Author for correspondence: Jon Millett Tel: +44 (0)151 291 2174 Fax: +44 (0)151 291 3100 Email: email@example.com
• Deciduous trees recycle nitrogen within their tissues. The aim of this study was to test the hypothesis that reductions in plant growth, caused by competition and herbivory, reduce the sink strength for N during autumn nutrient withdrawal, and reduce the storage capacity and hence the amount of N remobilized in the following spring.
• We used 15N-labelled fertilizer to quantify N uptake, leaf N withdrawal and remobilization. Betula pubescens saplings were grown with either Molinia caerulea or Calluna vulgaris, and subjected to simulated browsing damage.
• Competition reduced B. pubescens leaf N withdrawal and remobilization, with C. vulgaris having a greater effect than M. caerulea. However, simulated browsing had no significant effect on sapling N dynamics. The patterns of leaf N withdrawal and remobilization closely followed sapling dry mass.
• We conclude that the effect of competition on sapling mass reduces their N-storage capacity. This reduces sink strength for leaf N withdrawal and the source strength for remobilized N. The ability of saplings to compensate for browsing damage removed any potential effect of browsing on N dynamics.
Nitrogen remobilization by silver birch is a source-driven process because the amount remobilized depends on the size of the storage pool and is unaffected by sink strength (Millard et al., 1998). However, while the effect of N availability on N remobilization in trees is well understood, other biotic and abiotic factors that may influence N uptake, withdrawal and remobilization are only partly understood.
Herbivory can influence internal N cycling through the removal of N, and thus a reduction in the amount of N available for storage (Millard et al., 2001). The effect is species-specific, and depends on the timing of herbivory because of the low N concentrations in browsed tissues during winter dormancy. Millard et al. (2001) found no effect of simulated mammalian browsing on the amount of N remobilized in Betula pendula saplings because the N storage sites are in tissues that remain unbrowsed. Moreover, fast-growing deciduous species such as B. pendula are generally able to compensate for the loss of tissue (and N) caused by herbivory (Hester et al., 2004, 2006).
Plant responses to competition may also be mediated by the effects on internal N cycling of interactions with other plants. Such interactions can influence all plant life-cycle stages, and can reduce resource availability and thus growth and development (Grace & Tilman, 1990; Tilman, 1997). Trees in temperate systems normally experience both intra- and interspecific competition, often resulting in growth inhibition (Berkowitz et al., 1995) and changes in morphology (Aphalo et al., 1999). It seems likely that interactions with other plants will also influence the pattern of internal nutrient cycling, because of the effects of resource limitation and the growth responses of the plant to competition. However, despite these factors being fundamentally important influences on plant development and growth in natural systems, the effects of competition on internal N cycling, and possible interactions with herbivory, have not been studied as far as we can ascertain.
In this experiment we measured the effects of competition (belowground or above- and belowground) and simulated herbivory on the internal N cycling of Betula pubescens saplings. Saplings were subjected to competition from either Molinia caerulea or Calluna vulgaris and simulated mammalian browsing at either bud-burst or pre-leaf senescence. Competition effects were separated to enable the relative contribution of above- and belowground processes to be quantified. Betula pubescens is a temperate deciduous tree; M. caerulea is a temperate deciduous grass; and C. vulgaris is an evergreen woody shrub. These three species are widespread throughout Europe and are found in many heath and woodland systems (Gimingham, 1960; Atkinson, 1992; Taylor et al., 2001). They often co-occur, and provide an interesting and ecologically important model system for studying plant interactions because of differences in their life-history traits. The aims of this study were to test the hypothesis that reductions in sapling size, caused by competition and herbivory, will reduce the sink strength for N during autumn nutrient withdrawal, and reduce N-storage capacity and hence the amount of N remobilized during the following spring. Specifically, this study aimed to determine whether reductions in B. pubescens sapling size, caused by competition with C. vulgaris or M. caerulea and simulated browsing, result in: (1) a reduction in leaf N withdrawal during senescence because of a reduction in the sink for withdrawn N; and/or (2) a reduction in the size of the potential N store and therefore in the amount of N remobilized in the spring.
