Journal of Chemical Ecology, Vol. 18, No. 6, 1992
DO DEFOLIATION A N D SUBSEQUENT PHYTOCHEMICAL RESPONSES REDUCE FUTURE HERBIVORY ON OAK TREES?
STANLEY
H. F A E T H
Department of Zoology Arizona State University Tempe, Arizona, 85287-1501 (Received December 9, 1991; accepted February 3, 1992) Abstract--Perennial plants are thought to respond to partial or complete defoliation by producing new foliage that is less susceptible to herbivores because of induction of allelochemicals. Here, I tested this hypothesis by manually removing primary foliage from branches of Quercus emoryi (Fagaceae) at two different times in the season and monitoring changes in protein and tannin levels and the amount of herbivory relative to control branches. New, secondary leaves had 2.5 x greater hydrolyzable tannin content than mature foliage of control branches. Condensed tannins, which constitute a relatively low fraction of leaf mass, were lower, while protein content was temporarily greater, in new secondary leaves relative to mature leaves. Despite large increases in hydrolyzable tannins, herbivory levels were greater on refoliated branches than on control branches. New foliage is susceptible to herbivory regardless of when it is produced in the season, possibly because lower toughness and higher water content override any induced or developmentally related changes in allelochemistry. My results do not support the hypothesis that postherbivore changes in phytochemistry protect perennial plants from future herbivory, at least within a growing season. Key Words--Oak, Quercus emoryi, Fagaceae, defoliation, allelochemical induction, herbivory, tannins, defense.
INTRODUCTION T h e foliage o f m a n y perennial plants changes in nutrition, allelochemistry, phen o l o g y , and m o r p h o l o g y f o l l o w i n g h e r b i v o r y (for r e v i e w s , see Spencer, 1988; T a l l a m y and R a u p p , 1991). T h e s e induced responses to h e r b i v o r y h a v e b e e n v i e w e d as another l e v e l o f plant defense against h e r b i v o r e attack, in addition to 915 0098-0331/92/0600-0915506.50/0 9 1992 Plenum Publishing Corporation
916
FAETH
constitutive variation in phytochemistry, phenology, and morphology (cf. Tallamy and Raupp, 1991). Induced responses can be shown to decrease feeding, slow development, or increase mortality of some herbivores in both laboratory and field settings (cf. Tallamy and Raupp, 1991). The critical assumption underlying the notion of induced responses as plant defenses is that responses following herbivory reduce future tissue losses or increase fitness of perennial plants (Schultz, 1988). For example, induced chemical responses may increase mortality of herbivorous insects, but no fitness gain is realized by the plant because compensatory mortality factors maintain constant herbivore population size (Schultz, 1988; Faeth, 1991a; Karban, 1991). I tested the hypothesis that defoliation reduces future herbivory in Quercus emoryi (Fagaceae) by manually removing leaves from branches within trees and monitoring phytochemical changes and amount of arthropod herbivory on new, secondary foliage. Branches were defoliated at two different times in the growing season to assess how timing of defoliation affects regrowth, phytochemistry, and subsequent herbivory.
METHODS AND MATERIALS
Study Trees and Site. Quereus emoryi, Emory oak, is a dominant, semievergreen tree in central Arizona ranging from elevations of 1060 m to 2600 m (Keamey and Peebles, 1960). Emory oak usually grows determinately with new leaves produced in a single flush in mid-April to mid-May as year old leaves abscise. However, secondary, new leaves are produced later in the growing season if primary leaves are damaged physically or consumed by herbivores (Faeth, 1987). The study site is at the Sierra Ancha Experimental Station (USDA) in a riparian zone surrounding Parker Creek (Gila County, Arizona) at 1554 m elevation, approximately the midrange of Emory oak. Rainfall is high (62.7 cm/ year) and soils in the riparian zone are well-developed (Pase and Johnson, 1968) relative to other habitats of Emory oak in Arizona. Defoliation and Estimation of Herbivory. Ten trees were selected randomly in May 1988 at the study site from a pool of similar-sized trees within a 100m radius. Within each tree, four branches were chosen randomly from a pool of similar-sized branches at the same canopy height (1-3 m) and same aspect (i.e., cardinal direction) within a tree. One randomly-selected branch was defoliated early in the season, June 2, 1988, and paired with a randomly selected control branch (no defoliation). Paired branches were adjacent secondary branches on the same primary branch. A second randomly selected branch within each tree was defoliated later in the season (June 28, 1988) and also paired with a control branch. Branches within trees were selected as experimental units for
