Forest habitat, not leaf phenotype, predicts late-season folivory of Quercus alba saplings


Correspondence author. E-mail:


1. Natural sunlight gradients occur on multiple scales in space and time. However, the direct (via habitat) and indirect effects (via plant quality) of light environment are often confounded, obscuring the relative importance of each in influencing herbivore responses.

2. Potted saplings of Quercus alba (white oak, Fagaceae) were grown in either full sun or full shade from budburst to leaf hardening to manipulate leaf phenotype (creating sun or shade leaves), then placed in either sunny light gaps or adjacent shaded forest understorey habitats. This two-way factorial design isolated the effects of sunlight level during leaf expansion from light environment late in the growing season on leaf phenology, leaf traits associated with host plant quality for herbivores, herbivore density and folivory.

3. Sunlight level during leaf expansion and hardening had strong and persistent effects on Q. alba leaf phenology and phenotype. Shade saplings had later budburst (c. 4 days), and fewer but larger leaves, resulting in greater total leaf area compared with sun leaf saplings. Shade leaves had higher water content, specific leaf area and nitrogen content, and lower toughness, carbon content, C/N ratios and concentrations of hydrolysable and condensed tannins than sun leaves.

4. Despite the apparent higher quality of shade leaves, forest habitat better predicted damage by folivores than leaf type, suggesting that the direct effects of light environment predominate for herbivory. Potted saplings of both leaf types placed in the shaded understorey suffered almost two times more folivory, on average, than saplings in sunny light gaps, despite more than three times higher mid-season herbivore density on sun leaf saplings relative to shade leaf saplings.

5. Taken together, these results suggest that both leaf phenotype and forest habitat, two factors frequently confounded in nature, have significant but distinct effects on leaf quality and herbivory. These findings have implications for plant–herbivore interactions following disturbances such as treefalls, when shade leaves may be present in sunny habitats, and may help explain patterns of herbivory in understorey plants with early leaf flushing phenology relative to the canopy, when sun leaves are present in the shaded understorey.


Abiotic factors affect the quantity and quality of resources available to plants, herbivores and higher trophic levels (Preisser & Strong 2004; Yarnes & Boecklen 2006). One important and heterogeneous abiotic factor is the amount of sunlight available to plants. In nature, sunlight gradients commonly occur on multiple scales in both space (e.g. within and among plants) and time (e.g. within a day or within a season). In both temperate and tropical forests, natural light gaps caused by treefalls are a significant source of variation in the amount of sunlight plants receive and, as such, play an important role in forest dynamics and the maintenance of diversity (Runkle 1982; Denslow 1987).

Light level influences a suite of plant nutritional and resistance traits (e.g. leaf toughness, carbon-based secondary chemistry, leaf nitrogen levels and leaf water content; Larsson et al. 1986; Nichols-Orians 1991; Coley & Barone 1996; Forkner, Marquis & Lill 2004; Barber & Marquis 2011). In general, primary productivity tends to be higher in sun-grown plants, but concentrations of some defensive compounds, such as tannins, are also often higher, while water content is lower (Aide & Zimmerman 1990; Dudt & Shure 1994; Henriksson et al. 2003). These differences may lead to differential herbivory of sun and shade leaf types depending on herbivore preferences and physiological tolerance. Previous studies report greater herbivory on leaves in shaded environments (Maiorana 1981; Niesenbaum 1992), increased herbivore performance when fed shade leaves (Henriksson et al. 2003; Connor 2006) and herbivore preference for shade leaves (Crone & Jones 1999). In contrast, other studies have found increased herbivory of leaves in sunny environments (Chacon & Armesto 2006), higher herbivore performance when fed sun leaves (Fortin & Mauffette 2002) and herbivore preference for sun leaves (Nichols-Orians 1991). These inconsistent patterns in the effects of light level suggest that optimal foraging by herbivores may depend not only on leaf traits but also on the habitat in which host plants grow. Herbivores may be affected by sunlight via two mechanisms: (i) directly through microhabitat-related differences in temperature and humidity (Stamp & Casey 1993; Bale et al. 2002) and (ii) indirectly through sunlight-induced variation in host plant traits or via effects on natural enemies (Richards & Coley 2008; Barber & Marquis 2011). Further, sunlight is likely to modify leaf physical, chemical and nutrient properties in different ways, resulting in complex effects on overall leaf quality. Finally, herbivore species are likely to differ in their responses to and tolerance of variation in sunlight. Few previous studies, to our knowledge, have manipulated both the indirect and the direct components of sunlight level simultaneously and recorded the responses of the herbivore community (see also Connor 2006).

Oaks (Quercus spp.: Fagaceae) are ideal for examining the effects of light environment on herbivory as they exhibit variation in plant defence and tolerance (Feeny 1970; Hunter & Schultz 1995; Wold & Marquis 1997; Lill & Marquis 2001; Hochwender, Sork & Marquis 2003; Adams et al. 2009), resource allocation (Sork, Bramble & Sexton 1993), local adaptation to herbivory (Sork, Stowe & Hochwender 1993), trophic cascades and community structure (Marquis & Whelan 1994; Lill & Marquis 2003), the relative roles of top-down and bottom-up forces (Forkner & Hunter 2000), and insect population dynamics (Varley & Gradwell 1968; Varley, Gradwell & Hassell 1973; Schultz & Baldwin 1982). Therefore, it is important to understand how sunlight level both early and late in leaf development affects leaf traits and herbivore responses in oaks.

