1To determine if the response of trees to herbivory is related to their ability to store nitrogen, saplings of Pinus sylvestris L., Betula pendula Roth. and Sorbus aucuparia L. were clipped, either when dormant or at spring bud burst, to remove half the previous year’s shoot growth. The impact of clipping on N remobilization and uptake was quantified in relation to their growth responses.
2Pinus sylvestris stored N during the winter in needles grown the previous summer, whereas B. pendula and S. aucuparia used their woody roots and older stems. Therefore, N remobilization was unaffected by clipping the deciduous species, but was reduced by half in the evergreen.
3For both P. sylvestris and B. pendula, root uptake contributed N for leaf growth immediately after bud burst, concurrently with remobilization. Sorbus aucuparia had remobilized half the N from storage before any N was taken up by the roots.
4Pinus sylvestris, which has a fixed pattern of growth, produced a lower total needle mass when clipped, but individual needles were heavier. Nitrogen uptake during summer was reduced by 26 and 44% for winter- and spring-clipped saplings, respectively. Both deciduous species showed compensatory leaf growth such that, by the end of the summer, leaf mass (and B. pendula leaf area) were unaffected by clipping. Nitrogen uptake by both deciduous species was unaffected by clipping.
5We conclude that the site of N storage during winter is a crucial factor in determining the response of a sapling to herbivory, as interactions with the growth patterns of the three species can lead to different responses to damage. We suggest that for evergreen trees, the production of antiherbivory compounds serves primarily to protect N stored in their foliage, rather than the leaves themselves.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Pinus sylvestris L., Betula pendula Roth. and Sorbus aucuparia L. are common species in native, temperate woodlands (Rodwell 1991), and their natural regeneration in hill and upland areas is often limited due to damage caused by browsing herbivores (Gill 1992c; Humphrey et al. 1998). From the few published studies to date, browsing damage appears to be more detrimental to the growth of P. sylvestris, and saplings of this species appear more likely to be killed by browsing damage than either B. pendula or S. aucuparia (Miller et al. 1982). However, the reasons for these differences are not fully understood.
The physiological responses of P. sylvestris to browsing damage are variable, and have been shown to depend on factors such as the extent of biomass removed and the timing of the damage. Honkanen, Haukioja & Suomala (1994), for example, found that bud damage resulted in positive growth responses, whereas needle damage could result in stimulated or inhibited growth depending on the relative position of the shoot and the timing of defoliation. Furthermore, defoliation of 1-year-old needles (Honkanen, Haukioja & Kitunen 1999) reduced the mean length of needles in new shoots, whereas removal of 2-year-old needles had no effect upon current needle growth. Betula spp. also respond to herbivory in a variety of ways according to timing, severity and location of damage. For example, moderate browsing of Betula pubescens and B. pendula was reported by Danell & Huss-Danell (1985) to produce trees with fewer leaves, but which were larger, heavier and contained more nitrogen than leaves from control trees. Removal of apical buds in the winter from B. pubescens also resulted in the production of larger leaves (Senn & Haukioja 1994), but bud removal at bud burst in spring produced little or no such response. Hjaltén, Daniel & Ericson (1993) also found that removal of apical dominance by browsing can lead to overcompensation in growth. Sorbus aucuparia is widely quoted as being a highly preferred browse species (Bergström & Danell 1987; Gill 1992a,Gill 1992b; Miller et al. 1982), but the physiological responses of this species to browsing damage have been little studied (Gill 1992c).
In the experiment reported here, we examined the impact of simulated mammalian browsing damage on the internal cycling of N in saplings of P. sylvestris, B. pendula and S. aucuparia. Saplings below 1 m high were selected for study because at this stage all the above-ground biomass is vulnerable to damage by large mammals. Saplings were grown in sand culture for 2 years, during the first of which they were supplied with 15N. In the second year the trees received N at natural abundance and were clipped (to remove half of the previous year’s above-ground growth) either in January (when the trees were still dormant), or at bud break in March or April, or were left intact. During the spring and summer a series of destructive harvests were taken in order to: (i) determine the impact of clipping on the contribution of both N remobilization and N uptake for leaf growth; (ii) relate the growth responses of the three species following clipping to their sites of N storage; and (iii) determine the impact of the timing of damage in relation to bud burst on the amount of N remobilization.
