Effects of defoliation and girdling on fruit production in Ilex aquifolium



1. The effects of defoliation and girdling on fruit production in European holly Ilex aquifolium were studied in a northern Spanish population. Three defoliation treatments (control, 50% and 100% leaf removal) were applied at branch level in 10 trees. Six branches were used per treatment in each tree and half of these branches were girdled (a ring of bark and cambium was removed from the branch base).

2. Leaf removal from ungirdled branches had little effect on fruit set, fruit production and reproductive allocation at branch level. However, these variables decreased as the extent of defoliation increased on girdled branches.

3. Fruit production in undefoliated branches did not differ between girdled and ungirdled branches, showing that girdled branches were autonomous for fruit production when undamaged. Mean fruit mass of girdled-100% defoliated branches averaged 8% of fruit mass produced by control branches.

4. Fruit production of ungirdled-100% defoliated branches did not differ from control branches, showing the importance of resource importation from other branches. Hence the branches may be regarded as only partially autonomous for fruit production.

5. Fruit production of girdled-50% defoliated branches was 42% of fruit production of control branches, while ungirdled-50% defoliated branches amounted to 70% of controls. These results also demonstrated resource importation but did not support the existence of photosynthetic compensatory mechanisms.

6. The ability of resource importation developed by fruiting branches might be a mechanism to reduce the effect of reproductive costs in this species.


Plant tolerance to herbivory is the ability of a plant to maintain its fitness even after tissue removal by herbivores (Rosenthal & Kotanen 1994). After leaf removal by herbivores, plants may compensate for loss of tissue by a number of mechanisms which include morphological and physiological components, such as increased photosynthetic rates and mobilization of storage reserves (Rosenthal & Kotanen 1994). The compensatory response may include allocation to sexual reproduction (Hendrix 1988). However, in woody plants, natural or simulated herbivory has been shown to result in a decreased fruit set, diminished seed number or lower seed quality (Obeso 1993; but see McCarthy & Quinn 1992; Obeso & Grubb 1993).

Reproductive responses of branches after defoliation might be interpreted with regard to partitioning of assimilate within the tree, which may be determined by physiological sinks and vascular architecture (Gifford & Evans 1981; Watson & Casper 1984; Watson 1986). If assimilate is driven by gradients of concentration, compensation for reproduction after defoliation in reproductive branches should be possible. However, these compensatory responses depend on the ability to translocate assimilates, which depend in turn on the extent of branch autonomy.

The responses of trees to artificial or natural defoliations have been extensively studied (e.g. Sprugel, Hinckley & Schaap 1991; Trumble, Kolodny-Hirsch & Ting 1993); however, such studies have not usually taken account of the combined effect of defoliation and girdling on the reproductive output. Following experimental defoliation of branches it has been demonstrated that the extent of dependence on other parts of the tree for fruit maturation is a species-specific response. The extent of carbohydrate importation after defoliation may be trivial (e.g. Stephenson 1980 in Catalpa speciosa), intermediate (e.g. Janzen 1976 in Gymnocladus dioicus) or even the main contribution to fruit maturation (Tuomi et al. 1988 in Betula pubescens). Girdling experiments have shown that carbohydrates may be superabundant above the girdle (Chalmers et al. 1975 in Prunus persica). Newell (1991) demonstrated that girdled branches of Aesculus californica, which fills its fruits during the leafless season, obtained most of the energy for fruit maturation from reserves stored in fruit-bearing shoots. Gorchov (1988) found in Amelanchier arborea that the manipulation of resources in the shoot at the time of fruit initiation by defoliation and girdling had no significant effect on seed number per fruit, although fruit set and seed mass were affected.

To examine the extent of branch independence in Ilex aquifolium L. (Aquifoliaceae) the effects of experimental defoliations and branch girdling during flowering on the reproductive output of the branches were studied. This paper reports: (1) the effects of defoliation and girdling on fruit set, fruit production and reproductive allocation at branch level, and (2) the effects of these factors on fruit traits.