DEFOLIATION AND FUTURE HERBIVORY IN OAK
917
two reasons. First, complete defoliation of branches mimics natural berbivory events for this oak. Arthropod herbivores rarely defoliate entire trees, but early season defoliators such as the western tent caterpillar commonly defoliate whole branches and these branches reflush leaves later in the growing season. Second, the amount of simulated herbivory is < 1% of the total leaf area of the trees. This experimental level of damage is less than that typically removed by endemic herbivores (5-20%) (Faeth, 1987). Previous studies (Faeth, 1986, 1987, 1990) show that induction of allelochemicals is localized within Emory oak trees (i.e., there are no systemic responses in nondamaged branches or leaves within trees) even at much higher levels of simulated and insect damage. Using paired branches within trees permitted better control of biotic and abiotic factors that can influence rates of herbivory among Emory oak trees (Faeth, 1985, 1991b). Defoliation was accomplished by carefully plucking individual leaves until all leaves were removed from experimental branches. Artificial herbivory may not always produce the same phytochemical responses as does herbivory by arthropods (Baldwin, 1990). However, at least for Q. emoryi, artificial herbivory results in similar quantitative changes in tannin and protein content relative to damage by native arthropod herbivores (Faeth, 1986). I artificially defoliated branches to control both extent and timing of damage in the field, which is not feasible with native herbivores. All defoliated branches were monitored weekly for evidence of refoliation. Budbreak was defined as first appearance of unfurled, secondary leaves from swelled buds. Time of refoliation was measured in days from defoliation and difference between early and late defoliation branches was tested by a Student's t test. Entire branches were removed at the end of the growing season (April 4, 1989) before leaf abscission began and returned to the laboratory. All leaves were counted and examined for herbivory. Mean (+SD) number and range of leaves from branches were: control (early and late) X = 662.1 + 206.1, range = 331-957; early defoliation X = 169.4 + 117.2, range = 29-335, and late defoliation X = 262.8 + 208.3, range = 94-735. The amount of herbivory on new, secondary leaves was compared to leaves on control branches to test the hypothesis that induced responses in secondary leaves of Emory oak reduced further tissue loss. Amount of herbivory was measured in two ways. First, the percent of damaged leaves was calculated by dividing the number of damaged leaves by the total number of leaves for each branch. Second, total leaf area removed was estimated by either randomly selecting a subset of 50 damaged leaves from each branch or, if there were less than 50 damaged leaves within a branch, all of the damaged leaves. The area of each damaged leaf was measured with a leaf area meter, and this area was subtracted from the area of a reconstructed leaf (with opaque tape) to obtain area removed. Each branch was examined thoroughly for petiole scars, which indicate whole
918
FAETH
missing leaves. The area of missing leaves was estimated by using the average area of leaves remaining in a terminal cluster. The mean damaged area per leaf was calculated and multiplied by percent of damaged leaves to obtain the total area removed (cm2). The percent of damaged leaves (arcsine square root transformed), and percent total area removed (arcsine square root transformed) was compared statistically between early and late defoliation and control branches with paired t tests. I compared size of refoliated leaves to mature leaves by weighing 50 randomly selected, undamaged leaves from each branch and comparing means with a paired t test. PhytochemicalAnalyses. A small, haphazard sample of leaves (2-20 leaves) was removed from refoliated branches and paired control branches at three 30day intervals, June 30, July 29, and August 28, 1988, for phytochemical analyses. Leaves were immediately frozen on Dry Ice in the field and returned to the laboratory, where they were lyophilized and then examined for herbivory (see above). Leaves were then ground to a fine powder in a Wiley mill and stored in,airtight jars with a desiccant at 5 ~ until analyses of protein and tannins could be performed. Details of the phytochemical methods and standardization are in Faeth (1985, 1986). Briefly, relative protein content was determined by the BioRad method (Richmond, California), after washing samples 2 • with acetone to remove pigments and tannins. Condensed tannins were determined by a modified acidified vanillin method (Broadhurst and Jones, 