Oecologia (2000) 125:504Ð511 DOI 10.1007/s004420000482

Rainer Dolch á Teja Tscharntke Defoliation of alders (Alnus glutinosa) affects herbivory by leaf beetles on undamaged neighbours

Received: 2 November 1999 / Accepted: 27 June 2000 / Published online: 23 August 2000 © Springer-Verlag 2000

Abstract The effects of defoliation of alder (Alnus all changes after herbivory, “induced resistance” only glutinosa) on subsequent herbivory by alder leaf beetle changes that affect subsequent herbivory, and “induced (Agelastica alni) were studied in ten alder stands in defence” results in a fitness benefit to induced plants. northern Germany. At each site, one tree was manually Biochemical consequences of defoliation may affect defoliated (c. 20% of total foliage) to simulate herbivory. neighbouring plants (Tuomi et al. 1990), either by con- Subsequent damage by A. alni was assessed on ten alders nection via roots or mycorrhizae (Simard et al. 1997) or at each site on six different dates from May to September by reception of volatile substances such as terpenoids 1994. After defoliation, herbivory by A. alni increased (e.g. Dicke 1994). Volatiles implicated in signal transfer with distance from the defoliated tree. Laboratory exper- include ethylene and methyl jasmonate, which may acti- iments supported the field results. Not only leaf damage vate genes coding for plant resistance (Ecker and Davis in the field, but also the extent of leaf consumption in 1987; Farmer and Ryan 1990), and the latter has been laboratory feeding-preference tests and the number of shown to induce changes in the production of secondary eggs oviposited per leaf in another laboratory test were compounds (Shonle and Bergelson 1995). Such interplant positively correlated with distance from the defoliated communication may result in chemical changes in un- tree. Resistance was therefore induced not only in defoli- damaged plants adjacent to damaged ones, leading to ated alders, but also in their undamaged neighbours. lower susceptibility to herbivores of plants neighbouring Consequently, defoliation of alders may trigger inter- damaged conspecifics (Baldwin and Schultz 1983; plant resistance transfer, and therefore reduce herbivory Haukioja et al. 1985; Rhoades 1985). Evidence for this in whole alder stands. kind of interplant signal transfer Ð popularly referred to as “communication” – remains controversial (Bruin et al. Keywords Herbivory á Defoliation á Plant responses á 1995; Shonle and Bergelson 1995). Once harshly critici- Induced resistance á Talking trees sed by Fowler and Lawton (1985), the concept of inter- plant signal transfer and so-called “talking trees” has un- dergone a recent revival (e.g. Karban and Baldwin 1997). Introduction Here we present results from a field experiment using a bioassay in a natural setting involving black alders (Al- Plant responses following insect herbivory are known nus glutinosa) and alder leaf beetles (Agelastica alni). from a wide range of plants (Rosenthal and Janzen 1979; This system was chosen to study the consequences of Hedin 1983; Barbosa and Letourneau 1988; Schultz herbivory, since mass outbreaks of alder beetles and 1988; Haukioja 1990; Tallamy and Raupp 1991; Fritz complete defoliations of alders, triggering both rapid and and Simms 1992; Karban and Baldwin 1997). These in- delayed induced resistance, are well-known (Jeker 1981; duced responses may either be incidental physiological Baur et al. 1991; Seldal et al. 1994; R. Dolch and T. reactions (for instance to nutrient stress, Haukioja and Tscharntke, unpublished work). Neuvonen 1985) or purely defensive traits to prevent fur- This study addresses the following questions: ther insect damage (induced resistance). According to Karban and Baldwin (1997) “induced responses” means 1. How does manual defoliation affect herbivory of Agelastica alni on Alnus glutinosa? R. Dolch (✉) á T. Tscharntke 2. Does manual defoliation lead to induced resistance Agroecology, University of Göttingen, Waldweg 26, 37073 Göttingen, Germany in A. glutinosa? e-mail: [email protected] 3. Can resistance also be induced in non-defoliated Tel.: +49-551-392358, Fax: +49-551-398806 neighbours? 505 number of all leaves and the total number of leaves damaged by Materials and methods A. alni was counted for each of those alders, so that percentage of leaf area consumed could be calculated. We found that there was a We studied consequences of manual defoliation of alders (A. gluti- strong positive relationship between the percentage of total dam- nosa) for subsequent herbivory of the alder leaf beetle (Agelastica age and the estimated percentage of leaf damage on the lower alni; Coleoptera, Chrysomelidae) in the field. A. alni is a wide- branches (r2=0.675, n=100, P<0.001), and also between absolute spread univoltine specialist on alder and considered its most im- number and relative number (percentage) of damaged leaves portant herbivore. We examined 100 individuals of alder at ten (r2=0.908, n=10, P<0.001). This indicates that the