Materials and Methods
Experimental design and set-up
A total of 180 2-yr-old Betula pubescens Ehrh. seedlings (UK provenance region 202, north-east Scotland) were lifted from a nursery when dormant and planted in sterile sand in pots (300 mm diameter × 260 mm deep) in early spring 2002. Before lifting, seedlings were grown outside and received nutrients at a moderate level. The pots were placed in a covered, fenced enclosure at the Macaulay Institute in Aberdeen, Scotland (57° N, 2° W). All pots were watered with a complete nutrient solution containing 3 mm NH4NO3 and other nutrients, as described by Millard & Proe (1991). In 2002 each pot received 300 ml nutrient solution twice a week. In 2003 and 2004 this was increased to 500 ml to reflect the increasing size of plants (and therefore nutrient requirements). This was added only during the growing season (February–October). The pots were also watered with demineralized water to prevent drying. Pots were moved into a glasshouse during each winter and kept frost-free, having been washed through with demineralized water at the end of each growing season to remove any remaining N in the sand.
The timing of events is detailed in Fig. 1. At the beginning of 2003, half the saplings received 15NH415NO3 enriched to 1.91 at% excess until harvest (group 1). These saplings were harvested on 5–8 September 2003 (pre-senescence). The other half (group 2) continued to receive NH4NO3 with 15N at natural abundance during 2003. In 2004 these saplings received 15NH415NO3 enriched to 1.91 at% excess until harvest. This enabled the current N uptake (labelled) to be differentiated from N taken up in previous years.
At the start of the first growing season, all saplings were cut to 100 mm height, which removed most 2001 growth, leaving a single stem of primarily old growth with a number of lateral buds. The aim of this was twofold: to start the experiment with all saplings heavily ‘browsed’; and to standardize their initial height to that of the surrounding vegetation. Because of this pretreatment, none of the saplings can be considered to be truly ‘unbrowsed’. However, we have used the terms ‘browsed’ and ‘unbrowsed’ to enable our subsequent browsing treatments to be distinguished.
Two levels of competing species treatments [Calluna vulgaris (L.) Hull or Molinia caerulea (L.) Moench.] were crossed with two types of competition removal (above- and belowground competition, or aboveground competition only). These four treatments, plus saplings with no competition (control), gave a total of five competition treatments. Each treatment was subjected to one of three browsing treatments (detailed below). These competition and browsing treatments were combined in a 5 × 3 design. The pots were arranged in a randomized design in 12 blocks giving six replicates of each treatment combination, to be harvested at two time points.
Saplings received either no competition, competition from M. caerulea or competition from C. vulgaris. Calluna vulgaris plants (2 yr old) were purchased from a nursery (UK provenance zone 109, south-east Scotland). Rhizomes of dormant M. caerulea plants were collected from the field in Aberdeenshire, Scotland in early March 2002. Six plugs of C. vulgaris or six groups of three M. caerulea rhizomes (to give similarly sized plant groups) were planted in a circle at equal spacing around the B. pubescens saplings, 80 mm from the saplings. This design is equivalent to the ‘target’ technique described by McPhee & Aarssen (2001).
Competition was controlled as follows: no neighbours (NN) where saplings were grown in the pots with no competing vegetation; neighbour roots and shoots (NRS) where competing plants were allowed to interact fully with the saplings; neighbour roots (NR) where a wire mesh cone (height 150 mm, base diameter 50 mm, top diameter 150 mm) was placed around the base of the sapling. The surrounding plants were trained around this cone as they grew to remove the aboveground competition. It was assumed that training the shoots of the surrounding plants behind the mesh cone did not significantly affect root growth of the plants. Therefore aboveground interactions were effectively eliminated without altering the belowground interactions.