1978) after extraction in acetone. Monomeric phenols were first isolated and quantified by removing tannin polymers with a selective adsorbent (Sephadex LH-20). Condensed tannins then were determined by subtracting monomers from total vanillin positive (monomers plus condensed tannin polymers), using a catechin standard prepared for each run. Only results from condensed tannins are presented here since total vanillin positive shows similar trends and because monomeric fractions were very low and varied little among samples. Hydrolyzable tannins were determined by a modified potassium iodate technique (Bate-Smith, 1977) and expressed as percent dry mass tannic acid equivalents. The method is specific for only the galloyl group of hydrolyzable tannins, not ellagitannins, but it is commonly used as an estimate of hydrolyzable tannin content (Schultz and Baldwin, 1982; Baldwin and Schultz, 1983; Faeth, 1990). All phytochemical analyses were run in triplicate for each leaf branch sample at each date. All phytochemical methods have been standardized and have produced consistent results for Emory oak over an 8-year span (Faeth, 1985, 1986, 1988). The objective was to sample branches that had just refoliated and then to resample branches 30 days later to determine temporal changes relative to controis. However, since not all branches had refoliated by these dates, especially
DEFOLIATION AND FUTURE HERBIVORY IN OAK
919
the more variable late defoliated branches, and two branches did not refoliate due to insect damage to buds, two consecutive sampling periods were available from six of the early defoliated branches (June 30, July 29) and only three of the late defoliated ones (July 29, August 28). I used ANOVA to determine if phytochemical samples from late defoliated branches could be pooled with early defoliated ones. In other words, I statistically determined if the phytochemistry of new, secondary leaves and 30-day-old leaves on late defoliated branches was the same as that of new and 30-day-old leaves on early defoliated branches, but separated by a 30-day time period. None of the phytochemical parameters varied significantly (F < 5.10, df = 1,7, P > 0.10 for all comparisons) between secondary leaves of the same developmental age on early and late defoliated branches, so samples were pooled in subsequent statistical analyses. I used paired t tests to compare statistically phytochemical measures (arcsine square root transformed) from new, secondary leaves and 30-day-old leaves to their control branch counterparts.
RESULTS
Time to Refoliation. Branches defoliated early in the season produced secondary leaves earlier (X = 21.9 _ 5.7 days) than branches defoliated later in the season (~7 = 36.5 • 20.6 days, t = 2.05, df = 15, 0.05 < P < 0.10). Herbivory Levels. The percentage of leaves damaged by arthropods on branches defoliated early was not significantly different from control branches (Figure 1, t = 1.13, df = 7, P > 0.20). Similarly, the percentage of leaves damaged on branches defoliated late was not different from paired control branches (Figure 1, t = .37, df = 7, P > 0.40). Percent leaf area removed from branches defoliated early was not significantly different from paired control branches (Figure 2, t = 1.27, df = 7, P > 0.20). However, percent leaf area on branches defoliated late in the season was greater than that removed from control branches (Figure 2), although this difference was only marginally significant (t = 2.33, df = 7, 0.05 < P < 0.10). Generally, herbivory tended to be greater on experimentally defoliated branches than control branches (Figures 1 and 2), opposite to predictions that refoliated branches are more resistant to herbivory. Refoliated leaves on experimental branches were significantly smaller than mature leaves on control branches (~7 + SE, refoliated leaves = 19.16 mg • 4.92; ~7 • SE, mature leaves = 47.89 mg _ 11.97; t = 9.38, df = 15, P < 0.0001). Refoliated leaves were approximately 60% of the size of the mature leaves. Phytochemistry. Protein content of new, secondary leaves on defoliated branches was significantly greater (t = 3.32, df = 8, P < 0.05) than paired
920
FAETH 30
a bJ
25
~ 2,
I-.Z
w
10
b.J
5
EC
Eli
LC
LE
TREATMENT
FIG. 1. Mean (___SE)percent of leaves damaged by arthropod herbivores on early-season control (EC) and experimental (EE) and late-season control (LC) and experimental (LE) branches. Data shown are untransformed; percentages were arcsine transformed for statistical tests.
10 9 Z
8
0 I-Z IM O
9~ bJ n
4'
3-
21 EC
8[
LC
TREATMENT
FiG. 2. Mean (_SE) percent of leaf area consumed by arthropod herbivores on earlyseason control (EC) and experimental (EE) and late-season control (LC) and experimental (LE) branches. Data shown are untransformed; percentages were arcsine transformed for statistical tests.
