Ecosystems

, Volume 17, Issue 3, pp 442–457

Cascading Effects of Climate Change on Forest Ecosystems: Biogeochemical Links Between Trees and Moose in the Northeast USA

Authors

    • Vassar College
  • M. J. Mitchell
    • SUNY College of Environmental Science and Forestry
  • P. M. Groffman
    • Cary Institute of Ecosystem Studies
  • G. M. Lovett
    • Cary Institute of Ecosystem Studies
Article

DOI: 10.1007/s10021-013-9733-5

Cite this article as:
Christenson, L.M., Mitchell, M.J., Groffman, P.M. et al. Ecosystems (2014) 17: 442. doi:10.1007/s10021-013-9733-5
  • 460 Views

Abstract

The relationship between herbivores, plants and nutrient dynamics, has been investigated in many systems; however, how these relationships are influenced by changing climate has had much less attention. In the northeastern USA, both moose populations and winter climate have been changing. Moose, once extirpated from the region, have made a comeback; while locally, snow depth and duration of snow cover have declined. There is considerable uncertainty in how these changes will interact to influence forested systems. We used small experimental plots and transects along with snow removal (to elicit soil freezing and expose potential forage plants), mechanical browsing, and fecal additions (labeled with 15N) to examine ecosystem responses. We found that snow removal changed moose browsing behavior, with balsam fir more heavily browsed than sugar maple or Viburnum under low snow conditions. Soil freezing alone did not significantly alter N dynamics or selected plant responses, but there were significant interactions with moose activity. The combined effects of moose fecal additions, mechanical browsing, and soil freezing resulted in higher levels of NO3 leaching under fir and maple, whereas Viburnum had essentially no response to these multiple factors. Our results suggest that declines in snow depth can initiate a cascade of ecosystem responses, beginning with exposure of plants to increased browsing that then triggers a series of responses that can lead to higher N losses, precipitated by decreased N demand in plants compromised by soil freezing damage. Balsam fir may be particularly susceptible to this cascade of multiple stresses.

Key words

climate changesnow pack depthsoil freezingN cyclingherbivoresmoose15N stable isotopes

Introduction

A major focus of ecological research over the past two decades has been the investigation of climate change impacts on ecosystems of the world (Jonasson and others 1993; Melillo and others 2002; Wan and others 2005; Groffman and others 2012). An important factor associated with changing climate in the northeastern United States over the past few decades has been the alteration of snowpack development and duration (less snow on the ground for a shorter time), which can lead to large-scale shifts in ecosystem functioning and ultimately influence ecosystem structure (Campbell and others 2009; Kreyling 2010; Groffman and others 2012). Declines in snow depth and cover duration have raised concerns about a greater frequency and depth of soil freezing events across the USA, as these events have been linked to alterations in microbial activity and community composition (Schimel and Clein 1996; Schadt and others 2003; Schmidt and Lipson 2004) increased export of NO3 to streams (Mitchell and others 1996; Brooks and others 1998; Fitzhugh and others 2001) and N gas flux (Groffman and others 2006). Many of these observed changes appear to be driven by increases in fine root mortality and loss of root vitality (Tierney and others 2001; Cleavitt and others 2008; Kreyling 2010) and variation in decomposition (Christenson and others 2010).

Ecosystems do not experience climate change effects independently; rather, other factors are simultaneously interacting to generate novel ecosystem outcomes. As winter climate has been changing over the past 50 years in the lower New England states, another notable change has been the return of a once extant large herbivore, the moose (Alces alces) (Campbell and others 2005; Christenson and others 2010). These large herbivores can significantly affect forest ecosystems, producing complex changes in forest composition and nutrient dynamics (Krefting 1974; Brandner and others 1990; Pastor and others 1993; Pastor and others 1998; Cahoon and others 2012). Current models examining the impact of climate change on future forest composition in the Northeast do not consider the role that herbivores may play in structuring these ecosystems (Prasad and Iverson 1999). A major goal of our study is to link these two factors.

Much research has addressed the effects of herbivory on plants including changes in foliar N content (Matson 1980; Ritchie and others 1998; Singer and Schoenecker 2003), changes in plant chemical composition through production of secondary defensive compounds (Matson 1980), or increased growth rate (Bergstrom and Danell 1987; Bergman 2002). Alternatively, a plant may have less growth, be stunted by repeated herbivory (Brandner and others 1990) and ultimately die under excessive herbivore pressure (Bryant and others 1992; Molvar and others 1993). The responses of plant communities to herbivory can lead to diverse changes in nutrient cycling rates, including both stimulation of N cycling by increased N content in foliage and slowing of cycling through replacement of “fast” nutrient cycling plant species (low C:N litter) with “slow” nutrient cycling tree species (high C:N litter) (Belovsky and Slade 2000). Tree species strongly influence N cycling in forested ecosystems, primarily through differences in the chemical quality of their foliage and resultant litter that they produce (Mudrick and others 1994; Finzi and others 1998; Lovett and Mitchell 2004). Conifer species tend to have poor quality litter (high C:N ratios, high lignin, and tannin concentrations) and slow decomposition and are generally not preferred by large browsing ungulates. In contrast, deciduous species tend to have better quality litter (low C:N ratios, low lignin, and secondary compound concentrations) and faster decomposition (Aber and Melillo 1980; Melillo and others 1982) and are often the preferred forage.