Materials and methods
Two-year-old seedlings of Pinus sylvestris L., Betula pendula Roth. and Sorbus aucuparia L. (105 of each species) were lifted from a nursery while dormant, in March 1996, and planted in pots (30 cm diameter × 26 cm deep). The trees had previously received a moderate N supply. The pots were arranged randomly in a Latin square design in a greenhouse, and all trees were watered with 300 cm3 of a nutrient solution containing 3·0 mol N m−3 as 15NH415NO3 enriched with 15N to 2·8 atom percentage excess. Other nutrients were supplied as described by Millard & Proe (1991). Trees were watered every 2 days throughout the spring, summer and autumn of 1996. A natural photoperiod was used, and the greenhouse ventilated to provide temperatures close to ambient. Throughout the winter of 1996/97 the trees were kept frost-free (≥2 °C) and moist. In early January 1997 trees were carefully removed from the pots and all sand washed off the roots before being planted in fresh sand in a new, larger pot (40 cm diameter × 35 cm deep).
Morphological measurements to characterize sapling pre-clipping treatment variability were taken in January 1997, before application of the experimental clipping treatments, by measuring their height, the diameter of the main stem at the sand surface, the number of branches, the number of stems with lateral branches, the number of buds, and the number of shoots grown during 1996. Out of that list, a subset of measurements which, together, explained the highest proportion of between-sapling variability were then used as covariates in all subsequent data analyses, as described in the data analysis section below.
Thirty-five trees of each species were clipped on 14 January 1997 to remove half of the previous year’s (1996) shoot growth. This was designated the winter (dormant) clipping treatment. Clipping treatments were designed to mimic as closely as possible browsing damage by Red deer: removal of 1996 shoots (including buds/leaves) up to the maximum stem diameters normally eaten by Red deer for each of the three species (L.A. Shipley, unpublished data; Shipley et al. 1999). As the basal diameters of all 1996 shoots were within the ranges taken by Red deer, whole shoots were clipped on all species. If more than one 1996 shoot was present on a sapling, every second shoot was clipped to just above the point where 1996 growth was initiated, starting with the leader/top shoot. If only one 1996 shoot was present (some individuals of S. aucuparia), only the top 50% of that one shoot was removed. The dry weights and total number of buds removed in clippings from each tree were recorded. As the morphology of the three species is very different, the total number of shoots per sapling and the number and proportion of buds on old and new shoots differed between species. Thus, although the dry weight removed was similar for all species, the total numbers of shoots and buds removed and, importantly, the proportions of buds removed, differed between species.
From the January 1997 repotting onwards, trees were watered with 300 cm3 of nutrient solution, containing unlabelled N at natural abundance, once a week until the beginning of March, when the volume applied was increased to 500 cm3 and the frequency of application to every 2 days. A second clipping treatment was applied to a further 30 trees of each species on the date of their bud burst. This was designated as the spring (bud burst) clipping treatment. The timing was assessed by observing buds daily; for each individual tree, the day the first bud opened was noted and designated the date of bud burst. These observations were made on all trees in the experiment. The spring clipping treatment was applied in exactly the same way as the winter clip. Unlike the two deciduous species, in pine considerable bud elongation can occur before actual bud burst, which is likely to have physiological implications for this species, but it is not known how the response to damage might differ if this species were clipped at the time of first shoot elongation rather than bud burst. Therefore, for consistency between the three species we chose to clip all species at the time of emergence of the first needles/leaves.
Tree harvesting and analysis
At each of seven destructive harvests, five replicate saplings of each species were randomly selected. Saplings were removed from their pots and all remaining sand was gently washed from the roots. The P. sylvestris trees were then separated into 1995 needles; 1996 needles; current year’s (1997) needles; 1995 (and older) stem; 1996 stem; current year’s (1997) stem; woody roots; and fine roots. Both B. pendula and S. aucuparia were separated into leaves (1997); 1995 (and older) stem; 1996 stem; 1997 stem; woody roots (including tap roots); and fine roots. All plant samples were then freeze-dried, weighed and milled prior to 15N analysis. A Tracer MAT continuous flow mass spectrometer was used for determinations of both 15N enrichment and the total N concentrations in samples. The 15N enrichment was used to calculate the uptake of labelled N (Millard & Neilsen 1989). Recovery of labelled N in leaves grown during 1997 gave a direct measure of N that had been taken up during 1996 and remobilized for leaf growth during spring 1997.