921
DEFOLIATION AND F U T U R E HERBIVORY IN OAK
control branches (Figure 3). Thirty days after reflushing, however, protein content of replacement leaves declined and was not different (t = 1.03, df = 8, P > 0.20) from mature foliage of paired control branches (Figure 3). Hydrolyzable tannin content of new, secondary leaves on defoliated branches was significantly greater (t -- 5.34, df = 8, P < 0.001) on defoliated branches relative to control foliage (Figure 4). Although declining after 30 days, hydrolyzable tannin content on experimental branches was still significantly greater (t = 6.68, df = 8, P < 0.001) than that of control branches. Condensed tannin content of new, secondary leaves on defoliated branches was significantly less (t = 3.67, df = 8, P < 0.01) than that of control branches (Figure 5). Condensed tannin content of defoliated branches remained significantly lower than control branches (t = 3.54, df = 8, P < 0.01) 30 days after reflushing (Figure 5).
DISCUSSION
New, secondary leaves contain 2.5 x more hydrolyzable tannins than mature foliage. Such increases in hydrolyzable tannins in replacement leaves after partial defoliation have been considered induced defenses in Q. rubrum against future herbivory (Schultz and Baldwin, 1982). However, it is unclear if changes in secondary foliage of Emory oak are induced responses or simply reflect 3.5 Z
3.0
CONTROL EXPERIMENTAL
0 2.5 W
_~
2.0
ILl 112 I--Z
1.5
1.0
0 W Q.
0.5 0.0 REFLUSH
+30 DAYS
TREATMENT
Fie. 3. Mean (_+SE) percent relative protein content of control and experimental branches at the time of reflushing of defoliated branches and 30 days later. Data from early and late control and experimental branches were pooled after ANOVA showed no significant differences (see text). Data shown are untransformed; percentages were arcsine transformed for statistical tests.
922
FAETH 60 55, Z Z Z
5O
._1 [13
35
~ CONTROL [ - - ' ] EXPERIMENTAL
45
S
:'5, 2s 121 n," a -r-
2O 15 10
5 0
REFLUSH
+30 DAYS
TREATMENT
FIG. 4. Mean ( _ SE) percent hydrolyzable tannin content of control and experimental branches at the time of reflushing of defoliated branches and 30 days later. Data from early and late control and experimental branches were pooled after ANOVA showed no significant differences (see text). Data shown are untransformed; percentages were arcsine transformed for statistical tests.
4.5
Z -.~
4.0
Z ~.
DO DEFOLIATION A N D SUBSEQUENT PHYTOCHEMICAL RESPONSES REDUCE FUTURE HERBIVORY ON OAK TREES?
STANLEY
H. F A E T H
Department of Zoology Arizona State University Tempe, Arizona, 85287-1501 (Received December 9, 1991; accepted February 3, 1992) Abstract--Perennial plants are thought to respond to partial or complete defoliation by producing new foliage that is less susceptible to herbivores because of induction of allelochemicals. Here, I tested this hypothesis by manually removing primary foliage from branches of Quercus emoryi (Fagaceae) at two different times in the season and monitoring changes in protein and tannin levels and the amount of herbivory relative to control branches. New, secondary leaves had 2.5 x greater hydrolyzable tannin content than mature foliage of control branches. Condensed tannins, which constitute a relatively low fraction of leaf mass, were lower, while protein content was temporarily greater, in new secondary leaves relative to mature leaves. Despite large increases in hydrolyzable tannins, herbivory levels were greater on refoliated branches than on control branches. New foliage is susceptible to herbivory regardless of when it is produced in the season, possibly because lower toughness and higher water content override any induced or developmentally related changes in allelochemistry. My results do not support the hypothesis that postherbivore changes in phytochemistry protect perennial plants from future herbivory, at least within a growing season. Key Words--Oak, Quercus emoryi, Fagaceae, defoliation, allelochemical induction, herbivory, tannins, defense.