estimated val- different sites (ten alders at each site) located near Göttingen in ues of leaf damage sufficiently reflect real damage levels. Leaf north-central Germany. Distances between the sites varied from damage caused by phytophagous arthropod species other than 300 to 3500 m. At all sites, alders were growing in rows along the A. alni (<5%) could be ignored as it was of minor importance bank on the same side of a creek at a similar distance (c. 1.5 m) (R. Dolch and T. Tscharntke, unpublished work). from the water. Individual alders can be regarded as genetic indi- We assessed stem diameter as a measure of tree age and related viduals, since the probability of alders growing along creeks to be it to subsequent leaf damage caused by A. alni. Distances between ramets of the same clone is almost zero (A. Bogenrieder and E. the defoliated tree and the non-defoliated trees at each site were Dister, personal communication). In order to standardize sampling also measured. By 1 week after manual defoliation, the ten defoli- and minimize possible effects of water or air currents, at each site ated alders had reflushed. Leaf flush of all 100 alders was assessed one randomly selected tree was chosen for defoliation. Nine alders in the field 7 and 37 days after defoliation as the percentage of to- located immediately downstream were not defoliated. Additional- tal leaf area accounted for by freshly flushed leaves on the three ly, we chose sites with trees whose diameters were distributed ran- lowest branches. domly (no relationship between position and diameter) and varied In the laboratory, oviposition and feeding preference tests were only slightly. carried out, using equal-aged leaves from each site (n=10 repli- Manual defoliation involved one randomly selected tree at cates). Six leaves from the manually defoliated tree, six from its each site in order to simulate damage by herbivory. Manual defoli- nearest neighbour and six from the farthest tree (all from the low- ation of trees is known to induce less resistance than herbivore est branches) were taken at each site. They were put into six petri damage (Haukioja and Neuvonen 1985; Hartley and Lawton dishes with one leaf of each tree per dish. Female beetles (six per 1987), but is useful for simulating herbivory. Damage was caused dish) could choose between those leaves for oviposition. In the by stripping 20% of the tree's leaves from the lower branches of feeding experiments, adult beetles (6 per dish) or larvae (12 per the canopy, up to a height of about 2.5 m. The majority of the dish) could choose between the leaves for consumption. leaves were crunched and torn, but care was taken not to affect In an additional feeding experiment (n=10 replicates) adult buds, damage to which might mask effects purely due to defolia- beetles and larvae had the choice between a young and an old leaf tion (e.g. Haukioja et al. 1990). Distance between the defoliated per petri dish. In all experiments, beetles were removed after 24 h alder and the farthest non-defoliated one averaged 10.8 m, and ac- and consumed leaf area was assessed using templates of known cordingly, average distance between the trees was 1 m. surface area. Manual defoliation took place in early May 1994, before adult beetles had colonized the alders. Resprouting occurred before the beetles started ovipositing. Oviposition usually takes place from May to June and larvae hatch after 9Ð15 days. After three larval stages, adults of the new generation emerge at the end of July, feed Results until September and begin hibernating in the ground near their host trees soon after. Damage pattern in the field Leaf damage by A. alni was estimated as percentage of total leaf area consumed on all trees at each of the ten sites (n=100 trees) at six dates between May and September 1994 (0, 7, 37, 52, Leaf damage was best explained by distance from the 81 and 133 days after manual defoliation). One site became inac- manually defoliated tree. Until 37 days after defoliation, cessible during the last two dates (81 and 133 days after defolia- distance and leaf damage were significantly positively tion), so that estimation of leaf damage was restricted to the re- maining 9 sites on these dates (n=90 trees). Results were also correlated in eight of ten alder stands (Table 1). When compared with estimated leaf damage at all ten sites before defoli- data from all sites were pooled, this correlation held until ation. 81 days after defoliation (Fig. 1). When regression was Since unconscious biases in the estimation of leaf damage performed without considering the manually defoliated might affect the detected pattern, we tested the fidelity of field es- timates for leaf damage. We first estimated the percentage of leaf trees, slopes of the graphs were lower and regression co- damage on ten randomly chosen alders that were not included in efficients were smaller, but there was almost no effect on our experiment. For all branches up to a height of 2.5 m, the total significance levels (Table 2).

