While the importance of plant secondary metabolites to belowground functioning is gaining recognition, the perception remains that secondary metabolites are produced for herbivore defense, whereas their belowground impacts are ecological by-products, or ‘afterlife’ effects. However, plants invest a significant amount of resources into production of secondary metabolites that have minimal effects on herbivore resistance (e.g. condensed tannins and insect herbivores). We show that genetically mediated variation in condensed tannin concentration is correlated with plant nitrogen recovery following a severe defoliation event. We used single-tree mesocosms labeled with 15N to track nitrogen through both the frass and litter cycling pathways. High concentrations of leaf tannins in Populus tremuloides were correlated with 15N recovery from frass within the same growing season and in the following growing season. Likewise, leaf tannin concentrations were also correlated with 15N recovery from the litter of defoliated trees in the growing season following the defoliation event. Conversely, tannins were not well correlated with nitrogen uptake under conditions of nominal herbivory. Our results suggest that tannins may confer benefits in response to herbivore pressure through conserved belowground nitrogen cycling, rather than via defensive properties. Consequently, tannins may be considered as chemical mediators of tolerance rather than resistance.
Introduction Tannins comprise a wide variety of high molecular mass polyphenolics that can constitute large portions of leaf and litter dry mass (Cadisch & Giller, 1997). They are defined, in part, by their ability to bind proteins (Bate-Smith, 1975). Historically considered as herbivore defense compounds, they have also been investigated for their effects on belowground nutrient cycling (Northup et al., 1995, 1998; Schimel et al., 1998; Kraus et al., 2003; Schweitzer et al., 2004; Meier & Bowman, 2008; Joanisse et al., 2009; H€attenschwiler et al., 2011). Variation in tannin production is often tied tightly to nutrient availability, with low nutrient availability during plant growth correlated with high leaf condensed tannin production (Cadisch & Giller, 1997). Tannin production can vary a hundred-fold both among and within species (Dalzell & Shelton, 2002), and much of this variation is genetically mediated (Klaper et al., 2001; Donaldson & Lindroth, 2007). Although tannins constitute significant portions of green leaf biomass, they are frequently ineffective as defensive compounds, particularly against insect herbivores (Ayres et al., 1997; Barbehenn & Constabel, 2011). Tannins vary widely in chemical structure, with important functional consequences. For instance, 410 New Phytologist (2015) 208: 410–420 www.newphytologist.com
hydrolysable tannins may defend against insect herbivores while condensed tannins are rendered ineffective by high-pH conditions inside insect guts (Barbehenn et al., 2006). Trembling aspen (Populus tremuloides) plants contain condensed tannins, and, although tannins can comprise nearly one-quarter of leaf biomass, they do not negatively affect the performance of aspenfeeding Lepidoptera (Lindroth & St Clair, 2013). Nonetheless, despite being ineffective against Lepidopteran herbivore attacks, condensed tannins are often up-regulated in aspen following foliar damage (Osier & Lindroth, 2001, 2004; Stevens et al., 2007). Moreover, tannin production comes at a cost: tree growth rates are negatively correlated with condensed tannin production in Populus species (Kosola et al., 2004; Stevens et al., 2007). Why would aspen invest in expensive ‘defensive’ tannin compounds that do not defend well? Tannins may be produced as a means through which plants exert control over nutrient cycling. Northup et al. (1995, 1998) argued that polyphenolics facilitate nitrogen (N) recovery in some ecosystems. Tannins may, therefore, be produced primarily for their ‘afterlife’ effects on litter decomposition and nutrient cycling (H€attenschwiler & Vitousek, 2000; Kraus et al., 2003; Joanisse et al., 2009). Tannins have the potential to influence nutrient cycling through their effects on litter decomposition. Tannins generally retard decomposition Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
New Phytologist and N mineralization, either via direct microbial toxicity or by forming recalcitrant complexes with organic N (Cadisch & Giller, 1997; Kraus et al., 2003). These ‘afterlife’ effects are not inconsequential. For instance, the majority of dissolved organic N in forest soils can be tied up in protein–tannin complexes (Wu, 2011). Verkaik et al. (2006) demonstrated that leaf litter with high tannin concentrations contributes significantly to longterm N storage in soils. Protein–tannin complexes have generally been considered inaccessible to plants and most microbes (H€attenschwiler & Vitousek, 2000). However, a growing body of evidence suggests that N stored in protein–tannin complexes can be used as a temporary storage pool for later access by plants. For example, plant species that produce N-binding tannins have associations with mycorrhizal fungi that can grow well on tanninamended substrates (Joanisse et al., 2009), and some mycorrhizas can make organic N sources directly accessible to plant symbionts (Wurzburger & Hendrick, 2009). H€attenschwiler et al. (2011) argued that tropical decomposition proceeds much more slowly than expected given temperature and moisture conditions. In tropical environments, tannins may retard decomposition as a means to combat nutrient scavenging by microbes or nutrient loss via leaching, and thereby decelerate nutrient cycling to a rate better suited for plant, rather than microbial, uptake. Following insect herbivory, tannins can influence belowground processes through frass and litter decomposition, both of which are important pathways for forest nutrient cycling (Lovett et al., 2002). For example, insect herbivory can induce high tannin production in damaged leaves, neighboring undamaged leaves, and reflushed leaves within the same growing season (Osier & Lindroth, 2001; Frost & Hunter, 2008; K. F. Rubert-Nason et al., unpublished) as well as in new leaf growth during subsequent growing seasons (Osier & Lindroth, 2004). Elevated concentrations of tannins persist until senescence and the resulting high-tannin leaf litter decomposes more slowly than does leaf litter from undefoliated trees (Findlay et al., 1996; Schweitzer et al., 2005). In addition, frass chemistry mirrors green leaf chemistry and can induce microbial respiration accordingly (Madritch et al., 2007a). Consequently, variation in plant chemistry, and tannins in particular, can influence both the ‘slow’, litter-driven, and ‘fast’, frass-driven, nutrient cycles (sensu Lovett & Ruesink, 1995). We explore the possibility that genetically determined variation in tannin production may provide significant benefits by facilitating N recovery from insect frass and senesced leaf litter. We employed a mesocosm experiment to determine how variation in leaf tannin concentrations influences both the slow (litter-based) and fast (frass-based) N cycles in the presence and absence of severe herbivory. Using trembling aspen as a model species, we manipulated both the genetic and environmental contributors to tannin plasticity, and subsequently evaluated the consequences of variation in tannin concentration for plant N uptake. We used trembling aspen as a model species because tannin concentration is highly variable, with both plant genotype and nutrient availability strongly affecting tannin concentrations (Osier & Lindroth, 2004; Madritch et al., 2006). We hypothesized that trees with high tannin concentrations would recover Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
higher amounts of labeled N from frass and litter than would trees with low tannin concentrations.
