Effects of Arctic Shrub Expansion on Biophysical Vs. Biogeochemical Drivers of Litter Decomposition

Ecology, 95(7), 2014, pp. 1861–1875 Ó 2014 by the Ecological Society of America

Effects of arctic shrub expansion on biophysical vs. biogeochemical drivers of litter decomposition

1,4 2 3 JENNIE DEMARCO, MICHELLE C. MACK, AND M. SYNDONIA BRET-HARTE 1Department of Biology, New Mexico State University, Las Cruces, New Mexico 88003 USA 2Department of Biology, University of Florida, Gainesville, Florida 32611 USA 3Institute of Arctic Biology, University of Alaska, Fairbanks, Alaska 99775 USA

Abstract. Climate warming in arctic tundra may shift dominant vegetation from graminoids to deciduous shrubs, whose functional traits could, in turn, alter biotic and abiotic controls over biogeochemical cycling of carbon (C) and nitrogen (N). We investigated whether shrub-induced changes in microclimate have stronger effects on litter decomposition and nutrient release than changes in litter quality and quantity. In arctic tundra near Toolik Lake, Alaska, USA, we incubated a common substrate in a snow-addition experiment to test whether snow accumulation around arctic deciduous shrubs altered the environment enough to increase litter decomposition rates. We compared the influence of litter quality on the rate of litter and N loss by decomposing litter from four different plant functional types in a common site. We used aboveground net primary production values and estimated decay constant (k) values from our decomposition experiments to calculate community-weighted mass loss for each site. Snow addition had no effect on decomposition of the common substrate, and the site with the highest abundance of shrubs had the lowest decomposition rates. Species varied in their decomposition rates, with species from the same functional type not always following similar patterns. Community-weighted mass loss was 1.5 times greater in the high shrub site, and only slightly decreased when adjusted for soil environment, suggesting that litter quality and quantity are the primary drivers of community decomposition. Our findings suggest that on a short time scale, the changes in soil environment associated with snow trapping by shrubs are unlikely to influence litter nutrient turnover enough to drive positive snow–shrub feedbacks. The mechanisms driving shrub expansion are more likely to do with shrub–litter feedbacks, where the higher growth rates and N uptake by shrubs allows them to produce more leaves, resulting in a larger litter N pool and faster internal cycling of nutrients. Key words: Arctic; climate change; deciduous shrubs; decomposition; litter quality; snow manipulation.