Data for N uptake and remobilization were related to the stage of leaf development by using the number of days from bud burst as a measure of time. Rates of uptake and remobilization with time were analysed using exponential curve-fitting, with treatment as a grouping factor (genstat 5; Genstat 5 Committee 1993). Comparisons of proportional 15N data per sapling were made using paired t-tests (see Table 3). In view of the large variation in sapling sizes and morphology, all analyses of actual values (see Tables 4), rather than proportions, used residual maximum likelihood (REML) (genstat 5) with: (i) treatment (timing of clip) and harvest date as fixed factors; (ii) sapling number as a random factor; and (iii) three or four pretreatment morphological measurements (described below) as covariates. Wald tests were used to test for significance of main effects and interactions, and least significance difference tests were used to test differences between tables of means (genstat 5). Principal component analyses (genstat 5) were used to select the covariates by determining which pretreatment (January 1996) morphological measurements for each species made the most substantial contribution to the first two principal components and were also relatively uncorrelated. The following four measurements were identified for B. pendula and P. sylvestris: height; diameter of main stem; number of 1996 shoots; number of overwintering 1996/97 buds; for S. aucuparia measurements were the same except for height. The use of covariates in the REML analyses allowed explicit examination of the variation between individuals not caused by the treatments, together with analyses of treatment effects on the residual variation allowing for the covariates. This removed the risk of false ‘treatment differences’ caused by groups of larger or smaller saplings randomly occurring within any one harvest.
Table 3. Effect of winter or spring clipping on leaf growth characteristics at final harvest§
†SED value for loge-transformed means.
SED value for square root-transformed means.
Final harvest 8 September for P. sylvestris; 3 September for B. pendula and S. aucuparia.
All leaf/needle masses in g; leaf areas in cm2. Data are REML-adjusted means of five replicates. Where data were transformed for analysis (see superscript on SED), means have been back-transformed for clarity. Within each row means with the same superscript letter are not significantly different at P < 0·05. Significance values (least significant difference tests): ns, not significant; *, P < 0·05; **, P < 0·01; ***, P < 0·001.
Table 4. Uptake and partitioning of nitrogen by Pinus sylvestris, Betula pendula and Sorbus acuparia following winter or spring clipping
Unlabelled N content
†SED value for loge-transformed means.
SED value for square root-transformed means.
Values are given for unlabelled N content (mg per tree) of tissues from final harvest of the experiment (8 September for P. sylvestris; 3 September for B. pendula and S. aucuparia). Data presented are REML-adjusted means of five replicates. Where data were transformed for analysis (see superscripts on SED), means have been back-transformed for clarity. Within each row, means with the same superscript letter are not significantly different at P < 0·05. Significance values (least significant difference tests): ns, not significant; *, P < 0·05; **, P < 0·01; ***, P < 0·001.
The total biomass of 1996 shoots removed was approximately 50% for all species, but the number and proportion of buds removed per sapling differed between the three species, reflecting their different morphology (Table 1). Because of the architecture of the S. aucuparia trees, with many buds on older stems, removing half the previous year’s growth removed only about a quarter of the buds, as compared to about half the buds removed from other two species (Table 1). In addition to removing just over half the buds from the P. sylvestris trees, clipping removed 55% of the needle pairs from the previous year’s growth.
Table 1. Number and proportion of buds removed in shoot clippings from each species studied
Number of buds
Percentage removed by clipping
Total per sapling
Removed by clipping
Values are mean ± SE of 30 replicates of spring-clipped trees.
31·9 ± 2·35
18·5 ± 1·49
58·5 ± 1·98
288·5 ± 13·56
125·0 ± 6·11
43·4 ± 0·92
17·1 ± 1·05
5·1 ± 0·98
28·7 ± 1·92
Site of nitrogen storage
To determine which tissues of each species were used to store N during the winter, the 15N content of perennial tissues of control trees of each species was measured before bud break when the trees were dormant (14 January), and again after remobilization had finished (66 ± 1·4 days from bud burst for P. sylvestris; 56 ± 1·1 days from bud burst for B. pendula; 59 ± 1·5 days from bud burst for S. aucuparia), and expressed as a proportion of the total 15N content of the whole tree (Table 2). Pinus sylvestris stored N predominantly in the 1996 needles, although there were also smaller proportional decreases in the 15N content of 1995 needles and stem (formed before 15N addition) following remobilization. In contrast, B. pendula stored N predominantly in the woody roots and 1995 stem, and S. aucuparia in these tissues plus the 1996 stem (Table 2).