INTRODUCTION T h e foliage o f m a n y perennial plants changes in nutrition, allelochemistry, phen o l o g y , and m o r p h o l o g y f o l l o w i n g h e r b i v o r y (for r e v i e w s , see Spencer, 1988; T a l l a m y and R a u p p , 1991). T h e s e induced responses to h e r b i v o r y h a v e b e e n v i e w e d as another l e v e l o f plant defense against h e r b i v o r e attack, in addition to 915 0098-0331/92/0600-0915506.50/0 9 1992 Plenum Publishing Corporation
916
FAETH
constitutive variation in phytochemistry, phenology, and morphology (cf. Tallamy and Raupp, 1991). Induced responses can be shown to decrease feeding, slow development, or increase mortality of some herbivores in both laboratory and field settings (cf. Tallamy and Raupp, 1991). The critical assumption underlying the notion of induced responses as plant defenses is that responses following herbivory reduce future tissue losses or increase fitness of perennial plants (Schultz, 1988). For example, induced chemical responses may increase mortality of herbivorous insects, but no fitness gain is realized by the plant because compensatory mortality factors maintain constant herbivore population size (Schultz, 1988; Faeth, 1991a; Karban, 1991). I tested the hypothesis that defoliation reduces future herbivory in Quercus emoryi (Fagaceae) by manually removing leaves from branches within trees and monitoring phytochemical changes and amount of arthropod herbivory on new, secondary foliage. Branches were defoliated at two different times in the growing season to assess how timing of defoliation affects regrowth, phytochemistry, and subsequent herbivory.
METHODS AND MATERIALS
Study Trees and Site. Quereus emoryi, Emory oak, is a dominant, semievergreen tree in central Arizona ranging from elevations of 1060 m to 2600 m (Keamey and Peebles, 1960). Emory oak usually grows determinately with new leaves produced in a single flush in mid-April to mid-May as year old leaves abscise. However, secondary, new leaves are produced later in the growing season if primary leaves are damaged physically or consumed by herbivores (Faeth, 1987). The study site is at the Sierra Ancha Experimental Station (USDA) in a riparian zone surrounding Parker Creek (Gila County, Arizona) at 1554 m elevation, approximately the midrange of Emory oak. Rainfall is high (62.7 cm/ year) and soils in the riparian zone are well-developed (Pase and Johnson, 1968) relative to other habitats of Emory oak in Arizona. Defoliation and Estimation of Herbivory. Ten trees were selected randomly in May 1988 at the study site from a pool of similar-sized trees within a 100m radius. Within each tree, four branches were chosen randomly from a pool of similar-sized branches at the same canopy height (1-3 m) and same aspect (i.e., cardinal direction) within a tree. One randomly-selected branch was defoliated early in the season, June 2, 1988, and paired with a randomly selected control branch (no defoliation). Paired branches were adjacent secondary branches on the same primary branch. A second randomly selected branch within each tree was defoliated later in the season (June 28, 1988) and also paired with a control branch. Branches within trees were selected as experimental units for
DEFOLIATION AND FUTURE HERBIVORY IN OAK
917
two reasons. First, complete defoliation of branches mimics natural berbivory events for this oak. Arthropod herbivores rarely defoliate entire trees, but early season defoliators such as the western tent caterpillar commonly defoliate whole branches and these branches reflush leaves later in the growing season. Second, the amount of simulated herbivory is < 1% of the total leaf area of the trees. This experimental level of damage is less than that typically removed by endemic herbivores (5-20%) (Faeth, 1987). Previous studies (Faeth, 1986, 1987, 1990) show that induction of allelochemicals is localized within Emory oak trees (i.e., there are no systemic responses in nondamaged branches or leaves within trees) even at much higher levels of simulated and insect damage. Using paired branches within trees permitted better control of biotic and abiotic factors that can influence rates of herbivory among Emory oak trees (Faeth, 1985, 1991b). Defoliation was accomplished by carefully plucking individual leaves until all leaves were removed from experimental branches. Artificial herbivory may not always produce the same phytochemical responses as does herbivory by arthropods (Baldwin, 1990). However, at least for Q. emoryi, artificial herbivory results in similar quantitative changes in tannin and protein content relative to damage by native arthropod herbivores (Faeth, 1986). I artificially defoliated branches to control both extent and timing of damage in the field, which is not feasible with native herbivores. All defoliated branches were monitored weekly for evidence of refoliation. Budbreak was defined as first appearance of unfurled, secondary leaves from swelled buds. Time of refoliation was measured in days from defoliation and difference between early and late defoliation branches was tested by a Student's t test. Entire branches were removed at the end of the growing season (April 4, 1989) before leaf abscission began and returned