INTRODUCTION mycorrhizal fungi (Clemmensen et al. 2006). In addi- Temperatures in the Arctic region are warming at tion, more woody stems associated with shrubs could twice the rate of the rest of the globe (Hassol 2004, mean more C stored in coarse woody debris and stems, Serreze and Francis 2006, Kaufman et al. 2009, Screen which decompose slowly (Hobbie 1996). Taken togeth- and Simmonds 2010), altering the structure and function er, these effects of shrub expansion have been identified of ecosystems (Chapin et al. 1995, Chapin and Shaver as a negative carbon cycling feedback to climate 1996, Shaver et al. 2000, Mack et al. 2004) by causing a warming. Biophysical feedbacks to regional climate, shift in species composition from one previously however, may be in the opposite direction. Shrublands dominated by graminoids to an increase in the have a lower albedo than tundra, which can lead to an abundance and expansion of shrubs (Jia and Epstein increase in absorbed solar radiation during the snow- 2003, Stowe et al. 2004, Tape et al. 2006, Jia et al. 2009, free period, resulting in a positive feedback to regional Forbes et al. 2010, Elmendorf et al. 2012). Increasing warming (Chapin et al. 2005). In addition, deciduous shrub dominance will increase plant productivity and shrubs in fertilized tundra cycled C and N faster than in unfertilized tundra, resulting in a net loss of deep soil C biomass, resulting in more uptake of atmospheric CO2 (Cahoon et al. 2012) and the storage of more C (Mack et al. 2004). Soils in naturally occurring shrub aboveground in woody tissue or belowground in tundra similarly store less C than those in graminoid rhizomes (Shaver and Chapin 1991), roots, and ecto- tundra (Ping et al. 1997). Thus, net positive feedbacks to warming are possible if increased shrub cover alters ecosystem structure and function so that soil C stocks Manuscript received 3 December 2013; accepted 17 decrease. Since the Arctic stores 20–30% of the total December 2013; final version received 8 January 2014. Corresponding Editor: R. W. Ruess. amount of terrestrial soil-bound C (McGuire et al. 2009) 4 E-mail: [email protected] and woody vegetation is predicted to increase by as 1861 1862 JENNIE DEMARCO ET AL. Ecology, Vol. 95, No. 7 much as 52% by 2050 (Pearson et al. 2013), it is inputs of slowly decomposing litter (Hobbie 1992, important to better understand the mechanisms behind Buckeridge et al. 2010). By slowing nutrient turnover these potential feedbacks to climate change. and decreasing plant-available N, this litter could drive a The mechanisms that drive shrub expansion are not negative plant–soil feedback that would slow shrub well understood. It has been hypothesized that shrubs expansion. may enhance their dominance and growth through These two competing mechanisms lead to the question increased nutrient availability associated with shrub- of whether changes in microclimate or changes in litter induced changes in the abiotic environment. The snow– quality and quantity have stronger effects on litter shrub hypothesis suggests that taller and more abundant decomposition and nutrient release. This question has shrubs accumulated greater snow depth due to greater been studied in mesic (Hector et al. 2000, Knops et al. retention of snowfall (e.g., less snow lost to wind events) 2001, Scherer-Lorenzen 2008) and dry (McLaren and and trapping of wind-distributed snow than tundra Turkington 2010) grasslands and temperate forests areas with fewer shrubs (Sturm et al. 2001, 2005, (Hobbie et al. 2006, Vivanco and Austin 2008), but Pomeroy et al. 2006). This deeper snow cover insulates has not been as well studied in Arctic tundra systems, the soil, maintaining warmer temperatures through the particularly in the context of shrub expansion. Will winter. By altering the abiotic environment through positive or negative plant–soil feedbacks prevail as trapping snow and maintaining warmer soil tempera- shrubs expand in the Arctic? Although there is much tures, shrubs can influence the biogeochemical processes evidence to suggest that shrubs can influence their that drive nutrient cycling. Decomposition of litter is the environment to alter key biogeochemical processes that major pathway by which nutrients are recycled and control plant nutrient supply, there have been no made available for plant uptake in these nutrient-limited published studies to date that have directly tested the systems. Changes in the environment can influence rates effect of added snow (at the depth that would be trapped of litter decomposition. However, it has not been by shrubs) or increased shrub cover on litter decompo- directly tested whether the amount of snow trapped by sition. In addition, within the Alaskan Arctic, most deciduous shrubs alters the environment enough to decomposition studies have largely occurred in grami- stimulate litter decomposition and increase litter nutri- noid-dominated moist acidic or nonacidic tundra. We ent release. know relatively little about how the environment of Previous studies have shown that snow acts as an shrub tundra may influence litter decomposition. insulator that can increase soil temperature (Brooks et The goal of our study was to understand the relative al. 1996, 1998, Grogan and Jonasson 2003, Schimel et al. importance of mechanisms through which arctic decid- 2004, Wahren et al. 2005) and the availability of water uous shrubs affect litter decomposition and N dynamics. to soil microorganisms (Romanovsky and Osterkamp Our objectives were threefold: (1) to test whether snow 2000, Mikan et al. 2002), potentially regulating the rate addition, at a rate realistic for increasing shrub at which microbes and fungi can break down litter over abundance, altered the environment for decomposition the winter. Indeed, litter decomposition has been found enough to stimulate rates of litter mass and N loss, (2) to to occur in the winter and under snow (Stark 1972, compare how litter C quality and the relative availability Hobbie and Chapin 1996, McLaren and Turkington of C to N influenced the rate of litter mass and N loss, 2010, Saccone et al. 2013), and to be higher in areas that and (3) to better understand how the changes in plant have deeper snow cover (Baptist et al. 2010, Saccone et species composition associated with the shift from al. 2013). If faster turnover of litter N results in an graminoid- to shrub-dominated tundra influence com- increase in plant-available N, then snow accumulation munity decomposition. We hypothesized that across by shrubs could indirectly influence N availability by three sites that represented a gradient in shrub maintaining warmer soil temperatures in fall and winter, abundance and height, (1) the addition of snow would and allowing a longer window for microbial breakdown slow temperature decline in the winter and lead to faster of litter substrates, increased N release, and higher rates decomposition and net N release from litter, (2) of N supply to plants, resulting in a positive plant–soil decomposition and net N release would covary posi- feedback that promotes further shrub expansion. tively with lignin : N ratios, and (3) community-weighted Shrubs may also have effects on decomposition and N mass loss would be lowest in the shrub-dominated sites, release that are independent and opposite of their effects because of the greater abundance of woody litter. on winter soil temperatures. For example, in moist To test our hypotheses, we measured litter quality, acidic tundra, the deciduous shrub Betula nana allocates quantity, decomposition, and net N release across three 79% of its total annual aboveground biomass to new plant communities that represented natural variation in and old stems (Shaver et al. 2001) that decompose three shrub abundance across the landscape (DeMarco et al. times slower than leaves and one to eight times slower 2011). In each community, we manipulated snow depth than leaves and stems from graminoids and evergreen using snow fences. To test for the effects of site and shrubs (Hobbie 1996). A compositional shift to more snow on decomposition and net N release, we decom- deciduous shrub dominance may thus alter nutrient posed a common substrate in all sites and treatments. turnover through biotic controls by a shift toward larger We decomposed litters from multiple plant functional July 2014 SHRUB EXPANSION AND DECOMPOSITION 1863 groups in a common environment, hereafter referred to tundra where the deciduous shrubs are of intermediate as a common garden, to control for differences in size. The vegetation consists of graminoids (primarily C. microclimate and directly test the effect of litter quality bigelowii), deciduous shrubs (B. nana, V. uliginosum, S. on decomposition. Finally, we combined these studies pulchra, and S. richardsonii), and mosses (H. splendens with aboveground net primary production estimates to and Dicranum spp.). Our high shrub site was located in calculate community-weighted mass loss rates. riparian tundra dominated by tall deciduous shrubs and has predominantly deciduous shrubs (B. nana, S. METHODS pulchra,andsomePotentilla fruticosa), with some Study area evergreen or wintergreen shrubs (V. vitis-idaea and All sites are located near Toolik Field Station at the Linnaea borealis), forbs (Polygonum bistorta, Petasites Arctic Long Term Ecological Research (LTER) site frigidus, Stellaria longipes, Valeriana capitata,and Artemisia alaskana), graminoids (Poa arctica, C. bigelo- (688380 N, 1498380 W, elevation 760 m) in the foothills wii, and Calamagrostis canadensis), and mosses (Sphag- region on the North Slope of the Brooks Range, Alaska, num spp. and H. splendens). USA. This area is a younger landscape glaciated during the late Pleistocene. It includes large areas of the Itkillik Snow manipulation I (deglaciated ;60 000 yr) and Itkillik II (deglaciated To determine the influence of increased snow depth ;10 000 yr) glacial drifts (Hamilton 1986). The entire on litter decomposition, snow fences that represented foothills region of the Brooks Range is treeless and maximum regional shrub height (1.5 m high) were set up underlain by continuous permafrost, 250–300 m thick in the fall of 2005 at all three sites to manipulate snow (Osterkamp and Payne 1981). Mean annual air temper- depth (DeMarco et al. 2011). Two treatments, control ature is around 108C, with average summer tempera- À (ambient snow) and drift (manipulated snow), were set tures from 7–128C. Annual precipitation is 318 mm, with up at each site. For all sites, subplots on the drift side of 43% falling as snow (data available online).5 Average the fences were located in the zone of maximum snow snow depth is 50 cm, although snow distribution can be accumulation, which was relatively uniform. Within variable due to redistribution by wind. Snowmelt occurs each treatment, 18 2 3 10 m plots, with 1-m buffer strips in early May. between, were established. For this study, six plots per In the fall of 2005, three sites were selected for the treatment (n 6) were randomly assigned to measure snow manipulation experiment that varied primarily in litter decomposition.¼ Remaining plots were used for deciduous shrub abundance, hereafter referred to as low, other experiments not described here. medium, and high shrub sites. These sites are described Soil temperature at 5 cm within the organic layer was in detail in DeMarco et al. (2011). In short, sites were measured continuously (1–3 h intervals) from July 2006– chosen to have similar state factors (climate, parent May 2009 in each study plot (n 3–4 plots within each material, time since deglaciation) but varied in the treatment and site) using I-button¼ temperature data abundance of deciduous shrubs (Jenny 1994). The same loggers (IButtonLink, East Troy, Wisconsin, USA). species of deciduous shrubs (Betula nana and Salix Mean daily soil temperatures were calculated for all pulchra)arefoundatallthreesites(exceptS. plots within each treatment and site for each year. Mean richardsonii, which is found only at the medium shrub growing season and winter soil temperatures were site). However, percent cover of deciduous shrubs calculated from the mean daily temperatures from each increases from 15% to 94%, and their canopy height plot within each treatment and site. I-buttons were not increases from 4 cm to 50 cm across sites. All sites are always installed or removed on the same day; therefore within 1 km of each other, and have similar parent we only analyzed data from days in which we had data material, time since last glaciation (Itkillik I, 60 000 yrs), for all sites and treatments. The growing season included and regional climate, although microclimates vary measurements taken from 1 July to 1 August of that across sites due to differences in slope and aspect. year and included years 2006, 2007, and 2008. Winter Elevation changes from about 764 m at the low shrub growing season includes measurements taken from 1 site to 741 m at the medium and high shrub sites. September through 1 May of the following year and Our low shrub site was located in moist acidic tussock includes winters from 2006–2007, 2007–2008, and 2008– tundra, where the vegetation consists of approximately 2009. equal biomass of graminoids (Eriophorum vaginatum The snow fences produced snow packs in treatment and Carex bigelowii), dwarf deciduous shrubs (B. nana, plots that were, on average, 87, 96, and 104 cm deeper Vaccinium uliginosum, and S. pulchra), evergreen shrubs than ambient snow depth for the low, medium, and high (Ledum palustre ssp. decumbens and V. vitis-idea), and shrub sites, respectively. Snow addition increased mosses (Hylocomium splendens, Aulacomnium turgidum, average winter soil temperatures by 38C and summer Dicranum spp., and Sphagnum spp.; Shaver and Chapin soil temperatures by 28C in the high shrub site; the 1991). Our medium shrub site was located in riparian medium and low shrub sites showed similar trends, although the differences between treatments were 5 http://ecosystems.mbl.edu/ARC smaller in magnitude (DeMarco et al. 2011). 1864 JENNIE DEMARCO ET AL. Ecology, Vol. 95, No. 7