Table 1 Dependence of leaf damage (%) by alder leaf beetle Time after defoliation (days) on distance (m) to the manually defoliated tree at all different Site 0 7 37 52 81 133 sites. Correlation coefficients (r) are given for six different 1 Ð0.213n.s. +0.719*1 +0.534n.s. Ð0.284n.s. +0.126n.s. +0.188n.s. dates. Regressions were calcu- 2 Ð0.113n.s. +0.857*3 +0.788*2 +0.409n.s. +0.314n.s. +0.189n.s. lated with original data (n=10 3 No damage +0.809*3 +0.899*4 +0.718*1 +0.444n.s. +0.230n.s. sites, with 10 trees at each site) 4 +0.524n.s. +0.701*1 +0.757*1 +0.264n.s. no data no data 5 Ð0.243n.s. +0.795*2 +0.769*2 +0.790*2 +0.881*4 +0.844*2 6 Ð0.612n.s. +0.849*3 +0.811*3 +0.808*3 +0.852*3 Ð0.153n.s. 7 Ð0.427n.s. +0.363n.s. +0.663*1 +0.760*1 +0.466n.s. +0.132n.s. n.s. 3 3 2 n.s. n.s. 1 2 8 +0.713 +0.858* +0.827* +0.766* Ð0.508 Ð0.069 * P<0.05, * P<0.01, n.s. 3 3 1 1 n.s. 3 4 9 +0.616 +0.864* +0.829* +0.670* +0.653* +0.454 * P<0.005, * P<0.001, n.s. not 10 Ð0.450n.s. +0.255n.s. +0.558n.s. +0.449n.s. +0.282n.s. +0.331n.s. significant 506 Fig. 1AÐF Relationship be- tween leaf damage by Agelasti- ca alni to Alnus glutinosa and distance from the manually de- foliated tree in the field. Data are pooled, damage and dis- tances averaged for all sites. Analysis of variance was per- formed for different distances. SEs are shown to indicate vari- ation in leaf damage between sites. Leaf damage is arcsine- transformed. Regression analy- sis: A control: before defolia- tion, the amount of leaf damage within each plot was randomly distributed (F=1.16, n=100, P=0.28). B 7 days after defolia- tion: ln y=2.22+0.09x; F=126.8, r2=0.941, n=10, P<0.001 (original data: ln y=0.61+0.18x; F=42.53, r2=0.303, n=100, P<0.001). C 37 days after defoliation: ln y=2.31+0.09x; F=55.80, r2=0.875, n=10, P<0.001 (ln y=0.90+0.16x; F=41.47, r2=0.297, n=100, P<0.001). D 52 days after defoliation: ln y=2.91+0.06x; F=14.34, r2=0.642, n=10, P=0.005 (ln y=1.97+0.14x, F=38.94, r2=0.284, n=100, P<0.001). E 81 days after defoliation: ln y=3.38+0.03x; F=17.24, r2=0.683, P=0.003 (ln y=2.96+0.07x; F=12.21, r2=0.122, n=90, P<0.001). F 133 days after defoliation; the distribution pattern of leaf damage could no longer be ex- plained by tree distance (F=1.28, n=90, P=0.26)

Table 2 Dependence of leaf 2 damage (%) by alder leaf beetle Time after defoliation Equation FrP on distance (m) to the manually (days) defoliated tree, including (a) or not including (b) the manually 0 (a) Ð 1.16 Ð 0.28 defoliated tree. Regressions are (b) Ð 0.40 Ð 0.53 based on (a) n=100 trees (10 7 (a) ln y=2.22+0.09x 126.8 0.941 <0.001 sites with 10 trees each) and (b) (b) ln y=2.15+0.09x 79.67 0.919 <0.001 n=90 trees (10 sites with 9 trees 37 (a) ln y=2.31+0.09x 55.80 0.875 <0.001 each) (b) ln y=2.30+0.08x 38.74 0.847 <0.001 52 (a) ln y=2.91+0.06x 14.34 0.642 0.005 (b) ln y=3.02+0.04x 8.39 0.055 0.023 81 (a) ln y=3.38+0.03x 17.24 0.683 0.003 (b) ln y=3.38+0.03x 9.14 0.057 0.019 133 (a) Ð 1.28 Ð 0.26 (b) Ð 0.45 Ð 0.50 507 Table 3 Dependence of leaf damage (%) by alder leaf beetle on (%). Regressions were calculated with original data (n=100, i.e. 10 tree characteristics in the field. Correlation coefficients (r) are giv- sites with 10 trees each). Log-transformed and untransformed en for six different dates. Effects of distance from the defoliated variables were tested tree (m), stem diameter (cm), and leaf flush (%) on leaf damage

Factor Regression model Time after defoliation (days)

0 7 375281133

Distance from defoliated tree Exponential Ð0.089n.s. +0.550*4 +0.545*4 +0.533*4 +0.349*4 +0.118n.s. Stem diameter Multiplicative Ð0.316*4 Ð0.329*4 Ð0.293*3 +0.063n.s. Ð0.141n.s. +0.045n.s. Leaf flush Linear No data No data Ð0.233*1 No data No data No data

*1P<0.05, *2P<0.01, *3P<0.005, *4P<0.001, n.s. not significant