Materials and methods


The holly is a dioecious broad-leaved evergreen tree native to Europe and North Africa (Peterken & Lloyd 1967). It flowers in April–May and the fruit, which ripens in autumn, is a drupe containing up to four pyrenes, which are typically bird-dispersed (Snow & Snow 1988). Fruit and pyrene production are not pollen-limited because some fruits are parthenocarpic and some pyrenes are blank (Richards 1988; Obeso 1996). Here fruit traits according to Lee, Grubb & Wilson (1991) are used: ‘pulp’ (soft tissue external to pyrenes), ‘fibrous coat’ (stony endocarp) and ‘seed’ (embryo and endosperm).

Vegetative spread by suckers arising from shallow lateral roots is also common. Under heavy browsing, holly is maintained as a low shrub.


The study was conducted at Pome Forest, located in the Picos de Europa National Park (950 m a.s.l., 43°16’N, 4°59’W), Cantabrian Range, Asturias province, northern Spain.

Browsing ungulates (mainly cattle and horses) are abundant. They consume most of the saplings (despite seedling production being high) and browse on the lower branches of the trees (despite the fact that they are defended by spiny leaves).


The effects of leaf area reduction on components of the reproductive output of I. aquifolium were assessed by means of an artificial defoliation experiment. To determine whether individual branches can supply the carbohydrate resources required for fruit maturation, a subset of branches was girdled: a ring of bark and cambium ≈ 1 cm wide was removed from the base of the branch. This procedure interrupts phloem transport but does not affect xylem transport.

Twenty female trees were selected in April 1995 and 20 branches per tree were tagged with numbered plastic tape. These plants were randomly assigned to control plants (n = 8) or to experimental treatments (n = 10). Within these experimental plants six tagged branches were randomly assigned to the following defoliation treatments: control, 50% and 100% of leaf area removed. The branches used were unbranched second-order branches. Three branches per defoliation treatment were randomly selected and girdled at their bases. Defoliation treatments consisted of the removal of entire leaves or half a leaf and were applied on 21–23 May, during flowering.

The number of flowers and buds were counted at the dates of treatment application. Flowering finished at the end of May and the green fruits were counted on 28 May to determine the initial fruit set. Current year shoot growth begins at the end of flowering and the new leaves growing on the new shoots were removed at these dates when they were still small.

The tagged branches with ripe fruits were harvested on 27 September and the following variables were determined after oven drying: dry mass of the shoot produced in the current season, dry mass of the whole shoot, dry mass of leaves and dry mass of the fruits. Reproductive allocation at branch level was calculated as the ratio of fruit mass to shoot mass.

To examine the effect of assimilate limitation on fruit traits the following variables were determined for fruits from four experimental plants (some plants did not produce fruits in branches which were both girdled and defoliated): individual fruit mass, mass of the individual pyrenes, proportion of blank pyrenes and mass of the fibrous seed coat. Individual seed mass was calculated by the difference between pyrene mass and fibrous-coat mass. Because some trees did not produce fruits on defoliated and girdled branches, the only four trees that produced fruits on all experimental branches were selected for fruit analyses. Three fruits were randomly selected per branch on these trees and, taking into account the fact that traits from fruits within the same branch may be dependent on each other, only mean values per branch were used in the analyses.

Triplicate subsamples of a single bulked sample of each fruit component were analysed for nitrogen concentration by combustion using a standard automated CNH method (PE 2400 Series II, CNHS/O).


The experimental design consisted of a three-factor ANOVA with tree as a random factor and defoliation and girdling as fixed factors. Fruit production was dependent on shoot mass and so this was included as a covariate. The variable fruit mass per branch was log-transformed, the reproductive allocation was square root-transformed and the proportions (initial and final fruit set, and proportion of seeded pyrenes per fruit) were arcsin-transformed in order to obtain the maximum fit to normality and variance homogeneity.

Mean values are given ± 1 SD with sample size in parentheses.



Mean number of leaves on control branches at the end of the experiment was 9·6 ± 4·1 (60) and their dry mass averaged 2·4 ± 41·05 (60) g. Mean number of fruits shoot –1 was 28 ± 18 (60) and their average mass was 3·47 ± 2·45 (60) g. Fruit mass per shoot was related to shoot mass (r = 0·479, P = 0·0001); however, neither leaf number nor leaf mass were correlated with fruit production.