Common substrate experiment site was in the control plots in a previously set up To directly test the effect of microclimate and snow experiment in moist acidic tundra (referred to as species addition on litter decomposition rates, we incubated the removal; Bret-Harte et al. 2008). All litter was collected senesced leaves from a common substrate, B. neo- and processed using the same methods as described in the common substrate experiment, except that only 1.6- alaskana (Sarg.), in the ambient and snow-manipulated mm mesh bags, 4 3 8 cm in area, were used, and leaf plots across all three sites. Senesced leaves were collected samples were replicated six times, while stem litter was from trees growing near Fairbanks, Alaska, USA. replicated three times. For the litter collected at the Leaves were still attached to the trees, but the petiole species removal site, litter bags were installed in July of had already started to abscise. This common substrate 2003 and removed in July of 2004, 2005, 2007, and 2008. was used because the large leaf size and relative Litter collected from the other sites was installed in the abundance allowed us to collect enough material for field as litter bags in early June of 2006 and replicate our study. Leaves were air dried, well mixed, and then bags were removed in July of 2007, 2008, and 2009 and subsampled for litter bags. One gram of leaves was sewn processed as described previously. into 2-mm mesh bags, 8 3 8 cm in size. Litter bags were incubated beneath the live moss and litter layer starting Calculations in early June of 2006. The moss and litter in this system The exponential decay constant, k, was determined by are well mixed, so bags were inserted in this layer. Four assuming a single exponential decay model (Olson identical bags were strung together for four separate kt 1963): M M eÀ , where M is litter mass at time t, annual harvests. Bags were placed in six treatment plots t 0 t and M is¼ initial mass. The slope of the regression of in each site, with three sub-replicates within each plot. 0 proportion of initial mass remaining against time was Bags were removed in July of 2007, 2008, and 2009, and used to determine the decay constant for each substrate were kept frozen until they could be processed. at each site. At time of processing, bags were thawed and then gently rinsed with deionized (DI) water to remove soil Initial litter quality and loose litter attached to the outside of the bag. All A subsample of each leaf and stem collection was original leaf litter was removed, dried at 458C for a analyzed for percentage of C, percentage of N, and C minimum of 48 h, and weighed. To determine the quality to determine the quality of the litter substrates percentage of C and N of the litter, samples were ground prior to decomposition. Percentage of C and N was to a fine powder on a Wiley mill, with a no. 40 mesh determined from samples that had been ground to a fine screen, and then analyzed using an ECS 4010 elemental powder on a Wiley mill, with a no. 40 mesh screen, and analyzer (Costech Analytical, Valencia, California, then analyzed using an ECS 4010 elemental analyzer. C USA). Percentage of initial mass remaining was quality measurements were carried out on an ANKOM calculated by dividing the incubated mass by the initial fiber analyzer (Ankom Technology, Macedon, New mass and multiplying by 100. Percent of initial C (ICR) York, USA) and included determination of (1) soluble and initial N remaining (INR) was calculated by the cell contents (carbohydrates, lipids, pectin, starch, and following equations: soluble protein), (2) hemicelluloses plus bound proteins, (3) cellulose, and (4) lignin plus other recalcitrants t1mass 3 t1Carbon ICR ð Þ 3 100; (Ryan et al. 1990). ¼ t0mass 3 t0Carbon ð Þ Community-weighted mass loss t1mass 3 t1Nitrogen INR ð Þ 3 100: To separate out the effect of changes in species ¼ t0mass 3 t0Nitrogen ð Þ composition vs. changes in the microenvironment associated with shrub expansion on community level Common garden experiment decomposition, we used measured aboveground net To compare differences in litter decomposition rates primary productivity (ANPP) and k values to calculate among species, we incubated senesced leaf litter from 10 a community-level mass loss for each of the three shrub vascular plant and three moss species, and stem litter sites (Appendix A: Table A1). We followed the from four shrub species collected across eight sites procedure outlined in Hobbie and Gough (2004). We located in the Arctic Foothills region on the North Slope harvested biomass in July of 2007 from the control of the Brooks Range (Table 1). Three of the sites were treatments of the three snow fence sites (DeMarco et al. 2 1 the control plots at the low, medium, and high shrub 2011). Total ANPP (g mÀ yrÀ )perplot(n 8 Á Á sites. Four sites were located in alder-dominated (Alnus replicates per site) was calculated by summing¼ new viridis spp. fruticosa) tall deciduous shrub tundra; two apical biomass (g/m2) for that year, i.e., of each tissue near the Sagavanirktok River and two along the Dalton type from each species found within that plot. Our Highway, ;32 km north of Toolik Field Station. Alders ANPP calculations were only for vascular plants and are one of the deciduous shrubs that have been thus do not include mosses, lichens, or belowground documented as expanding in many arctic regions. One parts such as rhizomes or roots. We estimated the July 2014 SHRUB EXPANSION AND DECOMPOSITION 1865

TABLE 1. List of all the species decomposed in the common garden site.

Group Species Graminoids Carex bigelowii, Eriophorum vaginatum Deciduous shrubs Alnus crispa viridis spp. fruiticosa, Betula nana, Betula neoalaskana (Sarg.), Rubus chamaemorus, Salix pulchra, Vaccinium uliginosum Evergreen shrubs Ledum decumbens, Vaccinium vitis-idaea Mosses Aulacomnium turgidum, Hylocomium splendens, Sphagnum spp. contribution of secondary stems to ANPP by multiply- most decomposition studies do not conduct full factorial ing the ANPP of species likely to produce woody tissue incubations where all species are incubated in all by a proportion determined by Bret-Harte et al. (2002). environments. In moist acidic and moist nonacidic These were 0.158, 0.181, and 0.079 for B. nana, S. tundra, B. nana stem decomposition followed the pulchra, and L. palustre ssp. decumbens, respectively. We opposite pattern between the two sites compared to followed the methods outlined in Hobbie and Gough the other species decomposed (Hobbie and Gough (2004) and assumed that C. tetragona and V. uligonosum 2004). resembled L. palustre in their proportional secondary Species decomposed in our common garden contrib- growth, and S. reticulate, S. glauca, and S. richardsonii uted 93% of ANPP. For the species that contributed the all resembled S. pulchra in their proportional secondary remaining 7% of ANPP, we used k values from other growth. We also assumed that Andromeda polifolia, published studies within the region or substituted k Arctostaphylos alpina, Dryas integrifolia, Empetrum values for species with similar growth forms. For all nigrum, Rubus chamaemorus, V. vitis-idaea, and Linnaea forbs we used the decay constant for P. bistorta reported borealis had negligible secondary growth. We assumed by Hobbie and Gough (2004), which was decomposed in inflorescences would have similar k values as foliar litter moist acidic tundra not far from our common garden from the same species and thus used foliar k values for site. For Calamagrostis spp., Poa arctica, and Juncus inflorescences. This method assumes all new production spp., we used an average of the k values for Carex spp. that year ends up as litter. For deciduous shrub leaves, and E. vaginatum (k 0.18) from our common garden. ¼ this is a valid assumption, as they lose all of their leaves For Empetrum spp., L. borealis, and Cassiope spp. at the end of the growing season. Evergreen shrubs and leaves, we used an average of k values from Ledum spp. graminoids can retain leaves through the winter, and and V. vitis-idaea from our common garden. For stems thus we may be overestimating litter production for of these three species, we used an average of k values these species. from B. nana and S. pulchra stems (k 0.08). ¼ The mass loss for each plant species and their parts Statistical analysis (inflorescences, leaves, and stems) was calculated by multiplying the species’ ANPP by that species’ decay To test our hypotheses about the effects of site and constant, k, from the common garden experiment to snow depth on decomposition of the common substrate, produce an ANPP-weighted mass loss for each species. we used two-way ANOVA (JMP 7.0, SAS Institute, ANPP-weighted mass loss values were summed across Cary, North Carolina, USA). Relationships between k all species in a plot to produce community-weighted and initial litter quality indices for 10 vascular plant mass loss values (see Appendix A: Table A5 for full list species decomposed in the common garden were tested of ANPP and k values used for each species present). using single regression analysis with k as the dependent To determine how differences in site environments variable and the litter quality indices of interest as the contributed to community mass loss, we adjusted the independent variable. The effect of changes in litter community-weighted k values for site-specific differences quality/quantity on community mass loss and the effect in decomposition rates. We divided site mean k values of changes in environment on community mass loss were for the common substrate, B. neoalaskana, by the mean tested using two separate one-way ANOVAs with site as k value for B. neoalaskana that was decomposed in the the main effect. Tukey’s HSD test was used as a post hoc common garden (k 0.256). The proportional difference test when ANOVAs were significant at P , 0.05. Data between the k value¼ from the common garden and the were tested for normality (Shapiro-Wilks), and ln- shrub sites was multiplied by the species-specific k values transformed when necessary to achieve homogeneity of to correct for differences in decomposition found across variance. When homogeneity of variance could not be the shrub sites. The adjusted species-specific k values achieved, data was analyzed using a Kruskal-Wallis were summed to obtain a decay k constant for the entire nonparametric test. community. This method assumes that all species and RESULTS tissues respond the same way as B. neoalaskana when decomposed in different environments, and does not Effect of added snow on soil temperature take into account any site by species interactions. Snow addition significantly increased winter soil Evidence of site by species interactions are rare, because temperatures by an average of 1–48C across all three 1866 JENNIE DEMARCO ET AL. Ecology, Vol. 95, No. 7

TABLE 2. Three-way ANOVA comparing differences in soil temperature between vegetation type, treatment, and year.