Large herbivores can influence ecosystem processes through feedbacks in N cycling dynamics (Bardgett and Wardle 2003; Bakker and others 2004). Two major pathways of influence are through selective browsing (indirect) and the contribution of fecal and urine inputs to the system (direct) (Frank and others 1998). Many researchers have concluded that indirect herbivore effects (that is, changes in litter quality and quantity through selective herbivory) influence nutrient cycling more than direct fecal inputs (McInnes and others 1992; Pastor and others 1993; Kielland and Bryant 1998). However, these studies have generally not investigated the actual contribution of feces as a source of N within the ecosystem. Investigations into the contribution of insect frass (feces) to ecosystems have shown that feces N is directly available to plants and may also be stored in soils (Christenson and others 2002; Frost and Hunter 2007), whereas other studies have shown increased leaching loss of N from systems experiencing insect herbivore defoliation (Swank and others 1981; Eshelman and others 1998). This considerable variation both within and among ecosystems on the movement of N into plants and/or loss via soil leaching stemming from herbivore feces led us to question how an important herbivore, such as the returning moose, would impact forest dynamics after a long absence from these systems.

Recent research has focused on complex relationships between herbivores and their environment, including changes in the environment induced by the herbivore with feedbacks to above- and belowground communities (Hobbs 1996; Bardgett and Wardle 2003) or to coupled herbivore/climate change ecosystem responses (Cahoon and others 2012; Sjögersten and others 2012; Stevnbak and others 2012; Wang and others 2012). Work by Post and Stenseth (1999), in Norway, related climatic change to plant phenology and ungulate demographics; whereas Bardgett and Wardle (2003) reported on the indirect effects of herbivores on decomposer organisms, the activities of which affect plant communities. Many of these studies (Cahoon and others 2012; Sjögersten and others 2012; Stevnbak and others 2012; Wang and others 2012) tested relationships between herbivores and warming, including moisture conditions; however, relatively few studies have investigated the impacts of changes in winter climate via changes in snowpack and soil freezing (Martin and Maron 2012).

The purpose of our study was to investigate the relationships between moose activity, nitrogen dynamics, and plant response under modified conditions of snowpack. We hypothesized that declines in snowpack, as projected by Hayhoe and others (2006), would lead to changes in moose browsing behavior. To test this hypothesis, transects dominated by either balsam fir, sugar maple, or Viburnum, all palatable species to moose, were established and treated with experimental manipulations of snow depth (to expose plants). Secondly, we hypothesized that browsing of plants would increase N uptake and that moose fecal N would be available to these plants. We also predicted that reduced snow depth would negatively affect plant N uptake and moose fecal N availability through soil freezing damage to roots (belowground stress) and that browsing would amplify that effect. To test these hypotheses, we used small experimental plots treated with snow removal (to elicit soil freezing), mechanical browsing, and fecal addition (labeled with 15N) and dominated by balsam fir or sugar maple or Viburnum. This study was conducted at the Hubbard Brook long term ecological research (LTER) site in the White Mountains of NH, USA. This work is part of ongoing efforts to evaluate the effects of winter climate change on the structure and function of the northern hardwood forest ecosystems that dominate this study site (Campbell and others 2005; Christenson and others 2010; Groffman and others 2012).

Site Description

This research was conducted at the Hubbard Brook Experimental Forest (HBEF), located in central New Hampshire, USA (43°56′N, 71°45′W). The area is classified as a mature northern hardwood forest with a greater abundance of red spruce and balsam fir at higher elevations (Schwarz and others 2003). The major overstory species found in the area of the experimental plots were sugar maple (Acer saccharum Marsh.), yellow birch (Betula alleghaniensis Britt.), red spruce (Picea rubens Sarg.), American beech (Fagus grandifolia Ehrh.), white ash (Fraxinus americana L.), and red maple (Acer rubrum L.). Balsam fir (Abies balsamea L. Mill.) is also present and Viburnum alnifolium Marsh. (hobble bush) is an important understory shrub species. The snowpack is generally present from mid-November to mid-April (165 days, 30-year average) with average January air temperatures of −9°C and average winter (December–March) temperature of −4.7°C (Hardy and others 2001). Soil freezing is spatially and temporally variable depending on snowdepth and aboveground temperatures, but can occur from the end of November to the end of March (Hardy and others 2001; Campbell and others 2010).

Experimental Design

To investigate how moose activity and soil freezing impact N dynamics and tree species response, 48 experimental plots (2 × 2 m2) were established on relatively level terrain on the north facing slope of Kineo Mt. (43.918686N 71.778091W) at the HBEF at an elevation of approximately 640 m. Each plot was selected to contain three naturally growing saplings of the same species (sugar maple, balsam fir, or Viburnum) with 16 plots for each sapling species. These three species were selected based on their known palatability to moose and observations at HBEF indicating that fir was being browsed heavily (Christenson 2007) that Viburnum was filling forest gaps and that maple is susceptible to soil freezing through root mortality (Tierney and others 2001). The heights of these saplings ranged from 0.5 to 1.5 m. Each plot was randomly assigned one of the following treatments with two replicates: snow shoveling/no shoveling; mechanical clipping of saplings/no clipping; addition of 15N labeled moose feces/no addition; in a complete factorial design to include each possible combination. Two reference plots were also established for each of the sapling species. We assumed that mechanical clipping resembled herbivory based on results found by McLaren (1996) and Bergman (2002). Snow was removed in each of the snow manipulation plots leaving approximately 5 cm of snow on the plots to protect the ground and maintain surface albedo. This depth of snow was maintained from December through mid-February in 2003/2004 and 2004/2005.

Transects with snow manipulation were established to test the hypothesis that snow depth would modify browsing behavior by moose in winter. Twelve, 15 × 1 m2 moose browse transects were established at two locations (six transects at each location). These transects were located at a distance greater than 100 m from the small experimental plots. Location 1 transects were established at approximately 600 m elevation, and location 2 transects were established at approximately 700 m elevation on Kineo Mt. At each location, a pair of transects each dominated by fir, sugar maple or Viburnum were selected (six total transects/location). One transect of each pair had snow maintained (by shoveling) at a 0.5 m depth for the duration of winter 2004–2005, whereas the other transect remained un-manipulated.