Here, we grew potted Quercus alba L. (white oak, Fagaceae) saplings in either full sun or full shade during the entirety of leaf expansion to manipulate leaf phenotype (creating sun and shade leaves) and then placed these saplings in sun or shade (light gap or understorey) forest habitats in a two-way factorial design. In so doing, we determined the relative importance of sunlight treatment during leaf development and forest habitat type in determining plant phenology, leaf physical and chemical traits, arthropod density and folivory. Specifically, we addressed three questions: (i) How do the herbivore and arthropod predator communities respond to leaf phenotype and forest habitat? (ii) How does the amount of sunlight received during budbreak and leaf expansion affect plant phenology and leaf phenotype? (iii) What effect does forest habitat (light gap or shaded understorey) have on leaf phenotype? We predicted that herbivore density would be influenced by both leaf type and forest habitat, with herbivore density greatest on shade leaf saplings within shaded forest understorey because of both the predicted higher nutritional quality of shade leaves and the more favourable ambient conditions provided by shaded understorey. We predicted that this would lead to both higher folivory and higher arthropod predator density in these treatments (if predator foraging is density-dependent). Finally, we predicted that sun and shade leaves would differ in mechanical, chemical and nutritional properties and that these phenotypes would remain relatively unaltered after placement in alternative forest light environments.

Materials and Methods

We focused on one canopy-dominant plant species, Q. alba, which naturally occurs along a gradient of light levels, from full shade to full sun, and which shows strong phenotypic responses to light (Le Corff & Marquis 1999; Forkner, Marquis & Lill 2004; Barber 2009). Potted Q. alba saplings (c. 2 m tall, 4–6 years old, n = 60) were obtained from the Maryland Department of Natural Resources. Saplings were grown in 5-gallon pots with drainage holes in a custom-blend soil of aged pine bark, peanut hulls, agricultural compost and sand. Prior to purchase, saplings had been fertilized with a 3-3-3 N-P-K granular fertilizer. All saplings were provided with an equal amount of supplemental water as needed to avoid water stress.

Leaf Phenotype: Sun or Shade Leaves

The leaf phenotype of the potted Q. alba saplings was manipulated by growing the plants in either full sun (n = 30) or full shade (under black 90% light reduction shade cloth, n = 30) on the roof of Bell Hall, a four-storey building located at The George Washington University, in Washington, DC. Saplings were kept under these light conditions for 3 months (30 March–30 June, 2010), which encompassed the entire developmental process from pre-budbreak through full leaf expansion and hardening, to produce sun and shade leaf saplings, respectively. Sapling leaves were not fully hardened until mid-June (T. M. Stoepler pers. obs.). We acknowledge that by growing saplings under these leaf treatments on the roof until late June, we miss early-season herbivory. In the shade leaf treatment, shade cloth reduced photosynthetically active radiation (PAR) by 90·2% based on hourly, paired measurements above and below shade cloth on three clear, sunny days during the peak hours of 10 am–2 pm (mean ± SD 4493 ± 100 μmol s−1 m−2 reduction; quantum photometer, Li-190SB; Li-cor Inc, Lincoln, NE, USA). Temperatures were elevated in the sun compared with the shade leaf treatment based on 20 paired measurements April–June (sun treatment mean: 26·8 °C; shade treatment mean: 24·4 °C, paired t-test: t19 = 4·24, = 0·0004). Saplings were rotated weekly within each light treatment to minimize any unintended position effects. We acknowledge that artificial shade treatments may not entirely reproduce the changes in the spectral balance of leaves (red: far-red radiation or UV/PAR) that typically accompany natural shading, for example, by forest canopies (reviewed in Roberts & Paul 2006). Importantly, Q. alba generally flushes only one set of leaves for the season (Barber & Marquis 2011; T. M. Stoepler, pers. obs.). Therefore, it is assumed that these sun and shade leaf phenotypes and their associated chemical and physical properties remain relatively fixed throughout the season, even when placed in alternative light environments.

To quantify these sun and shade leaf type differences, the following traits were measured for all or a subset of potted saplings prior to their transport to the forest in June 2010: (i) budbreak date (date of first bud opening), (ii) leaf greening date (date when ≥50% of a sapling's leaves first turned from red to green), (iii) sapling height (cm), (iv) trunk diameter (cm; at 12 cm above soil level), (v) total number of leaves per sapling, (vi) individual leaf area (cm2; n = 30 leaves/sapling), (vii) total leaf area/sapling (number of leaves/sapling × mean individual leaf area; m2), (viii) number of shoots per sapling, (ix) new shoot length (cm; all shoots), (x) leaf density (leaves per cm of shoot), (xi) leaf fracture toughness (J m−2; see below and Appendix S1, Supporting information), (xii) leaf thickness (mm), (xiii) leaf water content (%, measured from standardized leaf punches, avoiding veins; n = 3 punches/leaf, one leaf/sapling), (xiv) specific leaf area (SLA; leaf area/dry mass measured from standardized leaf punches) and (xv) initial folivory before being placed in the forest (% leaf area consumed by chewing herbivores while on the roof either under the shade cloth or adjacent unshaded rooftop). Folivory was assessed by measuring the amount of leaf area removed using a clear acetate grid of 1-cm2 squares laid over each leaf (n = 30 leaves/tree).