The B. pubescens saplings were subjected to one of three simulated browsing treatments (Fig. 1). These were: no browsing (control); late-summer browsing (August 2002 at first sign of leaf senescence); or browsing at bud-burst (March 2003). The simulated browsing treatment was applied by clipping off 50% of the current year's shoots from each sapling, starting with the leader and then cutting every alternate shoot (as described by Hester et al., 2004).
Measurements and harvesting
To give an estimate of the amount of N remobilized during spring 2003 without destructively harvesting the saplings, a random sample of the leaves of saplings in group 1 was taken in spring 2003 (Fig. 1). The sample was stratified to ensure that as much within-sapling variation as possible was accounted for. Samples were taken from the first flush of leaves, and a separate sample was taken from leaves that developed later. Ten per cent of the leaves were removed (restricted to a minimum of two and a maximum of 10). Samples were dried and weighed, and 15N enrichment was measured. The results were pooled and scaled up to give an approximation of the mean for each sapling.
Plants in group 1 were harvested destructively at the onset of leaf senescence on 8 September 2003 (pre-senescence 2003), and plants in group 2 were harvested sequentially 64 d after bud-burst in 2004 (bud-burst 2004). This was considered to be sufficient time for N remobilization to have ceased, but was not so long that any remobilized N would have been retranslocated out of the leaves (Millard et al., 1998). Each sapling was separated into leaves, roots, old shoots (previous years’ growth), and new shoots (current year's growth). Abscised leaves were collected from the saplings in group 2 in autumn 2003 by placing netting around each sapling. All plant material was freeze-dried and weighed.
The 15N and total N content of the dried material (except for the abscised leaves from group 2, for which only N content was calculated) was determined using a FlashEA 1112 series Elemental Analyser, interfaced with a ConFlow III, to a Delta plus Advantage IRMS (Thermo Electron Corporation, Waltham, MA, USA). Each of the two groups of saplings received only labelled N in the year of harvest. The recovery of this labelled N in the plant tissues enabled the amount of N remobilized from the previous years’ uptake (unlabelled) (N stored over winter) and N taken up during the year of harvest (labelled) to be quantified. The contribution of this stored N and new uptake to the current year's growth could then be calculated. The proportion of the previous years’ N in plant tissues was calculated as follows (after Millard & Neilsen, 1989):
A = CD/E
B = (1 − C/E) × D
where A = N uptake in the year of harvest (g); B = remobilized N in tissue (g); C = at%15N excess in tissue; D = N content of tissue (g); and E = at%15N excess in fertilizer. For values of E, the natural abundance of 15N was taken to be 0.37 at% (International Atomic Energy Agency, 1983).
Data were analysed in genstat 7th edn (VSN International, 2004) using the linear mixed model with residual maximum likelihood (REML) estimation (Patterson & Thompson, 1971). The use of REML provides a more powerful technique for analysing data that are unbalanced, or where some data are missing. For data that are balanced and where no data are missing, analysis using REML gives the same result as anova. The test statistic for this procedure is the Wald Statistic (WS), which can be considered analogous to the F statistic in anova. Start-of-experiment measurements of stem diameter and fresh mass were used as covariates. The fixed model used was: browsing × (presence of competition + competing species) × competition treatment × year. The use of presence of competition + competing species enabled the effect of competition and species-specific differences to be disentangled. Block and sapling number were used as random factors. Comparisons between treatments were made using an adjusted Fisher's LSD test (Snedecor & Cochran, 1980). Regression analyses were carried out in spss (SPSS, 2003).
Effect of competition
Competition reduced sapling dry mass (Fig. 2a) and N uptake in 2003 (Fig. 2b), with C. vulgaris reducing uptake to a greater extent than M. caerulea (Table 1). This translated into differences in the labelled N content in the leaves of the saplings. Additionally, saplings growing with M. caerulea with both above- and belowground interactions had a higher dry mass (though not statistically significant) and took up more N during 2003 than those with only belowground interactions. However, for saplings growing in competition with C. vulgaris there was no difference in the amount of N taken up, whether there were both above- and belowground, or only belowground interactions (Table 1).