The amounts of leaves removed at 50% and 100% treatments were used to determine the accuracy of the defoliation treatments. The mean mass removed was 2·23 ± 1·15 g for the 50% and 4·06 ± 2·25 g for the 100% treatment. At the end of the experiment, leaves on 50% treatment branches had a mass of 1·12 ± 0·93 (60) g. Leaf mass on 100% treatment branches was negligible and consisted of a few small leaves on regrowth points at the end of the summer. Tissue removal in the 100% treatment was 1·8 times greater than in 50% defoliation treatment. At the end of the experiment, leaves on control branches weighed 2·2 times more than those on 50% defoliation treatment.

Fruit set and fruit production per branch on control trees did not differ significantly from control branches on experimental trees.

Nitrogen concentration was highest in the seed fraction (4·0% of dry mass) and lowest in the fibrous coat (0·38% of dry mass); the concentration in the pulp amounted to 0·74% of dry mass. According to the mean mass per fruit of the fruit components, four-seeded fruits had 14·6% of total fruit nitrogen in the pulp, 17·1% in the fibrous coats and 68·3% in the seeds.


Both initial and final fruit set were relatively high in control branches (Table 1) and were affected by defoliation and girdling treatments (Table 2). The defoliation by girdling interaction was also significant for both variables (Table 2). Defoliation had little effect on the fruit set of the ungirdled branches, but reduced the fruit set on girdled branches, especially the final fruit set.

Table 1.  . Mean (± SD, n = 30 branches) initial and final fruit set (fruits/flowers), fruit mass (g) and reproductive allocation per branch (mass of fruits/shoot mass) at different defoliation and girdling treatments Thumbnail image of
Table 2.  .ANOVA results for arcsin-transformed initial fruit set and final fruit set (fruits/flowers), log-transformed mass of fruits and reproductive allocation per branch (square-root transformed). Shoot mass is a covariate. *P < 0·05; **P < 0·01 after sequential Bonferroni correction Thumbnail image of

Fruit production was significantly affected by defoliation, girdling and the interaction between both factors. Branch mass was a significant covariate (Table 2). Considering the ungirdled branches alone, 100% defoliation treatment showed no reduced fruit production compared with control branches, despite the fact that 50% defoliation treatment lowered fruit production. When girdled branches are considered, fruit production decreased as the extent of defoliation increased (Table 1). Reproductive allocation per branch followed the same pattern as fruit production.


Individual fruit mass and their components (individual seed mass and the mass of its fibrous coat) followed similar patterns (Tables 3 and 4). Defoliation and the girdling by tree interaction showed significant effects on individual fruit mass and on fibrous-coat mass. The extent of defoliation reduced fruit mass and fibrous-coat mass but the detrimental effect of girdling was dependent on the tree. Neither the mass of the pulp nor individual seed mass were affected by either defoliation or girdling.

Table 3.  . Mean (± SD, n = 12 branches) mass (mg) of individual fruits, pulp, individual seeds and fibrous coats (pyrene without seed) at different defoliation and girdling treatments Thumbnail image of
Table 4.  .ANOVA results for arcsin-transformed initial fruit set and final fruit set (fruits/flowers), log-transformed mass of fruits and reproductive allocation per branch (square-root transformed). *P < 0·05; **P < 0·01 after sequential Bonferroni correction Thumbnail image of

Most of the fruits had four pyrenes; however, about 25% of all pyrenes produced were blank. Neither defoliation nor girdling had a significant effect on the number of pyrenes produced per fruit nor on the proportion of seeded pyrenes (Tables 3 and 5).

Table 5.  .ANOVA results for the number of pyrenes per fruit and arcsin-transformed proportion of seeded pyrenes. **P < 0·01 after sequential Bonferroni correction Thumbnail image of


Experimental defoliation and girdling in I. aquifolium can reduce the amount of resources available for current reproduction in individual branches. These branches may be regarded as only partially autonomous in their reproduction because there may be some resource importation, and the experimental treatments enabled an estimation of the importance of resource importation from other branches for fruit mass gain.