Forest Habitat: Light Gap or Shaded Understorey

Naturally occurring light gap and adjacent shaded forest understorey sites were chosen within Little Bennett Regional Park, Montgomery County, MD (39°15·9′N, 77°16·7′ W; Elev: 500 m), a 13-km2 secondary forest in the Piedmont region dominated by beech, oak and hickory. After leaves were fully expanded and hardened, saplings of each leaf phenotype were placed in the forest in both sun (light gap) and shade (shaded forest understorey) habitats as potted plants in 15 blocks on 30 June 2010 (Fig. 1). Pots were dug into the ground c. 30 cm to aid in moisture retention and to prevent saplings from falling over. We used a two-way factorial randomized complete block design with two sunlight factors: leaf type (leaf expansion and hardening in either full sun or full shade conditions) and habitat (subsequent placement in either sunny light gap or shaded forest understorey light environments). Using potted plants allowed us to control for leaf age, plant size and growth stage, each of which has been shown to affect leaf palatability to herbivores in previous studies (e.g. Coley 1980; Boege & Marquis 2006). Naturally occurring size-matched Q. alba saplings (or branches of larger saplings when a sapling was not available) in each light gap and shaded understorey site were also studied to account for any ‘pot’ effects (‘control’ saplings hereafter, n = 1/habitat type/block; Fig. 1). Potted saplings in each block were provided with an equal amount of supplemental water in the field as needed to prevent desiccation within pots (c. 4 L once per week per sapling); potted saplings in sun and shade habitats received the same amount of water. Control saplings were not watered.

Figure 1.

One of the 15 experimental blocks, which all contained a light gap (foreground) and shaded understorey habitat (background) at the study site, Little Bennett Regional Park, Clarksburg, MD. Each block contained six trees total, as each habitat contained one potted Quercus alba sapling of each of the two leaf phenotypes, sun leaf or shade leaf (indicated by arrows), and one ‘control’ Q. alba sapling or branch naturally occurring in each habitat. Photograph credit: Teresa Stoepler.

To quantify light level in sun and shade habitats, the pre-marked location chosen for each sapling was monitored diurnally and scored for the amount of sunlight received each hour for the peak hours of 9 am–4 pm on one clear, sunny day immediately prior to the start of the field experiment (23 June 2010). Saplings in designated ‘shade’ habitats never received more than 1 h of full sun or 2 h of partial sun and on average received 6·3 h (range 5–7 h) of shade per day. Saplings in designated ‘sun’ habitats received at least 2 h of full sun each day plus 2 h of partial sun (or 3 h full sun plus 1 h partial sun) for a total of 4 h of direct sunlight, and they never received more than 3 h of shade during these peak hours. On average, saplings in sun habitats received 5·4 h total sun per day. Per cent canopy openness was also measured for each sun and shade habitat using a concave spherical densiometer. Four measurements were taken at 1·5 m height in each habitat (once facing each cardinal direction), and these were averaged for each habitat in each block. Across all 15 blocks, mean ± SE per cent canopy openness in sun habitats was 57·9 ± 6·2%; in the shade, it was 2·4 ± 0·6%, a significant difference (paired t-test: t14 = 8·88, < 0·0001). Per cent canopy openness was highly correlated with the mean number of hours of sunlight received by each plot (R229 = 0·818, < 0·0001).

To compare chemical and nutritional differences between leaf types, we measured leaf hydrolysable and condensed tannin content, total per cent nitrogen, per cent carbon and carbon: nitrogen ratio on leaves collected in mid-summer (July 2010). To quantify arthropod community response to each of the four leaf type × habitat treatment combinations, all leaf surfaces and shoots of each potted and control sapling were visually censused for herbivorous and predaceous arthropods on 28 July 2010. All arthropods excluding mites were counted and identified to class, order or family and classified into functional groups (herbivore, predator, other). We acknowledge that observer presence may have caused some mobile arthropods to flee prior to being recorded, as in all arthropod censuses. To document seasonal changes in leaf type, late in the season (September), we again measured leaf fracture toughness, leaf water content and SLA, and additionally measured late-season herbivory (folivory: % leaf area consumed by chewing herbivores, and skeletonization: % leaf area skeletonized; n = 45 leaves/sapling × 40 trees). Leaves for which folivory could not be assessed, either because of leaf abscission or because missing leaf area was not estimable (i.e. only a very small leaf fragment remained), were avoided. Therefore, late-season folivory estimates are conservative. Damage by piercing and sucking insects was not assessed; however, chewing damage and skeletonization are the primary forms of folivory observed on white oak leaves at the study site (T. M. Stoepler, pers. obs.). To understand the distinct effects of leaf treatment and forest habitat on late-season phenology, leaf colour change was additionally measured as the per cent of leaves that had turned from green to red/orange on each potted sapling in eight of the blocks (n = 32 saplings) during weekly censuses 7 October–4 November 2010. These data are presented as the mean per cent of leaves/sapling that had changed colour by 22 October 2010 for simplicity. Leaf drop phenology was not measured because many oak species, including Q. alba, retain marcescent leaves into the winter and the following spring (Berkley 1931; T. M. Stoepler pers. obs.). All variables described previously were also measured on the control saplings except for phenology (i.e. budbreak, leaf greening and late-season leaf colour change).

