Abstract
Riparian willows provide important ecosystem functions, however, in some locations they have been degraded due to over-browsing by wild and domestic ungulates. Fires can also have a detrimental effect on riparian vegetation, but have positive effects by creating bare soil necessary for willow seed germination. Both active and passive methods of restoration are common, but questions related to willow seed production are not well studied. We examined two questions: 1) Are bare soil patches created by wildfire suitable for willow seedling establishment? and 2) How has aerial seed rain changed since the establishment of exclosures to limit access of ungulates to willow stands? Working in montane riparian areas in Rocky Mountain National Park (USA), we seeded 1 m2 plots on bare soil patches created by a wildfire and monitored germination and survival. Seeds germinated in 46% of the plots and the number of seedlings ranged from 1 to 591. Plots with higher average growing season soil moisture had greater numbers of seedlings and seedling survival rates. In a nearby unburned valley, we assessed aerial seed rain biweekly using traps established at the same locations as a previous study in 2000. Five years after the establishment of exclosures, aerial seed rain was lower than in the pre-fence year 2000. We conclude that fire could trigger willow seeding establishment and controlled burns used as a viable option for restoration. In addition, five years of fencing is insufficient to reestablish willow seed rain after a period of heavy browsing.
Restoration Recap
Reducing browsing by constructing exclosures around riparian willow stands is common practice, but it is unclear how long it may take for willow seed rain to recover.
We found that five years of fencing is insufficient to reestablish willow seed rain after periods of heavy browsing.
Fires can have a detrimental effect on riparian vegetation but can create bare soil that is an ideal substrate for willow seed germination.
We found that fire could trigger willow seeding establishment and controlled burns could be used as a viable option for restoration.
Willows (Salix spp.) are the dominant woody plant in many montane riparian areas throughout the western United States. They provide important ecosystem functions such as stream bank stabilization, moderation of stream water temperature, and habitat for many species of mammals, amphibians, birds, and insects (Naiman and Decamps 1997). They are utilized by wild and domestic ungulate species as a food source, but in many areas, riparian willows have been degraded due to over-browsing (Singer et al. 1998, Brookshire et al. 2002, Peinetti et al. 2002). Techniques for restoration of riparian willow stands include building exclosures to remove browsers (Sarr 2002), planting willow stakes (Gage and Cooper 2004a), or rooted cuttings (Cooper et al. 2017), and seeding (Cooper and Van Haveren 1994).
Willows can reproduce sexually from aerially dispersed seed and asexually from adventitious rooting of stem fragments (Karrenberg et al. 2002). A single mature female willow can produce millions of wind dispersed seeds, with release timed with the decline of river stage (Karrenberg and Suter 2003). The aerial deposition of willow seeds is termed “seed rain.” Willow seeds are short-lived and do not form a seedbank (Karrenberg et al. 2002). High seed germination rates are possible under ideal conditions, but germination and seedling survival rates in the field are very low if seeds do not land on a suitable substrate (Gage and Cooper 2004b). While many active restoration projects use willow stakes to establish new populations, further colonization will be limited without a suitable seed source.
Willow seeds require bare mineral soils for germination, which along river corridors, may be exposed by flood erosion or deposition or a beaver dam blowout (Naiman and Decamps 1997, Karrenberg et al. 2002, Gage and Cooper 2005, Cooper et al. 2006). Ecosystems lacking flood disturbances or beavers may provide few suitable locations for seed germination. However, other disturbances, such as fire, may create suitable locations by exposing bare soil. After the 1988 fires in Yellowstone National Park willow seedlings established at sites more than 40 meters away from and greater than two meters above the active stream channel suggesting the availability of open burned patches can be suitable locations for establishment (Wolf et al. 2007). Seed sources in close proximity to bare sites are needed as seed rain density declines exponentially with distance. In one study, 90% of total seed rain occurred within 200 m of seed sources and the highest seed rain density was within 15 m of seed producing willows (Gage and Cooper 2005). In the absence of mature seed-producing willows, seeding could be used to establish new populations of willows, but this technique has rarely been tested (Cooper and Van Haveren 1994).
Willow catkins are formed on the previous year's stems, and in areas with heavy browsing, there is a low probability of stems surviving through the next season resulting in low catkin production and low aerial seed rain (Kay and Chadde 1992). Measurement of aerial seed rain densities can be a good indicator of the browsing status of a willow stand (Gage and Cooper 2005). Studies have examined the long term recovery of riparian willows using grazing exclosures and demonstrate increases in heights of unbrowsed willows (Marshall et al. 2013, Batchelor et al. 2015), but an increase in height may not equate to an increase in aerial seed production. There could be a reproductive lag preventing new willow recruitment (Hough-Snee et al. 2013) and it is unclear how long it takes for seed rain to recover after removal or reduction of heavy browsing.