Fig. 2A,B Effects of defoliation on leaf damage by beetles and distance from the defoliated tree alone explained 26.73% leaf flush in alders in the field. A Relationship between the coeffi- of the variance 7 days after defoliation. Adding stem di- cient of determination (r2 of leaf damage vs. distance from the de- foliated tree) and time after defoliation. Coefficients obtained ameter to the model, both parameters taken together ac- from correlations performed with original data (Fig. 1) were used: counted for 29.72% of the variance. Values for the assess- y=36.38-0.26x; F=26.74, r2=0.899, n=5, P<0.01. B Relation be- ment 37 days after defoliation are 22.49% and 27.27%, tween leaf flush and distance from the defoliated tree, 37 days af- respectively. Leaf flush was also measured 37 days after ter defoliation: y=23.96-1.93x; F=11.53, r2=0.591, n=10, P<0.01. Data are pooled for all ten sites, distances and leaf flush are aver- defoliation. In a simple regression it was negatively cor- aged (original data: F=13.43; r2=0.132; n=100; P<0.001). Leaf related with leaf damage (Table 3), but if included in the flush (%) is arcsine-transformed. multiple model, leaf flush had no effect on leaf damage at all (F=0.62; P=0.43). Leaf flush decreased with distance from the manually defoliated tree (Fig. 2B). Correlation between leaf damage and distance was best shortly after defoliation and then progressively de- clined (Fig. 1BÐF). Accordingly, coefficients of determi- Oviposition and feeding experiments nation of the correlation between leaf damage and dis- tance from the defoliated tree were negatively correlated Laboratory feeding and oviposition experiments support- with time (Fig. 2A). Likewise, regression slopes signifi- ed our findings from the field. Females laid the fewest cantly declined (from 0.18 to 0.07, Fig. 1). Absolute leaf eggs on leaves of the manually damaged tree (Fig. 3A). damage steadily increased during the year, as can be no- Leaves of the immediately neighbouring tree were more ticed by the increase of the intercepts with time (Fig. 1). often selected for oviposition, while most eggs were laid Damage caused by A. alni to alders was also negative- on leaves from the most distant alder. A similar pattern ly correlated with tree diameter until 37 days after defoli- was found for clutch size (Fig. 3B). ation (Table 3), i.e. younger trees were more heavily Beetles did not discriminate between equal-aged alder damaged by the beetles than older ones. Later than leaves of different origin before defoliation (Table 4). In 37 days after defoliation, leaf damage caused by the bee- contrast, 7 and 37 days after defoliation, they showed a tles was not significantly affected by tree diameter. Stem significant preference for leaves from the farthest tree diameter explained much less of the variance than the compared to leaves of close neighbours of the manually distance from the defoliated tree alone, and distance of al- defoliated tree, whereas leaves of the defoliated tree ders to the manually defoliated tree was a much better were consumed least. predictor for leaf damage than stem diameter. Based on By 52 and 81 days after defoliation, adults had com- multiple linear models using original data (n=100 trees), pletely disappeared in the field in favour of larvae. Adult 508 Fig. 3A,B Number of eggs laid by A. alni on leaves of different origins, i.e. from trees at differ- ent distances from the defoliat- ed tree, in laboratory oviposi- tion trials (37 days after defoli- ation). ANOVA: arithmetic means and SEs are shown. A Number of eggs per leaf: F=15.0, n=10 sites, P<0.001. B Number of eggs per clutch: F=5.92, n=10 sites, P<0.005

Table 4 Laboratory feeding experiments comparing leaf con- defoliated, nearest neighbour, most distant alder) for 6 different sumption of defoliated, nearby, and distant alders. Analysis of dates. Values marked with identical superscript letters are not sig- variance for consumed leaf area (cm2) of three origins (manually nificantly different

Time after defoliation Mean leaf area consumed (cm2) (days) Manually defoliated Nearest neighbour Most distant df F rs 0 0.47 0.38 0.27 9 0.78 Ð0.135 7 0.58a 1.08a 7.92b 9 22.61*4 +0.533*4 37 0.15a 0.58ab 0.81b 9 4.35*2 +0.216*3 52 1.77a 0.78b 0.51b 9 5.36*2 Ð0.189*2 81 9.00a 4.61b 3.54b 9 5.93*3 Ð0.165*2 133 3.57 2.54 3.01 9 0.77 Ð0.070

