Herbivore-mediated material fluxes in a northern deciduous forest under elevated carbon dioxide and ozone concentrations

Authors


Summary

  • Anthropogenic changes in atmospheric carbon dioxide (CO2) and ozone (O3) are known to alter tree physiology and growth, but the cascading effects on herbivore communities and herbivore-mediated nutrient cycling are poorly understood.
  • We sampled herbivore frass, herbivore-mediated greenfall, and leaf-litter deposition in temperate forest stands under elevated CO2 (c. 560 ppm) and O3 (c. 1.5× ambient), analyzed substrate chemical composition, and compared the quality and quantity of fluxes under multiple atmospheric treatments.
  • Leaf-chewing herbivores fluxed 6.2 g m−2 yr−1 of frass and greenfall from the canopy to the forest floor, with a carbon : nitrogen (C : N) ratio 32% lower than that of leaf litter. Herbivore fluxes of dry matter, C, condensed tannins, and N increased under elevated CO2 (35, 32, 63 and 39%, respectively), while fluxes of N decreased (18%) under elevated O3. Herbivore-mediated dry matter inputs scaled across atmospheric treatments as a constant proportion of leaf-litter inputs.
  • Increased fluxes under elevated CO2 were consistent with increased herbivore consumption and abundance, and with increased plant growth and soil respiration, previously reported for this experimental site. Results suggest that insect herbivory will reinforce other factors, such as photosynthetic rate and fine-root production, impacting C sequestration by forests in future environments.

Introduction

Human activities are altering the composition of the earth's atmosphere. Fossil fuel consumption and land-use change have caused carbon dioxide (CO2) concentrations to reach their highest levels in over 800 000 yr (IPCC, 2013). Similarly, tropospheric ozone (O3) concentrations have doubled over the Northern Hemisphere during the past century, due to anthropogenic increases in O3 precursors such as nitrogen (N) oxides and methane (Vingarzan, 2004). These dramatic changes are impacting many aspects of forest biology, from leaf-level physiology to community interactions and ecosystem processes.

For example, it is now well established that the chemical characteristics of leaves are influenced by atmospheric concentrations of CO2 and O3 (Stiling & Cornelissen, 2007; Valkama et al., 2007; Lindroth, 2012; Couture & Lindroth, 2013). Increased CO2 concentrations commonly lead to decreased concentrations of foliar N, and increased concentrations of foliar starches, tannins, and simple phenolics, especially in younger trees (Lindroth, 2012; Couture & Lindroth, 2013). Increased O3 concentrations occasionally lead to decreased concentrations of foliar N and starch, and typically increase concentrations of foliar phenolics, flavonoids, and lignins, especially after several years of exposure (Lindroth, 2012; Couture & Lindroth, 2013). Moreover, changes in foliar chemistry have implications for the herbivores that eat those leaves. For instance, increased CO2 concentrations commonly lead to decreased growth rates, increased development times, and, notably, increased consumption rates in leaf-chewing insects (Lindroth, 2012; Couture & Lindroth, 2013). Increased consumption rates are probably attributable to compensatory feeding by herbivores, which must consume more low-quality food in order to meet dietary requirements (Lindroth et al., 1993). By contrast, increased O3 concentrations often lead to decreased development times and increased pupal masses, although they do not appear to impact herbivore consumption rates (Lindroth, 2012; Couture & Lindroth, 2013).

While the effects of atmospheric change on leaf chemistry and individual herbivore performance have been fairly well studied, effects on populations and communities of herbivores remain poorly understood. The available information suggests that herbivore populations can increase or decrease under elevated CO2, depending on the herbivore species, host tree species, and measurement approach (Lindroth, 2012; Couture & Lindroth, 2013; Hillstrom et al., 2014). Species-specific responses to altered atmospheric chemistry lead to changes in herbivore community structure that are generally difficult to predict (Hillstrom et al., 2014). Recent results from two long-term studies, however, show that under elevated CO2 the total number of leaf-chewing herbivores decreases on a per-leaf basis, but increases per unit ground area, as a result of the increase in leaf area index that typically accompanies elevated CO2 (Stiling et al., 2009; Hillstrom et al., 2014). The opposite trend has been reported for herbivorous insect communities under elevated O3 (Hillstrom et al., 2014), as a result of the negative effects of O3 on plant production (Karnosky et al., 2003; but see also Zak et al., 2011).