Growing season Winter Source df FPdf FP Vegetation type 2 12.0 ,0.0001 2 26.2 ,0.0001 Treatment 1 6.0 0.02 1 24.5 ,0.0001 Year 2 18.6 ,0.0001 2 28.4 ,0.0001 Vegetation type 3 Treatment 2 0.7 0.51 2 3.3 0.05 Vegetation type 3 Year 4 1.2 0.30 4 2.4 0.07 Treatment 3 Year 2 0.1 0.88 2 0.1 0.90 Vegetation type 3 Treatment 3 Year 4 0.2 0.95 4 1.4 0.25 Note: Error degrees of freedom are 49 for growing season and 35 for winter. sites (Table 2, Fig. 1). There were significant differences for greater mass loss in the ambient treatment at the low in soil temperature across sites with snow addition and shrub site, but greater mass loss occurred in the snow- across years, but there were no significant interactions addition treatment at the medium and high shrub sites among the main effects. Soil temperatures were highly (Appendix B: Table B1). Decomposition rates (k), N variable across years. In general, the low shrub site had remaining, or C remaining were also not significantly both lower summer and winter soil temperatures different under snow addition, although there was a compared to the medium and high shrub sites. trend for higher k rates and less N and C remaining in the snow-addition treatment (Figs. 2 and 3). However, Common substrate experiment initial N remaining showed a significant site by After three years, snow addition did not significantly treatment effect. When the main effect of treatment alter mass remaining of the common substrate (two-way was tested for each site separately, only the low shrub ANOVA; treatment; F1,36 1.4, P 0.24 [Fig. 2]) but site showed a significant effect of treatment on N there was a significant difference¼ across¼ sites (site; F remaining after three years (treatment; F 122.2, P 2,36 ¼ 1,11 ¼ ¼ 13.6, P , 0.001) and a significant interaction between 0.03). In contrast to the effect of snow addition, site and treatment (site 3 treatment; F2,36 3.5, P decomposition rates and N and C remaining varied 0.04). When the effects of treatment were tested¼ for each¼ significantly across sites (Fig. 3). The litter decay rate, k, site independently, treatment was not significant at the was highest at the low shrub site, losing 10% and 6% low shrub site (one-way ANOVA; treatment; F1,11 1.9, more mass and 8% and 6% more C than in the medium P 0.20) and was only marginally significant at¼ the and high shrub sites, respectively. Initial N remaining ¼ medium (treatment; F1,11 3.4, P 0.10) and high shrub followed a different pattern, with mineralization of litter sites (treatment; F 3.6,¼ P 0.09).¼ There was a trend N occurring at the low shrub site only, but immobili- 1,12 ¼ ¼

FIG. 1. Soil temperature at 0–5 cm soil depth (means 6 SE) during the growing season and winter over three years (2006, 2007, and 2008), with two snow treatments (ambient and snow addition) and across three shrub sites (low, medium, and high). Data points that share a lowercase letter are not signiﬁcantly different at the P , 0.05 level. July 2014 SHRUB EXPANSION AND DECOMPOSITION 1867

FIG. 2. Initial mass, C, N, and proportion of C:N (means 6 SE) remaining of the common substrate (Betula neoalaskana) from litter bags incubated over three years in the ambient and snow-addition treatments at the low, medium, and high shrub sites. zation occurring at both the medium and high shrub hemicellulose in their leaves, compared to evergreen and sites. The proportion of initial C:N remaining decreased deciduous shrub leaves (species: F 90.1, P , 6,32 ¼ with an increase in shrub abundance (Fig. 3). 0.0001). Graminoids also had the highest percentage of cellulose in their leaves, two to four times more than in Common garden experiment evergreen and deciduous shrubs. B. nana had the least After three years of incubation, neither site of origin percentage of cellulose in its leaves compared to all six of nor species had an effect on the percentage of initial the other species (species: F 157.3, P , 0.001). B. 6,32 ¼ mass remaining or decay rate of leaf and stem litter from nana also had the highest percentage of lignin, followed B. nana and S. pulchra, despite significant differences in by evergreen shrubs and the deciduous shrubs, R. their initial litter quality (Appendix C: Tables C1 and chamaemorus and V. uliginosum. Graminoids had the C2). least amount of lignin in their leaves (species: v2 27.4, ¼ Initial leaf litter quality significantly differed across df 6, P , 0.0001). ¼ species for all indices measured, although species within After five years of decay, R. chamaemorus leaf litter the same growth form were not always similar in their lost 1.5–6 times more mass than leaf litter from the other initial litter quality (Table 3). Deciduous shrubs had up six species of vascular plants and three species of mosses to two times more N in their leaves than evergreen collected at the same moist acidic tundra site and shrubs, graminoids, and mosses (species: v2 38.3, df decomposed in the same common garden (Fig. 4; IMR ¼ ¼ 2 7, P , 0.0001). Evergreen shrubs had the highest species: v 36.5, df 7, P , 0.0001). Species within the ¼ ¼ percentage of C in their leaves, followed by deciduous same functional group did not always follow the same shrubs (except R. chamaemorus), graminoids, mosses, pattern in their rates of decomposition (Fig. 4). Leaf and R. chamaemorus (species: v2 40.0, df 7, P , litter from the deciduous shrubs B. nana and V. ¼ ¼ 0.0001). Evergreen shrubs, mosses, and the graminoid E. uliginosum had decay constants that were similar to that vaginatum had high C:N ratios in their leaves, while the of the evergreen shrub L. decumbens, and the graminoid deciduous shrubs R. chamaemorus and V. uliginosum E. vaginatum had a higher decay constant than those of had the lowest C:N ratios (species: F 72.2, P , the evergreen shrub V. vitis-idaea and the graminoid C. 9,53 ¼ 0.0001). The evergreen shrubs and the deciduous shrub bigelowii.Mosseshadthelowestdecayconstants B. nana had lignin : N ratios that were two to four times compared to the other seven vascular plant species higher than the deciduous shrubs, R. chamaemorus and decomposed in our experiment. V. uliginosum, and the graminoids (species: v2 27.1, df In a comparison with all 10 vascular species decom- ¼ 6, P , 0.0001). Graminoids had 1.5–4 times more posed in our common garden, leaf litter decay rates were ¼ 1868 JENNIE DEMARCO ET AL. Ecology, Vol. 95, No. 7

FIG. 3. Decay constants, initial C and N remaining, and the proportion of initial C:N remaining for the common substrate (means 6 SE), B. neoalaskana, decomposed across three sites (low, medium, and high) and two snow fence treatments (ambient and snow addition) and calculated for the entire 3-yr decomposition time period. Results of two-way ANOVAs are displayed for each variable with site and treatment as the main effects and site by treatment (S 3 T ) interaction. In each panel, data points that share a lowercase letter are not signiﬁcantly different at the P , 0.05 level.

TABLE 3. Initial litter quality for senesced leaves of seven species of vascular plants collected from plots in a moist acidic tundra community and incubated in a common site.

Species N (%)C(%) C:N Graminoids Carex bigelowii 1.34 6 0.04 42.39 6 0.20 31.91c 6 1.14 Eriophorum vaginatum 0.87 6 0.04 42.15 6 0.13 8.90b 6 2.11 Deciduous shrub Betula nana 1.67 6 0.06 44.59 6 0.15 26.94c 6 0.93 Rubus chamaemorus 1.90 6 0.04 40.53 6 0.57 21.32d 6 0.38 Vaccinium uliginosum 1.74 6 0.05 43.63 6 0.10 25.15c,d 6 0.79 Evergreen shrub Ledum decumbens 1.03 6 0.03 47.83 6 0.30 46.64b 6 1.25 Vaccinium vitis-idaea 0.81 6 0.03 45.15 6 0.16 56.05a 6 2.31 Mosses Aulacomnium turgidum 0.81 6 0.07 37.36 6 3.32 46.10b 6 1.01 Hylocomium splendens 0.87 6 0.16 40.71 6 0.23 46.61b 6 0.70 Sphagnum spp. 0.88 6 0.02 38.96 6 0.17 44.40b 6 0.71 Notes: Data are presented as means 6 SE. Different letters within the same foliar trait indicate that values are signiﬁcantly different at the P , 0.05 level from post-hoc tests after running one-way ANOVAs comparing each variable across species (n 6 individual replicates). Cell soluble refers to soluble cell contents (carbohydrates, lipids, pectin, starch, and soluble protein). ¼ July 2014 SHRUB EXPANSION AND DECOMPOSITION 1869