to the laboratory. All leaves were counted and examined for herbivory. Mean (+SD) number and range of leaves from branches were: control (early and late) X = 662.1 + 206.1, range = 331-957; early defoliation X = 169.4 + 117.2, range = 29-335, and late defoliation X = 262.8 + 208.3, range = 94-735. The amount of herbivory on new, secondary leaves was compared to leaves on control branches to test the hypothesis that induced responses in secondary leaves of Emory oak reduced further tissue loss. Amount of herbivory was measured in two ways. First, the percent of damaged leaves was calculated by dividing the number of damaged leaves by the total number of leaves for each branch. Second, total leaf area removed was estimated by either randomly selecting a subset of 50 damaged leaves from each branch or, if there were less than 50 damaged leaves within a branch, all of the damaged leaves. The area of each damaged leaf was measured with a leaf area meter, and this area was subtracted from the area of a reconstructed leaf (with opaque tape) to obtain area removed. Each branch was examined thoroughly for petiole scars, which indicate whole
918
FAETH
missing leaves. The area of missing leaves was estimated by using the average area of leaves remaining in a terminal cluster. The mean damaged area per leaf was calculated and multiplied by percent of damaged leaves to obtain the total area removed (cm2). The percent of damaged leaves (arcsine square root transformed), and percent total area removed (arcsine square root transformed) was compared statistically between early and late defoliation and control branches with paired t tests. I compared size of refoliated leaves to mature leaves by weighing 50 randomly selected, undamaged leaves from each branch and comparing means with a paired t test. PhytochemicalAnalyses. A small, haphazard sample of leaves (2-20 leaves) was removed from refoliated branches and paired control branches at three 30day intervals, June 30, July 29, and August 28, 1988, for phytochemical analyses. Leaves were immediately frozen on Dry Ice in the field and returned to the laboratory, where they were lyophilized and then examined for herbivory (see above). Leaves were then ground to a fine powder in a Wiley mill and stored in,airtight jars with a desiccant at 5 ~ until analyses of protein and tannins could be performed. Details of the phytochemical methods and standardization are in Faeth (1985, 1986). Briefly, relative protein content was determined by the BioRad method (Richmond, California), after washing samples 2 • with acetone to remove pigments and tannins. Condensed tannins were determined by a modified acidified vanillin method (Broadhurst and Jones, 1978) after extraction in acetone. Monomeric phenols were first isolated and quantified by removing tannin polymers with a selective adsorbent (Sephadex LH-20). Condensed tannins then were determined by subtracting monomers from total vanillin positive (monomers plus condensed tannin polymers), using a catechin standard prepared for each run. Only results from condensed tannins are presented here since total vanillin positive shows similar trends and because monomeric fractions were very low and varied little among samples. Hydrolyzable tannins were determined by a modified potassium iodate technique (Bate-Smith, 1977) and expressed as percent dry mass tannic acid equivalents. The method is specific for only the galloyl group of hydrolyzable tannins, not ellagitannins, but it is commonly used as an estimate of hydrolyzable tannin content (Schultz and Baldwin, 1982; Baldwin and Schultz, 1983; Faeth, 1990). All phytochemical analyses were run in triplicate for each leaf branch sample at each date. All phytochemical methods have been standardized and have produced consistent results for Emory oak over an 8-year span (Faeth, 1985, 1986, 1988). The objective was to sample branches that had just refoliated and then to resample branches 30 days later to determine temporal changes relative to controis. However, since not all branches had refoliated by these dates, especially
DEFOLIATION AND FUTURE HERBIVORY IN OAK
919
the more variable late defoliated branches, and two branches did not refoliate due to insect damage to buds, two consecutive sampling periods were available from six of the early defoliated branches (June 30, July 29) and only three of the late defoliated ones (July 29, August 28). I used ANOVA to determine if phytochemical samples from late defoliated branches could be pooled with early defoliated ones. In other words, I statistically determined if the phytochemistry of new, secondary leaves and 30-day-old leaves on late defoliated branches was the same as that of new and 30-day-old leaves on early defoliated branches, but separated by a 30-day time period. None of the phytochemical parameters varied significantly (F < 5.10, df = 1,7, P > 0.10 for all comparisons) between secondary leaves of the same developmental age on early and late defoliated branches, so samples were pooled in subsequent statistical analyses. I used paired t tests to compare statistically phytochemical measures (arcsine square root transformed) from new, secondary leaves and 30-day-old leaves to their control branch counterparts.