Both initial and final fruit set were high as had been previously found by Richards (1988) and according to general patterns in woody dioecious species (Sutherland & Delph 1984). Initial fruit set was scarcely affected by experimental manipulations (there was not enough time to show any effect), except in girdled-100% defoliated branches in which it was reduced about 65%. Final fruit set was mainly affected by the combined application of defoliation and girdling. Hence, defoliation alone had little effect on fruit abortion probably owing to the ability of branches to import assimilates.

Fruit production in undefoliated branches did not vary between girdled and ungirdled branches. There are two alternative explanations for this result: either control ungirdled branches did not export assimilates to the main branch or, if they did, the superabundant assimilates on control girdled branches were not employed in increasing fruit production.

Mature fruits were produced in 40% of the girdled-100% defoliated branches and mean fruit mass for this treatment averaged 8% of mass produced on control branches. The only sources of assimilates for these branches were the reserves stored in the shoots and/or photosynthesis on green fruits.

There was no reduction in fruit production in the ungirdled-100% defoliated branches when compared with control branches. Thus, contrary to most published results for woody plants, there was no detrimental effect on fruit production after leaf removal. If these branches produced about 8% when they were girdled, most of the assimilates (> 90%) were imported from other branches.

When girdling and 50% defoliation were applied, the branches decreased their fruit production (42% of control) in proportion to the leaf area removed. However, when 50% defoliated branches were ungirdled they produced 70% of fruit production in control branches, which means that they imported some assimilates from other branches. Otherwise, these results did not support the existence of increased photosynthetic rates as a compensating mechanism.

The detrimental effect of defoliation and girdling consisted in a reduction of individual pyrene mass, which was the result of a reduction in the mass of the fibrous coat rather than seed mass reduction. Pulp mass was not significantly affected by experimental treatments and the proportion of the fruit made up by the pulp only varied among treatments from 21 to 24%, which means that fruit architecture was maintained despite whole fruit mass reduction. Nevertheless, fruit mass reduction may have a limit because 60% of girdled-100% defoliated branches aborted all their fruits.

The production of hormonal signals by developing embryos enables resource importation from other sources and thus subsequent fruit maturation (Lee 1988). Hence, it is probable that importation from distant sources (other branches) needs a larger proportion of developing embryos per fruit. Sprugel et al. (1991) proposed that branches may be autonomous for carbon but not for nitrogen. In this sense, the number of pyrenes per fruit was affected by defoliation but was not influenced by girdling, and blank pyrenes were mainly based on carbon. The seeds have a higher proportion of nitrogen than other fruit portions (Lee et al. 1991). Four-seeded holly fruits contain 1·64 mg of nitrogen in the seeds, while the pulp contains 0·35 mg of nitrogen (see also Herrera 1987). The fibrous coats, which amount to 55% of total dry fruit mass, contain only 0·41 mg of nitrogen and are made up mainly by carbon. However, neither the proportion of seeded pyrenes nor the seed mass was affected by girdling.

The current view of branch autonomy is that, under normal conditions, most branches are autonomous during the growing season. However, when local sources of photosynthate are eliminated there may be some import of photosynthate (Sprugel et al. 1991). Furthermore, experimental defoliation of branches routinely demonstrates their functional independence because the induced effects are invariably highly localized within the tree (Gill et al. 1995). The importance of photosynthate or reserves interchange between branches differs between species and functional groups (Sprugel et al. 1991).

Fruiting branches of European holly showed a high ability to import resources from other, probably non-fruiting, adjacent branches. This particular response might be a mechanism developed at tree level to meet the cost of fruit production. The existence of cost of the reproduction in females compared with males, and in some cases in fruiting compared with non-fruiting branches within females, has been demonstrated in this species (Obeso 1997). A mechanism which minimizes the consequences of this cost may be the import of resources from other branches rather than the increase of photosynthetic rates in partially defoliated branches.


The staff of the Picos de Europa National Park granted permission to work at Pome Forest. I thank Esteban Cabal, Nacho Fdez-Calvo and Manuel Alvarez-Santullano for helping in field work and Pedro Jordano for reviewing an earlier version of the manuscript. Furthermore, Esteban Cabal performed the analyses of nitrogen and dissected and weighed most of the fruits and seeds. Penelope Metcalfe corrected the English. This research was supported by a DGICYT grant number PB-94–1538.