Competition as an amplifier of short-term vegetation responses to climate: an experimental test



1. Responses to a spring warming treatment were measured on five common herbaceous species grown in outdoor microcosms in Northern England. Although elevated temperature had a beneficial effect on canopy height and plant cover in monospecific cultures of all species, strongly divergent responses to warming occurred in mixtures.

2. We show that the effect of interspecific competition was to modify and amplify the vegetation response to the warming treatment through effects on phenology and morphological development.

3. The observed responses between the species to warming are consistent with predicted differential responses linked to genome size.

4. We conclude that the mechanism of competitive interaction proposed by Boysen-Jensen (1929) coupled with the insights related to genome size provide a basis for explaining and predicting the role of interannual variation in temperature in determining year to year fluctuations in the relative abundance of species in productive perennial herbaceous vegetation.


The principle is widely accepted in plant ecology that measurements of the edaphic and climatic tolerances of plant populations and species are imperfect predictors of their geographical and ecological distributions. Many factors dictate that realized distributions are usually much narrower than those corresponding to physiological limits determined in the laboratory. Among these factors it is important to recognize the role of interventions by historical events, vegetation management and selective attack by herbivores and pathogens. However, it has been argued (Ellenberg 1958; Harper 1961; Rorison 1969) that the most pervasive factor confining field distributions to skeletons of those predicted from the laboratory is interspecific competition. A logical extension of this theory is the notion that interspecific competition may influence the relative abundance of plant species within communities over short and long time-scales (Al-Mufti et al. 1977; Grubb, Kelly & Mitchley 1982; Silander & Antonovics 1982).

Various experiments have confirmed the potential of interspecific competition to restrict the spatial distribution and abundance of individual species (Ellenberg 1953; Austin & Austin 1980; Parrish & Bazzaz 1982; Wilson & Keddy 1986; Campbell & Grime 1992). In marked contrast, however, there has been little experimental enquiry into the role of interspecific competition in year to year variation in the composition of plant communities and the performance of individual species. From a large number of long-term monitoring studies on field plots (Brenchley & Warington 1958; Watt 1960; Muller & Foerster 1974; Herbern et al. 1995; Willis et al. 1995) it is well-established that marked fluctuations in the relative abundance of species can occur on a yearly basis. Some of these changes are correlated with variation in climate (e.g. Silvertown et al. 1994; Willis et al. 1995), but it is not possible from the available data to ascertain to what extent such responses represent direct effects of climate on the species or depend upon a modulating influence of interspecific competition.

In this paper we describe an experiment designed to measure the contribution of interspecific competition as a determinant of a plant community response to an imposed modification of climate. Our method was to expose five perennial herbaceous plants to controlled winter warming and to quantify their responses in monoculture and in an additive mixture. The results suggest that interspecific competition can act as a powerful amplifier of the climate signal through changes in the relative contribution of species to primary production.


Although discussion continues about the factors that control the intensity with which plants compete for resources in various environments (Welden & Slauson 1986; Tilman 1987; Thompson & Grime 1988; Campbell & Grime 1992; Grime et al. 1997) it is now possible to advance a tentative hypothesis concerning the way in which interspecific competition modulates impacts of climatic events. Following the arguments advanced in Grime (1974) we suggest that climatic impacts will be most strongly amplified by interspecific competition in circumstances of high productivity. This prediction does not rest on the assumption that differential responses to climate will become more evident as a simple consequence of the faster growth rates of the species of productive vegetation. Whilst this phenomenon may occur it should be noted that it does not suggest an explicit involvement of competition for resources. The basis for predicting that competition will amplify climatic effects most strongly in productive vegetation can be traced back to the work of Boysen-Jensen (1929). He drew attention to the sharp unidirectional spatial gradient of light within rapidly ascending mixed canopies and pointed out that small initial differences in the capacity to project shoots out of shade were rapidly transformed into marked disparities in the carbon economies of competing species. Boysen-Jensen (1929) suggested that even a marginal failure to sustain a superior position within an ascending canopy could quickly lead to an inferior rate of photosynthesis, reduced rate of shoot growth, submergence in shade and competitive suppression.