*1P<0.05, *2P<0.01, *3P<0.005, *4P<0.001 beetles significantly preferred young leaves (brighter, ter defoliation younger trees appeared to be more prone softer) to older leaves (darker, harder) (F=10.0, n=10, to herbivore damage than older ones. This supports the P<0.005), while larvae of A. alni showed a reverse pref- idea that fast-growing young trees invest more energy in erence (F=17.0, n=10, P<0.001). We therefore tested growth rather than in defence (Coley et al. 1985; Karban leaves of equal age in all experiments to meet these pref- 1990). erences. In agreement with these laboratory bioassays, The major factor explaining leaf damage caused by the larvae in the field (52Ð81 days after defoliation) pre- herbivory was clearly distance from the manually defoli- ferred leaves of the manually defoliated tree, resulting in ated tree. We hypothesize that manual defoliation of negative correlations between percentage of consumed Alnus glutinosa may have led to specific plant responses leaf area and distance from the defoliated tree (Table 4). that are reflected in the spatial and temporal damage pat- In conclusion, laboratory feeding experiments sup- tern found. ported the notion of A. alni feeding preference for trees First, we suspected that direct plant compensation had farthest away from the defoliated tree up to 37 days after influenced the outcome, since leaf flush was negatively defoliation, whereas in the field this pattern was found correlated with distance from the manually defoliated until 81 days after manual damage was applied (Fig. 1). tree. Alders are known to sprout constantly during the year and show no seasonal differences in the timing of leaf flush, but after defoliation plants usually reallocate Discussion reserves from roots to shoots to make up for their loss of leaf area (Caldwell et al. 1981), i.e. they are expected to Our results indicate that manual defoliation of alders have higher rates of leaf flush. However, even the imme- leads to a reduction in subsequent herbivory by their ma- diate neighbours of defoliated alders had higher rates of jor antagonist, the leaf beetle A. alni, not only on the de- leaf flush than trees further away Ð despite much lower foliated trees themselves but also on their undamaged levels of leaf damage. It could be argued that, due to the neighbours. Variation in leaf damage in the field was higher leaf flush on nearby trees, the percentage of leaf best explained by distance from the defoliated tree, damage would be lower on those than on the distant al- whereas a weak negative correlation of leaf damage with ders, all other factors being equal. We therefore calculat- alder stem diameter (and leaf flush) explained only a lit- ed leaf damage without taking the freshly grown leaves tle of the variance (Table 3). During the first 5 weeks af- into account, to examine whether the damage pattern is 509 simply because of the reflushing effect. The regression pattern of reduced herbivore preference seem plausible: for leaf damage on old leaves only (ln y=1.15+0.15x; alders may react to defoliation either by (1) growth com- F=28.90, r2=0.247, n=100, P<0.001) is very similar to pensation only, or (2) defensive traits (Ruess et al. 1983; the previous result (Fig. 1C). In addition, leaf flush had Van der Meijden et al. 1988), termed “civilian” and “de- no effect on leaf damage in a multiple model. Thus, we fensive” responses, respectively (Karban and Baldwin can rule out the hypothesis that the damage pattern was 1997). It is difficult to know where to draw the line be- caused by direct reflushing effects. As hardly any other tween these two types of plant responses. Whether the phytophagous insects were present, this pattern can al- characteristic damage pattern is a mere effect of defolia- most exclusively be attributed to herbivory of A. alni. tion followed by inevitable biochemical processes or an This conclusionis corroborated by the pattern found active response to herbivory is interesting from an evolu- for both beetle feeding and ovipositing, as leaves of the tionary perspective, but the effects on the herbivores re- most distant tree were preferred to those of the neigh- main the same. bours of the damaged one (Fig. 3). As many insect herbi- Biochemical rearrangement directed to compensation vores are able to discriminate among plants of different after defoliation appears to be one explanation for the quality, these results support findings of Reznik (1991), detected damage pattern. Defoliation obviously alters the who found that female chrysomelids laid fewer eggs on chemistry of a plant by changes in photosynthetic activi- damaged host plants. Results of our laboratory experi- ty, activation of new meristems or reallocation of re- ments suggested that preference for leaves from the most sources from roots to shoots (Caldwell et al. 1981). Nu- distant trees may have led to augmented oviposition