FIG. 4. Leaf litter decay constants (k; means 6 SE) from (a) leaf litter collected at the moist acidic tundra species removal site and decomposed for 5 yr in a common site, and (b) leaf and stem litter of different deciduous shrubs species collected across seven different sites and decomposed for 3 yr in a common site. B. neoalaskana, the litter used in the common substrate experiment, was also incubated in the common garden and is included in (b). Species are presented by functional group; deciduous shrubs (DESH), evergreen shrubs (EVSH), graminoids (GRAM), and mosses. Full genus and species names are given in Table 1. Abbreviations within the parentheses refer to the site where the litter was originally collected: Dalton Highway site 1 (D1), Dalton Highway site 2 (D2), Sagavanirktok River site 1 (S1), Sagavanirktok River site 2 (S2), low shrub site (L), medium shrub site (M), and high shrub site (H). ANOVA results are from comparisons across species for samples collected at the species removal site. In panel (a), data points that share a lowercase letter are not signiﬁcantly different at the P , 0.05 level. only weakly related to some of the litter quality indices 0.04 [Appendix D: Fig. D1]). For stem litter, the measured. The percentages of C and cell soluble percentage of cellulose was the only litter quality index contents were positively correlated with decay rates (C; that correlated positively with decay rates, explaining r2 0.23, n 21, P 0.03, cell soluble contents; r2 55% of the variation in stem decay rate (Appendix D: ¼ ¼ ¼ ¼ 0.23, n 21, P 0.03, cellulose; r2 0.23, n 21, P Fig. D2; r2 0.55, n 9, P 0.01). ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼

TABLE 3. Extended.

Cell soluble (%) Hemicellulose (%) Cellulose (%) Lignin (%) Lignin : N

41.87b 6 2.61 28.39a 6 1.25 24.47b 6 0.90 4.81 6 0.93 3.60 6 0.64 35.79b 6 2.38 30.82a 6 0.66 27.16a 6 0.46 4.30 6 0.37 4.91 6 0.43

62.76a 6 0.75 10.15c,d 6 0.47 7.39e 6 0.47 19.44 6 0.31 11.72 6 0.29 64.98a 6 1.54 17.87b 6 1.12 9.82d,e 6 0.58 6.78 6 0.57 3.54 6 0.22 69.71a 6 0.11 9.49c,d 6 0.23 10.52c,d,e 6 0.09 8.35 6 0.36 4.83 6 0.33

63.29a 6 0.99 8.14d 6 0.41 12.14c,d 6 0.30 16.08 6 0.36 15.66 6 0.38 60.93a 6 0.75 13.29c 6 0.85 13.22c 6 0.27 11.19 6 1.45 14.09 6 2.26 1870 JENNIE DEMARCO ET AL. Ecology, Vol. 95, No. 7

DISCUSSION Microenvironment controls over litter decomposition Surprisingly, we were unable to detect any effect of added snow on decomposition of the common litter substrate after three years of incubation, despite a 28C increase in soil temperature during the growing season and a 48C increase during the winter. Walker et al. (1999) also found no effect of deeper snow on litter decomposition after two years of decomposing Betula nana leaf litter under ambient and up to 3 m of added snow in tussock tundra near Toolik Lake. In contrast, in an alpine tundra community, Baptist et al. (2010) found a trend for greater litter mass loss in late snowmelt sites, presumably due to warmer soil temperatures in the spring prior to snowmelt. Using the same species of litter as our common litter experiment, Hobbie and Chapin (1996) found differences in litter mass loss between arctic tundra microsites whose summer soil temperatures differed by 48C, with greater mass loss occurring in the warmer microsites. In addition, litter mass loss in lab incubations that included warming treatments of either 28C or 68C above the ambient growing season temperatures showed an increase in mass loss with increased temperature (Hobbie 1996, Jonasson et al. 2004). Two out of the three studies had temperature differences that were twice as high as ours, which may explain why they found significant differences in litter mass loss with change in temperature, while our study did not. At these sites, temperature may not be the main control over FIG. 5. (a) Aboveground net primary productivity (ANPP; decomposition. Site-specific differences in soil moisture, means 6 SE) across sites, and (b) mass loss (means 6 SE) nutrient availability, litter quality and quantity, and the across sites with changes in litter quality and quantity alone and with changes in litter quality, litter quantity, and soil decomposer community may play a larger role in microenvironment. In each panel, data points that share a decomposition in these arctic and alpine communities. letter are not significantly different at the P , 0.05 level, with When comparing decomposition of our common uppercase vs. lowercase letters in panel (b) indicating significant substrate across our three sites, the common litter differences across sites within the same variable. decomposed faster in the low shrub site compared to the medium and high shrub sites, even though ambient soil temperatures at the medium and high shrubs sites Community-weighted mass loss were actually warmer than at the low shrub site both during the growing season and over the winter. Soil Community-weighted mass loss that took into ac- temperature differences across these sites are of the same count differences in litter quality and quantity across magnitude as the differences in soil temperature we saw sites was greatest in the high shrub site compared to the when we added snow. Since we did not see strong medium and low shrub sites (site: F 3.5, P 0.05). 2,30 ¼ ¼ differences in decomposition when we added snow and This was primarily driven by the higher overall ANPP elevated soil temperatures, this suggests that other found at the high shrub sites compared to the medium factors such as moisture, soil nutrients, or the decom- and low shrub sites (site: F 7.0, P , 0.01), because poser communities may be more important than small 2,30 ¼ the proportion that ANPP contributed to community (,48C) changes in temperature for driving decomposi- mass loss (mass loss/ANPP) was similar across sites (low tion at our sites. Soil moisture was not measured over 0.16, medium 0.15, and high 0.15). Community- the 3-yr incubation period, but measurements in June of ¼ ¼ ¼ 2006 showed no difference in soil moisture across sites weighted mass loss that took into account both (DeMarco et al. 2011) and could not explain the differences in litter quality and quantity and in the differences we found in litter decomposition across the environment was greater in the low and high shrub sites sites. than the medium shrub site (site: F 4.4, P 0.03; 2,30 ¼ ¼ Previous research from these sites has shown that Fig. 5) suggesting that different mechanisms may be percentage of N of the top 10 cm of soil at the medium driving these differences at each of the sites. and high shrub sites is twice as high as the percentage of July 2014 SHRUB EXPANSION AND DECOMPOSITION 1871

FIG. 6. Soil N (means 6 SE) for each site vs. decay constant (k), initial C remaining, initial N remaining, and the proportion of initial C to N remaining of the common substrate, B. neoalaskana, decomposed at each site over a 3-yr period.