Table 2. Storage of nitrogen in perennial tissues of Pinus sylvestris, Betula pendula and Sorbus aucuparia trees during winter
Values are given for the proportional 15N content (± SEM) of tissues of control trees from the first harvest when the trees were dormant and in trees harvested after the majority of remobilization had finished (30 June for P. sylvestris; 12 May for B. pendula and S. aucuparia). Data are means ± SEM of five replicates. Significance values (t-test): ns, not significant, *, P < 0·05; **, P < 0·01; ***, P < 0·001.
Nitrogen remobilization and uptake for leaf growth
The provision of labelled N to trees during 1996 allowed the recovery of labelled N in the needles or leaves which grew in 1997 to be used as a direct measure of remobilization (Fig. 1). Remobilization started straight after bud burst in all three species, but continued for significantly longer in P. sylvestris than in the deciduous trees. Clipping significantly reduced the amount of remobilization by P. sylvestris (P < 0·001), with the spring-clipped trees being more affected than those clipped in the winter. The reduced remobilization following clipping presumably reflected the fact that the clipping treatments removed over half the 1996 needles, which were the main site of N storage during the winter (Table 2). Neither the amount nor the duration of N remobilization were affected by clipping of B. pendula or S. aucuparia (Fig. 1), because the N had been stored in older, woody tissues unaffected by the clipping.
The recovery of unlabelled N in the needles and leaves grown during 1997 represented N that had been taken up by the roots after bud burst in 1997. For both P. sylvestris and B. pendula, root uptake contributed N for leaf growth immediately following bud burst, concurrently with N remobilization (Fig. 1). In contrast, no unlabelled N was recovered in the leaves of S. aucuparia until 22 days from bud burst (Fig. 1), by which time some 53% of the N remobilization had already occurred. The rates and total amounts of uptake of unlabelled N by S. aucuparia differed slightly between treatments (P < 0·01), with winter-clipped saplings having the highest total uptake of unlabelled N (Fig. 1). The parameters from the curves in Fig. 1 were used to calculate when the amounts of N in leaves from remobilization and root uptake were equal. For P. sylvestris needles this was 65 days from bud burst in the control trees, and 36 and 24 days for the winter- and spring-clipped trees. For B. pendula the comparable value was after 61 days’ leaf growth irrespective of clipping treatment. Leaf growth by S. aucuparia was supported predominantly by remobilization of N for longer than in the other two species, up to 103 days for the control trees, and 91 and 108 days for the winter- and spring-clipped trees, respectively.
The effects of clipping treatments on individual leaf growth differed among the three species. Removal of half the previous year’s growth (including half the buds) from P. sylvestris resulted in a significantly lower mass of 1997 needles at the end of the growing season (P < 0·001; Table 3), reflecting the lower number of needle pairs produced (P < 0·001; Table 3) and the reduced amount of N remobilized for needle growth (Fig. 1). However, both clipping treatments significantly increased the mass per needle pair (P < 0·001; Table 3). The impacts on total 1997 needle mass and number of needle pairs were more severe for trees clipped at bud burst than trees clipped in winter (both P < 0·05; Table 3), also reflecting the reduced amounts of N remobilized for needle growth (Fig. 1). However, the clipping treatments had no significant effect on the number of needle pairs that grew per bud remaining on the tree after the spring clipping treatment was applied at bud burst (shown as the needle/bud ratio in Table 3). In contrast, clipping the B. pendula trees resulted in fewer leaves being produced (P < 0·001), but the mass per leaf and area per leaf were significantly greater (both P < 0·05). Thus there were no significant differences in total leaf mass per tree (Table 3) or total N uptake for leaf growth (Fig. 1) by the time of the last harvest in September. Betula pendula also responded to clipping by altering the leaf/bud ratio, with the spring-clipped trees producing significantly more leaves per bud than the control or winter-clipped trees (P < 0·01; Table 3). The clipped S. aucuparia trees produced significantly fewer leaves (P < 0·01), but area per leaf and mass per leaf were highly variable and did not differ significantly between treatments. Total leaf area per sapling was thus also highly variable, and was significantly smaller only in spring-clipped trees (P < 0·05). Total leaf mass did not differ significantly between treatments by the time of last harvest in September, as per the few significant differences in total N uptake for leaf growth (Fig. 1). Leaf/bud ratios did not differ significantly between treatments for this species, although values for the clipped trees were greater than for the controls.