Table 1. Results of univariate residual maximum likelihood analysis of Betula pubescens sapling growth and tissue nitrogen content
Wald Statistic (WS) and significance (NS, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001) for main effects of, and interactions between: T, time; C, competition (present/absent); S, competing species (Calluna/Molinia); L, competition type (belowground/above- and belowground); B, browsing (late summer/spring/none).
When saplings were growing with no competition or root-and-shoot interactions from M. caerulea, their leaves contained less N following abscission than before senescence (Table 1; Fig. 3a). However, for saplings growing with C. vulgaris or with only belowground interactions from M. caerulea, there was no statistically significant difference in leaf N content pre-senescence or post-abscission. Saplings growing with no competition withdrew a larger proportion of the N from their leaves before abscission, while those growing with C. vulgaris withdrew the least. On average the saplings withdrew the following amounts of N from their leaves before abscission: no competition = 160 mg; competition from M. caerulea = 79 mg; competition from C. vulgaris = 32 mg. These correspond to 40, 34 and 25% of leaf N, respectively. Consistent with these patterns were the differences in leaf percentage N pre-senescence and post-abscission (Fig. 3b). Additionally, saplings growing with no competition, or with above- and belowground competition from M. caerulea, reduced the N concentration of abscised leaves to a lower level than those with competition from C. vulgaris or belowground competition from M. caerulea.
In 2004, the saplings growing with no competition remobilized more N into their leaves than those growing with M. caerulea which, in turn, remobilized more than those growing with C. vulgaris (Table 1; Fig. 4). However, there were smaller differences in the amount of leaf N from new uptake, with only the saplings growing with no competition obtaining significantly more N from new uptake. These different patterns of N remobilization and uptake resulted in differences (caused by competition) in the proportion of N that the sapling leaves obtained from remobilization. Those growing with no competition, or competition from M. caerulea, contained the largest proportion of N from remobilization (74 ± 2 and 73 ± 1%, respectively), while for those growing with C. vulgaris, 50 ± 2% of their N was from remobilization (effect of competition, P < 0.001; difference between species, P < 0.001). There was no difference in the proportion of N from remobilization between saplings with above- and belowground interactions or with just belowground interactions within the competing species.
The saplings remobilized more N in spring 2003 than was withdrawn from the leaves during senescence in 2003. On average, saplings withdrew a total of 75.5 mg N from their leaves before abscission. In spring 2004, a mean of 131.3 mg N in the new leaves came from N uptake in previous years. Therefore withdrawn leaf N accounted for, on average, 58% of the N remobilized in the spring. This amount was affected by the type of competition. When only belowground competition from C. vulgaris or M. caerulea was present, withdrawn N accounted for c. 87% of the N remobilized in the spring. However, when both above- and belowground competition were present, withdrawn leaf N accounted for only 34% of N remobilized in the spring. There was no difference between C. vulgaris and M. caerulea as competitors in this respect.
There was a strong significantly positive relationship between sapling size and N uptake in 2003 (Fig. 5). Furthermore, during spring 2004 the larger saplings remobilized more N into their leaves than smaller saplings (Fig. 6a). This resulted in remobilized N making a greater contribution to leaf N in larger than in smaller saplings (Fig. 6b).
Effect of browsing
There were very few significant effects of simulated browsing on the N uptake and remobilization of the saplings. In spring, directly after browsing damage was imposed, leaves of saplings that had been browsed contained less N from remobilization and more N per leaf than those that remained unbrowsed (Table 2). However, in spring 2004 these differences were no longer apparent. There were no other significant effects of browsing on any aspects of N dynamics within the saplings (Table 1). Additionally, browsing did not alter the effect of competition on sapling N dynamics.