Materials and Methods Aspen propagation and treatments Trembling aspen (Populus tremuloides Michx.) is a widespread, genetically diverse, and frequently dominant tree species throughout northern and western North America (Mitton & Grant, 1996). We created single-tree mesocosms that varied by genotype, nutrient availability, and defoliation. We limited our design to a single tree species to control for confounding effects of species identity on tannin quality, as both the amount and chemical structure of tannins can vary widely (Barbehenn et al., 2006). We employed multiple genotypes of aspen because, while the effects of nutrient availability on tannin concentrations can be large, no one single genotype is sufficient for representing the extent of natural variation in tannin concentrations; that is, both genotype and nutrient availability are important drivers of variation in foliar chemistry, and our goal was to manipulate both in order to create a range of leaf tannin concentrations representative of trees in natural environments. Aspen trees were propagated following the protocols of Donaldson & Lindroth (2004). We chose four aspen genotypes based on plant secondary chemistry from a set of 12 genotypes collected randomly from south-central Wisconsin, USA. In early May 2006, 1-yr-old trees were transplanted into 38-l pots filled with 30 l of a 40 : 40 : 20 mix of torpedo sand, silt-loam soil, and perlite. Potted trees were grown outside in Madison, WI, USA, and watered as necessary. Trees were subjected to fertilization treatments with 18–6–12 (N–P–K + micronutrients) Osmocote (Scotts Co., Marysville, OH, USA) 8–9-month slow-release fertilizer at low, medium and high rates (0, 2 and 4.5 g l1 soil, corresponding to 0, 10.8 and 24.3 g N per tree, respectively). We selected these fertilization rates because they have elicited large variation in condensed tannin production in previous studies (Hemming & Lindroth, 1999; Donaldson & Lindroth, 2007). Osmocote applications were repeated yearly in late April to early May. Herbivory treatments were imposed by applying third- and fourth-instar gypsy moths (Lymantria dispar) to individual trees enclosed with fine mesh. Gypsy moth larvae were reared on gypsy moth diet (MP Biomedicals, Santa Ana, CA, USA) in the laboratory, and were then applied at numbers sufficient to defoliate the entire canopy, as our goal was to simulate severe herbivory conditions. Therefore, some large and/or well-defended trees required many gypsy moth larvae in order to cause near-total defoliation (> 95%). Mesh enclosures served to contain gypsy moths as well as to collect gypsy moth frass. Nondefoliated trees experienced < 5% damage from natural herbivory. Frass and litter transplants To separate the ‘fast’ and ‘slow’ nutrient cycles, we used stable N isotopes to track N contained in insect frass and in leaf litter (Christenson et al., 2002). We employed 15N-labeled ‘donor’ New Phytologist (2015) 208: 410–420 www.newphytologist.com
412 Research 15
N-labeled frass from the donor trees into their corresponding herbivory-frass recipient mesocosms in amounts equal to 75% (by mass per tree) of frass removed from the recipient mesocosms (mean frass added: 7.8 0.6 g of frass). Because donor trees did not consistently produce the same amount of frass as did recipient trees under identical treatment conditions, we used 75% of the frass produced by the herbivory-frass recipient trees as a standard application rate across all treatments. We allowed the mass of frass transplants to vary according to genotype and fertilization treatment in order to reflect in situ variation in net primary productivity, plant chemistry, and herbivory. Because frass and litter chemistry reflects variation in green leaf chemistry (Madritch et al., 2007a), higher tannin trees probably produce frass that has higher amounts of tannin-bound N (adding a uniform mass of frass across all genotype and nutrient treatments would have ignored variation known to be important under natural conditions). A similar procedure was carried out with 15N-labeled litter. Defoliation treatments were carried out early in the growing season, allowing sufficient time for leaf reflush to occur. During leaf senescence, we used mesh enclosures to collect leaf litter from both the donor and recipient mesocosms. Litter was transplanted from the 15N donor mesocosms to the herbivory-litter recipient mesocosms according to 75% of the mass of litter actually produced by the recipient trees (mean litter added: 20.5 1.2 g). ‘Herbivory-litter’ treatment trees received reflushed, 15N-labeled litter from defoliated trees. ‘Undefoliated-litter’ treatment trees received 15N-labeled litter from undefoliated donor trees (Fig. 1). Fig. 1 Schematic of the experimental design, showing how ‘herbivoryfrass’, ‘herbivory-litter’ and ‘undefoliated-litter’ treatments were created. Donor trees with identical genotype, herbivory, and nutrient treatments supplied 15N-labeled frass and litter for recipient mesocosms. Defoliated treatments were divided into frass and litter treatments to separately track fast and slow nitrogen cycles, respectively. Undefoliated treatments supplied 15N-labeled litter to recipient mesocosms.