Nitrogen uptake by whole trees
The amounts of unlabelled N recovered in the various tissues of the trees by final harvest are shown in Table 4. Clipping had significantly (P < 0·05; Table 4) reduced the amount of unlabelled N in P. sylvestris trees and, as a consequence, reduced recovery in both needles (P < 0·05) and stem (P < 0·01) which had grown during the year of clipping (1997), and in woody and fine roots (P < 0·05; P < 0·01). However, the proportion of total unlabelled N recovered in the 1997 needles was not greatly altered by clipping, being 40% for the control and 45% for both the clipped treatments. In contrast, clipping had no effect on total N uptake by B. pendula, but very slightly increased the amount of unlabelled N allocated to the growth of leaves (P < 0·05). Total N uptake by S. aucuparia was not significantly affected by clipping, but clipping in winter significantly reduced the amount of unlabelled N allocated to 1996 stem (P < 0·01) and new buds (P < 0·05).
Many studies have considered the consequences of herbivory on the carbon physiology of trees (Heichel & Turner 1983; Reich et al. 1993; Honkanen et al. 1999; Vanderklein & Reich 1999). Fewer studies have considered the impacts of herbivory on their N physiology (Holopainen et al. 1995; Lovett & Tobiessen 1993). Our data suggest that the different growth response to herbivory of the evergreen species compared to the deciduous trees was due to two main contrasting aspects of their physiology. First, storage of N in needles that are susceptible to browsing resulted in P. sylvestris trees having less N available for remobilization after they had been clipped, whereas the deciduous species were unaffected. Second, P. sylvestris has a fixed pattern of growth whereas both B. pendula and S. aucuparia had an indeterminate growth form, allowing compensatory growth.
In P. sylvestris, remobilization and root uptake were concurrent. By the time remobilization had finished, labelled N accounted for nearly half the N in the needles of control trees. The fact that P. sylvestris relies to such an extent on remobilization to provide N for needle growth might explain why Honkanen et al. (1999) found that defoliation of 1-year-old but not 2-year-old needles reduced the subsequent mass and length of needles growing on new shoots. These authors interpreted this response as being due to an alteration of shoot sink strength for carbon as a consequence of defoliation (Honkanen et al. 1999). Our data suggest that the removal of N stored in the 1-year-old needles could also be an important factor in determining the subsequent growth response of new foliage.
Clipping also reduced N uptake by P. sylvestris during the summer, while allocation patterns within the tree were unaltered. This effect on N uptake might be further exacerbated under field conditions, as above-ground herbivory can reduce both fine root productions (Ruess, Hendrick & Bryant 1998) and the extent of root colonization by mycorrhizal fungi (Rossow, Bryant & Kielland 1997). As a consequence, less N was recovered in the current year’s needles at the end of the summer, which in turn would probably mean that less N would be available for remobilization the following year.
Remobilization of N is unaffected by the current N supply to a range of tree species, being dependent only on the amount of N in store (Millard 1996). This suggests that the process of remobilization is source- rather than sink-driven. The present study provided further evidence for remobilization by B. pendula being source-driven. Betula pendula is heterophyllous, and winter buds contain only a limited number of leaf primordia which form the first population of ‘early’ leaves (Maillette 1982). We found that this initial population of early leaves remained constant until after the harvests taken on 12 May, 99 ± 2·3 days from bud burst. This corresponded with the time that remobilization finished. The control trees had produced 606 ± 3·2 leaves per tree, and the winter- and spring-clipped trees 329 ± 2·4 and 348 ± 4·1, respectively (control versus clipped, P < 0·001). The amount of N remobilized per leaf was 0·23 mg per leaf in the control trees, and 0·43 and 0·41 mg per leaf for the winter- and spring-clipped trees, respectively (control versus clipped; P < 0·01). No such relationship between bud and initial leaf numbers and the amount of N remobilized was found for S. aucuparia. However, the fact that leaf growth proceeded for some 22 days before N from root uptake was recovered in the leaves, and remobilization was unaltered by clipping, suggests that remobilization is likely to be source-driven in this species as well.
Nitrogen was stored by the deciduous trees in organs which were unaffected by clipping. In addition, as a source-driven process, remobilization provided the same amount of N in total for the buds remaining after clipping, allowing the fewer leaves that grew to contain more N and to have a greater area per leaf compared to the control trees (except for the spring-clipped S. aucuparia). The compensatory growth which both species were able to produce also meant that clipping had no effect on N uptake during the summer, with the amount of N allocated to storage also being unaffected. It is likely therefore that the amount of N available for remobilization the following year after damage would have been unaffected by the clipping.