The effects of atmospheric change on herbivorous insect communities are of considerable importance, because of the functional roles of insects in forest ecosystems (Schowalter, 2006). Herbivore communities regularly consume a substantial fraction of net primary production in forests (Cyr & Pace, 1993; Cebrian, 1999). In doing so, they move organic substrates from the canopy to the soil in the form of frass (arthropod excrement), greenfall (unconsumed leaf fragments that fall from trees), and insect carcasses (Hunter, 2001). These materials fall throughout the growing season and are of higher nutritional quality than senesced leaf litter (hereafter, ‘leaf litter’) deposited at the end of the growing season (Hunter, 2001; Lovett et al., 2002a). This pulse of labile, nutrient-rich material moves through numerous avenues in forest ecosystems, including uptake by trees and assimilation into canopy foliage (Frost & Hunter, 2007), immobilization by microbial communities (Christenson et al., 2002; Hillstrom et al., 2010), and export from the system via surface and ground water (Hunter et al., 2003; Frost & Hunter, 2004, 2007; Townsend et al., 2004). The timing and chemical composition of herbivore-mediated material fluxes (hereafter, ‘herbivore fluxes’) have a strong impact on forest nutrient cycles (Hunter, 2001; Lovett et al., 2002b), even when herbivores occur at endemic population levels (Hunter et al., 2003), and especially in highly nutrient-limited forests (Metcalfe et al., 2013). Unfortunately, poor understanding of the relationships between atmospheric change and herbivore community structure severely limits our capacity to predict how atmospheric change will alter herbivore fluxes. This limitation further constrains our ability to predict carbon (C) sequestration by forests in the future, given that C uptake by forests is contingent on photosynthetic area and soil nutrient availability (Reich et al., 2006; Norby et al., 2010; Drake et al., 2011; Zak et al., 2011).

To address this challenge, we assessed the impacts of atmospheric change on material fluxes from the canopy to the soil by monitoring frass deposition, greenfall, and leaf-litter fall in forest plots exposed to elevated CO2 and O3. Our study, conducted at the Aspen Free Air Carbon Dioxide Enrichment (Aspen FACE) site in the Great Lakes region of North America, had two main objectives: to describe patterns in the overall quantity and quality of materials deposited via the herbivore and leaf-litter pathways; and to evaluate predictions for how the quality and quantity of those materials will change when stands are exposed to elevated CO2 and O3 concentrations.

Given the documented effects of atmospheric change on individual plants and insects, we predicted that elevated CO2 concentrations would lead to compensatory feeding and increased individual-herbivore fluxes, ultimately leading to increased community-level fluxes. And given recent suggestions that elevated CO2 concentrations increase herbivore abundance per unit ground area, we predicted additional increases in community-level fluxes. These predictions are based on recognized plant-mediated effects of enriched CO2 on insects, as the direct effects of CO2 on insect herbivores are expected to be nil (Lindroth, 2010). Regarding O3, previous research suggests that increased concentrations will not alter individual feeding rates. However, as O3 can decrease the abundance of herbivores per unit ground area, as a result of both the negative effects of O3 on plant production and the direct effects on insect abundance (Lindroth, 2010), we predicted an overall decrease in herbivore community fluxes under elevated O3 concentrations.

Materials and Methods

Our study was conducted in early-succession temperate deciduous forest stands at the Aspen FACE facility near Rhinelander, Wisconsin, USA, before its decommissioning in 2010. The Aspen FACE site consisted of twelve 30-m-diameter FACE rings, each receiving one of four treatment combinations: (1) ambient CO2 and ambient O3 (control), (2) elevated CO2 (c. 560 ppm) and ambient O3 (+CO2), (3) ambient CO2 and elevated O3 (c. 1.5× ambient O3 concentration; +O3), and (4) elevated CO2 and elevated O3 (+CO2+O3). The target atmospheric concentrations at this FACE experiment corresponded with those predicted for the region in 2050 under a business-as-usual greenhouse gas emissions scenario. Aspen (Populus tremuloides Michx.) and birch (Betula papyrifera Marshall) saplings were planted in 1997 and fumigation with CO2 and O3 began in 1998. When our study began, in 2006, trees were 10 yr old, and all stands had closed canopies. More details about the study system can be found in Dickson et al. (2000).

Within each FACE ring, we sampled herbivore fluxes and leaf-litter deposition in pure aspen stands with five different aspen genotypes (genotypes 8, 42, 216, 259 and 271), and also in mixed aspen and birch stands with a single aspen genotype (216). We sampled herbivore fluxes during either five or six sampling campaigns between late May and late August of each year from 2006 through 2008. Each herbivore flux campaign lasted 9–10 d and was separated by c. 2 wk. We sampled leaf litter over a 10-d campaign during leaf drop in late September and early October of each year from 2006 through 2008. Sampling of herbivore fluxes and leaf litter was accomplished using plastic baskets staked to the forest floor (Reynolds et al., 2000). Each basket was fitted with muslin cloth, stretched across the opening and weighted in the center, to collect inputs while allowing precipitation to pass through. During each campaign, four baskets were placed in each of the two stand types, totaling eight baskets per FACE ring.