Leaf Toughness

Leaf fracture toughness (R; J m−2) was quantified using a scissors ‘cutting test’ following Lucas & Pereira (1990). This leaf toughness metric is highly relevant for chewing insect herbivores as it measures the energy required to fracture (cut) the leaf (Lucas & Pereira 1990; Sanson 2006; see Appendix S1 for more detail on this method, Supporting information). Three cutting tests were performed on one haphazardly selected leaf per sapling on each of the potted saplings and control saplings; these three tests were averaged to produce one toughness value per sapling. Leaves were sampled on two collection dates (i) ‘early’ season: 17 June (n = 17/leaf type; 38 total) and (ii) ‘late’ season: 14 September 2010 (n = 8/leaf type × habitat combination; 48 total).

Leaf Chemistry

Three leaves per potted or control sapling were systematically collected over the height of the plant from all six plants in each of eight blocks (n = 48 saplings total) on 30 July 2010. Leaf samples were kept on ice in the field, immediately dried in a drying oven for 3 days at 60 °C and then stored in a freezer at −20 °C. Before extraction, leaves were ground with a Wiley Mill, and the ground leaf powder was freeze-dried overnight. Leaf samples were bulked to provide one sample per sapling. Samples were extracted exhaustively by sonication for 30 min, centrifugation at 3900 g RCF and 4 °C for 15 min, decanting the supernatant into a separate microcentrifuge tube, and addition of 0·5 mL of 70% acetone with 1 mm ascorbate for a total of four cycles (Adams et al. 2009). After extraction, the acetone was removed by rotary evaporation, and samples were made up to a constant volume of 0·5 mL.

Foliar hydrolysable and condensed tannins were measured using the potassium iodate assay (Bate-Smith 1977; Schultz & Baldwin 1982) and the acid–butanol assay (Rossiter, Schultz & Baldwin 1988; Rehill et al. 2006), respectively. Following a modification of the method of Hagerman & Butler (1980) as discussed in Rehill et al. (2006), we used a previously prepared standard created from equal amounts of lyophilized, ground leaf powder from lower-level canopy Q. alba leaves collected along a latitudinal gradient from Georgia to Maine in late May–late July (Adams et al. 2009). The results of these assays are quantified as percentage tannin equivalents, indicating reactivity relative to the purified standard (Adams et al. 2009).

Elemental analysis was used to determine foliar per cent nitrogen, per cent carbon and carbon: nitrogen from the same leaf samples collected for the tannin analyses described previously (n = 48); samples were sent to the Cornell Stable Isotope Laboratory for analysis on a CE Elantech Flash 1112 elemental analyzer.

Data Analysis

Where applicable, for each analysis, we first tested for a ‘pot’ effect by comparing control saplings to pots of the same treatment combination, that is, sun habitat + sun leaf and shade habitat + shade leaf, excluding the ‘sun habitat + shade leaf’ and ‘shade habitat + sun leaf’ treatment combinations. If ‘pot’ effects were significant, we performed two separate analyses: one comparing controls and pots of the same treatment combination as above and a second excluding all controls, comparing only potted saplings in the four treatment combinations to one another. If the effect of ‘pot’ was not significant, controls and pots in the same treatment combination were combined in final models. Full models were constructed with block (random factor) and the fully crossed fixed factors habitat, leaf type and habitat × leaf type using mixed models (Proc Mixed, REML estimation, variance components (VC) covariance structure, SAS Institute 2008). To account for unbalanced replication because of combining potted and naturally occurring trees of the same treatment type, type III sum of squares were used in all analyses, and the Satterthwaite option was used for all degrees-of-freedom calculations in Proc Mixed (Littell et al. 1996). While we tested for ‘pot’ and ‘block’ effects in all models, ‘pot’ effects were only evident in folivory and tannin analyses, and ‘block’ was only significant in the herbivore density analysis, but was retained in all final models. Folivory (mean proportion of leaf area consumed by chewing herbivores per sapling), tannin content (% tannin equivalents) and leaf colour change (% leaves/sapling turned to red on 22 October 2010) were arcsine-square-root-transformed prior to analyses, and herbivore and arthropod predator densities (number of herbivores or arthropod predators per m2 foliage) were square-root-transformed to meet model assumptions. For the folivory analysis, data on individual leaves were pooled to provide a single mean value per sapling. To determine whether herbivore density predicted folivory, the relationship between mid-season chewing herbivore density and late-season folivory was tested using linear regression. To test for differences in leaf type traits between sun and shade leaf potted saplings prior to placement in forest habitats, we performed separate unpaired t-tests. Data were analysed using JMP 9.1.2 (SAS Institute 2010) or sas 9.2 (SAS Institute 2008).