It is clear therefore that disturbance of the mechanism whereby fast-growing plants compete for light provides a possible amplifier of vegetation response to climate variation. However, there is no reason why the amplifying mechanism should be confined to the struggle for light. When plants compete for light, and particularly when shoots become engulfed in the shaded stratum, there is a tendency for photosynthate to be retained in the shoot. This may cause starvation of the root and a risk that its absorptive surfaces may become confined to the zones of nutrient depletion which, even in fertile soils, surround the roots of fast-growing plants (Bhat & Nye 1973). There is strong interdependence therefore between competition above and below ground (Donald 1958; Grime et al. 1997) and we can predict that the disadvantage described for shaded shoots by Boysen-Jensen (1929) will coincide with an analogous and parallel phenomenon in the rhizosphere resulting in a sharply declining rate of nutrient capture by a carbon- and energy-starved root system.

In the finely balanced circumstances where fast-growing perennials are in competition it is not difficult to envisage conditions in which a controlling effect of climate early in the growing season could generate initial differences between plants in the effectiveness of shoot and root foraging which would then become subject to amplification as the main period of growth ensued.

Against this theoretical background we set out to measure the effect of a controlled climate manipulation (warming) applied early in the growing season to transplants of five common herbaceous plants. We expected that the species would differ in response when growing in isolation but it was predicted that a selective amplification of the effect of warming would occur when the species were grown in additive mixture.

Materials and methods


The transplants used in the experiment were collected as rooted shoots from established individuals in roadside verge grassland at Bibury in Gloucestershire, UK. Records were available from permanent plots in these verges (Dunnett et al. 1998) concerning year to year fluctuations in above-ground biomass over the period 1958–present in a range of common herbaceous plants. Five of the species were selected to represent groups of plants which had shown contrasted responses to climate at Bibury (Dunnett 1996). Selection was also based upon their availability in large quantity in the grassland vegetation closely adjacent to the permanent plots. Dactylis glomerata is a dominant of the vegetation and Festuca rubra and Poa pratensis are locally abundant. Cirsium arvense and Achillea millefolium are subordinate contributors to the biomass but they are of widespread occurrence at Bibury.


Approximately 100 rooted shoots of each species were collected at random from within the vicinity of the permanent plots at Bibury in April 1992. Turves containing the required plants were lifted and transported to Sheffield in plastic bags where they were washed free from soil and the shoots of each species were trimmed to a uniform size. Individuals were then selected at random for planting into large plastic containers 450 mm diameter × 600 mm depth. The base of each container was provided with 15 drainage holes each 5 mm in diameter and was covered with a 30 mm layer of coarse gravel, above which there was a nylon mesh. Most of the container was then filled with silty sand but the topmost 300 mm consisted of topsoil of moderate fertility and pH 6·5 removed from an agricultural field.

An additive design was employed to create monospecific cultures and mixtures. An additive design, in which the density of a species in mixture is equal to its density in monoculture, allows the direct effects of experimental treatments on plant performance to be distinguished from those mediated through competition (Mahmoud & Grime 1976, Austin et al. 1988), as opposed to a substitutive design in which the density of a species in mixture is a fraction of its monoculture density. Three individual transplants of each species were allocated at random to three of 15 predetermined, equidistant planting positions, arranged in an outer ring of nine positions, an inner ring of five and a central position (Fig. 1). Three plants only of one species were planted in each monospecific culture, again by random selection among the 15 predetermined positions. Ten containers of both mixtures and monocultures were planted. In April 1992 the containers were positioned in an outdoor sheltered south-facing position in the experimental gardens of the University of Sheffield. The containers were part of a larger experiment: the containers designated for warming and the controls were randomly allocated positions within four adjacent rows, each of 50 containers. The containers were re-positioned randomly within the overall experimental block again in the springs of 1993 and 1994. The communities were allowed to establish for one full growing season before climate manipulations were applied. At the end of 1992 the vegetation in each container was clipped at 100 mm simulating the management procedure applied at Bibury.