trees to generate the frass and litter for distribution to unlabeled ‘recipient’ trees. Four genotypes of donor trees varied by three levels of nutrient availability and two levels of herbivory (nominal and outbreak): 4 genotypes 9 3 nutrient treatments 9 2 herbivory treatments 9 3–4 replicates per treatment combination = 87 donor trees. We created three sets of nonisotopically labeled recipient trees with identical fertilization and genotype treatments as those experienced by donor trees: 4 genotypes 9 3 fertilization treatments 9 3–4 replicates = c. 40 trees in each of three sets, for a total of 120 recipient trees. Two of these three sets experienced herbivory treatments identical to those of the defoliated donor trees. Each set of recipient trees was allocated to a different recipient treatment (Fig. 1) and received: (1) 15N-labeled frass from the defoliated donor trees and their own, unlabeled litter (herbivory-frass treatment), (2) 15N-labeled litter from the defoliated donor trees and their own, unlabeled frass (herbivory-litter treatment), or (3) 15N-labeled litter from the undefoliated donor trees and no frass (undefoliated treatment). We collected, weighed, and discarded frass produced by larvae in the herbivory-frass recipient mesocosms. We then transplanted New Phytologist (2015) 208: 410–420 www.newphytologist.com
N labeling and mesocosm sampling schedule
We labeled each donor tree in May 2007 using 500 ml of 0.25 g l1 K15NO3 (99% enriched) and applied identical unlabeled KNO3 treatments to recipient trees. We collected 12–15 leaves from each tree for chemical analysis in June 2007, immediately before gypsy moth application. Removing 12–15 leaves had little impact on the canopy as trees were 2 yr old at the time of collection (mean number of leaves per tree: 473 15). Gypsy moth defoliation treatments were continued until July in order to effect near-total defoliation. Frass was collected periodically throughout the defoliation period to prevent leaching during rainfall events, and transferred to recipient mesocosms in July of the same growing season. This timing of frass deposition approximates that experienced by trees during outbreaks of early-season defoliators such as gypsy moths. We sampled leaf litter at the end of the 2007 (‘year 1’) growing season. Leaf litter from all treatments was sorted and re-deployed in November–December of 2007. We sampled green leaves again in June 2008 (‘year 2’), after leaves had fully expanded. Chemical analyses Soluble condensed tannin concentrations (% dry mass) of leaves and litter were determined as in Madritch et al. (2007a) by extracting plant material with 70 : 30 acetone : water (containing 10 mM ascorbic acid) and assaying extracts via the N-butanol Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
New Phytologist method of Porter et al. (1986). Aspen tannin standards were purified by the method of Hagerman & Butler (1989). Leaves and litter were analyzed for N and 15N by the Colorado Plateau Stable Isotope Laboratory (Flagstaff, AZ, USA). We report 15N values as d15N where:
Research 413 (a)
d15 Nsample ¼ ðð15 N=14 Nsample Þ=ð15 N=14 Nair Þ 1Þ 1000 Eqn 1 We report d15N rather than Atm%N because of the low levels of 15N enrichment in our focal tissue. Statistical analyses Data were transformed as necessary to meet assumptions of normality. For clarity, we display nontransformed data in figures, but report statistics for transformed data. We employed analysis of variance (ANOVA) procedures to reveal treatment effects. We used genotype and nutrient availability, and their interaction, to identify treatment effects on tannin and N concentrations before defoliation treatments. Post-defoliation models that describe leaf tannin and N concentrations included genotype, nutrient availability, defoliation, and all possible interactions. Post-defoliation models describing frass N concentrations necessarily omitted defoliation effect terms as no frass was generated from undefoliated treatments. To separate the fast and slow cycles, we also performed ANOVA on post-defoliation N values using genotype, nutrient availability, and herbivory-frass/herbivory-litter/undefoliated-litter as the main effects and included all interactions. Herbivory-frass/herbivory-litter/undefoliated-litter refers to whether recipient mesocosms received labeled frass from defoliated trees, labeled litter from defoliated trees, or labeled litter from undefoliated trees (see Fig. 1). We used simple correlations when estimating the relationship between N recovery and tannin concentration across and within nutrient treatments. Our analysis of fast-cycle recovery of 15N was limited to herbivore-frass treatments, as no 15N-labeled litter from donor mesocosms was deployed to herbivory-frass mesocosms. We used 1st-year litter d15N for the recipient mesocosms as an indicator of within-season recovery of N from frass during the first growing season. As a result of phenological variation among treatments, calendar dates for the end of the growing season varied widely among individual trees. Consequently, we used freshly fallen litter as a definitive end of the 1st-year growing season. We evaluated slow-cycle recovery of 15 N that occurred across growing seasons through the undefoliated and defoliated litter treatments by measuring green leaf d15N of trees in recipient mesocosms in the 2nd year of the experiment, after frass and litter additions. We focus in this report on d15N values; we could not calculate 15N recovery rates as all trees in our experimental design received some source of labeled 15N, either as fertilizer, frass, or litter. In all analyses, we consider the single-tree mesocosms to be the unit of replication. P values < 0.05 are considered significant. All statistical analyses were performed with JMP 9 (SAS Institute, Cary, NC, USA). Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
Fig. 2 Green leaf tannin concentrations. (a) Condensed tannin concentrations in green leaves were influenced by both aspen (Populus tremuloides) genotype and nutrient availability treatments before herbivory treatments in the first year of the experiment (n = 119 recipient mesocosms). (b) Main effects of aspen genotype, herbivory, and nutrient availability on leaf tannin levels in the second year of the experiment, 1 yr after defoliation for recipient mesocosms only (n = 94 recipient mesocosms; 25 trees lost to overwintering mortality). ‘Low’, ‘medium’, and ‘high’ labels refer to the nutrient treatment. Bars represent means ( 1 SE). ns, not significant.