Quercus alba sun and shade leaf type treatments significantly altered phenology, sapling architecture and leaf traits, and in general, these sun and shade leaf phenotypes remained constant throughout the season, irrespective of subsequent placement in sun and shade forest habitats (light gaps or shaded forest understorey) (Table 1). The date of first budbreak was 4 days later on average in shade leaf saplings than in sun leaf saplings (Table 1). Shade leaf saplings had 21% fewer leaves, but 37% greater total leaf area than sun leaf saplings. Individual shade leaves were 78% larger in area than sun leaves in potted saplings. Interestingly, several sapling architectural traits induced by the sun and shade leaf treatments were maintained the following year, despite the fact that all saplings were transferred to the ‘shade leaf’ treatment in that year (see Appendix S2, Supporting information). Early to mid-season, shade leaves were 8% higher in water content, 12% higher in nitrogen, 17% lower in thickness, 27% lower in toughness and 14% lower in carbon: nitrogen than sun leaves (Table 1). Late-season SLA of shade leaves was 52% higher than sun leaves, regardless of forest habitat (Table 1). In the fall, the leaves of shade leaf saplings in the shaded understorey were significantly slower to change from green to red than leaves of sun leaf saplings (Table 1). In general, the foliar phenotype and plant architecture of control saplings were similar to potted saplings of the same leaf type (Table 1). Although pre-experiment leaf damage (incurred during leaf type treatments on the roof) was significantly higher on potted sun leaf saplings than on potted shade leaf saplings (t29 = 6·583, < 0·0001), the mean proportion of leaf damage/leaf/sapling was <4% for both leaf types. Control sapling early-season damage did not differ between leaf types (t28 = 0·0302, = 0·9761), and the mean proportion of leaf damage/leaf/sapling was 6% for both sun and shade control saplings.

Table 1. Potted Quercus alba (white oak) sapling traits by leaf phenotype (sun or shade) during leaf treatments on the roof before and after placement in sun and shade forest habitats. ‘Controls’ are naturally occurring Q. alba saplings or branches of larger trees in the same sun and shade forest habitats included to account for any ‘pot’ effects
Sampling period/traitRoof Shade leafRoof Sun leafControl ShadeControl Sun
Budbreak date (SD in days) 17 Apr ± 6·9 a 13 Apr ± 4·5 b Not measuredNot measured
Leaf greening date (SD in days)1 May ± 5·2a28 Apr ± 5·5aNot measuredNot measured
Initial sapling height (cm)180·9 ± 11·9a183·6 ± 8·4a204·7 ± 33·8x186·3 ± 36·0x
Trunk diameter (cm)1·44 ± 0·12a1·48 ± 0·10aNot measuredNot measured
Early Season (late June)
N leaves/sapling 218·7 ± 69·2 b 278·1 ± 101·7 a 192·2 ± 70·4x219·7 ± 60·8x
Single leaf area (cm2) 53·6 ± 30·3 a 30·0 ± 16·1 b 42·4 ± 8·6x41·2 ± 8·7x
Total leaf area/sapling (m2) 1·12 ± 0·29 a 0·82 ± 0·27 b 0·82 ± 0·36x0·90 ± 0·30x
N shoots/sapling40·5 ± 17·6a46·7 ± 10·7a70·4 ± 37·6x79·0 ± 43·1x
New shoot length (cm)7·2 ± 2·8a6·0 ± 2·3a 2·5 ± 0·65 y 3·7 ± 1·4 x
Leaf density (N leaves per cm shoot)1·44 ± 0·64a1·67 ± 0·64a 1·36 ± 0·37 x 1·08 ± 0·32 y
Leaf toughness (J m−2) 480·5 ± 46·0 b 653·5 ± 70·6 a 391·2 ± 63·6 y 517·4 ± 107·9 x
Leaf thickness (mm) 0·15 ± 0·01 b 0·18 ± 0·01 a 0·16 ± 0·02 y 0·18 ± 0·02 x
Leaf water content (%) 55·4 ± 2·5 a 51·1 ± 1·8 b 56·8 ± 3·6 x 53·8 ± 2·5 y
Specific leaf area (leaf area/dry mass) 169·8 ± 15·0 a 91·7 ± 9·5 b 165·0 ± 38·8 x 126·7 ± 25·1 y
 Forest Shade habitatForest Sun habitatControl ShadeControl Sunanova: Controls vs. Like pots only
Shade leafSun leafShade leafSun leaf
  1. Values are shown as mean ± 1 SD. Statistical significance in leaf phenotype differences prior to placement in forest habitats (‘Spring’ and ‘early-season’, below) is based on unpaired two-tailed t-tests (pots and controls tested separately). Statistical significance in leaf type, habitat (fully crossed, fixed factors), leaf type × habitat interaction and block (random) following placement in forest habitats on 30 June 2010 (‘mid-season’ and ‘late-season’, below) was tested with mixed-model anova. However, leaf type was the only factor that was ever significant in these analyses. In these ‘mid-’ and ‘late-season’ analyses, control saplings were combined with potted saplings of the same treatment type (when no ‘pot’ effects were evident). When ‘pot’ effects were evident (indicated with *), they were tested separately (see ‘Data Analysis’ for more detail). Statistically significant differences are in bold; different sequential subscript letters denote significant differences at α = 0·05. Per cent values were arcsine-square-root-transformed prior to analysis to improve normality of residuals. See Materials and Methods for details regarding the measurement of each trait.