Studies investigating questions related to willow seeds are lacking. When the long-term goal is a self-sustaining population, this topic can be important in both active and passive riparian restoration. In this paper we investigate two questions: 1) How suitable are bare soil patches created by wildfire for willow seedling establishment? and 2) How has aerial seed rain production and density changed in an unburned valley since the establishment of exclosures to limit access of ungulates to willow stands?
Methods
Study Area
Research was conducted in two similarly sized riparian valleys in Rocky Mountain National Park (RMNP), Colorado, USA: Horseshoe Park and Moraine Park (2600 and 2480 m a.s.l.) (Figure 1). Fall River, a second order stream, flows east, meandering through Horseshoe Park. The Big Thompson River, also a second order stream, flows east through Moraine Park in several channels that converge on the park's eastern end. Both streams are unregulated and have snowmelt-driven peak flows occurring in late May/ early June. Average annual precipitation is 35 cm, with most falling as winter snow (Western Regional Climate Center 2012). Floodplain soils in both parks are dominated by the Kawuneeche mucky peat (low precipitation, 0−1% slopes type), with pockets of Kawuneeche loam (0−1% slopes) in Horseshoe Park (Soil Survey Staff).
The riparian vegetation in Moraine Park is dominated by Betula occidentalis (Western river birch), Alnus incana subsp. tenuifolia (thinleaf alder), and willows, including Salix geyeriana (Geyer willow), S. monticola (park willow), and S. planifolia (diamondleaf willow). The same species of willows dominate the riparian vegetation of Horseshoe Park. Nomenclature follows Weber and Wittmann (2012).
In both parks, ungulate exclosures were constructed by RMNP to reduce or eliminate ungulate browsing, primarily by elk, and restore riparian vegetation. Twelve small exclosures were erected in both parks in 1995 and each covers approximately 0.14 hectares. Due to heavy browsing of riparian willows, a total of three large exclosures were built in Horseshoe Park in 2007 and 2008, averaging 10.26 ha in size, and a total of five exclosures were built in Moraine Park in 2008, 2010 and 2014, averaging 8.84 ha. Exclosures consist of wire fencing approximately 1.93 meters in height, and more recent constructions (2007 to present) have a gap of approximately 0.4 meters above the ground surface to allow small mammals to enter.
A fire burned through the majority of the Moraine Park riparian zone in December 2012, severely burning overstory woody riparian shrubs (Kaczynski and Cooper 2015). The fire burned approximately 150 ha of riparian area, with the majority classified as low severity burn. Horseshoe Park did not burn.
A large flood occurred in September 2013 affecting both Horseshoe and Moraine Parks. Peak flows on Fall River, just above Estes Park were 108 m3∕sec, an estimated event probability of 0.50% (Gochis et al. 2015). Flooding decimated the willow seedling burn plots therefore we only have data from the first growing season.
Map of study sites located in Rocky Mountain National Park, Colorado, USA. Horseshoe Park (A) and Moraine Park (B). Hatched locations are ungulate exclosure sites. Dots in Horseshoe Park are seed rain trap locations for both 2000 and 2013.
Field Methods
Seeding experiment. In early June 2013, we established 30, 1 m2 plots, in groups of five, throughout the recently burned riparian area in Moraine Park (Figure 1). Most plots were bare soil, however if sparse vegetation was present it was removed from the plot. We raked the plots to loosen the soil. We collected mature S. monticola catkins from the far-western side of Horseshoe Park and allowed the capsules to open in brown paper bags (2−3 catkins per bag). Plots were seeded in late June 2013 by pressing seeds onto the wet soil. Each plot was watered once with a total 19 L of water: 7.6 L of water was applied immediately before seeding and an additional 11.4 L was applied after seeding to saturate the soil. Cumulative precipitation from June through August 2013 was 13.51 cm, which is slightly lower than average (15.9 cm; NOAA Station ID: GHCND: US1COLR0767). Soil volumetric water content (VWC) was assessed weekly using a handheld Campbell Scientific Hydrosense TDR meter with 12 cm probes. Soil volumetric water content ranged from dry (0%) to saturated (50%). Five samples were averaged to obtain a mean soil moisture per plot. Plots were monitored weekly for germination and the number of seedlings were counted biweekly. We assessed seedling survival in late August 2013.