Results As expected, condensed tannin concentrations varied among aspen genotypes and decreased with increasing nutrient availability in recipient mesocosms in the 1st year before defoliation treatments (Fig. 2a). In the 2nd year, aspen genotype, nutrient availability, and defoliation, but not their interactions, contributed to variation in condensed tannin concentrations (Fig. 2b). Low nutrient availability correlates with high leaf tannin concentrations in aspen (Osier & Lindroth, 2001, 2006). Important to our subsequent analysis and interpretation is the fact that relative New Phytologist (2015) 208: 410–420 www.newphytologist.com
variation of leaf tannin concentrations among genotypes was much higher under high nutrient availability (five-fold) than under low nutrient availability (two-fold) (Fig. 2a). Before herbivory treatments and early in the 1st-year growing season, green leaf d15N and N both varied widely within treatments for recipient trees, and the only significant treatment effect was that of nutrient availability on leaf N (Fig. 3a,b; Table 1). Conversely, aspen genotype and nutrient availability influenced both frass d15N and, to a lesser extent, frass N in donor trees (Fig. 3c,d; Table 1). By the end of the first growing season, aspen genotype and nutrient availability influenced both litter d15N and N in donor trees (Fig. 3e,f). For both litter and frass, high nutrient availability decreased the concentration of 15N, probably by diluting 15N with increased uptake of total mineral N from applied fertilizer. In the growing season following defoliation and frass and/or litter applications (i.e. the 2nd year), green leaf d15N was
New Phytologist (2015) 208: 410–420 www.newphytologist.com
influenced by aspen genotype, nutrient availability, and herbivory-frass/herbivory-litter/undefoliated-litter application (Fig. 4; Table 2). Recipient mesocosms that received labeled frass resulted in green leaves with higher levels of d15N than did treatments that received labeled litter (Fig. 4). In addition, low-nutrient treatments generally had higher d15N than did higher nutrient treatments (Fig. 4). Green leaf d15N values varied widely with genotype and defoliation treatments, with the herbivory-frass treatment of genotype W1 having green leaf d15N values double that of any other treatment (Fig. 4; note difference in y-axis scale). Given that variation in condensed tannin concentrations were explained by genotype, nutrient availability, and defoliation treatments in the 2nd year (Fig. 2), we evaluated correlations between green leaf d15N and condensed tannin concentrations. However, because fertilization decreased both tannin production and green leaf d15N (Figs 2, 3), we performed correlations both across and
Fig. 3 Leaf, frass, and litter d15N and nitrogen (N) values varied with aspen (Populus tremuloides) genotype and nutrient availability. (a, c, e) d15N values; (b, d, f) N values. (a, b) Data from all recipient mesocosms before herbivory treatments were applied (n = 120). (c, d) Data from defoliated donor mesocosms that were used to provide frass for recipient mesocosms (n = 40). (e, f). Data from donor mesocosms that provided litter for recipient mesocosms (averaged across herbivory and undefoliated treatments; n = 80). ‘Low’, ‘medium’, and ‘high’ labels refer to the nutrient treatment. Bars represent means ( 1 SE). See Table 1 for ANOVA model results. Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
New Phytologist Table 1 Statistical results corresponding with Fig. 3 describing the effects of aspen (Populus tremuloides) genotype, nutrient availability, and their interactions on initial leaf nitrogen in recipient mesocosms before defoliation treatment (Fig. 3a,b), and on frass nitrogen from defoliated trees (Fig. 3c,d); and the effects of aspen genotype, nutrient availability, defoliation, and their interaction on litter chemistry in donor mesocosms following defoliation treatments (Fig. 3e,f)
within each nutrient treatment. Green leaf N was not correlated with tannin concentrations early in the 1st year’s growing season (Fig. 5a,b). Within-season 15N recovery from frass, as indicated by late 1st-year leaf litter d15N (see the Materials and Methods section), was correlated with foliar condensed tannin concentrations that were measured earlier in the same growing season in the high-nutrient treatments (Fig. 5d), and marginally so in the low-nutrient treatments (Fig. 5c). Conversely, green leaf tannin was not correlated with litter d15N in the medium-nutrient treatment or when considered across all nutrient treatments (data not shown). The relationships between litter N and green leaf tannin paralleled those of litter d15N and green leaf tannin (Fig. 5e,f). These data indicate that the positive relationship between lateseason leaf N concentration and plant tannin production was attributable to increased within-season uptake of N from frass by high-tannin genotypes. In the 2nd year, 1 yr following the addition of frass and litter that occurred in the 1st year, green leaf d15N concentrations were positively correlated, albeit weakly, with condensed tannin concentrations across all nutrient and defoliation treatments in recipient mesocosms (r = 0.28; P = 0.006; data not shown). When considered by herbivory-frass/herbivory-litter/undefoliated-litter treatments, green leaf d15N values were positively correlated with tannin concentrations in herbivory-frass and herbivory-litter treatments across all nutrient levels (Fig. 6a,c, respectively). The correlation between green leaf d15N and tannin in herbivore treatments was strongest under high nutrient availability for both herbivory-frass (Fig. 6b) and herbivory-litter treatments (Fig. 6d). In undefoliated litter treatments, however, green leaf d15N values were not correlated with condensed tannin concentrations (Fig. 6e,f). Despite the correlations between leaf tannin and green leaf d15N, we did not find positive correlations between leaf tannin concentration and total leaf N in the 2nd year of the experiment. Rather, and as expected, 2nd-year leaf tannin concentrations were negatively correlated with N concentrations when considered across all nutrient treatments: r = 0.84 (P < 0.001) for defoliated frass treatments, r = 0.88 (P < 0.001) for defoliated litter Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
treatments, and r= 0.69 (P < 0.001) for undefoliated litter treatments (data not shown). Leaf tannin and N values are generally negatively correlated with each other in aspen (Osier & Lindroth, 2001, 2006); as N availability declines, tannin production increases.