In August 2004, 10 saplings (~2 m height) each of Viburnum, fir, and sugar maple were identified on each transect (120 total saplings) and marked with a metal tree tag tied to a major branch. Each sapling was surveyed in August 2004 for lowest height of browsing from the ground using a folding forestry ruler. Saplings were monitored for browsing during the snow removal period and a final survey of the saplings was made after the snow manipulation (June 2005), to measure new browsing and lowest height of browsing from the ground. Browsing was quantified as the removal of plant material on each individual plant. We did not consider the intensity of browse rather a “browsed” plant was the one that had at least one branch eaten.

Methods

Production of 15N-labeled moose feces is described in detail elsewhere (Christenson and others 2010). Briefly, in May 2003, a 30 × 30 m2 vegetation exclosure plot at the Kenai Moose Research Center, Alaska was labeled with 15N as 15NH4Cl (99 at.%). In October 2003, after leaf senescence, a captive bred, 7-year-old female moose was introduced to the exclosure and allowed to eat the labeled browse, supplemented with pelleted food (aspen sawdust pellets) that were labeled with 15N-enriched urea (99 at.%).

Fecal pellet groupings were collected individually, weighed and stored at 0°C. Approximately 25 kg of moose feces were collected and packed in a cooler with ice packs and flown back to the Cary Institute of Ecosystem Studies in Millbrook, N.Y. Sub-samples of each day’s collection were dried at 60°C for 48 h, ground and sent to the stable isotope laboratory at UC Davis for 15N analysis. Labeled feces were highly enriched in 15N (0.493 atom %).

Sapling Plot Instrumentation and Sampling

Each plot was fitted with a frost gauge, tension lysimeter, and two tall (3 m) PVC posts to help locate plots during the winter snow season. Frost gages were constructed following methods used by the Cold Regions Research and Engineering Lab (CRREL), US Army Corp. of Engineers (Ricard and others 1976). PVC casings (1.27 cm diameter, 100 cm length) were installed in each plot to a minimal depth of 20 cm below the surface to a maximum depth of 60 cm. Flexible PVC tubing was filled with water and methylene blue dye and plugged on either end to form the gauge. As soil temperature decreases below freezing (<0°C) the “blue” indicator dye turns clear, allowing for sensitive demarcation of soil frost depth. Frost gauges were installed in December 2003 and soil frost depth was recorded monthly during winter 2003/2004 and 2004/2005. The snow removal manipulation resulted in significantly (P ≤ 0.02) greater soil freezing depth for the months of January through April (data not shown).

Porous cup lysimeters (Mitchell and others 2001) were installed below the major rooting zone (20–40 cm depth) in each plot to monitor dissolved organic nitrogen (DON), NH4+ and NO3 concentrations and the 15N fraction in total inorganic N (TIN). Lysimeter samples were collected monthly from January 2004 through December 2005. Tension was set at 0.276 MPa 24 h prior to collection of samples. Samples were analyzed for pH on the day of collection, and sub-samples for further chemical analyses were stored at 4°C. A second sub-sample was frozen for later 15N analysis. NH4+ (salicylate method #10-107-06-2) and NO3 (cadmium reduction method #10-107-04-1-A) were analyzed on a Lachat Quikchem 8100 Flow Injection Analyzer (Lachat Instruments 2012). A persulfate digest method (modified from Cabrera and Beare 1993) was used to determine DON concentration. Frozen samples were thawed and bulked into growing season (May–September) and non-growing season (October–April) composite samples for each sapling plot (96 total samples). A modified diffusion technique (Stark and Hart 1996) was used to determine 15N values in the TIN pool. DON concentration was low to non-detectable (<0.02 ppm); therefore, 15N analyses were not performed for the DON pool.

All sapling plot soils were sampled in May and August of 2004 and 2005 (four sampling periods). On each sampling date, the surface litter (discernible litter plus some of the Oi horizon incorporated in this litter layer) was gently moved aside and a PVC soil core (~5 cm diameter) was used to extract soils from each plot. A minimum of two cores were collected in each plot and composited, ensuring an adequate mass of soil for analysis. Soils were sampled to 10 cm depth, and the cores separated into organic, comprising the Oe and Oa horizons (forest floor), and mineral horizons. The forest floor depth ranged from 3 to 12 cm. If the organic horizon depth exceeded 10-cm depth, no mineral soil was collected. Soils were transported to the laboratory within 4–5 h of collection and stored at 4°C for further analyses. All incubations and soil extracts were performed within 24 h of field collection.

To ensure that all saplings were subject to similar light conditions, light availability to the saplings was measured in August 2005 using canopy fisheye photography and photos were analyzed using GLA Version 2® software (Frazer and others 1999). No significant differences in light conditions were observed across the experimental plots.

Green and senescent leaf tissue was analyzed for total N, 15N, and C content (Christenson and others 2002). Total C and N content were determined with a Carlo Erba Elemental Analyzer®. 15N analysis was done by isotope ratio mass spectrometry at the Stable Isotope Facility at UC Davis, CA. Moose fecal pellets from Alaska and Hubbard Brook were analyzed for N, 15N, and C using the same methods as those used for soil.

Labeled fecal pellets were applied to designated sapling plots in January 2004 (one time application). Fresh native moose fecal pellet groupings (20) were surveyed at Hubbard Brook during fall 2003 to determine the mass of feces to add to the experimental plots (499 g wet ± 12.89 std err). Approximately 500 g (exact mass recorded) of labeled pellets were placed in mesh (2-cm opening) bags, covering an approximately 30 × 20 cm2 area (as per field observations) and secured on the designated treatment plots with metal ground staples. Fine mesh (2 mm) screen was placed around these bags to prevent pellets from washing away during spring snowmelt. Screens were removed after snowmelt. Manual clipping of designated saplings took place in April 2004. Pruning sheers were used, and about 50% of the total sapling above ground biomass was clipped. All saplings were protected from natural moose browsing with fiberglass mesh screening (2-mm mesh) to cover the saplings during the non-leaf period (~October to April) for the duration of the study. Saplings in the plots were harvested in May 2006 as described by Christenson and others (2010).