and trient stress may also profoundly shift the carbon/nutri- to correspondingly higher larval densities and higher ent balance of a plant, so that nitrogen-deficiency caused amounts of damage on those trees in the field. by defoliation could lead to an increase in the production More than 52 days after defoliation, positive correla- of carbon-based allelochemicals that render plant re- tions between leaf damage and distance from the defoli- sponses defensive (Tuomi et al. 1984, 1990; Honkanen ated tree decreased greatly and regression slopes of the et al. 1994). Shifts in carbon and nutrient reserves after corresponding equations steadily declined (Fig. 2A). Per- defoliation of alders are thought to drive induced resis- centage of leaf area consumed and distance from the de- tance (Williams and Myers 1984). From this study, one foliated tree were negatively correlated, after larvae be- might conclude that induction of resistance depended on gan to prefer leaves of the manually defoliated tree proximity to the manually defoliated alders, since subse- (Table 4). Evaluation of the relative importance of these quent beetle herbivory was reduced on both their own results should take into account the different feeding and their neighbours' leaves. preferences of larvae and adult beetles and consider two Again, there are two possible explanations for that re- points: sult: (a). Larvae, which are known to be relatively inactive (Barker et al. 1995) and which were more abundant on 1. Alders may have compensated for loss of foliage by the most distant trees, were not able to switch to more intensified absorption of mineral nutrients. In this preferred leaves in the field. Forced to remain on the case, they would reduce the availability of those nu- most distant trees, they presumably consumed less (indi- trients for trees in the immediate vicinity. This effect cated by the bioassays in the laboratory). would even be more pronounced if root connections (b). Adults of the new generation were able to switch were present, since proximal trees are likely to be to the manually defoliated trees. Since young leaves Ð more connected to the damaged alder than distant their preferred food Ð were more available there trees. As a consequence, neighbouring alders could (Fig. 2b), they consumed more.Both decreased feeding react by producing lower-quality foliage, leading to of larvae on the most distant trees, and increased feeding reduced levels of herbivory depending on their dis- of adults on the manually defoliated trees from the tance. Defoliation-triggered increases in resource ac- 52nd day onwards, may have added to the decreasing quisition by roots and intensified nutrient absorption correlation between leaf damage and distance from the of defoliated trees may reduce foliage quality (and manually defoliated tree in the field: 133 days after defo- thus herbivory) of undamaged neighbours (Tuomi et liation, A. alni no longer discriminated between different al. 1990). leaves. Therefore, manual defoliation clearly had an ef- 2. Our results could indicate signal transfer between fect on subsequent herbivory Ð at least within 37 days af- woody plants in the field. Concentration of possible ter the procedure, when damage was found to increase chemical cues involved in this transfer would de- with distance from the defoliated tree in both field crease with distance and therefore lose effect, result- (Fig. 1) and laboratory situations (Table 4). In the field, ing in a pattern similar to the detected one. This this effect even held within 81 days after defoliation. would suggest that alders communicate as pointed out In this study, herbivory by A. alni after manual defoli- in the talking tree hypothesis (Fowler and Lawton ation was found to be reduced and, following the defini- 1985). Instead of speaking of “talking trees”, the tion of Karban and Baldwin (1997), reduced herbivore short-hand phrase “listening trees” may be more ap- preference for a plant should be named induced resis- propriate, emphasizing that individual trees benefit by tance. Two principal interpretations in explaining this “listening” to the transmitted signals without answer- 510 ing. In this context, several ways of communicating References (signalling) seem possible. Signal transfer may either occur through the soil or via airborne chemical cues. Baldwin