Discussion We show that variation in tannin concentration was determined by plant genotype, nutrient availability, and herbivory; and that green leaf d15N after severe herbivory was positively correlated with tannin production. Green leaf d15N and tannin concentrations were positively correlated within the same growing season, and during the 2nd year of growth after the defoliation event. Furthermore, the positive relationship between tannin concentration and leaf d15N persisted through both the fast (frass-based) and slow (litter-based) N cycles. Under nominal herbivory, the patterns were weaker but still suggestive of positive relationship between d15N and foliar tannin concentrations. These results show that condensed tannins can provide important nutrient cycling benefits to plants and that these benefits are larger after severe herbivory than under nominal herbivory. Despite the high d15N values of frass and litter (Fig. 3), the 15 d N values of post-treatment green leaves were relatively low (Figs 4–6). Leaves often have negative d15N values and the recovery rates of 15N-labeled fertilizer into tree foliage are frequently low (Nadelhoffer et al., 1999). For example, in a similar mesocosm experiment with slightly smaller pots, Frost & Hunter (2007) found that most 15N applied as labeled frass was retained by the soil, and only c. 10% was recovered in leaves. Given the large soil volume we used, it is not surprising that our leaf d15N values were low. Nonetheless, they are indicative of N transfers from frass and litter, through the soil–root interface, and into leaf tissue. Nutrient availability, defoliation, and their interaction had the largest effects on green leaf d15N values. d15N was highest under low nutrient availability herbivory-frass treatments. The addition of 15N was minor compared with the addition of N via New Phytologist (2015) 208: 410–420 www.newphytologist.com
Table 2 ANOVA results describing the effects of aspen (Populus tremuloides) genotype, nutrient availability, frass/litter/undefoliated litter application, and their interaction on leaf 15N content in the following growing season
Fig. 4 Leaf d15N values across genotypes 1 yr following herbivory treatments and subsequent 15N-labeled frass and litter application. Alphanumeric codes refer to four different aspen (Populus tremuloides) genotypes. ‘Low’, ‘medium’, and ‘high’ x-axis labels refer to nutrient availability treatments. Herbivory-frass, recipient mesocosms that received 15N-labeled frass from defoliated donor trees; Herbivory-litter, recipient mesocosms that received 15 N-labeled litter (and no labeled frass) from defoliated donor trees; Undefoliated-litter, recipient mesocosms that received 15N-labeled litter from undefoliated donor trees. Bars represent means ( 1 SE). See Table 2 for ANOVA model results. Note different y-axis scales.
New Phytologist (2015) 208: 410–420 www.newphytologist.com
fertilization, and nonlabeled trees received identical amounts of unlabeled KNO3. Thus, the increase in d15N by low-nutrient trees was not attributable to a 15N fertilization effect. Rather, trees under low-nutrient conditions probably scavenged organic N from labeled frass and litter, resulting in higher 15N sequestration. Conversely, the organically bound 15N in frass and litter was probably less important to fertilized trees that had relatively large amounts of readily available mineral N. Trees recovered more 15N from frass treatments than they did from litter treatments probably because N in frass was relatively labile compared with N in litter (Lovett et al., 2002). Genetically mediated variation in tannin concentrations persisted across different nutrient availability treatments; genotypes that produced high amounts of tannins under high nutrient availability also produced high amounts of tannins under low nutrient availability. The cost of tannin production (Donaldson et al., 2006a) may be offset by the ability to resorb N following defoliation events. Although tannin concentrations were positively correlated with green leaf d15N values, final tannin concentrations were not correlated with final N concentrations because of the large negative effect of nutrient availability on tannin production. In our experiment, decreasing nutrient availability increased both tannin production and d15N. The negative effect of nutrient availability on condensed tannin production is well documented in aspen, and is probably a consequence of several mass balance mechanisms (Osier & Lindroth, 2006). In our study, there was also an apparent negative effect of nutrient availability on 15N uptake; under high nutrient availability plants may preferentially take up mineral N forms rather than the 15N that is bound in organic frass or litter pools. Consequently, positive relationships between tannin and d15N values across nutrient treatments may have been solely an unintended consequence of our nutrient treatments. We addressed this potentially confounding relationship by performing correlation analyses between tannin and 15N within nutrient treatments. Within the high-nutrient treatment, variation in leaf tannin concentration was not driven by variation in nutrient availability and was positively correlated with 15N Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
Fig. 5 Correlations of leaf and litter nitrogen (N) with leaf tannin concentrations for defoliation treatments that received labeled frass in year 1 (‘herbivory-frass’ treatments). (a, c, e) Data from low nutrient availability mesocosms; (b, d, f) data from high nutrient availability mesocosms. (a, b) Early-season, pre-herbivory leaf N data. (c–f) Postherbivory, end of year 1 growing season data are represented by (c, d) litter d15N and (e, f) litter N. Each point represents a mesocosm. Dashed lines represent 95% confidence intervals. Note the different x-axis scales between the left and right columns, indicating high tannin production under low nutrient availability. ns, not significant.