Figure 1.

. Schematic diagrams of the standard template for assigning planting positions in the container (left) and the arrangement of the heating cable on the soil surface (right).

Between 26 February and 24 May 1993, five replicates of the monospecific and mixture treatments were subjected to a warming episode using continuously operating heating cables laid directly on the soil surface in rows 80 mm apart (Fig. 1). The use of heating cables at the soil surface allows heat to be applied directly at the point of heat exchange in open swards in spring, resulting in air temperature being raised in the vicinity of the plants without physical disruption to the plants or soil (Hillier, Sutton & Grime 1994; Grime et al. 1999). The soil surface was situated 100 mm below the rim of the container.

The temperature of the heated and control bins was monitored throughout the treatment period. Recordings were made at the soil surface at 5, 15 and 30 mm from cables and the treated and control bins were read simultaneously. It was inevitable that, when using a heating grid, temperature gradients would develop between the rows of cable. For example, Table 1 shows temperatures recorded on 8 April 1993.

Table 1.  . Mean temperatures (°C) at different distances from the heating cables, recorded from three replicates of each monoculture and mixture on 8 April 1993. The control (ambient) value is a mean of values recorded in a single replicate of each monoculture and mixture. Figures in brackets are standard errors Thumbnail image of

In experiments of this type it is not necessary for climate manipulations to attempt a precise simulation of a predicted future climate (Grime et al. 1999). Rather, it may be beneficial to apply exaggerated manipulations to establish the mechanisms governing community change. We therefore aimed to achieve a general increase in temperature of between 4 and 6 °C between the cables. This was achieved throughout the warming period, as illustrated in Table 2.

Table 2.  . Mean temperatures (°C) in warmed and control mixtures at midday on different dates throughout the warming period Thumbnail image of

The response of the vegetation to warming was measured in two ways. At weekly intervals throughout April 1993, the mean height of the leaf canopy of each species in each container was measured. From these data it was possible to compare the performance of each species in the warmed replicates with those in unwarmed control replicates, in both monospecific culture and mixture. Because the vegetation had been trimmed to a standard height in all the replicates prior to warming, the ratio of canopy height in warmed replicates to that in control replicates can be equated with growth rate in canopy height for each species.

Vegetation composition was assessed by point quadrat analysis prior to warming, one month after the end of the warming episode in June 1993 and again in August 1994 to detect any longer-term changes in vegetation composition. A point quadrat frame was built that would fit the experimental containers. Two Perspex sheets were attached, one 100 mm above the other, by nuts to four threaded steel legs. A grid of 5 mm diameter holes was made in each Perspex sheet. The holes were 50 mm apart in the rows and the rows were 50 mm apart. The frame was placed in the same position over the container at each recording. The holes in the two perspex plates were perfectly aligned so that a pin could be directed to a specific position. All touches on a pin were recorded. To avoid edge-effects, pins were not recorded within 75 mm of the rim of the containers. The performance of each species was determined by comparing above-ground biomass, as estimated by point quadrat, in warmed mixtures with that in the unwarmed mixtures. Furthermore, above-ground biomass of each species in warmed and unwarmed mixture was compared directly with above-ground biomass in the comparable monocultures to determine the influence of interspecific competition on species performance following warming. This was calculated as the percentage reduction in yield potential (CX) of each species in mixture (Campbell et al. 1991):

CX = (YP YM) × 100/YP,

where YP is the yield in monoculture (the maximum potential yield) and YM is the yield in mixture. Yield here is the above-ground biomass as estimated by point quadrat survey.


During 1992 all the transplanted material grew vigorously and flowering occurred in each species and in the majority of containers. In the monocultures the tussocks of D. glomerata expanded to form a continuous cover and the rhizomes of P. pratensis and F. rubra allowed these two species to form a continuous low sward. Extension and branching below the soil surface occurred in C. arvense and A. millefolium and tall flowering shoots appeared; it was noted, however, that neither of these species was able to produce a continuous cover within the first growing season.