N at the low shrub site at the same soil depth (DeMarco In a meta-analysis of 24 litter decomposition studies et al. 2011) and is highly positively correlated with litter in which external N was experimentally manipulated, initial N remaining (%) and initial C:N remaining (%), Knorr et al. (2005) found that external N availability but negatively correlated with decay constants (Fig. 6). and litter quality interact to influence decay rates, with This suggests that soil nutrients may play an important N additions stimulating decomposition of high-quality role in controlling litter decomposition and nutrient litters (,10% lignin content), while inhibiting decay of release at our sites. Over our 3-yr incubation period, low-quality (.20% lignin content) litters. The ‘‘micro- litter decomposed at the low shrub site mineralized litter bial N mining’’ hypothesis suggests that this pattern N, while litter decomposed at the medium and high occurs because some microbes use labile C to decompose shrub sites immobilized N. Thus, sites with greater soil recalcitrant organic matter in order to acquire N Nhavelowerratesofdecompositionandhigher (Fontaine and Barot 2005, Moorhead and Sinsabaugh retention of N on litter. In our study, sites that had 2006). Therefore, we would expect microbial N mining greater bulk soil N also had higher rates of net N to increase decomposition of low-quality litter when it is mineralization, suggesting greater N availability (De- incubated in soils with low soil N. Our highest Marco et al. 2011). Greater soil N availability has been decomposition rates were seen at the low shrub site found to stimulate (Hobbie 1996, Aerts et al. 2006), where soil N (DeMarco et al. 2011) is low and soil repress (Prescott 1995, Magill and Aber 1998, Aerts et microbial activity is N-limited (Lavoie et al. 2011, Sistla al. 2006), or have no effect (McClaugherty et al. 1985, et al. 2012), suggesting that decomposition at this site Prescott 1995, Hobbie 1996, Aerts et al. 2003, 2006) on was driven by microbes mining for N. In contrast, in a litter decomposition rates, and can lead to N immobi- high soil N environment, nitrogen is readily available to lization in some systems (Gallardo and Merino 1992, microbes, therefore it is not necessary for them to Magill and Aber 1998, Hobbie 2005, Aerts et al. 2006) mineralize litter to acquire N, which results in suppres- but see McClaugherty et al. (1985). These varying sion of litter decomposition. Indeed there is evidence responses of litter decomposition to external N may that this can occur in laboratory incubations using leaf have been attributed to interactions between the initial litter from a range of plant functional types and multiple quality of the litter and the availability of nitrogen in the soil types collected in southern Africa (Craine et al. soil. 2007) and in field studies using leaf litter from the same 1872 JENNIE DEMARCO ET AL. Ecology, Vol. 95, No. 7 species but with varying litter quality and soil nutrient deciduous shrub abundance will likely decrease the rate availability (Talbot and Treseder 2012). The B. neo- at which litter decomposes, assuming that the native alaskana litter we used for our study was of low quality litter responds similarly to the common substrate we with both a high C:N ratio (72 6 0.60; mean 6 SE) and used in our experiment. The litter of B. neoalaskana had percentage of lignin (19 6 2) and was decomposed a higher C:N ratio than the deciduous shrub litter native across sites that varied in soil N availability (DeMarco to these sites but similar lignin content; thus, we assume et al. 2011). Therefore the microbial N mining hypoth- that when native litter is decomposed at the high shrub esis may help explain why we saw a linear pattern of site it would respond in the same way as our common decreased rates of decomposition and an increase in N substrate. immobilization with increasing soil N at our sites. The N immobilization we found at our high shrub site Litter quality controls over litter decomposition may also be explained by interactions with soil nutrient Members of a plant functional group (i.e., deciduous, availability and the high lignin content of our litter. evergreen, graminoid, etc.) are often similar in their litter High nutrient content in soils can suppress the chemistry and decomposition rates (Hobbie 1996, production of fungal ligninase (Carreiro et al. 2000, Hector et al. 2000, Hobbie and Gough 2004). Compar- Sinsabaugh et al. 2002), which is induced by low N isons of litter decomposition among species and availability (Keyser et al. 1978), resulting in low rates of functional groups in this experiment suggest that decomposition. Fungal communities between tussock decomposition rates of arctic plants cannot be general- tundra and shrub tundra soils sampled near Toolik Lake ized using functional group designations, because differ at the phyla and subphyla levels; however, we do species within the same functional group did not always not know whether the species responsible for breaking follow the same pattern of decomposition. For example, down lignin or the production of ligninase differs the evergreen shrub, Ledum decumbens, and the grami- between these two plant communities (Wallenstein et noid, Eriophorum vaginatum, had decomposition rates al. 2007). These tussock tundra soils are dominated by that were similar to those of the deciduous shrubs, B. slow-growing microbes that have high affinities for C nana and Vaccinium uliginosum. Their decomposition substrates of low quality and quantity. In contrast, rates were higher than other species within their shrub tundra soils are dominated by microbes that have functional groups; Ledum decumbens had higher rates high growth rates with high nutritional requirements for than the evergreen shrub Vaccinium vitis-idaea, and C substrates of higher quality and quantity (Fierer et al. Eriophorum vaginatum had higher rates than the 2007, Wallenstein et al. 2007). It is possible that graminoid Carex bigelowii. This is in contrast with microbes at the low shrub site are better at decomposing other decomposition studies within this region (Hobbie litter that is of low quality than the microbes at the high 1996, Hobbie and Gough 2004), perhaps because our shrub site, and that microbes at the high shrub site study included more species of deciduous and evergreen immobilize more N because they have a higher nutrient shrubs than were used in previous studies. Decomposi- demand when mineralizing C. Low-quality litters can tion rates varied significantly among species. Of the 10 also contain high tannin contents, which can bind to N vascular plant species we decomposed, Rubus chamae- and become incorporated in the lignin fraction, decreas- morus had the highest quality litter and the fastest rate ing decomposition and increasing immobilization of N of decomposition, losing about 70% of its mass over a 3- (Gallardo and Merino 1992, Aerts et al. 2003). Although yr period. Aerts et al. (2006) also found that R. we did not measure tannins in our litter, others have chamaemorus decomposed more quickly than three found that Betula spp. leaves can have more polyphenols other subarctic bog species. In contrast, mosses had than Salix spp. and Populus spp. leaves (Palo 1984). In the lowest decomposition rates, only losing about 30% addition, litter of B. papyrifera grown under elevated of their mass over the same period. Differences among CO2 increased tannin content by as much as 81%. When the rest of the species were relatively small. A change in decomposed in a common garden, this high tannin species composition in the Arctic could lead to content litter had lower rates of decomposition and alterations in community decomposition rates and higher N immobilization compared to litter from nutrient turnover if the change includes an increase in ambient CO2 conditions, which had lower tannin levels the relative abundance of R. chamaemorus and a (Parsons et al. 2004). decrease in mosses as seen in fertilized tussock tundra Results from our common substrate experiment in Alaska where R. chamaemorus dominates the suggest that, at least on a short time scale, any effect understory and moss cover is reduced (Chapin et al. shrubs have on soil microclimate via their ability to trap 1995). snow will have little influence over nutrient turnover through the decomposition of aboveground litter. How Litter quality and quantity vs. microclimate controls shrubs influence decomposition of belowground litter over litter decomposition (roots and rhizomes) remains uncertain. In contrast, Based on our results, changes in litter quality and changes in soil nutrient availability and/or changes in quantity via shrub expansion will increase total com- soil microbial community associated with an increase in munity-weighted mass loss. This appears to be primarily July 2014 SHRUB EXPANSION AND DECOMPOSITION 1873 driven by changes in litter quantity, rather than litter CONCLUSIONS quality and microenvironment. For example, the low Our study suggests that on a short time scale, the and medium shrub sites had similar ANPP rates and changes in soil temperature and moisture associated similar mass loss rates, even when differences in litter with additional snow trapping by shrubs are unlikely to quality among the sites were taken into account. This influence litter nutrient turnover enough to drive the suggests that either the quality of litter among the two positive snow–shrub feedbacks proposed by Sturm et al. sites is not different enough to cause differences in (2001). The mechanisms driving shrub expansion are community mass loss, or that litter quality is not the more likely to do with shrub–litter feedbacks, where the main driver of decomposition at these two sites. Of the higher growth rates and N uptake by shrubs allows them three sites, the high shrub site had the highest ANPP and to produce more leaves, resulting in a larger litter N pool the highest mass loss, suggesting that the high quantity and faster internal cycling of nutrients (DeMarco et al. of litter produced at the high shrub site is driving the 2011). Retention of N in litter during the early stages of large mass loss from this site. The magnitude of decomposition in shrub sites may be beneficial for soil difference in mass loss among the sites, however, is organic matter decomposition and could help explain slightly minimized if changes in the soil microenviron- why we see more soil N and greater N mineralization at ment are taken into account, with a trend for an increase the medium and high shrub sites. in mass loss at the low shrub site and a decrease in mass loss with increasing shrub abundance. This suggests that ACKNOWLEDGMENTS community-weighted mass loss is more sensitive to We thank Charmagne Wasykowski, Grace Crummer, Anne shrub-induced changes in the amount of litter inputs Baker, Rafael Mendoza, Yi Wei Chang, Hanna Lee, Caitlin than to shrub-induced changes in the soil microenviron- Hicks-Pries, and the many undergraduates at the University of Florida for their assistance in the field and in the lab. This ment. research was supported by NSF grants DEB-0516041, DEB- Carbon loss of our litter closely followed that of mass 0516509, OPP-6737545, and the Arctic LTER (DEB-0423385). loss (Fig. 2). Initial percentage of C of our litter was LITERATURE CITED between 40% and 45%. Therefore, we can assume that community C loss would follow a similar pattern as Aerts, R., H. D. Caluwe, and B. Beltman. 2003. Plant community mediated vs. nutritional controls on litter mass loss, with the absolute values of C loss being about decomposition rates in grasslands. Ecology 84:1398–3208. half those of mass loss. Based on these assumptions and Aerts, R., R. S. P. van Logtestijn, and P. S. Karlsson. 2006. our data, shrub expansion will lead to about a 55% Nitrogen supply differentially affects litter decomposition increase in C loss from litter. However, we do not know rates and nitrogen dynamics of sub-arctic bog species. the exact fate of the C lost from the litter, which can be Oecologia 146:652–658. Baptist, F., N. G. Yoccoz, and P. Choler. 2010. Direct and released back to the atmosphere as carbon dioxide, indirect control by snow cover over decomposition in alpine leached into the soil as dissolved organic C, or be tundra along a snowmelt gradient. Plant Soil 328:397–410. incorporated as soil organic matter and microbial Bret-Harte, M. S., M. C. Mack, G. R. Goldsmith, D. B. Sloan, biomass. Previous research at these sites show that soil J. DeMarco, G. R. Shaver, P. M. Ray, Z. Biesinger, and F. S. C pools in the top 10 cm at the medium and high shrub Chapin, III. 2008. Plant functional types do not predict biomass responses to removal and fertilization in Alaskan sites have 40–60% more C stored in them compared to tussock tundra. Journal of Ecology 96:713–726. the soil C pool at the low shrub site (DeMarco et al. Bret-Harte, M. S., G. R. Shaver, and F. S. Chapin, III. 2002. 2011). This suggests that a larger portion of the C lost Primary and secondary stem growth in arctic shrubs: from litter is most likely remaining in the soil at the implications for community response to environmental change. Journal of Ecology 90:251–267. medium and high shrub sites compared to the low shrub Brooks, P. D., M. W. Williams, and S. K. Schmidt. 1996. site. Microbial activity under snow packs, Niwot Ridge, Colo- We assumed that litter from all species within shrub rado. Biogeochemistry 32:93–113. communities will respond to the environment in the Brooks, P. D., M. W. Williams, and S. K. Schmidt. 1998. same ways as the common litter we incubated. However, Inorganic nitrogen and microbial biomass dynamics before and during spring snowmelt. Biogeochemistry 43:1–15. it is possible that differences in litter quality may interact Buckeridge, K. M., E. Zufelt, H. Chu, and P. Grogan. 2010. with site environment at the high nutrient site, resulting Soil nitrogen cycling rates in low arctic shrub tundra are in a different pattern than what we saw with our enhanced by litter feedbacks. Plant Soil 330:407–421. common litter. Litter quality and site interactions are Cahoon, S. M., P. F. Sullivan, G. R. Shaver, J. M. Welker, and not always reported, as many decomposition studies do E. Post. 2012. Interactions among shrub cover and the soil microclimate may determine future Arctic carbon budgets. not conduct a full factorial experiment where all litter Ecology Letters 15:1415–1422. types are incubated in all environments. However, Carreiro, M. M., R. L. Sinsabaugh, D. A. Repert, and D. F. Hobbie and Gough (2004) did show that B. nana stem Parkhurst. 2000. Microbial enzyme shifts explain litter decay decomposition followed the opposite pattern when responses to simulated nitrogen deposition. Ecology 81:2359– 2365. incubated in moist acidic and moist nonacidic tundra Chapin, F. S., III, and G. R. Shaver. 1996. Physiological and compared to the other species decomposed at the same growth responses of arctic plants to a field experiment sites. simulating climate change. Ecology 77:822–840. 1874 JENNIE DEMARCO ET AL. Ecology, Vol. 95, No. 7