IT, Schultz JC (1983) Rapid changes in tree leaf chemis- Soil-borne transfer might be linked to root connec- try induced by damage: evidence for communication between plants. Science 221:227Ð279 tions or mycorrhizae (e.g. Simard et al. 1997), where- Barbosa P, Letourneau D (1988) Novel aspects of insect-plant in- as airborne transfer has to involve emission of volatile teractions. Wiley, New York cues. Rhoades (1983) found induced resistance in un- Barker AM, Wratten SD, Edwards PJ (1995) Wound-induced damaged neighbours of defoliated trees, that were not changes in tomato leaves and their effects on the feeding pat- connected via their roots. As systemic signal transfer terns of larval lepidoptera. Oecologia 101:251Ð257 Baur R, Binder S, Benz G (1991) Nonglandular leaf trichomes as within the vascular system of a plant may be limited short-term inducible defense of grey alder, Alnus incana, (Jones et al. 1993 and references therein), aerial dis- against the chrysomelid beetle, Agelastica alni. Oecologia persion of volatiles may lead to faster transfer to other 87:219Ð226 tissues of the same tree. Selection pressures for the Bruin J, Sabelis MW, Dicke M (1995) Do plants tap SOS signals from their infested neighbours? Trend Ecol Evolut 10:167Ð production of volatiles may include (a) within-plant 170 induction of resistance, but also (b) masking of defo- Caldwell MM, Richards JH, Johnson DA, Nowak RS, Dzurec RS liated trees by non-nutritious neighbours and (c) at- (1981) Coping with herbivory: photosynthetic capacity and re- traction of parasitoids or predators (e.g. Dicke 1994; source allocation in two semiarid Agropyron bunchgrasses. Oecologia 50:14Ð24 Bruin et al. 1995; Turlings and Benray 1998). Parasi- Coley PD, Bryant JP, Chapin FS (1985) Resource availability and toid attraction by possible chemical cues seems to be plant antiherbivore defence. Science 230:895Ð899 very specific (Demoraes et al. 1998), and is improba- Demoraes CM, Lewis WJ, Pare PW, Alborn HT, Tumlinson JH ble in this study, since rearing of A. alni in the labora- (1998) Herbivore-infested plants selectively attract parasito- tory did not produce a single parasitoid, and neither ids. Nature 393:570Ð573 Dicke M (1994) Local and systemic production of volatile herbi- parasitoids nor predators were found to be more abun- vore-induced terpenoids: their role in plant-carnivore mutual- dant near the source of damage in the field (R. Dolch ism. J Plant Physiol 143:465Ð472 and T. Tscharntke, unpublished work). Ecker JR, Davis RW (1987) Plant defense genes are regulated by ethylene. Proc Natl Acad Sci 84:5202Ð5206 Farmer EE, Ryan CA (1990) Interplant communication Ð airborne In conclusion, we have shown that manual defoliation methyl jasmonate induces synthesis of proteinase-inhibitors in affects herbivory of A. alni on Alnus glutinosa. Subse- plant leaves. Proc Natl Acad Sci 87:7713Ð7716 quent herbivory by Agelastica alni was reduced. Since Fowler SV, Lawton JH (1985) Rapidly induced defenses and talk- “responses that reduce herbivore [...] preference for a ing trees: the devil's advocate position. Am Nat 126:181Ð195 Fritz RS, Simms EL (1992) Plant resistance to herbivores and plant, are termed induced resistance” (Karban and Bald- pathogens Ð ecology, evolution, and genetics. University of win 1997), we conclude that manual defoliation leads to Chicago Press, Chicago induced resistance in Alnus glutinosa. In addition, up to Hartley SE, Lawton JH (1987) Effects of different types of dam- 37 days after defoliation, distance from the defoliated age on the chemistry of birch foliage and the responses of birch-feeding insects. Oecologia 74:432Ð437 tree was highly positively correlated with (1) leaf dam- Haukioja E (1990) Induction of defenses in trees. Annu Rev Ento- age in the field, (2) extent of leaf consumption in labora- mol 36:25Ð42 tory feeding preference tests, and (3) number of oviposit- Haukioja E, Neuvonen S (1985) Induced long-term resistance of ed eggs per leaf in another laboratory test. Thus, suitabil- birch foliage against defoliators, defensive or incidental? Ecol- ogy 66:1303Ð1308 ity for A. alni was not only reduced in the defoliated al- Haukioja E, Suomela J, Neuvonen S (1985) Long-term inducible ders themselves but also in their undamaged neighbours. resistance in birch foliage: triggering cues and efficacy