A composite soil sample comprised of three subsamples collected from the upper 10 cm of each plot was collected in August 2013 and analyzed to determine % organic matter (OM) by loss on ignition (Shulte and Hopkins 1996). Samples were oven dried to a constant weight at 105°C for 24 hours, weighed and then burned in a muffle furnace at 550°C for 7 hours. The remaining sample was weighed to determine the remaining mineral fraction.
We used logistic regression to examine seedling germination and survival in each plot at the end of the growing season (August). Average soil moisture throughout the growing season and percent OM were used as explanatory variables. All analyses were completed using R v3.0.2 (R core development project).
Seed rain. To quantify aerial seed rain, twenty 900 cm2 sticky traps were attached to t-posts approximately one meter above and parallel to the ground surface along transects in Horseshoe Park originally established in 2000 (Gage and Cooper 2005). Traps had adhesive Tanglefoot™ applied weekly. We counted seeds weekly from late May through mid–July, 2013 to capture the entire seed rain period. Seeds of different willow species are visually indistinguishable, so we did not assess the contribution of individual species to the overall seed rain. The inverse distance weighted metric in ArcGIS 10.1 (ESRI Redlands, CA) was used to interpolate seed rain values between traps. Using the interpolated surface, aerial seed rain in 2013 was compared with seed rain measured in 2000 (Gage and Cooper 2005). We performed a repeated measures generalized linear model analysis in R v3.0.2 (R core development project) to examine valley-wide differences in seed rain, incorporating year, location of the seed rain trap (inside or outside of exclosures) and their interaction. Seed rain was previously measured in Moraine Park and the majority of the study area had less than 10% of what was found previously due to the fire (Kaczynski and Cooper 2015).
A) Probability of willow seed germination at varying soil moisture levels, as measured by average growing season % volumetric water content in the soil. B) Probability of willow seed survival at varying soil moisture levels, as measured by average growing season % volumetric water content in the soil.
Interpolated aerial seed rain across Horseshoe Park in 2000 (top) and 2013 (bottom). Gray areas throughout the study area depict riparian willow communities. Points are the locations of seed rain traps. Large exclosures were constructed in 2007 and 2008, therefore are shown as dotted lines in 2000.
Results
Seeding Experiment
Willow seeds germinated in 14 of the 30 plots. The number of seedlings per plot ranged from 1 to 591. Germination in each plot was significantly related to average growing season soil moisture (%VWC; logistic regression, z = 2.684, p = 0.007), with higher germination in plots with higher average soil moisture (Figure 2A). Percent OM in plots ranged from 2−27% and was not related to germination (logistic regression, z = 0.918, p = 0.359). Percent OM and soil moisture were not correlated (R2 = 0.400).
Willow seedlings survived through late August 2013 in five plots and counts ranged from 5 to 33. Survival was significantly related to average growing season soil moisture (logistic regression, z = 2.537, p = 0.011) and was more likely to occur in plots with higher soil moisture (Figure 2B). Average growing season soil moisture in plots that had seedling survival through late August 2013 was 31% VWC compared with 21% VWC for plots that had no surviving seedlings. Survival was not related to percent OM of the plot (logistic regression, z = 1.343, p = 0.179).
Seed Rain
Seed rain was much lower in 2013 than in 2000 in most of Horseshoe Park. The greatest differences were in the western parts of the park, where our measured seed rain was ~1500 seeds/m2 compared with highs of over 4500 seeds/ m2 in 2000 (Figure 3). In eastern portions of the valley, seed rain was lower in most areas, but slightly higher in areas within the most eastern exclosure. Aerial seed rain also was lower in areas lacking exclosures (Figure 3). There was a highly significant difference in seed rain when comparing year and exclosed versus unexclosed (generalized linear model, z = −4.299, p < 0.001), where seed rain outside exclosures averaged 462 seeds/m2 in 2000 and 27 seeds/m2 in 2013. Inside exclosures, seed rain averaged 1560 seeds/ m2 in 2000 and 66 seeds/m2 in 2013.
Discussion
Riparian restoration in mountainous regions of the western United States typically involves a combination of planting willow stakes or fencing willow stands to reduce browsing. Here we examined two topics related to active and passive restoration: 1) the potential for willow seedling establishment and survival on post-fire bare soil patches; and 2) changes in aerial seed rain after the establishment of ungulate exclosures. Willow seeding may be a suitable alternative to staking where bare soil is abundant and soil moisture remains high through the summer. In addition, while exclosures reduced browsing pressure and allowed willows to gain height over a five-year period, the time frame for increased seed production appears to be considerably longer.