Pinus sylvestris produces only preformed buds, and so has a fixed pattern of growth (Rook 1985). However, P. sylvestris is capable of compensatory growth in response to herbivory, although these responses depend on the timing and severity of the defoliation. For example, Honkanen et al. (1994) found that branch defoliation retarded the growth of foliage in whorls above the defoliated branches, but produced a positive effect on growth of needles below. Debudding resulted in no shoot extension and, as a consequence, existing needles increased in both length and mass (Honkanen et al. 1994). Edenius, Danell & Bergström (1993) reported that P. sylvestris was able to compensate for needle loss, but that regrowth was delayed in the year following intense clipping, which they suggested was a mechanism by which trees were temporarily released from further herbivory. We found that in the growing season following clipping P. sylvestris was unable to increase the number of needle pairs produced per bud, but that there was an increase in the mass of needles growing from the remaining buds.
Betula pendula saplings responded to removal of nearly half their buds by increasing both the area of individual leaves and the number of leaves produced per bud. However, the timing of our clipping treatments (at dormancy and bud burst) was important in determining these responses. Ovaska, Walls & Vapaavuori (1993a) found that saplings of B. pendula partially defoliated after some 4 weeks’ leaf growth were incapable of compensatory growth, which they ascribed to a decreased total carbon gain, despite upregulation of photosynthesis in the remaining leaf area (Ovaska et al. 1993b). In contrast Danell & Bergström (1989) applied clipping treatments to B. pendula before bud burst and, as in the present study, found no effect on shoot biomass.
A proportion of the buds produced by B. pendula remain dormant during spring and summer, and contribute to the population of buds for the next year (Maillette 1982). Our clipping treatments removed apical buds and so will have reduced apical dominance (Senn & Haukioja 1994), potentially affecting the subsequent growth of axillary buds, as found by Collin et al. (2000) for Fraxinus excelsior. Clipping to remove either buds or whole shoots of Fraxinus pennsylvanica was shown by Davidson & Remphrey (1994) to increase neoformed leaf production relative to control trees, thereby allowing them to re-establish their leaf area quickly after injury. Our data do not show if the compensatory growth we found in B. pendula was due to: (i) neoformed growth within existing buds; (ii) a decrease in the proportion of buds remaining dormant; or (iii) the production of new buds which, in turn, grew new leaves. However, given that compensatory growth by B. pendula is apparently not possible if defoliation occurs after the initial phase of leaf growth (Ovaska et al. 1993a), this suggests that the response we measured was unlikely to be regulated by the carbon balance of the plants. The same was probably true for S. aucuparia saplings, which also exhibited compensatory growth in response to clipping.
Consequences for tree growth
It is likely that the consequences of large mammal herbivory on P. sylvestris will last for several years, because: (i) clipping reduced growth that year and resulted in fewer buds being set in late summer (Rook 1985) to provide for growth next year; and (ii) the slower rate of N uptake after clipping resulted in there being less N available for remobilization the following year. In contrast, the two deciduous species showed considerable plasticity in their growth response to clipping, such that neither bud numbers nor N allocated to storage the following winter was much affected. Both species had almost recovered from the effects of simulated herbivory by the end of the summer. The site of N storage during the winter therefore appears to be an important factor in determining the response of a tree species to herbivory.
The carbon/nutrient balance hypothesis (Bryant et al. 1983; Bryant et al. 1991) suggests that species adapted to fertile sites will respond to herbivory by utilizing stored resources for compensatory growth, whereas slower-growing species adapted to nutrient-poor sites will instead protect their leaves by investing a greater proportion of their carbon in antiherbivory defence compounds. Trees whose growth is limited by nutrient availability rely on the internal cycling of nutrients for growth to a greater extent than those well supplied with nutrients (Millard 1996). Long-lived, slow-growing evergreen species are commonly found in the most nutrient-deficient sites, and because they store N in their foliage it is susceptible to loss by herbivory. In addition to protecting the leaves per se, we suggest that antiherbivory compounds may serve primarily to protect the nutrients stored within them.
We thank A. Midwood for the 15N analyses and the Scottish Executive Rural Affairs Department for funding the research.