Chapin, F. S., III, G. R. Shaver, A. E. Giblin, K. J. Jia, G. J., H. E. Epstein, and D. A. Walker. 2009. Vegetation Nadelhoffer, and J. A. Laundre. 1995. Responses of Arctic greening in the Canadian Arctic related to decadal warming. tundra to experimental and observed changes in climate. Journal of Environmental Monitoring 11:2231–2238. Ecology 76:694–711. Jonasson, S., J. Castro, and A. Michelson. 2004. Litter, Chapin, F. S., III, et al. 2005. Role of land-surface changes in warming and plants affect respiration and allocation of soil Arctic summer warming. Science 310:657–660. microbial and plant C, N and P in arctic mesocosms. Soil Clemmensen, K. E., A. Michelsen, S. Jonasson, and G. R. Biology and Biochemistry 36:1129–1139. Shaver. 2006. Increased ectomycorrhizal fungal abundance Kaufman, D. S., et al. 2009. Recent warming reverses long-term after long-term fertilization and warming of two Arctic Arctic cooling. Science 325:1236–1239. tundra ecosystems. New Phytologist 171:391–404. Keyser, P., K. Kirk, and J. G. Zeikus. 1978. Ligninolytic Craine, J. M., C. Morrow, and N. Fierer. 2007. Microbial enzyme system of Phanerochaete chrysosporium: synthesized nitrogen limitation increases decomposition. Ecology 88: in the absence of lignin in response to nitrogen starvation. 2105–2113. Journal of Bacteriology 135:790–797. DeMarco, J., M. C. Mack, and M. S. Bret-Harte. 2011. The Knops, J. M. H., D. A. Wedin, and D. Tilman. 2001. effects of snow, soil microenvironment, and soil organic Biodiversity and decomposition in experimental grassland matter quality on N availability in three Alaskan Arctic plant ecosystems. Oecologia 126:429–433. communities. Ecosystems 14:804–817. Knorr, M., S. D. Frey, and P. S. Curtis. 2005. Nitrogen Elmendorf, S. C., et al. 2012. Plot-scale evidence of tundra additions and litter decomposition: a meta-analysis. Ecology vegetation change and links to recent summer warming. 86:3252–3257. Nature Climate Change 2:453–457. Lavoie, M., M. C. Mack, and E. A. G. Schuur. 2011. Effects of Fierer, N., M. A. Bradford, and R. B. Jackson. 2007. Toward elevated nitrogen and temperature on carbon and nitrogen an ecological classiﬁcation of soil bacteria. Ecology 88:1354– dynamics in Alaskan arctic and boreal soils. Journal of 1364. Geophysical Research 116:1–14. Fontaine, S., and S. Barot. 2005. Size and functional diversity Mack, M. C., E. A. G. Schuur, M. S. Bret-Harte, G. R. Shaver, of microbe populations control plant persistence and long- and F. S. Chapin, III. 2004. Ecosystem carbon storage in term soil carbon accumulation. Ecology Letters 8:1075–1087. arctic tundra reduced by long-term nutrient fertilization. Forbes, B. C., M. M. Fauria, and P. Zetterberg. 2010. Russian Nature 431:440–443. Arctic warming and ‘greening’ are closely tracked by tundra Magill, A. H., and J. D. Aber. 1998. Long-term effects of shrub willows. Global Change Biology 16:1542–1554. experimental nitrogen additions on foliar litter decay and Gallardo, A., and J. Merino. 1992. Nitrogen immobilization in humus formation in forest ecosystems. Plant and Soil 203: leaf litter at two Mediterranean ecosystems of SW Spain. 301–311. McClaugherty, C. A., J. Pastor, J. D. Aber, and J. M. Melillo. Biogeochemistry 15:213–228. 1985. Forest litter decomposition in relation to soil nitrogen Grogan, P., and S. Jonasson. 2003. Controls on annual dynamics and litter quality. Ecology 66:266–275. nitrogen cycling in the understory of a subarctic birch forest. McGuire, D., L. G. Anderson, T. R. Christensen, S. Dallimore, Ecology 84:202–218. L. Guo, D. J. Hayes, M. Heimann, T. D. Lorenson, R. W. Hamilton, T. D. 1986. Late Cenozoic glaciation of the central Macdonald, and N. Roulet. 2009. Sensitivity of the carbon Brooks Range. Pages 9–49 in T. D. Hamilton, K. M. Reed, cycle in the Arctic to climate change. Ecological Monographs and R. M. Thorson, editors. Glaciation in Alaska—the 79:523–555. geological record. Alaska Geological Society, Anchorage, McLaren, J. R., and R. Turkington. 2010. Plant functional Alaska, USA. group identity differentially affects leaf and root decompo- Hassol, S. J. 2004. Arctic climate impact assessment: impacts of sition. Global Change Biology 16:3075–3084. a warming Arctic. Cambridge University Press, Cambridge, Mikan, C. J., J. P. Schimel, and A. P. Doyle. 2002. Temperature UK. controls of microbial respiration in arctic tundra soils above Hector, A., A. J. Beale, A. Minns, S. J. Otway, and J. H. and below freezing. Soil Biology and Biochemistry 34:1785– Lawton. 2000. Consequences of the reduction of plant 1795. diversity for litter decomposition: effects through litter Moorhead, D. L., and R. L. Sinsabaugh. 2006. A theoretical quality and microenvironment. Oikos 90:357–371. model of litter decay and microbial interaction. Ecological Hobbie, S. E. 1992. Effects of plant species on nutrient cycling. Monographs 76:151–174. Trends in Ecology and Evolution 7:336–339. Olson, J. S. 1963. Energy storage and the balance of producers Hobbie, S. E. 1996. Temperature and plant species control over and decomposers in ecological systems. Ecology 44:322–331. litter decomposition in Alaskan tundra. Ecological Mono- Osterkamp, T. E., and M. W. Payne. 1981. Estimates of graphs 66:503–522. permafrost thickness from well logs in northern Alaska. Cold Hobbie, S. E. 2005. Contrasting effects of substrate and Regions Science and Technology 5:13–27. fertilizer nitrogen on the early stages of litter decomposition. Palo, R. T. 1984. Distribution of birch (Betula spp.), willow Ecosystems 8:644–656. (Salix spp.), and poplar (Populus spp.) secondary metabolites Hobbie, S. E., and F. S. Chapin, III. 1996. Winter regulation of and their potential role as chemical defense against herbi- tundra litter carbon and nitrogen dynamics. Biogeochemistry vores. Journal of Chemical Ecology 10:499–520. 35:327–338. Parsons, W. F. J., R. L. Lindroth, and J. Bockheim. 2004. Hobbie, S. E., and L. Gough. 2004. Litter decomposition in Decomposition of Betula papyrifera leaf litter under the moist acidic and non-acidic tundra with different glacial independent and interactive effects of elevated CO2 and O3. histories. Oecologia 140:113–124. Global Change Biology 10:1666–1677. Hobbie, S. E., P. B. Reich, J. Oleksyn, M. Ogdahl, R. Pearson, R. G., S. J. Phillips, M. M. Loranty, P. S. A. Beck, T. Zytkowiak, C. Hale, and P. Karolewski. 2006. Species effects Damoulas, S. J. Knight, and S. J. Goetz. 2013. Shifts in on decomposition and forest ﬂoor dynamics in a common Arctic vegetation and associated feedbacks under climate garden. Ecology 87:2288–2297. change. Nature Climate Change 3:673–677. Jenny, H. 1994. Factors of soil formation: a system of Ping, C. L., G. J. Michaelson, and J. M. Kimble. 1997. Carbon quantitative pedology. Dover Publications, New York, New storage along a latitudinal transect in Alaska. Nutrient York, USA. Cycling in Agroecosystems 49:235–242. Jia, G. J., and H. E. Epstein. 2003. Greening of Arctic Alaska, Pomeroy, J. W., D. S. Bewley, R. I. H. Essery, N. R. Hedstrom, 1981–2001. Geophysical Research Letters 30:2067. T. Link, R. J. Granger, J. E. Sicart, C. R. Ellis, and J. R. July 2014 SHRUB EXPANSION AND DECOMPOSITION 1875