Soil Analyses

All soils were brought directly back to the laboratory after collection and hand-sorted to remove roots, woody debris, and stones and gently hand-homogenized. Sub-samples of fresh soil were removed for extraction of inorganic N and for measurements of microbial biomass C and N content, potential net C and N mineralization and nitrification and pH. The remainder of the sample was dried for 48 h at 60°C to determine moisture content, and stored in a paper coin envelope placed in a sealed plastic bag.

pH

To determine soil pH, a soil slurry was created in a 2:1 ratio of ultrapure water to field moist soil. The suspension was stirred intermittently for 30 min then allowed to stand for 1 h (Carter 1993). An electrode (Fisher Accumet Model 610A®) was used to measure pH of the supernatant.

Moisture

Moisture contents were determined gravimetrically after drying at 60°C for 48 h. The % moisture was calculated by:
$$ \left( {{{\left( {{\text{wet}}\;{\text{wt}} .\; - \;{\text{dry}}\;{\text{wt}}.} \right)} \mathord{\left/ {\vphantom {{\left( {{\text{wet}}\;{\text{wt}} .\; - \;{\text{dry}}\;{\text{wt}}.} \right)} {{\text{dry}}\;{\text{wt}}.}}} \right. \kern-0pt} {{\text{dry}}\;{\text{wt}}.}}} \right)\; \times \; 100 $$

Total C and N

After moisture determination, the dried soils were ground to a fine powder in a KLECO® pulverizer and stored in paper coin envelopes. Analysis for total element % C and N were determined on a Carlo-Erba NA1500® analyzer.

Extractable N

Ten grams of sieved, field moist soil were weighed into plastic specimen cups and 50 mL of 2 mol/L KCl was added. Samples were placed on a shaker table at 125 rpm for 1 h and then allowed to stand for 1 h. The supernatant was filtered through Whatman® (42) ashless filter paper into polyethylene sample bottles. Samples were stored at 4°C until analysis. A Lachat Quikchem 8100 Flow Injection Analyzer® was used to analyze the samples for NH4+ (salicylate method) and NO3 (cadmium reduction method). All results are reported in μg N/gDW (dry weight) soil.

Potential Net C and N Mineralization and Nitrification

Unamended soil samples were incubated in glass quart (946 mL) canning jars fit with airtight lids fitted with butyl rubber septa to allow for gas sampling. Jars were incubated in the dark at room temperature for 10 days. After incubation, gas samples were taken by syringe for analysis of CO2 by thermal conductivity GC (Shimadzu ® GC 14) and inorganic N was extracted and analyzed as described above.

Daily potential C and N mineralization and nitrification rates were calculated by:
$$ \begin{aligned}{\text{C-mineralization}} = {\text{CO}}_{2} \ {\text{production}} \ {\text{days}} \\ {\text{N-mineralization}} = \left(({\text{final NH}}_{4}^{+} + {\text{NO}}_{3}^{-}) - ({\text{initial NH}}_{4}^{+} + {\text{NO}}_{3}^{-}) \right) \ {\text{days}} \\ {\text{Nitrification}} = ({\text{final NO}}_{3}^{-} - {\text{initial NO}}_{3}^{-}) \ {\text{days}} \end{aligned}$$
where CO2 production is the total amount of CO2 produced over the 10-day-incubation, final N is that found after incubation, initial N is the extractable N prior to incubation, and the 10 in the denominator represents the 10-day-incubation period. All N mineralization and nitrification rates are reported as μg N/(gDW soil/day).

Microbial C and N

Microbial biomass C and N content were determined using the chloroform fumigation-incubation method (Jenkinson and Powlson 1976). Ten grams of soil were fumigated for 14–16 h, inoculated with 0.1 g of fresh soil, and then incubated in 946 mL glass mason jars and sampled as described above. A proportionality constant (kc = 0.41) (Jenkinson and Powlson 1976) was used to calculate biomass C from the CO2 produced during the incubation. No constant was used in calculating biomass N.

To calculate the mass balance of added moose fecal 15N, soil mass values of 12.8 kg/m2 soil in the forest floor (Oe and Oa horizons) and 74.2 kg/m2 soil in the mineral horizon to a 10-cm depth from Bohlen and others (2001) were used.

Isotopic Analyses

To determine the abundance of 15N in KCl extracts and soil lysimeter solutions, a modified N-diffusion technique (Brookes and others 1989; Stark and Hart 1996) was used. Acidified glass-fiber disks were used to trap NH3 volatilized in the procedure, and these disks were packed in tin capsules, sealed and sent to the Stable Isotope Facility at the University of California at Davis. Recoveries of greater than 90% of known N diffused were accepted in data analysis. Samples with recoveries lower than 90% (less than 3% of total samples analyzed) were not used due to potential fractionation of N leading to underestimation of the 15N content in the pools measured (Stark and Hart 1996). Solid soil, plant, and fecal samples were dried at 60°C, ground to a powder, and packed in tin capsules and sent to the UC Davis facility for total 15N analysis.

Statistical Analyses

All data were tested for normality of distribution using the Shapiro–Wilk test (PROC UNIVARIATE; SAS 8.2 2000) and non-normally distributed dependent variables were log10 transformed prior to analysis. A two-way ANOVA followed by a Student–Newman–Keuls post hoc test was performed with the SAS statistical package to determine significant differences between treatments and species and to test for interactions between treatments and species (PROC GLM; SAS 8.2 2000). The independent variables were treatment (feces added vs. no feces added, snow removal vs. no snow removal, mechanical clipping of plants vs. no mechanical clipping or no manipulation) and species (fir, maple, Viburnum). If no interactions occurred between species and treatments, and if species differences in the independent variables tested were not significantly different, data were pooled across the species and a one-way ANOVA testing treatment effects was used. A repeated measures ANOVA (MANOVA JMP V10) with a sphericity test was used to determine significant differences for the response (dependent) variables measured over the four sampling dates (May 2004/2005, August 2004/2005) with species and treatment (independent variables) as the model effects.