Mid-season (late July)
Leaf nitrogen (%) 3·47 ± 0·27 a 3·05 ± 0·16 b 3·38 ± 0·20 a 3·04 ± 0·23 b 3·04 ± 0·19x2·95 ± 0·14x Pot: *
Leaf carbon (%) 48·02 ± 0·88 b 49·62 ± 0·50 a 47·91 ± 0·70 b 49·11 ± 0·90 a 48·70 ± 0·19x49·10 ± 0·75x 
Carbon: nitrogen 13·90 ± 0·97 b 16·28 ± 0·80 a 14·20 ± 0·76 b 16·20 ± 1·03 a 16·52 ± 0·84x16·20 ± 1·20x Pot: *
Late season (mid-September)
Leaf toughness (J m−2) 444·4 ± 56·0 b 596·6 ± 73·3 a 470·9 ± 70·5 b 627·2 ± 61·2 a 456·5 ± 75·5 b 554·8 ± 92·3 a  
Leaf thickness (mm) 0·15 ± 0·01 b 0·19 ± 0·02 a 0·15 ± 0·01 b 0·19 ± 0·01 a 0·15 ± 0·02 b 0·18 ± 0·01 a  
Leaf water content (%)56·3 ± 2·2a55·6 ± 1·2a54·3 ± 2·1a54·8 ± 2·2a 55·6 ± 1·8 x 51·7 ± 2·8 y Pot: *
Specific leaf area 157·8 ± 13·9 a 100·3 ± 12·5 b 147·8 ± 15·4 a 93·7 ± 11·7 b 163·4 ± 21·0 x 114·5 ± 16·1 y Pot: *
Leaf colour change (% of red leaves/sapling on Oct. 22, 2010) 29 ± 23 c 85 ± 23 ab 49 ± 44 bc 99 ± 4 a Not measuredNot measuredN/A

Mid-season insect herbivore density differed between treatments. Mean herbivore density was more than three times higher on sun leaf saplings compared with shade leaf saplings (leaf type, F1,73 = 5·99, = 0·0168; Fig. 2) and varied among blocks (Z14 = 2·17, = 0·0152), but notably, did not differ between the shaded and sunny habitats (habitat, F1,73 = 0·43, = 0·5147; leaf type × habitat, F1,73 = 0·401, = 0·9333). Herbivores recorded in the census included caterpillars (mainly Geometridae, Gracillariidae, Limacodidae, Lymantriidae, Noctuidae and others: Lepidoptera), Asiatic oak weevils (Cyrtepistomus castaneus, Curculionidae: Coleoptera) and other beetles (Coleoptera), aphids (Aphididae: Hemiptera), lace bugs (Tingidae: Hemiptera), leaf hoppers (Cicadellidae: Hemiptera), crickets and katydids (Orthoptera), walking sticks (Phasmida), sawfly larvae (Caliroa spp.: Tenthredinidae: Hymenoptera) and thrips (Thysanoptera). Mid-season chewing herbivore density (caterpillars, sawfly larvae, walking sticks and Orthopterans) was positively related to late-season folivory (R2 = 0·231, d.f. = 40, = 0·0017); however, this relationship was driven by two outlier trees that had high densities of chewing herbivores. When these two trees were removed, there was no relationship between herbivore density and folivory (R2 = 0·011, d.f. = 38, = 0·5277). In contrast, arthropod predator density did not differ between either sapling leaf types or habitats (GLM: F3,86 = 1·88, = 0·1388), but there was a non-significant trend towards greater predator density in the shade habitat compared with the sun habitat (45% increase). Arthropods classified as predators included spiders (Arachnida), ants (Formicidae: Hymenoptera), lacewings (Chrysopidae: Neuroptera), stink bugs and shield bugs (Pentatomidae: Hemiptera).

Figure 2.

Mean density of insect herbivores per sapling on potted and naturally occurring Quercus alba saplings combined by forest habitat (shaded understorey or sunny light gap) and leaf phenotype (shade or sun leaves) in a mid-season arthropod census on 28 July 2010 (leaf type, F1,73 = 5·99, = 0·0168). Density is measured as the number of herbivores per m2 foliage per sapling; each bar represents the mean of 15–30 saplings. Different letters above bars indicate significant differences based on Tukey's HSD tests on least square means of transformed data. Error bars indicate ± 1 SEM.