Table 2. Effects of simulated browsing on birch (Betula pubescens) leaf nitrogen content in the spring following browsing (2003) and the following spring (2004)
N from remobilization (mg)
N per leaf (mg)
N from remobilization (mg)
N per leaf (mg)
Means ± SEM for amount of leaf N from remobilization and amount of N per leaf for saplings that received three simulated browsing treatments.
23.6 ± 2.6
1.27 ± 0.08
132.2 ± 18.5
0.33 ± 0.03
17.6 ± 2.1
1.88 ± 0.09
130.6 ± 16.2
0.41 ± 0.04
18.6 ± 2.5
1.82 ± 0.11
131.0 ± 18.5
0.37 ± 0.03
Nitrogen withdrawal during leaf senescence has been shown to be influenced by a number of factors, including summer temperature (Nordell & Karlsson, 1995), leaf N content and time of leaf senescence. However, we are not aware of any studies that have shown a difference in leaf N withdrawal caused by the effects of competition. Mineral nutrient transport in plants is highly dependent on source–sink relationships (Marschner, 1995). Nambiar & Fife (1991) found that shoot production (sink strength) was closely related to nutrient withdrawal from Pinus radiata needles. Furthermore, Chapin & Moilanen (1991) considered the rate of phloem transport (source–sink interactions) to be the most important factor controlling nutrient resorption efficiency in Betula papyrifera. We found close relationships between sapling size and N uptake and remobilization. Additionally, the effect of competition on N uptake, withdrawal and remobilization closely followed the effects of competition on sapling dry mass. Therefore it is likely that the differences in N withdrawal and remobilization were the result of the effect of competition on sapling size. In birch trees, N is stored in the roots, bark and old wood (Millard et al., 1998). A reduction in sapling size will also result in a reduction in the size of these tissues (Millard et al., 2001). Consequently, the sink strength in the autumn could be reduced, reducing the proportion of available N that is withdrawn from the leaves. Additionally, the source strength for remobilized N in spring will also be reduced, resulting in a reduction in the amount of N remobilized.
Nordell & Karlsson (1995) found that variation between individuals (genetic variation) was the largest source of variation in leaf N withdrawal of B. pubescens ssp. czerepanovii. However, in the present study it is differences in sapling size that account for most of the variation in N withdrawal and remobilization. These apparently conflicting results can be explained by the ontogenetic stages of birch studied. Nordell & Karlsson (1995) studied adult trees while we studied saplings. Furthermore, the trees that Nordell & Karlsson (1995) studied differed in height by a factor of c. 1.5, whereas the final dry mass of the trees that we studied differed by a factor of 10–14. It appears that in our study the size of saplings was more important in determining leaf N withdrawal because of the relatively large variation in sapling dry mass.
Effects of competition
The reduction in growth and N uptake was caused by a negative effect of belowground interactions. This indicates that acquisition of N by competing plants reduced N availability for the saplings. Weih & Karlsson (1999) found that intraspecific competition reduced N concentration, productivity and uptake in B. pubescens ssp. czerepanovii. However, this effect was caused by a reduction in soil temperature and N uptake caused by aboveground competition. These differing results suggest that competitive effects on N uptake are dependent on a number of factors, the importance of which depends on the specific situation, including species, temperature and soil N availability. Furthermore, in the present study M. caerulea shoots increased growth and N uptake. Facilitation is a well documented (although apparently less common) result of plant interactions (Callaway & Walker, 1997; Callaway et al., 2002; Bruno et al., 2003). Aboveground facilitative interactions can be attributed to reducing temperature, water or nutrient stress (Callaway, 1995; Bruno et al., 2003). However, the mechanism in the present study is not clear and could conceivably be attributed to any one, or any combination of these.