To determine if snow removal on the moose browse transects increased browsing on selected tree species, a Fisher’s Exact Test was employed. A 2 × 2 table using browse and no browse (number of stems) against shoveled or no shovel treatment was used for each species independently and combined, and the 2-tailed P value reported.

Results

Snow Removal and Moose Browsing Behavior

The maintenance of snow at a 0.5 m depth during winter 2004–2005 resulted in more browsing of all plant species investigated, but the effect was more marked (P = 0.07) in fir plots than for the other sapling species (Figure 1). Analyzing the three sapling species together, snow manipulation resulted in significantly higher browsing compared to the non-manipulated transects (P = 0.05).
https://static-content.springer.com/image/art%3A10.1007%2Fs10021-013-9733-5/MediaObjects/10021_2013_9733_Fig1_HTML.gif
Figure 1

The percentage of sapling species browsed on shoveled (snow depth maintained at 0.5 m winter 2004/2005) and non-shoveled transects. Shoveling significantly increased the percentage of plants browsed (P = 0.057).

Ecosystem Response to Moose and Soil Freezing

Over a combined 2-year period, across species and soil horizons, the addition of moose feces did not significantly change microbial biomass, N mineralization, or nitrification rates compared to plots where no feces were added (Table 1). Saplings growing in these experimental plots also had no change in percent N in green leaf tissues when moose feces were added (Table 1). Addition of feces or mechanical clipping had no effect on C or N concentration in woody stems or roots of the saplings (data not shown).
Table 1

Microbial Biomass C and N, Potential Net N Mineralization, and Nitrification and Plant N Content in Plots With and Without Moose Feces Added

 

Moose feces added

No moose feces added

Microbial biomass C (μg C/gDW soil)

117.6 (14.7)

115.1 (12.7)

Microbial biomass N (μg N/gDW soil)

290 (22.7)

281 (21)

N mineralization rate (μg N/(gDW soil/day))

8.4 (1)

7.9 (1)

Nitrification rate (μg N/(gDW soil/day))

3.9 (0.3)

5 (0.6)

Plant–green leaf N (%N)

2.2 (0.18)

2.1 (0.2)

Values are means across species, soil horizon, and sampling dates. Standard errors are given in parentheses. N = 48 for soils data, N = 6 for plant %N. There were no significant differences between treatments for any variable.

Similarly, soil freezing did not affect N cycling on plots with moose feces added (Table 2). No changes in microbial biomass, N mineralization or nitrification or plant N concentrations were found in these plots (Table 2). As in the feces versus no-feces addition plots, soil freezing also had no effect on woody sapling tissue N and C concentrations (data not shown).
Table 2

Microbial Biomass C and N, Potential Net N Mineralization, and Plant N Content in Plots With and Without the Snow Removal Treatment

 

Soils frozen

Soils not frozen

Microbial biomass C (μg C/gDW soil)

129.7 (14.1)

117.6 (14.7)

Microbial biomass N (μg N/gDW soil)

302 (18.5)

290 (22.7)

N mineralization rate (μg N/gdw soil/day)

7 (0.8)

8.4 (1)

Nitrification rate (μg N/(gDW soil/day))

4 (0.4)

3.9 (0.3)

Plant–green leaf N (%N)

2.1 (0.1)

2.2 (0.18)

All plots had moose feces added. There were no significant differences between treatments for any variable. Values are means across species, soil horizon, and sampling dates. Standard errors are given in parentheses. N = 48 for soils data, N = 6 for plant %N.

Nitrification and microbial biomass C and N did differ in response to treatment for the species as measured over time. There was a significant time by species interaction for nitrification in the mineral soils (unadjusted F test; P < 0.02) (Supplemental Figure 1). Microbial biomass C and N responded to the multiple treatments (feces added plus clipping plus shoveling) in different ways as a function of tree species (Figure 3). There were significant time and species interactions for microbial biomass C (adjusted F test; P < 0.05) and microbial biomass N (unadjusted F test; P < 0.01).

15N Recovery

Generally, distribution of 15N within the pools measured did not differ significantly between the treatments within a species or among the species (Table 3), but there were notable exceptions. Over 80% of the 15N added as moose feces was recovered in all plots (Table 3). The major sink for mobilized fecal 15N was the mineral soil while a substantial proportion of 15N remained in the undecomposed moose feces (Table 3).
Table 3

The Fate of 15N-Labeled Moose Feces Added to Soils Under Balsam Fir, Sugar Maple, and Viburnum Subjected to Snow Removal and Clipping Treatments

 

Frozen soils

Non-frozen soils

Clipped/shoveled

No clipping/shoveled

Clipped/no shoveling

No clipping/no shoveling

Balsam fir

Sugar maple

Hobble bush

Balsam fir

Sugar maple

Hobble bush

Balsam fir

Sugar maple

Hobble bush

Balsam fir

Sugar maple

Hobble bush

mg 15N added

16.81

16.78

16.79

16.86

16.81

16.78

16.82

16.8

16.83

16.84

16.75

16.79

mg 15N left in undecomposed feces

5.77

5.35

4.66

5.60

5.57

4.20

5.58

5.14

5.78

4.24

5.48

7.07

mg 15N to trace

11.03

11.43

12.13

11.25

11.24

12.59

11.24

11.66

11.06

12.60

11.27

9.71

Organic soil (total mg 15N)

0

0

0

0

0

0

0

0

0

0

0

0

Mineral soil (total mg 15N)

8.02

11.08

9.19

13.15

22.85

11.72

12.66

10.25

8.91

9.35

8.57

8.83

 Extractable

0.02

0.02

0.01

0.01

0.01

0.02

0.01

0.03

0.01

0.01

0.01

0.03

 Mineralizable

0.07

0.04

0.02

0.08

0.06

0.06

0.06

0.02

0.06

0.04

0.04

0.15*

 Microbial

0.18

0.3

0.26

0.29

0.31

0.29

0.25

0.26

0.17

0.22

0.34

0.15

%Recovery of total added 15N

 Soil N pool

47.7

66.0

54.7

78.0

135.9

69.8

75.2

61.0

52.9

55.5

51.2

52.6

 Microbial N pool

1.6

2.6

2.1

2.6

2.8

2.3

2.3

2.2

1.5

1.8

3.0

1.6

 Mineralizable N pool

1.1

0.9

0.8

1.0

0.9

0.9

1.0

0.4

1.3

0.5

0.5

2.7

*Indicates statistically significant (P < 0.05) difference between species within a treatment.