Late-season folivory, that is, the mean proportion of leaf area consumed per sapling by chewing herbivores by September, did not differ between sun and shade leaf saplings, but did differ between sun and shade forest habitats. Comparing controls to pots of the same treatment combination only, control plants suffered 66% higher folivory than did potted plants (F1,31·8 = 39·89, P < 0·0001), and plants in the shade habitat suffered 22% higher folivory than those in the sun habitat (habitat: F1,31·9 = 14·36, = 0·0006; block: Z6 = 1·30, = 0·0966; Fig. 3). Excluding controls, potted plants in the shade habitat suffered almost two times higher folivory on average than those in the sun habitat (habitat, F1,16·6 = 8·97, = 0·0083; Fig. 3), while the main effect of leaf type, the habitat × leaf type interaction and block were not significant (leaf type: F1,16·6 = 0·94, = 0·3454; habitat × leaf type: F1,16·6 = 1·88, = 0·1891; block: Z6 = 0·70, = 0·2411). Skeletonization accounted for only 2–4% of late-season leaf damage and did not differ between treatments (F2,38 = 0·3147, = 0·7319).

Figure 3.

Late-season (September) mean proportion of leaf area consumed by chewing herbivores per sapling on potted and ‘control’ Quercus alba saplings by forest habitat (shaded understorey or sunny light gap) and leaf phenotype (shade or sun leaves). Solid bars: potted saplings (habitat, F1,16·6 = 8·97, = 0·0083); hatched bars: ‘control’ Q. alba saplings naturally occurring in shade and sun habitats used to control for ‘pot’ effects. Bars represent the mean proportion of leaf area consumed from 45 leaves per sapling, spread over the height of the plant (n = 6–7 saplings). Error bars indicate ± 1 SEM.

Sun leaves consistently had higher concentrations of both condensed and hydrolysable tannins compared with shade leaves. Among potted saplings, the sun leaves averaged 24% higher condensed tannins and 32% higher hydrolysable tannins than shade leaves (leaf type, F1,21 = 126·03, < 0·0001 and F1,21 = 53·24, < 0·0001, respectively; Fig. 4a,b). However, neither condensed nor hydrolysable tannins differed in potted saplings placed in light gaps vs. shaded forest understorey (habitat, F1,21 = 0·20, = 0·6615 and F1,21 = 0·24, = 0·6293, respectively) or by blocks (Z7 = 0·15, = 0·4387 and Z7 = 0·11, = 0·4547, respectively). Notably, condensed tannin levels increased in shade leaf saplings placed in the sun habitat relative to shade leaf saplings placed in the shade habitat (Fig. 4b). For condensed tannins, this habitat × leaf type interaction was highly significant (F1,21 = 16·04, = 0·0006), while for hydrolysable tannins, this interaction was marginally significant (F1,21 = 4·23, = 0·0522).

Figure 4.

Mean per cent tannin equivalents of leaves of potted and naturally occurring Quercus alba saplings as (a) hydrolysable tannins (potassium iodate assay) and (b) condensed tannins (N-butanol assay) by leaf phenotype (shade or sun) and forest light environment (shaded understorey or sunny light gap). Shaded bars indicate shade leaf phenotype; white bars indicate sun leaf phenotype; hatched bars indicate ‘control’ Q. alba saplings naturally occurring in sun and shade habitats used to control for pot effects. Although ‘pot’ effects were only significant in hydrolysable tannin levels, ‘pots’ and ‘controls’ are shown separately in both panels for clarity. Each bar represents the mean of one bulk sample of three leaves per sapling collected in mid-summer (30 July 2010); n = 8 saplings. Different letters above bars indicate significant differences based on Tukey's HSD tests on least square means of transformed data. Error bars indicate ± 1 SEM.


In a field experiment designed to manipulate Q. alba leaf phenotype (sun or shade leaves) and forest light environment (light gap or shaded understorey habitats) factorially, we found that late-season folivory was strongly affected by forest habitat and not leaf phenotype, despite the distinct and persistent physical and chemical phenotypes of sun and shade leaves. Potted saplings in shaded forest understorey habitats experienced almost two times higher folivory by chewing herbivores than those in light gaps, regardless of leaf type. The lack of leaf type effects on herbivory suggests that in this system, the primary late-season herbivores do not appear to be deterred by the higher toughness and lower water content of sun leaves as is generally predicted (Feeny 1970; Coley 1983).