The heating treatment applied in the late winter of 1993 promoted earlier shoot expansion in all species, an effect which was most conspicuous in C. arvense. The first shoots in the warmed monocultures emerged 3 days before those in the unwarmed controls. In mixture, shoot emergence was delayed by 2 weeks in the controls compared to those of the warmed replicates (Fig. 2). This indicates that C. arvense is both sensitive and responsive to temperature but also that interspecific competition retards shoot emergence. As shown in Table 3, the warmed plants in the monospecific cultures of all five species were significantly taller than their control counterparts at the beginning of the recording period in April. Dactylis glomerata, C. arvense and F. rubra were similarly promoted in mixture, but there was no significant difference between warmed and control plants of the other two species.

Figure 2.

. Mean number of shoots of Cirsium arvense in warmed and control mixtures throughout the warming period (• unwarmed plants; ▴ warmed plants). Bars represent standard errors.

Table 3.  . Mean heights of the five species in monospecific culture (mon) and mixture (mix) under both warming (W) and control (C) treatments Thumbnail image of

Again, 4 weeks later, all five species had greater mean heights in warmed monoculture than in unwarmed controls, although this difference was not statistically significant for P. pratensis. However, by this later stage of the experiment some disbenefits from warming had begun to become evident in the mixtures. The mean height of the warmed A. millefolium plants was now significantly less than that of the controls. The mean height of the warmed plants of F. rubra, which had initially been significantly greater than the unwarmed controls now appeared to be retarded, although the difference between the two sets of plants was not statistically significant.

These results are reflected in the ratios of mean canopy height in warmed replicates to that in control replicates, also shown in Table 3 (as noted above, these can be equated to relative rates of leaf extension or growth rate in canopy height given that all species were cut to a standard height prior to the start of the experiment). The results show that none of the five species exhibited reduced performance as a result of warming in monoculture and all but P. pratensis were promoted. In contrast, in mixture, both A. millefolium and F. rubra exhibit reduced performance as a result of warming, while C. arvense was strongly promoted.

Table 4 presents cover estimates from point quadrat surveys taken in June 1993. The results indicate that warming had shifted the competitive balance in the warmed replicates. Dactylis glomerata was promoted from occupying approximately one-third of the cover in the warmed replicates to approximately one-half. As shown in Fig. 3, D. glomerata tended to increase vegetative growth at the expense of flowering following warming. The two other grasses occupied significantly less cover in the warmed replicates than in the controls.

Table 4.  . Mean contribution of the five species to total above-ground biomass in mixture replicates, June 1993: Am, Achillea millefolium; Ca, Cirsium arvense; Dg, Dactylis glomerata; Fr, Festuca rubra; Pp, Poa pratensis; C, control treatment; W, warming treatment Thumbnail image of
Figure 3.

. Seed yield of Dactylis glomerata in relation to above-ground biomass. The points marked w represent the five warmed replicates, the unmarked points represent the unwarmed controls. Each replicate contained three individuals of D. glomerata. Biomass values were estimated by point quadrat in October 1993.

A similar pattern was evident in 1994 (Table 5), although the differences between the control and warmed replicates were no longer significant: the composition of the warmed replicates appeared to be reverting to that of the controls. None-the-less there are clear carry-over effects from the previous year in the warmed replicates. This is more apparent in the comparison of the influence of interspecific competition on competitive fitness for each species following warming shown in Fig. 4. The two vegetation dominants, Dactylis and Festuca, show divergent responses: Dactylis has been promoted with an increase of over 15% in competitve ability, while that of Festuca has been reduced by a similar amount.

Table 5.  . Mean above-ground biomass of the five species, August 1994: Am, Achillea millefolium; Ca, Cirsium arvense; Dg, Dactylis glomerata; Fr, Festuca rubra; Pp, Poa pratensis; C, control treatment; W, warming treatment; Mon, monospecific culture; Mix, mixture Thumbnail image of
Figure 4.