On returning to the laboratory, herbivore flux samples were air-dried and sorted under a dissecting scope into three components: frass, greenfall, and other. Herbivore frass was easily identified by its distinctive shape and texture. Herbivore-mediated greenfall was defined as green leaf fragments without a complete petiole. This included small leaf fragments dropped by caterpillars and larger leaf fragments with partial petioles, possibly clipped by caterpillars trying to avoid detection by predators (Heinrich & Collins, 1983). The third category included leaf fragments with complete petioles, bark fragments, branch fragments, and soil particles. Fragments with complete petioles were not included as herbivore-mediated greenfall because leaf abscission can be caused by a variety of factors other than herbivory (Risley, 1986). After sorting, frass and greenfall samples were freeze-dried and weighed, and frass was stored frozen until chemical analysis.

The C and N concentrations of frass and leaf-litter samples were measured using a Thermo-Finnigan Flash 1112 elemental analyzer (Thermo Scientific, Waltham, MA, USA). Condensed tannin concentrations of frass and leaf litter were quantified spectrophotometrically using the butanol–hydrochloric-acid method (Porter et al., 1985), with condensed tannin laboratory standards purified from aspen and birch leaves. C, N, and condensed tannin concentrations of greenfall were estimated using near-infrared reflectance spectroscopy, as described in Couture & Lindroth (2012). Finally, although we recognize that lignin concentrations also influence nutrient cycling dynamics, we were unable to quantify lignin levels in frass. (The small particle size of frass compromised efforts to quantify lignin via standard gravimetric acid-digestion methods.)

For each of the five or six sampling campaigns per year, the dry weights of frass and greenfall were divided by the campaign duration (i.e. 9 or 10 d) to attain a daily flux rate for each campaign. Annual deposition rates were then calculated by estimating the area under the curve of daily deposition rates versus calendar day, where deposition rates were set to zero on 15 May and 15 September of each year. In calculating the area under the curve, daily flux rates were linearly interpolated when empirical rates were not available. Annual leaf-litter deposition rates were estimated directly from the dry mass of leaf-litter samples. Flux rates for C, condensed tannins, and N via frass, greenfall, and leaf litter were calculated as total mass × proportional concentrations determined by chemical analyses.

The variables of interest in our analyses included dry mass, C, condensed tannin, and N fluxes via herbivore and leaf-litter pathways. Each response variable was averaged over the 3 yr of the study because we were interested in the mean response of aspen and birch stands to fumigation treatments, and not concerned with annual variation (annual data are available upon request). We pooled flux data from control plots and across tree community types to describe general patterns in the quantity and quality of materials deposited via each pathway at the Aspen FACE site (objective 1). To explore the effects of fumigation treatments on fluxes (objective 2), we analyzed variation in dependent variables using split-plot analysis of variance, with block, CO2, O3, and a CO2 × O3 interaction as whole-plot effects, and community type, community type × CO2, community type × O3, and community type × CO2 × O3 interactions as subplot effects. Upon finding significant fumigation effects on herbivore fluxes, we conducted additional analyses on frass and greenfall, separately, to determine the contributions of the two components to overall patterns. We observed relatively few statistically significant interactions between the fumigation treatments, and few three-way interactions between fumigation treatments and tree community type. Unless otherwise noted, we report the main, additive effects of the treatments. For example, a significant main effect of CO2 relates to both the +CO2 and +CO2+O3 treatment combinations, and a significant main effect of O3 involves both the +O3 and +CO2+O3 treatment combinations.

Results

Material fluxes under ambient CO2 and O3

Our first objective was to describe the quantity and quality of materials fluxed from the canopy to the soil via the leaf-chewing herbivore and leaf-litter pathways under ambient CO2 and O3 conditions. Across the 3 yr of the study, mean (1 SD) fluxes of dry matter, C, condensed tannin, and N via both pathways were 222.5 (18.5), 108.7 (8.2), 11.0 (3.0), and 3.2 (0.5) g m−2 yr−1, respectively (Fig. 1). Herbivore inputs (frass and greenfall, combined) represented a relatively small portion of the materials transferred from the canopy to the soil. For instance, the mean dry mass of pooled herbivore inputs was only 3% of that for leaf litter. The chemical characteristics of herbivore inputs and leaf litter also differed. For example, the C and condensed tannin concentrations of herbivore inputs were c. 3% and 96% higher, respectively, than those of leaf litter. Meanwhile, the N concentration of herbivore inputs was c. 36% higher than that of leaf litter, and the C : N ratio of herbivore inputs averaged 32% lower than that of leaf litter.

Figure 1.

Mean quantity and quality of material fluxes via herbivore and leaf-litter pathways under ambient CO2 and O3 at the Aspen Free Air Carbon Dioxide Enrichment (FACE) facility during 2006 through 2008.