The soil mineralizable N pool under Viburnum in the non-clipped, no-snow removal treatment had significantly higher 15N concentrations (P = 0.05) compared to the other three treatments (Table 3). Recovery of 15N in the microbial pool did not differ among treatments or species. However, soil beneath sugar maple always had a higher portion of 15N in this pool for all treatments compared to fir or Viburnum (Table 3).

To estimate the proportional plant uptake of fecal N, the ratio of the percentage of 15N in excess of background 15N to total 15N in plant tissue was calculated (Table 4). Fir had greater proportional uptake of fecal N compared to sugar maple and Viburnum (Table 4). The highest proportional uptake of N in the fir treatments occurred in plots that were not clipped and had no snow removal (0.277% of 15N added). Viburnum had the next highest proportional uptake of fecal N in non-clipped plants, regardless of snow removal. Sugar maple had the least amount of fecal N proportional uptake.
Table 4

Total and Excess (From 15N-labeled Moose Feces) 15N in Balsam Fir, Sugar Maple, and Viburnum Plant Parts in Plots Subjected to Snow Removal and Clipping Treatments

  

Frozen soil

Frozen soil

Non-frozen soil

Non-frozen soil

No clipping

Mechanical clip

No clipping

Mechanical clip

Total plant part 15N (mg 15N/g tissue)

Excess (mg 15N/g tissue)

Uptake of fecal N (% of total 15N)

Excess (mg 15N/g tissue)

Uptake of fecal N (% of total 15N)

Excess (mg 15N/g tissue)

Uptake of fecal N (% of total 15N)

Excess (mg 15N/g tissue)

Uptake of fecal N (% of total 15N)

Balsam fir

       

 Green needle

0.063

3.16E−05

0.05

1.80E−04

0.285

1.75E−04

0.277

1.65E−05

0.026

 Stem

0.032

0

0

4.96E−05

0.155

7.94E−05

0.248

9.97E−06

0.031

 Root

0.025

0

0

2.45E−05

0.1

3.10E−05

0.127

4.76E−06

0.019

Sugar maple

        

 Green leaf

0.09

0

0

0

0

0

0

0

0

 Stem

0.03

0

0

0

0

0

0

0

0

 Root

0.043

1.63E−05

0.038

0

0

0

0

3.00E−07

0.0007

Viburnum

      

 Green leaf

0.075

7.61E−05

0.102

0

0

3.72E−05

0.050

0

0

 Stem

0.026

1.50E−05

0.058

0

0

2.90E−05

0.113

0

0

 Root

0.037

1.68E−05

0.046

0

0

0

0

0

0

Soil Solution Chemistry

Fir plots that had multiple treatments (snow removal to freeze soil, feces added, and saplings mechanically clipped) had higher soil water (lysimeter) NO3 concentrations than maple or Viburnum plots (Figure 2). This pattern was observed in soil solutions during both the non-growing (P = 0.005) and growing (P < 0.0001) seasons (Figure 2).

The treatment effects differed among the sapling species. Fir plots with added moose feces had significantly higher NO3 (P = 0.02) in soil solution than fir plots with no feces added in both the growing (Figure 2A, P < 0.02) and non-growing season (Figure 2B, P < 0.0002). Fir plots with snow removal had higher NO3 concentrations during the growing season (Figure 2A, P < 0.028). Viburnum plots with mechanical clipping had higher NO3 (P = 0.001) compared to non-clipped treatment plots during the dormant season only (Figure 2B). Maple plots that were clipped (P = 0.018) and/or had snow removed (P = 0.004) had higher NO3 concentrations compared to plots with no clipping or shoveling (Figure 2A) during the growing season. Viburnum that were clipped had a higher concentration of NO3 (P = 0.03) in the soil solution compared to the other treatments (Figure 2A) during the growing season.
https://static-content.springer.com/image/art%3A10.1007%2Fs10021-013-9733-5/MediaObjects/10021_2013_9733_Fig2_HTML.gif
Figure 2

Mean concentration of NO3 collected in tension lysimeters in balsam fir, sugar maple, and Viburnum plots in the A growing season (May–September) and in the B non-growing season (October–April). Balsam fir has significantly higher NO in the non-growing (P < 0.02) and growing season (P < 0.001) in the multiple treatment plots.

Discussion

We found that moose can be an important regulator of ecosystem response to changing winter climate conditions at the HBEF. The decline of snow depth can initiate a cascade of ecosystem responses, beginning with exposure of forage plants to increased browsing that then triggers a series of responses that can ultimately lead to higher soil water NO3 concentrations. Our study indicates that balsam fir may be particularly susceptible to the combined impacts of moose activity, reduced snow depth, and resultant soil freezing. Snow removal increased moose browsing of balsam fir much more than the other two species (Figure 1), the addition of moose feces doubled soil water NO3 concentration under balsam fir; whereas the combination of moose feces, clipping, and snow removal treatment increased soil water NO3 tenfold (Figure 2). Therefore, decreased snow depth not only exposed balsam fir to greater browsing, the subsequent effects of addition of fecal material in conjunction with increased browsing and decreased snow depth greatly affected nitrogen cycling in balsam fir plots. This NO3 may then be lost from the system, precipitated by a decreased N demand in the plant community compromised by soil freezing damage (Tierney and others 2001; Cleavitt and others 2008).