RESULTS
Time to Refoliation. Branches defoliated early in the season produced secondary leaves earlier (X = 21.9 _ 5.7 days) than branches defoliated later in the season (~7 = 36.5 • 20.6 days, t = 2.05, df = 15, 0.05 < P < 0.10). Herbivory Levels. The percentage of leaves damaged by arthropods on branches defoliated early was not significantly different from control branches (Figure 1, t = 1.13, df = 7, P > 0.20). Similarly, the percentage of leaves damaged on branches defoliated late was not different from paired control branches (Figure 1, t = .37, df = 7, P > 0.40). Percent leaf area removed from branches defoliated early was not significantly different from paired control branches (Figure 2, t = 1.27, df = 7, P > 0.20). However, percent leaf area on branches defoliated late in the season was greater than that removed from control branches (Figure 2), although this difference was only marginally significant (t = 2.33, df = 7, 0.05 < P < 0.10). Generally, herbivory tended to be greater on experimentally defoliated branches than control branches (Figures 1 and 2), opposite to predictions that refoliated branches are more resistant to herbivory. Refoliated leaves on experimental branches were significantly smaller than mature leaves on control branches (~7 + SE, refoliated leaves = 19.16 mg • 4.92; ~7 • SE, mature leaves = 47.89 mg _ 11.97; t = 9.38, df = 15, P < 0.0001). Refoliated leaves were approximately 60% of the size of the mature leaves. Phytochemistry. Protein content of new, secondary leaves on defoliated branches was significantly greater (t = 3.32, df = 8, P < 0.05) than paired
920
FAETH 30
a bJ
25
~ 2,
I-.Z
w
10
b.J
5
EC
Eli
LC
LE
TREATMENT
FIG. 1. Mean (___SE)percent of leaves damaged by arthropod herbivores on early-season control (EC) and experimental (EE) and late-season control (LC) and experimental (LE) branches. Data shown are untransformed; percentages were arcsine transformed for statistical tests.
10 9 Z
8
0 I-Z IM O
9~ bJ n
4'
3-
21 EC
8[
LC
TREATMENT
FiG. 2. Mean (_SE) percent of leaf area consumed by arthropod herbivores on earlyseason control (EC) and experimental (EE) and late-season control (LC) and experimental (LE) branches. Data shown are untransformed; percentages were arcsine transformed for statistical tests.
921
DEFOLIATION AND F U T U R E HERBIVORY IN OAK
control branches (Figure 3). Thirty days after reflushing, however, protein content of replacement leaves declined and was not different (t = 1.03, df = 8, P > 0.20) from mature foliage of paired control branches (Figure 3). Hydrolyzable tannin content of new, secondary leaves on defoliated branches was significantly greater (t -- 5.34, df = 8, P < 0.001) on defoliated branches relative to control foliage (Figure 4). Although declining after 30 days, hydrolyzable tannin content on experimental branches was still significantly greater (t = 6.68, df = 8, P < 0.001) than that of control branches. Condensed tannin content of new, secondary leaves on defoliated branches was significantly less (t = 3.67, df = 8, P < 0.01) than that of control branches (Figure 5). Condensed tannin content of defoliated branches remained significantly lower than control branches (t = 3.54, df = 8, P < 0.01) 30 days after reflushing (Figure 5).
DISCUSSION
New, secondary leaves contain 2.5 x more hydrolyzable tannins than mature foliage. Such increases in hydrolyzable tannins in replacement leaves after partial defoliation have been considered induced defenses in Q. rubrum against future herbivory (Schultz and Baldwin, 1982). However, it is unclear if changes in secondary foliage of Emory oak are induced responses or simply reflect 3.5 Z
3.0
CONTROL EXPERIMENTAL
0 2.5 W
_~
2.0
ILl 112 I--Z
1.5
1.0
0 W Q.
0.5 0.0 REFLUSH
+30 DAYS
TREATMENT
Fie. 3. Mean (_+SE) percent relative protein content of control and experimental branches at the time of reflushing of defoliated branches and 30 days later. Data from early and late control and experimental branches were pooled after ANOVA showed no significant differences (see text). Data shown are untransformed; percentages were arcsine transformed for statistical tests.
922
FAETH 60 55, Z Z Z
5O
._1 [13
35
~ CONTROL [ - - ' ] EXPERIMENTAL
45
S
:'5, 2s 121 n," a -r-
2O 15 10
5 0
REFLUSH
+30 DAYS
TREATMENT
FIG. 4. Mean ( _ SE) percent hydrolyzable tannin content of control and experimental branches at the time of reflushing of defoliated branches and 30 days later. Data from early and late control and experimental branches were pooled after ANOVA showed no significant differences (see text). Data shown are untransformed; percentages were arcsine transformed for statistical tests.
4.5
Z -.~
4.0
Z ~.