Frass was the larger of the two herbivore inputs that we measured (Fig. 1); the dry mass of frass was more than twice that of greenfall. The chemical characteristics of frass and greenfall also differed; C concentration was 3% higher in frass than in greenfall, whereas condensed tannin and N concentrations were 54% and 34% lower, respectively, in frass than in greenfall. The C : N ratio averaged 55% higher in frass than in greenfall.

Input quality, fumigation treatment, and tree community

Our second objective was to explore, in more detail, how the quality and quantity of materials fluxed via the herbivore and leaf-litter pathways varied with atmospheric concentrations of CO2 and O3. Regarding input quality, we found a significant two-way interaction between CO2 and O3 treatments, where C concentrations of herbivore inputs were 6% higher under the ambient CO2 and elevated O3 treatment combination (i.e. +O3) than in other treatment combinations (Table 1). The condensed tannin concentrations of herbivore inputs were 18% higher under elevated CO2 than ambient CO2 treatments, but unaffected by O3. Nitrogen concentrations were 6% lower under elevated O3 treatments but unaffected by CO2. C : N ratios were 6% lower under elevated CO2 treatments and 10% higher under elevated O3 treatments. Finally, the quality of combined herbivore inputs did not vary according to community.

Table 1. Mean (1 SD) concentrations (% dry mass) of carbon (C), condensed tannins, and nitrogen (N), and ratios of C : N, per treatment combination for combined herbivore inputs (all frass and greenfall) and senesced leaf litter, with split-plot ANOVA results
Treatment combination Combined herbivore inputsLeaf litter
CarbonCondensed tanninsNitrogenC : NCarbonCondensed tanninsNitrogenC : N
  1. Bold text indicates statistical significance.

Aspen
Control 50.7 (1.3)9.4 (1.8)1.9 (0.1)26.9 (1.5)49.4 (0.1)4.9 (0.6)1.4 (0.2)34.5 (3.8)
+CO2 50.0 (0.5)11.7 (0.9)1.9 (0.1)26.4 (1.2)48.8 (1.1)7.8 (0.5)1.3 (0.2)39.4 (5.6)
+O3 53.3 (2.5)11.7 (0.7)1.8 (0.1)30.0 (1.4)48.8 (0.2)7.5 (1.4)1.4 (0.1)36.1 (2.5)
+CO2+O3 50.5 (3.0)12.5 (1.3)1.8 (0.2)28.1 (1.6)48.9 (0.5)7.0 (1.0)1.3 (0.2)39.1 (4.5)
Aspen and birch
Control 49.8 (1.4)9.42 (1.3)1.8 (0.1)27.5 (1.6)48.3 (0.1)4.7 (1.6)1.4 (0.3)35.4 (7.0)
+CO2 50.8 (0.6)12.23 (2.1)1.9 (0.1)26.6 (1.2)48.1 (1.0)8.1 (3.9)1.1 (0.1)45.6 (3.3)
+O3 53.6 (0.5)11.03 (1.1)1.7 (0.3)31.4 (4.7)48.6 (0.3)6.3 (0.7)1.3 (0.4)37.7 (9.0)
+CO2+O3 49.5 (1.0)12.95 (1.4)1.7 (0.1)28.4 (0.4)48.0 (0.2)6.0 (2.4)1.2 (0.4)42.0 (11.8)
ANOVA sourcedfP-value
CO21,6 0.036 0.033 0.394 0.033 0.3170.1850.1700.126
O31,60.0620.104 0.026 0.004 0.8260.7460.9840.998
CO2 × O31,6 0.026 0.4190.6400.1850.8020.0990.5400.555
Community1,80.7360.7600.4030.531 0.000 0.4390.4790.302
CO2 × community1,80.8520.1380.6770.7060.7360.7970.6390.547
O3 × community1,80.7670.5070.7580.8100.0910.4390.6390.812
CO2 × O3 × community1,80.2710.5500.5770.862 0.031 0.8960.7480.711

The chemical quality of leaf litter did not vary strongly across experimental factors (Table 1). We observed, however, one three-way interaction between CO2, O3, and tree community, where the C concentrations of litter were slightly lower in mixed aspen and birch stands than in pure aspen stands, except under the ambient CO2 and elevated O3 treatment combination (i.e. +O3).

Input quantity, fumigation treatment, and tree community

The total dry mass of herbivore inputs (frass and greenfall, combined) increased by 35% in elevated CO2 plots, decreased by 13% (a nonsignificant effect) in elevated O3 plots, and was 16% higher in aspen and birch stands than in pure aspen stands (Table 2; Fig. 2). The total amount of C fluxed via herbivores was 32% higher in elevated CO2 compared with ambient CO2, was not affected by O3 concentration, and was 16% higher in aspen and birch stands than in pure aspen stands. The mass of condensed tannins fluxed by herbivores was 63% higher in elevated CO2 plots, was not affected by O3 concentrations, and was 19% higher in aspen and birch stands than in pure aspen stands. The especially large increase in condensed tannin input under elevated CO2 was probably attributable in part to the positive effect of CO2 on the condensed tannin concentrations of herbivore inputs.