. The competitive fitness of the five species in mixture in the second growing season following warming. Competitive fitness is expressed as the ratio (expressed as a percentage) of the reduction in yield potential (CX) of plants in warmed mixture to that in unwarmed mixture. Values greater than 100 indicate an increase in fitness of the warmed plants compared with the controls: the horizontal lines represent increase or decrease in fitness of the warmed plants by 10%.


In Britain, temperature acts as the major limiting factor on the rate and timing of shoot growth during the spring. Consistent with this phenomenon, all five of the Bibury species responded to the spring warming treatment with increased height growth and cover in monospecific culture. However, this uniformity of response was not observed when the species were grown in mixture. The effect of interspecific competition was to sustain the benefit of the warming treatment in D. glomerata and C. arvense but to negate the potential benefit to A. millefolium and F. rubra. This provides clear evidence of the capacity of interspecific competition to modify and, in terms of species composition, to amplify vegetation responses to climatic fluctuation. Although, as would be expected, species responses to warming were most pronounced during the period of the treatment itself, the results show that the initial advantage gained by D. glomerata in both height and above-ground biomass demonstrated in Table 3 and Fig. 3 was still apparent in the following growing season, largely at the expense of the other abundant species in the system, F. rubra. While the results presented in this paper relate only to shoot response to warming, D. glomerata is equally likely to compete effectively below ground through its extensive root system (van den Bergh & Elberse 1970). It is well established (de Wit 1960; Harper 1977) that competition affects the relative abundance of species within plant communities by modifying access to resources but we are not aware of any previous experimental studies demonstrating the role of interspecific competition in generating marked and divergent responses to a change in temperature.

The temperature-dependant response and advance in shoot emergence of Cirsium in the warmed replicates, shown in Fig. 2, was particularly dramatic. This observation is consistent with predicted differential responses linked to genome size. Previous investigations (Grime & Mowforth 1982; Grime, Shacklock & Band 1985) have shown that plant communities in Britain contain species which differ considerably in genome size. It is established that under cold conditions higher rates of shoot growth are achieved by perennial plants with large genomes. Canopy expansion in species with small genomes tends to be delayed until the onset of warmer temperatures. It has been proposed (Grime & Mowforth 1982) that the rapid spring growth of many geophytes and grasses of high nuclear DNA amount and large cell size is achieved largely through the expansion of preformed but unexpanded cells, a process which is relatively insensitive to temperature. By comparison, the growth of small genome species is predicted to rely upon concurrent cell divisions and cell expansions with the temperature dependence of the former acting as a factor strongly delaying their phenology. However, the shorter cell cycles of the species with small genomes (Bennett 1971) are predicted (Grime 1989) to make the plants potentially very responsive to spring warming. The 2C nuclear DNA content of Cirsium is 3·1 pg, compared with 8·7 pg for Dactylis, 10·8 pg for Poa, 13·9 for Festuca and 15·3 for Achillea (Grime, Hodgson & Hunt 1988).

Monitoring studies (Brenchley & Warington 1958; Watt 1960; Muller & Foerster 1974; Grime et al. 1994; Herben et al. 1995) provide quantitative evidence of the ability of herbaceous vegetation to fluctuate in species relative abundance quite sharply from year to year. This capacity for rapid change has been implicated in the mechanisms maintaining diversity in certain types of relatively productive vegetation (Grime 1979; Silvertown et al. 1994) and is particularly evident in meadows and pastures where high rates of biomass removal by mowing or grazing reduce the carry-over effect of perennial plant dominance from one growing season to the next. In such circumstances where shoots and roots are largely renewed each year there is an opportunity for interannual variation in the spring climate to shift the competitive advantage between species and to determine patterns of relative abundance expressed later in the growing season. This hypothesis is supported by field evidence from the Bibury road verges. Correlation of interannual fluctuations of above-ground biomass (as recorded in the third week of July each year) of the 40 most abundant species at Bibury with interannual fluctuations in weather over the period 1959–96 showed that more species show a relation with spring (March–May) weather than that of any other season (Dunnett et al. 1998).


This research was conducted with the support of the Esmée Fairbairn Studentship.