The efficiency of N withdrawal from leaves before abscission can be defined in terms of the percentage of N removed (withdrawal efficiency), or the final level to which the plant is able to decrease leaf N (withdrawal proficiency) (Killingbeck, 1996). Withdrawal efficiency is highly dependent on the N content of leaves pre-senescence, whereas withdrawal proficiency is a measure of the ultimate ability of the plant to withdraw N. We found similar patterns in both withdrawal efficiency and proficiency. Saplings growing in M. caerulea or with no competition had both a greater N resorption efficiency and proficiency than those growing with C. vulgaris (they withdrew a greater proportion of leaf N and decreased the N content of abscised leaves to a greater extent). Killingbeck (1996) considered that woody perennials could potentially reduce the N concentration of abscised leaves to 0.3% (potential resorption proficiency). Chapin & Kedrowski (1983); Chapin & Moilanen (1991); and Escudero et al. (1992) measured the resorption potential of birch to be 0.50, 0.88 and 0.65%, respectively. The range of N concentrations in abscised leaves in our study was 0.78–2.79%. This suggests that some saplings achieved close to their ultimate resorption potential, while others achieved considerably below their potential. The efficiency of N withdrawal was also lower than may be expected for woody perennials (Aerts, 1996). Additionally, in a review of 60 studies Aerts (1996) found little evidence that N-resorption efficiency was related to N availability or leaf N content. It seems likely that, in the present study, differences in sapling size caused by the effects of competition, and therefore source–sink interactions, are the main driving mechanism determining N withdrawal and remobilization.
Nitrogen remobilization in the spring has been shown to be dependent on the amount of N in store (source strength). For example, Dyckmans & Flessa (2001) found that the amount of N remobilized for leaf growth in Fagus sylvatica saplings was strongly influenced by the previous year's N supply. Additionally, Millard & Proe (1992) showed that N remobilization in Acer pseudoplatinus was dependent on the previous year's N supply, and therefore the amount of N in store, and independent of the current year's N supply. Larger saplings will have a larger capacity for N storage. We conclude from this that the effect of competition on N remobilization was caused by the effects of competition on sapling size, and thus their ability to store N.
Effects of browsing
Fast-growing species are generally able to compensate for tissue loss caused by herbivory (Millard et al., 2001). Moderate levels of simulated browsing (50% of shoots) have also been shown to have little effect on N remobilization in B. pubescens (Millard et al., 2001). Additionally, Karlsson & Weih (2003) found that mature B. pubescens ssp. czerepanovii trees were able to compensate for severe defoliation within 2 yr of damage. In this study, we also found little effect of simulated browsing on sapling dry mass one growing season after the damage occurred. Furthermore, there was no significant effect of browsing on N uptake, withdrawal or remobilization. This corresponds with the response of B. pendula saplings (Millard et al., 2001), and is likely to be caused by the main storage tissues remaining undamaged. However, a delayed negative response to browsing cannot be ruled out. Weih (2000) found that B. pubescens ssp. czerepanovii seedlings exhibited a response to various environmental variables that was delayed until two seasons after treatments were imposed. The compensation for damage that the saplings in our study exhibited could potentially reduce their fitness for following growing seasons.
It has been suggested that the ability of plants to compete would be reduced by herbivory, but there was no evidence of this under the moderate levels of browsing applied in this study. This result concurs with that of Karlsson et al. (2005), who found no interaction between competition and herbivory in B. pubescens ssp. czerepanovii. We believe that these results are caused by the ability of birch saplings to compensate for such damage, with no further negative effect on their competitive ability.
We would like to thank the following for their assistance: Jim McGregor, Janet Woo, Ruth Gill, Jasmine Ross, Renate Wendler, Emily Green, Kirsi Neuavonon, Allan Simm and other staff at the Macaulay Institute and University of Aberdeen. BIOSS provided assistance with statistical analysis. Chemical analysis was carried out by Macaulay Analytical Services. Eric Paterson and five anonymous referees provided extremely useful comments on the manuscript. Jonathan Millett was funded by the Macaulay Development Trust. P.M. and A.H. are funded by SEERAD.