on a Yet, irrespective of underlying mechanisms (whether an defoliator. Oecologia 65:363Ð369 incidental consequence of defoliation or an active pro- Haukioja E, Ruohomäki K, Senn J, Suomela J, Walls M (1990) cess), the ecological outcome remains the same. Whether Consequences of herbivory in the mountain birch (Betula pu- bescens ssp tortuosa): importance of the functional organiza- warned by signals from their damaged neighbours, or tion of the tree. Oecologia 82:238Ð247 just affected by nutrient competition with them, undam- Hedin P (1983) Plant resistance to insects. American Chemical aged alders appeared to develop less attractive leaves, re- Society, Washington DC sulting in reduced herbivory, depending on the distance Honkanen T, Haukioja E, Suomela J (1994) Effects of simulated defoliation and debudding on needle and shoot growth in from damaged conspecifics. This may render alders un- Scots pine (Pinus sylvestris): implications of plant source/sink attractive for herbivores and enable them to reduce fur- relationships for plant-herbivore studies. Funct Ecol ther herbivore attack, resulting in decreased insect dam- 8:631Ð639 age. Thus, apart from significantly affecting the herbi- Jeker TB (1981) Durch Insektenfraß induzierte, resistenzähnliche Phänomene bei Pflanzen (Doctoral thesis 6895). Eidgenössi- vore, transfer of induced resistance may strongly con- sche Technische Hochschule, Zürich tribute to the plants' escape from herbivory. Jones CG, Hopper JH, Coleman JS, Krischik VA (1993) Control of systemically induced herbivore resistance by plant vascular Acknowledgements We are grateful for suggestions and helpful architecture. Oecologia 93:452Ð456 comments by Warren G. Abrahamson, Ian T. Baldwin, Arno Karban R (1990) Herbivore outbreaks on only young trees: testing Bogenrieder, Jan Bruin, Emil Dister, Erkki Haukioja, Monika hypotheses about aging and induced resistance. Oikos 59:27Ð Hilker, Andreas Kruess, Carsten Thies and one anonymous re- 32 viewer. We also thank Art Souther for further remarks and espe- Karban R, Baldwin IT (1997) Induced responses to herbivory. cially for assistance in linguistics. University of Chicago Press, Chicago 511 Reznik SY (1991) The effects of feeding damage in ragweed Am- Shonle I, Bergelson J (1995) Interplant communication revisited. brosia artemisiifolia (Asteraceae) on populations of Zyg- Ecology 76:2660Ð2663 ogramma suturalis (Coleoptera, Chrysomelidae). Oecologia Simard SW, Perry DA, Jones MA, Myrold DD, Durall DM, 88:204Ð210 Molina R (1997) Net transfer of carbon between ectomycor- Rhoades DF (1983) Responses of alder and willow to attack by rhizal tree species in the field. Nature 388:579Ð582 tent caterpillars and webworms: evidence for pheromonal sen- Tallamy DW, Raupp JH (1991) Phytochemical induction by herbi- sitivity of willows. In: Hedin P (ed) Plant resistance to insects. vores. Wiley, New York American Chemical Society, Washington, DC, pp 55Ð68 Tuomi J, Niemelä P, Haukioja E, Sirén S, Neuvonen S (1984) Nu- Rhoades DF (1985) Pheromonal communication between plants. trient stress: an explanation for plant anti-herbivore responses Rec Adv Phytochem 19:195Ð218 to defoliation. Oecologia 61:208Ð210 Rosenthal GA, Janzen DH (1979) Herbivores Ð their interaction Tuomi J, Niemelä P, Sirén S (1990) The Panglossian paradigm and with secondary plant metabolites. Academic Press, London delayed inducible accumulation of foliar phenolics in moun- Ruess RW, McNaughton SJ, Coughenour MB (1983) The effects tain birch. Oikos 59:399Ð410 of clipping, nitrogen source and nitrogen concentration on the Turlings TCJ, Benray B (1998) Effects of plant metabolites on the growth responses and nitrogen uptake of an East African behavior and development of parasitic wasps. Ecoscience sedge. Oecologia 59:253Ð261 5:321Ð333 Schultz JC (1988) Plant responses induced by herbivores. Trends Van der Meijden E, Wijn M, Verkaar H (1988) Defence and re- Ecol Evol 3:45Ð49 growth, alternative plant strategies in the struggle against her- Seldal T, Dybwad E, Andersen KJ, Högstedt G (1994) Wound-in- bivores. Oikos 51:355Ð363 duced proteinase inhibitors in grey alder (Alnus incana): a de- Williams KS, Myers JH (1984) Previous herbivore attack of red fense mechanism against attacking insects. Oikos 71:239Ð alder may improve food quality for fall webworm larvae. 245 Oecologia 63:166Ð170