recovery. In the field, we would expect leaf tannin and N concentrations to be negatively correlated, because soil nutrient availability varies across landscapes and can drive variation in condensed tannin production (Lindroth & St Clair, 2013). Thus, our data suggest that, second to the constraints on tannin production imposed by nutrient availability, genetically mediated increases in tannin production may increase N recovery even if they do not increase final N concentration. Several non-mutually exclusive mechanisms may explain how tannins increase N uptake. First, tannins may bind to nitrogenous substrates and thereby retain N that would otherwise be lost from the system via leaching (Bradely et al., 2000). Second, tannins may form complexes with organic N that can then be cycled directly back to the plant via mycorrhizal symbionts (Northup et al., 1995). Species that produce large amounts of tannins often associate with mycorrhizal symbionts that provide access to tannin-bound N (Joanisse et al., 2009; Wurzburger & Hendrick, 2009). Trembling aspen forms relationships predominantly with ectomycorrhizal Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
fungi (Cripps & Miller, 1993), yet it is unclear if aspen– mycorrhizal mutualisms are capable of acquiring N directly from protein–tannin complexes. Third, some tannin fractions may prime decomposition and N mineralization (H€attenschwiler et al., 2011). Simple tannin fractions can stimulate microbial activity, presumably by serving as a carbon (C) substrate, while larger fractions tend to retard decomposition and bind N (Kraus et al., 2004; Madritch et al., 2007b; Meier & Bowman, 2008). Fischer et al. (2006) demonstrated that fine-root production in Populus is correlated with tannin production, and may be a physiological response to recover tannin-bound N. Variation in green leaf d15N was more strongly correlated with tannin concentrations when trees were grown under high nutrient availability than when trees were grown under low nutrient availability. The relative variation of leaf tannin concentration within a fertilization treatment is determined by the genetic identity of the genotype (Osier & Lindroth, 2001). For instance, the variation in tannin concentration was greatest under high New Phytologist (2015) 208: 410–420 www.newphytologist.com
418 Research (a)
nutrient availability; tannins varied roughly two-fold by genotype under low nutrient availability, and nearly five-fold under high nutrient availability (Fig. 2a). The high variation in tannin concentrations under high nutrient availability corresponded with high variation in d15N. Under low nutrient availability, low variance in tannin concentrations precluded a relationship between d15N and tannin concentration. The lack of correlation between tannin and d15N does not necessarily indicate that tannins are unimportant to N recovery under low-nutrient conditions, but is rather a limitation of our experimental design: we were unable to produce large amounts of variation in leaf tannin concentrations under low-nutrient conditions. Nonetheless, genetically determined high tannin production may mitigate risks associated with growing under high-nutrient conditions, as Donaldson & Lindroth (2007) documented that gypsy moth larvae preferentially feed upon, and cause more damage to, high-nutrient trees. While the absolute amount of change in d15N shown in Fig. 6(b) is small, the relative change is large: a doubling of d15N corresponds to a 7% (dry weight) increase in leaf tannin concentration. The plasticity of tannin production, and the associated New Phytologist (2015) 208: 410–420 www.newphytologist.com
Fig. 6 Correlations of leaf 15N with leaf tannin concentrations in the year following defoliation treatments in recipient mesocosms across all fertilization levels. (a, c, e) Data from all nutrient availability treatments; (b, d, f) data from high nutrient availability mesocosms only. Leaf d15N was correlated with condensed tannin concentrations, indicating increased recovery of labeled nitrogen from both insect frass and leaf litter. Each point represents a mesocosm. Dashed lines represent 95% confidence intervals. Note the different y-axis scales. ns, not significant.
consequences for belowground N cycling, may be partially responsible for aspen’s ability to colonize and thrive under a wide range of environmental conditions. Herbivore frass inputs can constitute a labile, fast-cycling N source for belowground systems (Lovett et al., 2002). In our mesocosm experiment, tannin concentration was correlated with increased green leaf d15N values in the same year as well as in the year following defoliation. Frost & Hunter (2007) demonstrated that N contained in frass can be recycled back into leaf foliage within 1 month after defoliation. Herbivores such as gypsy moths, forest tent caterpillars, and white-marked tussock moths produce frass that mirrors variation in green leaf C : N and tannin concentration (Madritch et al., 2007a; Couture & Lindroth, 2014). The tannins that remain in insect frass may bind otherwise labile N and retain it for subsequent plant uptake. Leaf tannin concentrations were correlated with green leaf d15N values during the 2nd year, 1 yr following the defoliation event when labeled litter was transplanted. The positive correlation between d15N and leaf tannin concentrations in the defoliated litter treatments (Fig. 6c,d) demonstrates that tannins Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
New Phytologist influence the N cycle associated with the slow, litter-based decomposition cycle, in addition to the fast, frass-based N cycle as discussed above and shown in Figs 4 and 6(b). Tannins may have facilitated 15N recovery in the litter-based cycle through the independent or interactive effects of N retention, direct N cycling, and/or priming. Neither the frass- nor the litter-based N cycling necessitates mycorrhizal-facilitated N uptake. That potential, however, seems likely as mycorrhizal symbionts enable direct N cycling of tannin–N complexes (Wurzburger & Hendrick, 2009). Importantly, it remains unresolved from our study whether tannin production improved N recovery specifically by aspen, or simply N recovery by plants in general, as each pot contained only one individual aspen tree. While the mechanisms occurring in our system are unclear, the empirical evidence of increased d15N with increased tannin production suggests a potential benefit of tannin production realized through belowground nutrient cycling pathways. Finally, this research provides a new perspective on the role of secondary metabolites in plant defense. The two principal strategies of defense employed by plants are resistance (mechanisms that reduce damage) and tolerance (mechanisms that mitigate the negative effects of damage). Secondary metabolites have long been accorded roles as quintessential resistance traits. This study, however, reveals that some secondary compounds, such as condensed tannins, may also be considered as tolerance traits, as they may contribute to plant recovery following damage. That is, high tannin concentrations may benefit plants under herbivore pressure not through defense mechanisms, but rather through N cycling mechanisms. Our research covered only two growing seasons, and the effects of tannin on belowground nutrient cycling may become additive over longer time periods because tannin– protein complexes can persist for many years and because tannin concentrations in mature trees are generally higher than those of trees in this study (Donaldson et al., 2006b). The ontogenetic shift in aspen chemistry from phenolic glycosides to tannins (Donaldson et al., 2006b) may reflect a switch in herbivore defense strategy from resistance to tolerance. Consequently, the nutrient retention benefits of tannins may be especially important in long-lived, clonal species such as aspen, as older and larger clones are more likely to reap the nutrient cycling benefits of tannins than are small, isolated trees. How important tannins may be to N recovery in nonclonal tree species, where litter is often deposited out of the rooting zone, remains unclear. Nonetheless, tannins may still provide a benefit to N recovery if fine-root decomposition dynamics follow those of leaf litter dynamics, as fine-root tannin concentration can also vary widely (Kosola et al., 2004). Conclusions Tannins may provide benefits to plants through independent and interactive pathways of herbivore defense and N cycling. In aspen, condensed tannins are probably costly to synthesize and provide little defense against major lepidopteran defoliators (Osier & Lindroth, 2001), and, in outbreak years, genotypes that vary widely in tannin concentrations are all completely Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
defoliated by canopy herbivores (Donaldson & Lindroth, 2008). Consequently, up-regulation of tannins in Populus after exposure to herbivory (Osier & Lindroth, 2001, 2004; Stevens et al., 2007; Constabel & Lindroth, 2010) seems counterintuitive. However, condensed tannins may benefit plants under extreme herbivory not via defense chemistry, but rather through improved nutrient cycling; that is, tannins are up-regulated by defoliation, and genetically mediated variation in tannin production persists, because high-tannin phenotypes recover more N after defoliation than do low-tannin phenotypes. Thus, leaf tannins may serve roles as tolerance, rather than resistance, factors. These results support the hypothesis that the belowground effects of tannins may be more important than their green leaf effects (Joanisse et al., 2009; Coq et al., 2010; H€attenschwiler et al., 2011). Furthermore, the belowground effects of tannins may be one mechanism whereby foundation forest species such as aspen can create spatial mosaics of genetically mediated ecosystem processes (Madritch & Lindroth, 2011).
