaver habitat, they may all have been simultaneously occupied only during cooler, wetter intervals of the Holocene [Persico and Meyer, 2009, 2013]. The estimated carbon storage using 12% TOC and total sediment volume in extensive beaver meadows represents maximum potential carbon storage at the landscape level relative to contemporary, relatively low values. These low contemporary values may reflect some combination of elk competition, progressively warmer and drier late 20th century climate [Westerling et al., 2006], and other factors that limit beaver populations in the study area.

[18] These results illustrate the importance of beaver meadows at the landscape scale. In addition to site-specific changes well documented in the literature, ecosystem engineering by beaver results in significant landscape-scale storage of carbon. Restoration of beaver populations and beaver meadows thus assumes critical importance in headwater management strategies designed to enhance ecosystem services such as habitat provision and carbon sequestration. When beaver are removed, wet beaver meadows are replaced by the alternate stable state of relatively dry grasslands [Wolf et al., 2007; Green and Westbrook, 2009; Polvi and Wohl, 2012]. Dry grassland soils average 40–100 Mg C/ha [Buringh, 1984], as opposed to the values of 300–400 Mg C/ha in relict beaver meadows and 1150–1400 Mg C/ha in active beaver meadows estimated in this study.

[19] This valley-bottom metamorphosis likely has important implications for ecosystem resilience to climate change, drought, and wildfire. Beaver populations and their dam-induced hydrologic effects have varied during the Holocene in response to climate fluctuations [Persico and Meyer, 2009, 2013] and will presumably vary in future. The persistence of at least some beaver, however, and the maintenance of higher riparian water tables probably created Holocene refugia for more mesic communities during drier intervals. The buffering effects of wet valley bottoms during brief, episodic disturbances are illustrated by the 2012 Fern Lake Fire in RMNP, which burned a substantial portion of Moraine Park along the Big Thompson River. Moraine Park was an active beaver meadow into the 1960s [Polvi and Wohl, 2012], and the wet valley bottom had not burned prior to the 2012 fire, despite several historical fires in the catchment.

6 Conclusions

[20] The recent loss of beaver populations in mountainous headwater catchments in Rocky Mountain National Park in Colorado, USA, provides an opportunity to examine large-scale implications of older historical loss or reduction of beaver populations throughout Eurasia and North America. Absence of beaver and beaver dams drives the metamorphosis of wide valley-bottom segments from wet beaver meadows to drier grasslands, with associated loss of habitat, biodiversity, and carbon storage. First-order approximations of total carbon storage on the eastern side of RMNP suggest that the volume of organic carbon stored in beaver meadows represents 8 to 23% of the estimated total carbon storage across 27 watersheds. Estimated cumulative carbon storage in beaver meadows decreased by a factor of more than three with loss of beaver populations. Widespread alteration of unconfined valley segments in these mountainous headwater stream networks provides a starting point to understand the magnitude of cumulative effects associated with historical declines in beaver populations across headwater and lowland stream networks.