Table 2. Mean (1 SD) annual deposition (g m−2 yr−1) of dry matter, carbon (C), condensed tannin, and nitrogen (N) per treatment combination for combined herbivore inputs (all frass and greenfall) and senesced leaf litter, with split-plot ANOVA results
Treatment combination Combined herbivore inputsLeaf litter
Dry massCarbonCondensed tanninsNitrogenDry massCarbonCondensed tanninsNitrogen
  1. Bold text indicates statistical significance.

Aspen
Control 5.7 (0.5)2.9 (0.3)0.5 (0.1)0.1 (0.02)204.6 (8.6)101.2 (4.3)10.1 (1.6)3.0 (0.4)
+CO2 7.6 (1.0)3.8 (0.4)0.9 (0.2)0.2 (0.02)269.2 (8.1)131.3 (5.2)21.1 (1.3)3.4 (0.7)
+O3 4.4 (0.8)2.4 (0.5)0.5 (0.1)0.1 (0.02)171.0 (21.2)83.5 (10.6)12.8 (3.2)2.3 (0.4)
+CO2+O3 6.7 (0.6)3.4 (0.3)0.8 (0.1)0.1 (0.01)234.7 (4.8)114.7 (3.3)16.3 (2.1)3.0 (0.4)
Aspen and birch
Control 6.8 (0.5)3.4 (0.3)0.6 (0.1)0.1 (0.02)227.9 (18.0)110.0 (8.8)10.8 (4.5)3.2 (0.7)
+CO2 8.1 (1.5)4.1 (0.8)1.0 (0.4)0.2 (0.03)264.7 (46.7)127.6 (24.8)20.8 (8.1)2.8 (0.7)
+O3 5.6 (1.2)3.0 (0.6)0.6 (0.2)0.1 (0.03)171.3 (34.4)83.3 (17.1)10.8 (2.0)2.3 (0.8)
+CO2+O3 8.0 (1.3)4.0 (0.7)1.1 (0.3)0.1 (0.03)271.7 (29.5)130.5 (13.6)16.6 (7.4)3.3 (0.8)
ANOVA sourcedfP-value
CO21,6 0.007 0.008 0.014 0.003 0.001 0.002 0.021 0.179
O31,60.1270.1590.898 0.033 0.042 0.045 0.5440.222
CO2 × O31,60.4900.6920.9700.7850.2180.2310.2740.222
Community1,8 0.000 0.003 0.009 0.009 0.1100.2110.8200.935
CO2 × community1,80.5410.6710.4450.8620.7800.8300.8280.571
O3 × community1,80.2150.3650.4450.8620.5690.5110.7150.478
CO2 × O3 × community1,80.3400.6230.5200.3950.0730.0970.5890.215
Figure 2.

Mean (± 1 SE) mass of total dry matter (all frass and greenfall, a–c), carbon (d–f), condensed tannin (g–i), and nitrogen (j–l) fluxed annually by herbivores at the Aspen Free Air Carbon Dioxide Enrichment (FACE) facility as a function of CO2 and O3 concentrations and tree community type. Statistically significant differences: *, < 0.05.

The response of herbivore-mediated N deposition to fumigation treatments was slightly different from that of other inputs (Table 2; Fig. 2). Like other herbivore inputs, herbivore N flux was 39% higher in elevated CO2 plots than in ambient CO2 plots, and was 15% higher in aspen and birch stands than in pure aspen stands. Unlike other herbivore inputs, however, N input via greenfall and frass deposition was 18% lower in elevated O3 plots than in ambient O3 plots. This result was probably partly attributable to the negative effect of O3 on the N concentration of herbivore inputs.

Regarding leaf litter, the quantity of both total dry mass and C deposition was 34% higher under elevated CO2 and 12% lower under elevated O3 (Table 2; Fig. 3). The ratio of herbivore-derived deposition to canopy productivity (using leaf litter as a surrogate for productivity) was consistent across fumigation treatments, and was slightly higher for the aspen–birch community than for the aspen community (0.031 versus 0.028, respectively; = 0.021). The amount of condensed tannins deposited via leaf litter was 68% higher under elevated CO2, while the amount of N deposited in leaf litter was not influenced by experimental factors. Quantities of inputs via the leaf-litter pathway were not related to community type.

Figure 3.

Mean (± 1 SE) mass of dry matter (a–c), carbon (d–f), condensed tannin (g–i), and nitrogen (j–l) fluxed annually via autumn leaf fall at the Aspen Free Air Carbon Dioxide Enrichment (FACE) facility as a function of CO2 and O3 concentrations and tree community type. Statistically significant differences: *, < 0.05.