Acknowledgements Funding was provided by NSF DEB-0344019 to R.L.L. and M.D.M., NASA Biodiversity NNX09AK15G to M.D.M. and R.L.L., and NSF DEB-0841609 to R.L.L. A. Gusse and K. F. Rubert-Nason provided laboratory and field assistance. We thank Stephan H€attenschwiler, Amy Austin, and two anonymous reviewers for suggestions that improved the manuscript.
References Ayres MP, Clausen TP, MacLean SF Jr, Redman AM, Reichardt PB. 1997. Diversity of structure and antiherbivore activity in condensed tannins. Ecology 78: 1696–1712. Barbehenn RV, Constabel CP. 2011. Tannins in plant-herbivore interactions. Phytochemistry 72: 1551–1565. Barbehenn RV, Jones CP, Haberman AE, Karonen M, Salminen J-P. 2006. Ellagitannins have greater oxidative activities than condensed tannins and galloyl glucoses at high pH: potential impact on caterpillars. Journal of Chemical Ecology 32: 2253–2267. Bate-Smith EC. 1975. Phytochemistry of proanthrocyanidins. Phytochemistry 14: 1107–1113. Bradely RL, Titus BD, Preston CP. 2000. Changes to mineral N cycling and microbial communities in black spruce humus after additions of (NH4)2SO4 and condensed tannins extracted from Kalmia angustifolia and balsam fir. Soil Biology and Biochemistry 32: 1227–1240. Cadisch G, Giller KE. 1997. Driven by nature: plant litter quality and decomposition. Wallingford, UK: CAB International. Christenson LM, Lovett GM, Mitchell MJ, Groffman PM. 2002. The fate of nitrogen in gypsy moth frass deposited to an oak forest floor. Oecologia 131: 444–452. Constabel PC, Lindroth RL. 2010. The impact of genomics on advances in herbivore defense and secondary metabolism in Populus. In: Jansson S, Bhalerao R, Groover A, eds. The genetics and genomics of Populus. Plant Genetics and Genomics, vol. 8. New York, NY, USA: Springer-Verlag, Inc., 279–305. Coq S, Souquet J-M, Meudec E, Cheynier V, H€a ttenschwiler S. 2010. Interspecific variation in leaf litter tannins drives decomposition in a tropical rain forest of French Guiana. Ecology 91: 2080–2091. Couture JJ,Lindroth RL. 2014. Atmospheric change alters frass quality of forest canopy herbivores. Arthropod-Plant Interactions 8: 33–47. New Phytologist (2015) 208: 410–420 www.newphytologist.com
420 Research Cripps C, Miller OK. 1993. Ectomycorrhizal fungi associated with aspen on 3 sites in the north-central Rocky Mountains. Canadian Journal of Botany 71: 1414–1420. Dalzell SA, Shelton HM. 2002. Genotypic variation in proanthocyanidin status in the Leucaena genus. Journal of Agricultural Science 138: 209–220. Donaldson JR, Kruger EL, Lindroth RL. 2006a. Competition- and resourcemediated tradeoffs between growth and defensive chemistry in trembling aspen (Populus tremuloides). New Phytologist 169: 561–570. Donaldson JR, Lindroth RL. 2004. Cottonwood leaf beetle (Coleoptera: Chrysomelidae) performance in relation to variable phytochemistry in juvenile aspen (Populus tremuloides Michx.). Environmental Entomology 33: 1505–1511. Donaldson JR, Lindroth RL. 2007. Genetics, environment, and their interaction determine efficacy of chemical defense in trembling aspen. Ecology 88: 729–739. Donaldson JR, Stevens MT, Barnhill HR, Lindroth RL. 2006b. Age-related shifts in leaf chemistry of clonal aspen (Populus tremuloides). Journal of Chemical Ecology 32: 1415–1429. Findlay S, Carreiro M, Krischik V, Jones CG. 1996. Effects of damage to living plants on leaf litter quality. Ecological Applications 6: 269–275. Fischer DG, Hart SC, Rehill BJ, Lindroth RL, Keim P, Whitham TG. 2006. Do high-tannin leaves require more roots? Oecologia 149: 668–675. Frost CJ, Hunter MD. 2007. Recycling of nitrogen in herbivore feces: plant recovery, herbivore assimilation, soil retention, and leaching losses. Oecologia 151: 42–53. Frost CJ, Hunter MD. 2008. Insect herbivores and their frass affect Quercus rubra leaf quality and initial stages of subsequent litter decomposition. Oikos 117: 13–22. Hagerman AE, Butler CG. 1989. Choosing appropriate methods and standards for assaying tannin. Journal of Chemical Ecology 15: 1795–1810. H€a ttenschwiler S, Coq S, Barantal S, Handa IT. 2011. Leaf traits and decomposition in tropical rainforests: revisiting some commonly held views and towards a new hypothesis. New Phytologist 189: 950–965. H€a ttenschwiler S, Vitousek PM. 2000. The role of polyphenols in terrestrial ecosystem nutrient cycling. Trends in Ecology and Evolution 15: 238–243. Hemming JDC, Lindroth RL. 1999. Effects of light and nutrient availability on aspen: growth, phytochemistry, and insect performance. Journal of Chemical Ecology 25: 1687–1714. Joanisse GD, Bradley RL, Preston CM, Bending GD. 2009. Sequestration of soil nitrogen as tannin-protein complexes may improve the competitive ability of sheep laurel (Kalmia angustifolia) relative to black spruce (Picea mariana). New Phytologist 181: 187–198. Klaper R, Ritland K, Mousseau TA, Hunter MD. 2001. Heritability of phenolics in Quercus laevis inferred using molecular markers. Journal of Heredity 92: 421–426. Kosola KR, Dickmann DI, Hall RB, Workmaster BAA. 2004. Cottonwood growth rate and fine root condensed tannin concentration. Tree Physiology 24: 1063–1068. Kraus TEC, Dahlgren RA, Zasoski RJ. 2003. Tannins in nutrient dynamics of forest ecosystems – a review. Plant and Soil 256: 41–66. Kraus TEC, Zasoski RJ, Dahlgren RA, Horwath WR, Preston CM. 2004. Carbon and nitrogen dynamics in forest soil amended with purified tannins from different plant species. Soil Biology and Biochemistry 36: 309–321. Lindroth RL, St Clair SB. 2013. Adaptations of trembling aspen (Populus tremuloides Michx.) for defense against herbivores. Forest Ecology and Management 299: 14–21. Lovett GM, Christenson LM, Groffman PM, Jones CG, Hart JE, Mitchell MJ. 2002. Insect defoliation and nitrogen cycling in forests. BioScience 52: 335–341.