Greater herbivory in shaded relative to sunny habitats has been recorded in several forest types, including temperate, subtropical and tropical forests (Lowman 1992; Coley & Barone 1996; Muth et al. 2008). In addition, at least two studies have shown that although herbivory was higher in the shade, herbivore density did not differ between sun and shade habitats, consistent with the results of our study (Muth et al. 2008; Guerra, Becerra & Gianoli 2010). We found that although mid-season herbivore density did not differ between forest habitats, the density of herbivores was more than three times higher on sun leaf saplings than on shade leaf saplings. This result is somewhat unexpected as sun leaves appear to provide lower quality food for herbivores, based on the chemical and structural properties measured in our study and in others (e.g. Mattson 1980; Scriber & Slansky 1981; Coley 1983). However, because we only measured herbivore and predator density once, we do not know how arthropod densities vary throughout the summer. Mid-season chewing herbivore density was a poor predictor of late-season folivory, indicating that further study is needed to understand the specific mechanisms leading to greater late-season folivory in the shaded understorey. For example, differential herbivore densities in sun and shade habitats early (Hochwender, Sork & Marquis 2003) or late in the summer when herbivore densities are generally higher than in the mid-summer (Wold & Marquis 1997) could lead to greater accumulation of folivory in the shade by the end of summer.

Because potted saplings were not placed in the forest until late June, the two times higher late-season folivory of controls relative to potted saplings indicates that from spring to early summer, before oak leaf expansion and hardening are complete, is a critical period during which more than half of the seasonal folivory occurs. This difference in folivory between control and potted saplings is likely due in large part to increased opportunities for colonization and establishment of insects on control saplings during the spring, early summer and the previous year. Future studies should be conducted in the field for the entire season to determine whether early-season patterns of folivory mirror or negate late-season folivory.

As predicted, the light environment that saplings were exposed to during early leaf development determined leaf phenotype more than the light environment later in development. Most components of sunlight-induced foliar phenotypes remained fixed throughout the growing season, regardless of placement in reciprocal light environments, but several traits showed limited responses to forest habitat, including foliar tannin levels, leaf water content and leaf toughness. Hydrolysable and condensed tannin levels increased when placed in either reciprocal light environment, while water content and toughness shifted to match their light environments by late season (e.g. leaf water content increased while toughness decreased in saplings placed in shade habitats). Foliar phenotypes of control saplings were generally similar to potted saplings of the same treatment combination, suggesting that the potted saplings were a reasonable proxy for the effects of sunlight on Q. alba saplings in nature. Observed differences in leaf chemistry between pots and controls may have been due to ontogenetic effects (Boege & Marquis 2005), as controls were often branches of larger and older saplings, and differences in water content and SLA were expected, given that control saplings were not provided with supplemental water.

Condensed tannins showed the greatest increase when saplings were placed in reciprocal light environments, whereas hydrolysable tannins showed less response to forest habitat treatments. Increased sunlight intensity has been shown to increase foliar tannin concentrations (Hemming & Lindroth 1999; Yamasaki & Kikuzawa 2003), which can play an important role as photoprotectants via antioxidant properties (Close & McArthur 2002). Quercus alba shade leaf saplings placed in full sun habitats may have been under increased light stress compared with shade leaf saplings placed in the shade, leading to an increase in foliar tannins to prevent photodamage.

We conclude that sunlight level during leaf expansion and hardening has strong effects on Q. alba leaf phenotype, with shade leaf traits tending to be associated with higher nutritional quality for insect herbivores. However, despite the fact that shade leaves were larger, higher in water and nitrogen content, not as tough, and lower in tannins, late-season folivory did not differ between sun and shade leaf types. Instead, we show that forest habitat, not leaf phenotype, is the most important factor for folivory, suggesting that herbivore microhabitat preferences for the shaded understorey or physiological challenges associated with the full sun habitat may outweigh the effects of leaf quality.

The experimental design used in this study permitted decoupling the effects of leaf phenotype and forest habitat, but this decoupling may also occur because of various forms of disturbance. For example, shade leaves may be present in sunny habitats after new light gaps are created because of tree falls or because of forest fragmentation from logging, clearing or natural disasters such as fires or hurricanes (reviewed in Schowalter 2012). Fires and hurricanes can create patches of disturbed, open microhabitats adjacent to relatively undisturbed microhabitats that differ in herbivore assemblages and herbivory levels (Hunter & Forkner 1999; Knight & Holt 2005). Conversely, understorey plants that flush their leaves prior to the leaf flush of canopy trees will eventually have sun leaves in the shaded understorey (Harrington, Brown & Reich 1989; de la Cretaz & Kelty 2002). Taken together, these results suggest that both leaf phenotype and forest habitat, two factors that are frequently confounded in nature, affect leaf quality and herbivore pressure independently. When light environment changes because of disturbance or differences in phenology, leaf phenotype and forest habitat may be decoupled with consequences for herbivory.


We thank L. Bergner, R. Delzer, V. Fiorentino, K. Grenis, T. Hastings, J. Lill, P. Lill, S. Murphy and M. Rosati for field assistance; I. Forseth, P. Lucas, R. Perry and G. Wimp for equipment use; T. Galloway and the Treemendous program for facilitating obtaining saplings; W. Hanley for facilitating research at Little Bennett Regional Park; K. Sparks and the Cornell Stable Isotope Laboratory for elemental analysis; N. Barber for helpful discussions; and J. Lill, M. Abarca, E. Sigmon, K. Thompson and two anonymous reviewers for helpful comments on earlier drafts. This project was funded by Sigma Xi, Washington Biologists’ Field Club, and GWU Mortensen Fund grants to TMS. The authors have no conflicts of interest to declare.