When we explored the separate contributions of frass and greenfall to herbivore fluxes under different fumigation treatments, we found that increased dry matter deposition by herbivores under elevated CO2 was driven primarily by highly significant differences in greenfall, and secondarily by marginally significant differences in frass deposition, especially in aspen stands, but also in aspen and birch stands in the absence of elevated O3 (Table 3). Increased C, condensed tannin, and N flux under elevated CO2 appeared to arise from differences in total dry matter flux, and not from differences in the chemical composition of frass or greenfall. By contrast, decreased herbivore N flux under elevated O3 was probably attributable to both a nonsignificant decrease in dry matter deposition and a highly significant decrease in the N concentration of greenfall.

Table 3. Mean (1 SD) dry matter deposition (g m−2 yr−1), carbon (C) concentration (% dry mass), condensed tannin concentration (% dry mass), nitrogen (N) concentration (% dry mass), and C : N ratio per treatment combination for herbivore frass input and herbivore-mediated green leaf deposition, with split-plot ANOVA results
Treatment combination FrassGreenfall
Dry massCarbonCondensed tanninsNitrogenC : NDry massCarbonCondensed tanninsNitrogenC : N
  1. Bold text indicates statistical significance.

Aspen
Control 3.9 (0.4)51.2 (2.0)7.4 (1.5)1.6 (0.2)31.6 (2.2)1.8 (0.1)49.8 (0.2)13.7 (1.0)2.4 (0.1)20.4 (0.3)
+CO2 4.3 (0.4)50.1 (1.1)8.2 (0.9)1.6 (0.1)31.9 (2.0)3.4 (0.6)49.8 (0.1)16.1 (0.8)2.3 (0.1)21.6 (0.3)
+O3 3.0 (0.4)55.0 (3.7)9.5 (0.4)1.6 (0.1)34.3 (2.3)1.4 (0.4)49.8 (0.1)16.7 (0.5)2.2 (0.1)23.1 (0.5)
+CO2+O3 3.9 (0.2)50.8 (5.4)8.2 (1.6)1.5 (0.3)34.4 (4.2)2.8 (0.5)50.0 (0.3)18.1 (1.1)2.2 (0.1)22.5 (0.1)
Aspen and birch
Control 4.7 (0.3)50.2 (1.9)6.5 (1.4)1.6 (0.1)31.9 (1.9)2.0 (0.4)49.0 (0.2)16.2 (2.0)2.4 (0.1)20.6 (0.3)
+CO2 4.5 (0.4)52.2 (1.2)8.2 (1.6)1.7 (0.1)31.6 (2.6)3.6 (1.2)49.1 (0.2)17.3 (1.9)2.2 (0.1)22.0 (1.0)
+O3 4.0 (0.5)55.2 (1.1)8.4 (0.9)1.6 (0.3)34.9 (5.7)1.6 (0.7)49.2 (0.1)18.9 (1.9)2.1 (0.1)23.3 (0.2)
+CO2+O3 4.8 (0.3)49.7 (1.4)9.2 (0.4)1.5 (0.1)34.2 (1.4)3.2 (1.1)49.3 (0.1)18.5 (1.6)2.2 (0.1)22.6 (0.8)
ANOVA sourcedfP-value
CO21,60.0540.0750.3690.3430.848 0.005 0.1480.2030.1890.222
O31,60.0760.1390.0870.309 0.019 0.231 0.034 0.031 < 0.001 < 0.001
CO2 × O31,60.103 0.041 0.3120.2480.8700.9950.7770.454 0.006 0.006
Community1,8 < 0.001 0.9610.4440.9780.9460.159 < 0.001 0.007 0.009 0.230
CO2 × community1,8 0.005 0.6650.0850.7190.8380.6680.9740.1180.8820.864
O3 × community1,8 0.004 0.6350.6930.7400.9340.7540.3030.5180.5260.681
CO2 × O3 × community1,8 0.031 0.3140.4490.4450.9700.7320.5030.8000.8050.700

Discussion

Overall material fluxes

In forest ecosystems, materials are transferred from the canopy to the forest floor via a variety of pathways (Risley & Crossley, 1988; Hunter, 2001; Stadler et al., 2001). The first objective of our study was to document the quantity and quality of material transferred via two important pathways, summer herbivore activity and autumn leaf fall. Regarding input quantity, we found that a relatively small fraction of the canopy was fluxed through leaf-chewing herbivores in our system. Specifically, the ratio of material fluxed via herbivore and leaf-litter pathways was c. 0.03 : 1. This ratio is consistent with results from many studies showing that, at endemic population levels, leaf-chewing insects tend to remove between 3% and 20% of leaf material during the growing season (Landsberg & Ohmart, 1989; Schowalter, 2006).