New Phytologist (2015) 208: 410–420 www.newphytologist.com
New Phytologist Lovett GM, Ruesink AE. 1995. Carbon and nitrogen mineralization from decomposing gypsy moth frass. Oecologia 104: 133–138. Madritch MD, Donaldson JR, Lindroth RL. 2006. Genetic identity of Populus tremuloides litter influences decomposition and nutrient release in a mixed forest stand. Ecosystems 9: 528–537. Madritch MD, Donaldson JR, Lindroth RL. 2007a. Canopy herbivory can mediate the influence of plant genotype on soil processes through frass deposition. Soil Biology and Biochemistry 39: 1192–1201. Madritch MD, Jordan LM, Lindroth RL. 2007b. Interactive effects of condensed tannin and cellulose additions on soil respiration. Canadian Journal of Forest Research 37: 2063–2067. Madritch MD, Lindroth RL. 2011. Soil microbial communities adapt to genetic variation in leaf litter inputs. Oikos 120: 1696–1704. Meier CL, Bowman WD. 2008. Phenolic-rich leaf carbon fractions differentially influence microbial respiration and plant growth. Oecologia 158: 95–107. Mitton JB, Grant MC. 1996. Genetic variation and the natural history of quaking aspen. BioScience 46: 25–31. Nadelhoffer KJ, Downs MR, Fry B. 1999. Sinks for 15N-enriched additions to an oak forest and a red pine plantation. Ecological Applications 9: 72–86. Northup RR, Dahlgren RA, McColl JG. 1998. Polyphenols as regulators of plant-litter-soil interactions in northern California’s pygmy forest: a positive feedback? Biogeochemistry 42: 189–220. Northup RR, Yu Z, Dahlgren RA, Vogt KA. 1995. Polyphenol control of nitrogen release from pine litter. Nature 377: 227–229. Osier TL, Lindroth RL. 2001. Effects of genotype, nutrient availability, and defoliation on aspen phytochemistry and insect performance. Journal of Chemical Ecology 27: 1289–1313. Osier TL, Lindroth RL. 2004. Long-term effects of defoliation on quaking aspen in relation to genotype and nutrient availability: plant growth, phytochemistry and insect performance. Oecologia 139: 55–65. Osier TL, Lindroth RL. 2006. Genotype and environment determine allocation to and costs of resistance in quaking aspen. Oecologia 148: 293–303. Porter LJ, Hrstich LN, Chan BC. 1986. The conversion of procyanidins and prodelphinidins to cyaniding and delphinidin. Phytochemistry 25: 223–230. Schimel J, Cates RG, Ruess R. 1998. The role of balsam poplar secondary chemicals in controlling soil nutrient dynamics through succession in the Alaskan taiga. Biogeochemistry 42: 221–234. Schweitzer JA, Bailey JK, Hart SC, Wimp GM, Chapman SK, Whitham TG. 2005. The interaction of plant genotype and herbivory decelerate leaf litter decomposition and alter nutrient dynamics. Oikos 110: 133–145. Schweitzer JA, Bailey JK, Rehill BJ, Martinsen GD, Hart SC, Lindroth RL, Keim P, Whitham TG. 2004. Genetically based trait in a dominant tree affects ecosystem processes. Ecology Letters 7: 127–134. Stevens MT, Waller DM, Lindroth RL. 2007. Resistance and tolerance in Populus tremuloides: genetic variation, costs, and environmental dependency. Evolutionary Ecology 21: 829–847. Verkaik E, Jongkind AG, Berendse F. 2006. Short-term and long-term effects of tannins on nitrogen mineralization and litter decomposition in kauri (Agathis australis (D.Don) Lindl.) forests. Plant and Soil 287: 337–345. Wu T. 2011. Can ectomycorrhizal fungi circumvent the nitrogen mineralization for plant nutrition in temperate forest ecosystems? Soil Biology and Biochemistry 43: 1109–1117. Wurzburger T, Hendrick RL. 2009. Plant litter chemistry and mycorrhizal roots promote a nitrogen feedback in a temperate forest. Journal of Ecology 97: 528– 536.
Ó 2015 The Authors New Phytologist Ó 2015 New Phytologist Trust
Condensed tannins increase nitrogen recovery by trees following insect defoliation.
While the importance of plant secondary metabolites to belowground functioning is gaining recognition, the perception remains that secondary metabolit...