Although an uncommon disturbance in riparian areas, fire can have great ecological effects (Pettit and Naiman 2007), especially in areas where high intensity browsing occurs. After the 2012 fire in RMNP, browsed resprouting willow shoots had 64% less biomass than unbrowsed resprouting willow shoots (Kaczynski and Cooper 2015). Removing browsing by building exclosures is one method to restore post-fire willow populations. In addition, the newly exposed disturbed soil could be used for establishing new willow populations. Even at sites that experienced low to moderate burn severity, aerial seed rain was low following fire (Kaczynski and Cooper 2015), therefore broadcasting seeds collected from unburned riparian areas may enhance willow establishment. Working on terraces formed from placer mining in interior Alaska, Cooper and Van Haveren (1994) were able to establish willows from aerial seed rain when water was added, but soil organic matter negatively affected seedling survival by holding water from precipitation near the soil surface where it easily evaporated. In RMNP, we did not find a relationship between willow seedling survival and percent soil organic matter.
Researchers have investigated riparian ecosystem recovery after removal of wild and domestic ungulate browsing by assessing changes in stem height or biomass production (Bilyeu et al. 2008, Hough-Snee et al. 2013, Marshall et al. 2013, Kaczynski et al. 2014). Willow heights in Horseshoe Park declined from 2000 to 2008, but after installation of exclosures, some unbrowsed willows gained height. While height data were not recorded as part of this study, other published data show the trend. In 1998, Peinetti et al. (2002) documented willow heights in the far western portions of Horseshoe Park being greater than 200 cm. In the far eastern portion of the valley, where Gage and Cooper (2005) had found low willow seed rain in 2000, willows were heavily browsed and categorized as short old willows. By 2008, willows inside the newly constructed exclosures averaged 67 cm tall (se = 3 cm), while plants outside exclosures averaged 99 cm (se = 13 cm) (Zeigenfuss et al. 2011). Five years later, in 2013, plants inside exclosures had grown to an average of 120 cm (se = 6 cm) while plants outside decreased in height to an average of 74 cm (se = 9 cm) (Elk Vegetation Management, Rocky Mountain National Park, unpublished data). Willow stems greater than 250 cm in height are thought to be tall enough to escape most ungulate browsing (Keigley et al. 2002). Willows do not need to be this tall to produce catkins, but in heavily browsed ecosystems, stems that escape browsing have a higher probability of producing catkins. In a restored riparian area in Oregon, approximately 10% of willows produced catkins during the first year (Case and Kauffman 1997). Two years after exclosures were built, 34% of unbrowsed willows produced catkins, compared with only 2% of browsed willows outside exclosures. While we did not document the percent browsed and unbrowsed willows with catkins, our work suggests that five years may be insufficient for recovery of seed rain to recover to 2000 levels, when browsing intensity was lower.
A primary goal of restoration is to produce self-sustaining plant populations. Understanding the time scale involved is important. We demonstrate that despite the gains in height achieved by individual willows, seed rain increases very slowly after long periods of browsing. After five years, seed rain was low across the entire study area. Riparian plant regeneration is governed by both patterns of initial seedling establishment and long-term recruitment. While seedling recruitment was not evaluated in this study, low levels of seed rain likely produced few seedlings, reducing the probability of subsequent recruitment. Monitoring seed rain is an important aspect often overlooked in restoration studies. In addition, at sites with sufficient seed rain, there may be few locations suitable for seed germination. Controlled fires could be a useful tool for creating bare soils that are suitable for willow seedling establishment. Post-disturbance, willow seeding could be a low-cost alternative to staking where bare soil with sustained high water content is present. Because an historic large flood occurred in our study area in the fall following the seedling establishment, no individuals survived the disturbance and burial, highlighting the importance of flooding on seedling recruitment. Additional research is needed to determine the factors affecting long term seedling survival and the consequences for the success of seeding. Monitoring willow populations years after initial restoration is important to ensure successful recovery.
Acknowledgements
Research was funded through a grant from Rocky Mountain National Park. Amy Goodrich assisted with data collection and entry. We would like to thank Hanem Aboulezz and Therese Johnson at Rocky Mountain National Park for their support. Two anonymous reviewers provided useful comments that were incorporated into the final manuscript.