Regarding input quality, we found that the N concentrations decreased, and C : N ratios increased, from greenfall to frass, and from frass to leaf litter. We also found that condensed tannins were most concentrated in greenfall, followed by frass, and then leaf litter. These patterns are consistent with those reported in several previous studies (Risley & Crossley, 1988; Hunter, 2001; Hunter et al., 2003; Couture & Lindroth, 2014). Several studies have demonstrated that the relatively high quality of herbivore inputs stimulates plant growth and soil microbial activity (Belovsky & Slade, 2000; Hunter, 2001; Lovett et al., 2002a; Frost & Hunter, 2004), and we expect that the higher N concentrations and lower C : N ratios of herbivore inputs observed in our study would generate similar effects. It is not clear, however, how these stimulating effects would be tempered by the higher concentrations of condensed tannins observed in herbivore inputs. Condensed tannins can reduce microbial activity by binding organic N or acting as toxins (Schweitzer et al., 2008). Alternatively, tannin additions can stimulate soil respiration when more labile substrates are unavailable (Madritch et al., 2007).

Herbivore fluxes and atmospheric change

The second objective of our study was to evaluate how the quality, and ultimately the quantity, of herbivore and leaf-litter fluxes would be affected by atmospheric change. We found that elevated CO2 substantially increased fluxes of total dry mass (35%), elemental C (32%), condensed tannins (63%), and elemental N (39%) via the herbivore pathway. Insect-mediated fluxes are the mathematical product of numbers of insects multiplied by individual feeding rates. Hillstrom et al. (2014) provide a detailed analysis of numbers and species composition of canopy insects at Aspen FACE. In short, they found that enriched CO2 increased insect numbers in the aspen stands by 15%, whereas it had no effect on overall insect numbers in the aspen–birch stands. Thus, the increased herbivore flux rate in the aspen stands may have been linked to a larger number of herbivores per unit ground area (Stiling et al., 2009; Hillstrom et al., 2014). Lack of data on the individual feeding rates of the scores of canopy insects at Aspen FACE precludes assessment of how individual consumption may have affected fluxes at the whole-stand level. However, the increased flux rate observed is consistent both with increased per capita consumptions rates of several canopy insect species when fed foliage grown under elevated CO2 (Lindroth et al., 1993; Lindroth, 2012; Couture & Lindroth, 2013) and with elevated canopy damage rates that were observed at the Aspen FACE site under elevated CO2 (Couture, 2011).

The proportion of foliar N fluxed via insects was a small fraction of that fluxed via litter fall (Fig. 1; Table 2), and thus – at these levels of herbivory – unlikely to markedly affect forest productivity. Nonetheless, the activity of herbivorous insects alters the physical structure, chemical composition, and seasonal timing of nutrient inputs, and these changes may multiply the impact of the mass effects alone. The labile, high-nutrient nature of herbivore-mediated inputs makes this material readily available for re-assimilation into plant tissues or immobilization by soil microbes, potentially minimizing nutrient loss and reducing the negative effects of herbivory on tree growth. As suggested by Belovsky & Slade (2000), increased deposition of N by herbivores could stimulate saprotrophic microbe activity, having a positive effect on soil N availability during the growing season. Indeed, several FACE experiments have shown that soil respiration increases under elevated CO2 (King et al., 2004; Norby & Zak, 2011). While this result can be attributed mostly to increased fine-root production under elevated CO2, a portion of the increased respiration rates may be attributable to increased deposition of labile organic substrates by herbivores.

Elevated O3 did not significantly affect dry matter, C, or condensed tannin input, but had a moderate negative effect on N flux by herbivores. Dry matter deposition tended to decline (average of 13%) under elevated O3; although nonsignificant, that reduction is paralleled by reduced leaf area index (Karnosky et al., 2003) and reduced canopy insect densities (8%; Hillstrom et al., 2014) in O3-fumigated stands. The decrease in herbivore-mediated N fluxes was probably attributable to the combined effects of a slight decline in the mass of inputs, coupled with the observed decrease in the N concentration of inputs. Indeed, several other studies have shown that O3 exposure leads to lower concentrations of foliar N (Couture & Lindroth, 2013).

The moderate reductions in herbivore-mediated N fluxes under elevated O3 could have negative impacts on plant growth and soil microbial activity. The direct damaging effects of O3 on photosynthetic pathways are probably the dominant mechanism limiting stand production under elevated O3. Further reductions in productivity could result, however, from reduced N fluxes by herbivores, leading to reduced activity by saprotrophic microbes and reduced soil N availability during the growing season. Indeed, early studies at the Aspen FACE site showed modest decreases in soil respiration under elevated O3 (King et al., 2004), although this pattern did not persist as stands matured (Pregitzer & Talhelm, 2013).