Snow Depth and Browsing Behavior

The experimental transects provided evidence for browsing behavior changes in moose at HBEF under low snow conditions, especially for balsam fir (Figure 1). We evaluated changes in browsing behavior by comparing plants covered by snow versus plants not covered by snow adjacent to one another. If a moose eats more plants not covered by snow, we considered this evidence for “behavior modification,” that is, the moose takes advantage of the browse available. The behavior changes were not surprising for fir, which is known to be an important browse source for moose in winter (Brandner and others 1990). The low browsing for Viburnum with less snow depth was not expected. Both fir and Viburnum have large stems and buds, characteristics favored by winter browsing moose. Shipley and others (1998) reported that moose in northern Sweden selected browse species that had fewer, larger stems while not selecting species with many small stems. They also reported that moose did not select browse species based on nitrogen, fiber, or phenolic content. We have observed at the HBEF that moose browsing is very visible, where Viburnum density is high; both in terms of stem density (total stem count) and branching density (branch architecture). Our data could suggest that preferential browsing of fir may facilitate increases in Viburnum, a result that cascades through the system by affecting bird habitats at our site (Holmes and others 1996). Martin and Maron (2012) have identified a similar but reverse relationship in the montane region of Arizona, where declines in snow depth have increased browsing pressure by elk, which has led to declines in deciduous trees and songbirds.

Simulated Moose Browsing Effect

Removal of plant biomass through browsing has been argued to have a much larger effect on plant productivity and N uptake than fecal or urine inputs (Pastor and Naiman 1992; Persson and others 2005) or warming temperatures (Cahoon and others 2012; Sjögersten and others 2012; Wang and others 2012). We hypothesized that “browsing” would increase N uptake in plants, that mechanical clipping would result in higher N concentrations, and that more 15N from feces would be recovered in clipped plants. Total N concentrations in plant tissue was not influenced by the clipping treatment and uptake of fecal 15N varied with species (Table 4). In contrast to our hypothesis, clipping reduced recovery of moose fecal 15N in fir and Viburnum. Maple showed an opposite pattern, consistent with our hypothesis. Lovett and Tobiessen (1993) found that simulated insect defoliation did not increase uptake of N by red oak (applied as (NH4)2SO4), rather, availability of N drove plant N uptake. Nitrogen availability may drive differences between the three species in our study, as the response of fir (low N availability) was different than the response of maple (high N availability) (Supplemental Table 1).

Moose, Nitrogen, and Plants

From our understanding of moose influences on N cycling dynamics from past research (Pastor and others 1993), we hypothesized that moose feces would increase N cycling in soils at HBEF and that this N would be available to plants. In contrast, we found no significant changes in N dynamics (Table 1) but we did find that moose fecal N was available to plants under the varying experimental conditions (Table 4).

Our results may differ from past observations for a number of reasons. First, our study differs from the work of Pastor and others (1993) where moose fecal pellets were mixed with soils and incubated in the laboratory. They found increased N and C mineralization rates compared to soils incubated without moose fecal pellets. In our study, the C concentration of the feces added (% C = 50) was similar to that used by Pastor and others (1993) (% C = 52), but the N concentration was lower in the feces used in our study (1.6 vs. 2.5%). These differences are likely a function of seasonal changes in fecal composition (feces were collected in fall in our study versus in spring by Pastor and others 1993). The input of feces with relatively high N concentration and low C/N ratios is likely to stimulate N cycling, as reported by Pastor and others (1993). On the other hand, fecal addition with higher C/N ratios, as added in our experiment, would likely result in an N immobilization sink and may partially explain why we saw no stimulation in N cycling rates. Our results suggest that moose feces produced from a fall diet with a higher C/N could immobilize N, as evidenced by the consistent recovery of 15N in moose feces that remained undecomposed at the end of the experiment (Table 3), irrespective of treatment. Moose feces may thus play an important role in regulating N availability to plants, especially during snow melt, which is a critical period for dissolved N losses from these forested watersheds (Likens and Bormann 1995).

Second, the forest community surrounding our study plots was dominated by maple, birch, beech, ash, and spruce with sub-canopy balsam fir also being present. This community has developed in the absence of moose. The presence and subsequent increase of moose populations has only occurred at HBEF since the beginning of the 1990’s (Bontaites 2003). Our observed lack of response to moose fecal addition in the N cycling processes we measured may also be related to this short-term influence by moose at the HBEF, where there may be stronger control exerted by litter inputs from the current plant community.

Third, many of the studies investigating the role that large ungulate herbivores play in nutrient cycling attribute higher soil nutrient content and microbial biomass in areas affected by herbivores to fecal and urine inputs (Ruess and McNaughton 1987; Pastor and others 1993; Hobbs 1996). In our study, only moose feces from a “winter” browsing diet were added to the experimental plots and hence urine was not studied. Moose feces produced in the winter have lower N and higher C concentrations (Christenson and others 2010) than feces produced during the summer (Pastor and others 1993; Christenson 2007). We explicitly added known amounts of fecal N, and it appears that this source of N does not significantly contribute to changes in the measured soil N pools. Urine from herbivores can contribute a substantial amount of N (~0.619 mg NH4+ and 0.406 mg NO3 per moose urine event at HBEF in winter; Christenson 2007). The contribution of this additional N source should be investigated more fully.

Finally, by using 15N-labeled moose feces, we were able to determine that fecal N was available to all three of the sapling species studied. This was indicated by 15N above background levels in the plant materials analyzed and this availability was influenced by the different experimental treatments (Table 4). If we had only observed total N content in plants (Table 1; no plant N differences) we would have concluded that moose feces is not available to plants and may not contribute to NPP through fecal inputs. In the non-clipped, non-frozen soils, both fir and Viburnum had added 15N recovered in plant tissue, whereas sugar maple did not (Table 4). These differences among tree species suggest that there are species-specific responses to the form of N available from feces.