Janowicz. 2006. Shrub tundra snowmelt. Hydrological Sistla, S. A., S. Asao, and J. P. Schimel. 2012. Detecting Processes 20:923–941. microbial N-limitation in tussock tundra soil: implications Prescott, C. E. 1995. Does nitrogen availability control rates of for Arctic soil organic carbon cycling. Soil Biology and litter decomposition in forests? Plant and Soil 168–169:83–88. Biochemistry 55:78–84. Romanovsky, V. E., and T. E. Osterkamp. 2000. Effects of Stark, N. 1972. Nutrient cycling pathway and litter fungi. unfrozen water on heat and mass transport processes in the BioScience 22:355–360. active layer and permafrost. Permafrost and Periglacial Stowe, D. A., et al. 2004. Remote sensing of vegetation and Processes 11:219–239. land-cover change in Arctic tundra systems. Remote Sensing Ryan, M. G., J. M. Melillo, and A. Ricca. 1990. A comparison of Environment 89:281–308. of methods for determining proximate carbon fractions of Sturm, M., J. P. McFadden, G. E. Liston, F. S. Chapin, III, forest litter. Canadian Journal of Forest Research 20:166– C. H. Racine, and J. Holmgren. 2001. Snow-shrub interac- 171. Saccone, P., et al. 2013. The effects of snowpack properties and tions in Arctic tundra: a hypothesis with climatic implica- plant strategies on litter decomposition during winter in tions. Journal of Climate 14:336–344. subalpine meadows. Plant and Soil 363:215–229. Sturm, M., J. P. Schimel, G. Michaelson, J. M. Welker, S. F. Scherer-Lorenzen, M. 2008. Functional diversity affects de- Oberbauer, G. E. Liston, J. Fahnestock, and V. E. Roma- composition processes in experimental grasslands. Function- novsky. 2005. Winter biological processes could help convert al Ecology 22:547–555. Arctic tundra to shrubland. BioScience 55:17–26. Schimel, J. P., C. Bilbrough, and J. M. Welker. 2004. Increased Talbot, J. M., and K. K. Treseder. 2012. Interactions among snow depth affects microbial activity and nitrogen mineral- lignin, cellulose, and nitrogen drive litter chemistry-decay ization in two Arctic tundra communities. Soil Biology and relationships. Ecology 93:345–354. Biochemistry 36:217–227. Tape, K. D., M. Sturm, and C. Racine. 2006. The evidence for Screen, J. A., and I. Simmonds. 2010. The central role of shrub expansion in northern Alaska and the Pan-Arctic. diminishing sea ice in recent Arctic temperature ampliﬁca- Global Change Biology 12:686–702. tion. Nature 464:1334–1337. Vivanco, L. and A. T. Austin. 2008. Tree species identity alters Serreze, M. C., and J. A. Francis. 2006. The arctic ampliﬁcation forest litter decomposition through long-term plant and soil debate. Climate Change 76:241–264. interactions in Patagonia, Argentina. Journal of Ecology 96: Shaver, G. R., M. S. Bret-Harte, M. H. Jones, J. F. Johnstone, 727–736. L. Gough, J. A. Laundre, and F. S. Chapin, III. 2001. Species Wahren, C. H. A., M. D. Walker, and M. S. Bret-Harte. 2005. composition interacts with fertilizer to control long-term Vegetation responses in Alaskan Arctic tundra after 8 years change in tundra productivity. Ecology 82:3163–3181. of a summer warming and winter snow manipulation Shaver, G. R., et al. 2000. Global warming and terrestrial ecosystems: a conceptual framework for analysis. BioScience experiment. Global Change Biology 11:537–552. 50:871–882. Walker, M. D., et al. 1999. Long-term experimental manipu- Shaver, G. R., and F. S. Chapin, III. 1991. Production: biomass lation of winter snow regime and summer temperature in relationships and element cycling in contrasting Arctic arctic and alpine tundra. Hydrological Processes 13:2315– vegetation types. Ecological Monographs 61:1–31. 2330. Sinsabaugh, R. L., M. M. Carreiro, and D. A. Repert. 2002. Wallenstein, M. D., S. McMahon, and J. P. Schimel. 2007. Allocation of extracellular enzymatic activity in relation to Bacterial and fungal community structure in Arctic tundra litter composition, N deposition, and mass loss. Biogeo- tussock and shrub soils. Federation of European Microbio- chemistry 60:1–24. logical Societies Microbiology Ecology 59:428–435.

SUPPLEMENTAL MATERIAL

Appendix A Aboveground net primary productivity for each species within each shrub community and their corresponding k values (Ecological Archives E095-164-A1).

Appendix B Initial mass, carbon, and nitrogen remaining, and decay constants for the common substrate, Betula neoalaskana (Ecological Archives E095-164-A2).

Appendix C Initial litter quality of leaves and stems of Betula nana and Salix pulchra collected across the shrub gradient sites (Ecological Archives E095-164-A3).

Appendix D Regression analysis of initial litter quality and litter decay constants (Ecological Archives E095-164-A4).