Leaf-litter fluxes and atmospheric change

Elevated CO2 substantially increased deposition of dry matter, elemental C, and condensed tannins via the leaf-litter pathway. By contrast, elevated O3 decreased deposition of dry matter and elemental C via the leaf-litter pathway. These results are similar to those of Liu et al. (2007), who reported increases under elevated CO2 and decreases under elevated O3 in leaf-litter fluxes of several macro- and micro-nutrients. In our study, and that of Liu et al. (2007), changes in litter-based nutrient fluxes were attributable more to changes in total leaf-litter production and deposition, and less to changes in the chemical quality of leaf litter.

Interestingly, the ratio of herbivore-mediated dry-matter inputs to leaf-litter inputs was similar across fumigation treatments, although slightly lower in aspen relative to aspen–birch stands. These results suggest that, within stands, herbivore-mediated inputs scale directly with canopy productivity. That is, as atmospheric change increases (in the case of CO2) or decreases (in the case of O3) canopy productivity, the numbers or feeding rates (or both) of insects, as well as their substrate deposition rates, change in parallel.

Potential limitations

The results reported here are unique because they derive from in situ manipulations of atmospheric gases in relatively large deciduous forest stands. Also, herbivore flux measurements were conducted in parallel with several other studies, allowing us to compare our results with other forest ecosystem metrics. Despite these strengths, the study also necessarily incorporated several limitations that constrain interpretation of the results. First, the experimental plots at the Aspen FACE site were not enclosed, and insect movements into and out of the plots could confound our conclusions given that herbivore fluxes are affected by herbivore population densities. In this context, however, insect movement is only of concern with respect to ovipositing adults. The majority of herbivory that we quantified was caused by relatively sedentary insect larvae that probably hatched, developed, and pupated within a single forest stand. Secondly, our measurements of herbivore fluxes comprised only frass and greenfall, mostly from leaf-chewing herbivores. We did not include fluxes in the form of insect carcasses because we did not find carcasses in our collection baskets. Fluid excretions by phloem-feeding insects also contribute to herbivore fluxes from the canopy to the soil (Stadler et al., 2001), but these excretions were not monitored in our study. Thus, our herbivore flux estimates should be considered conservative measures. Thirdly, our herbivore flux measurements are based on the mass and chemical composition of herbivore frass and greenfall that remained on muslin sheets in sampling baskets over a 10-d sampling campaign. A portion of the nutrients in herbivore deposits might have leached out during rain events that occurred during a campaign. It does not appear that exposure to rain was an important factor in our study, as the nutrient concentrations found in herbivore frass were similar to, or only slightly lower than, those reported from a study of frass from herbivores raised in captivity on foliage from the Aspen FACE site (Couture & Lindroth, 2014). Nevertheless, possible leaching is another reason why our herbivore flux estimates should be considered conservative measures.

Conclusions

Our study, conducted in an aggrading temperate deciduous forest, shows that endemic populations of leaf-chewing herbivores flux meaningful quantities of nutrient-rich organic materials from the canopy to the forest floor throughout the growing season. Herbivore fluxes of dry matter, C, condensed tannins, and N increased considerably under elevated CO2, while fluxes of N decreased modestly under elevated O3. In general, herbivore-mediated dry-matter inputs scaled across fumigation treatments as a constant proportion of leaf-litter inputs. We observed few interactive effects of CO2 and O3 on the quality and quantity of herbivore inputs. Changes in herbivore fluxes were generally consistent with expectations derived from previous research on the consumption rates and overall abundance of herbivores under altered atmospheric conditions. Changes in herbivore fluxes were also broadly consistent with changes in plant growth and soil respiration under altered atmospheric conditions. Combined, these findings suggest that insect herbivory has the potential to reinforce other ecological processes expected to impact the productivity of, and C sequestration by, forests in future environments.

Acknowledgements

Aspen FACE was principally supported by the Office of Science (BER), US Department of Energy, grant DE-FG02-95ER62125 to Michigan Technological University; and contract DE-AC02-98CH10886 to Brookhaven National Laboratory; the US Forest Service Northern Global Change Program and North Central Research Station; Michigan Technological University; and Natural Resources Canada-Canadian Forest Service. This particular work was supported by National Science Foundation grant DEB-0129123 and Department of Energy (Office of Science, BER) grant DE-FG02-06ER64232 to R.L.L. We thank M. Hillstom and L. Vigue for field assistance; A. Gusse, K. Kelly, J. Doll, and G. Oates for lab assistance; and P. Townsend for field sampling advice. K. Pregitzer kindly provided additional data on annual leaf-litter deposition. We thank A. Austin and two anonymous reviewers for helpful comments that improved this manuscript.

Ancillary