Species differences in feces–15N uptake were likely related to differences in N availability. Fir had the lowest extractable soil NO3 compared to maple and Viburnum, whereas maple had the highest amount of available NO3 in mineral soils (Supplemental Table 1). Thus, our finding that fir took up more 15N from feces than maple (Table 4) is consistent with fir having greater demand for N than maple. It is interesting to note that approximately 37% of fecal N is in the form of amino acids (Christenson 2007). In soils with low N availability, trees may take up some forms of organic N directly (Hobbie and others 2000; Schimel and Bennett 2004; Finzi and Berthrong 2005), and it is possible that fir which evolved in forests occupied by moose is adapted to exploit this source of N. Wang and others (2012) postulated that organic N sources may have contributed to the lack of plant response in a study, where inorganic N fertilization under warming and/or grazing by sheep did not stimulate aboveground net primary production, counter to their original hypotheses. Results by Wang and others (2012) and our study indicate that all sources of N (inorganic and organic) need to be measured to understand ecosystem responses to changes in the systems. The importance of moose feces as a potential hotspot of organic N availability warrants further investigation.

Soil Freezing Effects on Plant Fecal N Uptake

Several studies have reported greater root mortality and loss of root vitality with soil freeze events, especially in sugar maple (Tierney and others 2001; Cleavitt and others 2008; Comerford and others 2013) and this loss of root functioning could lead to reduced N uptake and declines in plant productivity. We hypothesized that soil freezing induced by snow removal would result in lower % N and less 15N recovered in plants. Our results indicate a mixed response with differences among plant species. Foliar % N was not affected by the snow removal treatment in any species (Table 2). However, balsam fir and Viburnum took up fecal 15N in non-frozen plots, whereas sugar maple did not (Table 3). Soil freezing reduced uptake of fecal 15N by fir, whereas Viburnum uptake was not affected and sugar maple took up only a small amount of fecal N (Table 4). These results could suggest that fir, the most N limited of the three species, may be most susceptible to freeze disturbance of N cycling when compared to sugar maple and Viburnum.

Multiple Stresses: Interactions Between Moose and Snow

The most prominent effects observed in this study occurred in the dual or multiple treatment manipulations. Multiple stresses are often cited as the causal factor in tree death (Kolb and McCormick 1993; Lovett and Mitchell 2004). We hypothesized that a combination of browsing (aboveground pressure) and soil freezing (belowground pressure) would act to decrease plant uptake of N, resulting in greater leaching losses. We found no 15N in maple or Viburnum plants in plots subjected to both above- and belowground pressure (Table 3). Although this result may support this hypothesis, a lack of fecal 15N uptake did not necessarily lead to an increase in soil NO3 export (Figure 2). Plots with maple did have significantly higher NO3 in soil solution samples collected over the growing season, but these higher concentrations were not found during the non-growing season. Viburnum plots subjected to multiple treatments did not have elevated soil solution NO3 concentrations in either season (Figure 2). Fir plots showed the greatest leaching response to multiple treatments (Figure 2), but also had the highest or greatest proportional N uptake from feces (Table 4).

The increased soil solution export of NO3 from fir plots with clipping (simulated moose browse) and soil freezing was associated with a decrease in microbial biomass N (Figure 3), which was lower in the combined or multiple treatment plots compared to the single treatment plots in the forest floor, while the reverse trend was observed in the mineral soils (Figure 3). Soil microorganisms are strong N sinks for forest ecosystem N (Zak and others 1990; Lovett and Ruesink 1995) and a reduction in this sink caused by multiple treatments in our fir plots could explain increases in both plant uptake and leaching. In an alpine tundra snow removal experiment, Brooks and others (1998) attributed increased export of NO3 during the spring snowmelt to lack of uptake by the soil microbial pool.
https://static-content.springer.com/image/art%3A10.1007%2Fs10021-013-9733-5/MediaObjects/10021_2013_9733_Fig3_HTML.gif
Figure 3

Microbial biomass carbon (C) and nitrogen (N) in mineral soils in plots with feces added, clipping, and snow shoveling (multiple treatment plots) sampled in May 2004 and 2005. Different letters indicate significant differences (P < 0.05) between the species for each variable measured.

Conclusions

Understanding complex interactions among organisms with changing environmental conditions is a major challenge for ecologists. This work highlights the complexity of interactions through measures of biogeochemical and plant response. We found that N cycling was only modestly affected by individual stresses, but the combination of aboveground (clipping) and belowground (soil freezing) pressure increased leaching losses of N when moose feces were added to the plots. Moreover, these results varied among plots dominated by different species. Single-factor experimentation would not be able to reveal the complex interaction among stressors to this system. Although challenging to perform, multi-factor experiments are needed to capture the complexity of ecosystem response to multiple, simultaneous environmental changes. Results from our study as well as others that directly address multiple stresses, need to be incorporated into models of forest ecosystem dynamics so that we can build up a more comprehensive understanding of how these ecosystems are likely to function in a changing environment.

Acknowledgments

We would like to thank Lisa Martel and Jackie Wilson for extensive assistance in both the field and laboratory. We would also like to thank John Pastor and an anonymous reviewer for helpful comments that have significantly improved this paper. This project was funded by National Science Foundation (NSF) through Grant DEB 00-75387 (Ecosystem Studies) and Grant DEB 98-10221 (Long Term Ecological Research). This research was conducted at the HBEF, which is owned and operated by the Northeastern Research Station, USDA Forest Service. This paper is a contribution to the Hubbard Brook Ecosystem Study.

Supplementary material

10021_2013_9733_MOESM1_ESM.docx (390 kb)
Supplementary material 1 (DOCX 390 kb)

Copyright information

© Springer Science+Business Media New York 2013