Disturbance is the principal α-scale filter determining niche differentiation, coexistence and biodiversity in an alpine community

Authors

  • SIMON PIERCE,

    1. Dipartimento di Biologia Strutturale e Funzionale, Università degli Studi dell’Insubria, Via J.H. Dunant 3, I- 21100 Varese, *Dipartimento di Biologia, Sezione di Botanica Sistematica e Geobotanica, Università degli Studi di Milano, Via Celoria 26, I-20135 Milano, and †Centro Flora Autoctona, c/o Consorzio Parco Monte Barro, via Bertarelli 11, I-23851 Galbiate (LC), Italy
    Search for more papers by this author
  • ALESSANDRA LUZZARO,

    1. Dipartimento di Biologia Strutturale e Funzionale, Università degli Studi dell’Insubria, Via J.H. Dunant 3, I- 21100 Varese, *Dipartimento di Biologia, Sezione di Botanica Sistematica e Geobotanica, Università degli Studi di Milano, Via Celoria 26, I-20135 Milano, and †Centro Flora Autoctona, c/o Consorzio Parco Monte Barro, via Bertarelli 11, I-23851 Galbiate (LC), Italy
    Search for more papers by this author
  • MARCO CACCIANIGA,

    1. Dipartimento di Biologia Strutturale e Funzionale, Università degli Studi dell’Insubria, Via J.H. Dunant 3, I- 21100 Varese, *Dipartimento di Biologia, Sezione di Botanica Sistematica e Geobotanica, Università degli Studi di Milano, Via Celoria 26, I-20135 Milano, and †Centro Flora Autoctona, c/o Consorzio Parco Monte Barro, via Bertarelli 11, I-23851 Galbiate (LC), Italy
    Search for more papers by this author
  • * ROBERTA M. CERIANI,

    1. Dipartimento di Biologia Strutturale e Funzionale, Università degli Studi dell’Insubria, Via J.H. Dunant 3, I- 21100 Varese, *Dipartimento di Biologia, Sezione di Botanica Sistematica e Geobotanica, Università degli Studi di Milano, Via Celoria 26, I-20135 Milano, and †Centro Flora Autoctona, c/o Consorzio Parco Monte Barro, via Bertarelli 11, I-23851 Galbiate (LC), Italy
    Search for more papers by this author
  • BRUNO CERABOLINI

    1. Dipartimento di Biologia Strutturale e Funzionale, Università degli Studi dell’Insubria, Via J.H. Dunant 3, I- 21100 Varese, *Dipartimento di Biologia, Sezione di Botanica Sistematica e Geobotanica, Università degli Studi di Milano, Via Celoria 26, I-20135 Milano, and †Centro Flora Autoctona, c/o Consorzio Parco Monte Barro, via Bertarelli 11, I-23851 Galbiate (LC), Italy
    Search for more papers by this author

S. Pierce (tel. +39 332 421536; fax +39 332 421554; e-mail simon.pierce@uninsubria.it).

Summary

  • 1Many current biodiversity theories assume that resource competition determines niche segregation and thus coexistence within communities (i.e. at the α-scale). However, the action of disturbance, creating heterogeneous environments and suppressing potential dominants, may also be important for biodiversity maintenance.
  • 2Hypothesis: subordinate species exhibit primarily opportunistic (ruderal) survival strategies, with increasing disturbance intensity constraining dominant species – favouring opportunistic strategies and thus functional and species diversity.
  • 3The diversity, character and frequency of strategies in an alpine sedge-dominated vascular plant community were quantified in situ using CSR (competitor, stress-tolerator, ruderal) classification, and compared with a pasture in the same alpine vegetation belt (i.e. with additional disturbance). Adaptive trends were confirmed by independent multivariate analysis [detrended correspondence analysis (DCA) and non-metric multidimensional scaling (NMDS)].
  • 4The extremely stress-tolerant sedge Carex curvula (C : S : R = 17.2 : 72.9 : 9.9%) dominated the relatively undisturbed community (frequency = 52%), with 32 subordinates (typically < 5%) exhibiting a functional spectrum encompassing stress tolerance to ruderalism, but not competitive strategies. With grazing, the community exhibited weaker co-dominance by five species, greater biodiversity (76 species) and greater functional diversity, characterized by larger numbers of ruderals and some competitive-ruderals. The principal variation in both DCA1 and NMDS1 for both communities directly reflected CSR strategy spectra, confirmed by Spearman's correlation.
  • 5Dominance by stress-tolerators and restricted functional diversity demonstrates habitat-level (β-scale) functional convergence in response to stress. A spectrum of S to R strategies demonstrates α-scale functional divergence in response to differential stress and disturbance. Grazing suppresses potentially dominant species and favours diversity, with the additional presence of competitive-ruderals suggestive of a more intricate niche topology including more relaxed abiotic opportunities.
  • 6Natural communities are not necessarily structured according to the rules of resource competition models, as these fail to account for disturbance and facilitation processes.

Introduction

Functional divergence and niche segregation are often viewed as products of competition: a concept originating from Darwin's (1859) suggestion (for a turf community) that, ‘where the contest between individual and individual must be severe, we always find great diversity in its inhabitants’. Indeed, formal theories of niche differentiation are founded on resource competition, in which instantaneous resource availability is the sole environmental factor assumed to influence survival and adaptive radiation, e.g. the ‘resource ratio hypothesis’ (Tilman 1988), ‘stochastic niche theory’ (Tilman 2004) and ‘resource-based niche theory’ (Silvertown 2004). Experiments testing these theories are typically conducted on anthropogenic grasslands reminiscent of Darwin's.

However, competition is one of many selection pressures operating in nature, and does not necessarily dominate: abiotic selection pressures such as temperature extremes, drought, high radiation flux or salinity may induce stress (reduce metabolic performance) by imposing metabolic injury or inconsistency in resource availability (cf. consistently limited resource availability; reviewed by Pierce et al. 2005). For example, competitive ability is a minor component of survival strategies in alpine habitats, where conservative adaptations allow the resistance of suboptimal periods for growth, or ephemeral strategies allow the avoidance of biomass destruction (e.g. Caccianiga et al. 2006). Indeed, neighbour-removal experiments in alpine zones on four continents have demonstrated that the influence of competition diminishes with elevation and stress, with interactions in harsher abiotic conditions being mainly facilitative (Choler et al. 2001; Callaway et al. 2002). Additionally, the balance between competition and facilitation may change throughout the growth season (Kikvidze et al. 2006) and the magnitude of interspecific interactions may vary between communities (Lortie et al. 2004).

The exclusion of highly competitive species from alpine habitats (Callaway et al. 2002) indicates functional convergence at the habitat level. Cornwell et al. (2006) demonstrated such functional convergence for woody plant species in a number of communities, which occupy less functional trait space than expected from a random community structure, as specific combinations of traits are not viable in each habitat. Grime (2006) suggests that while the general abiotic regime may select for broad convergence of physiology and life history, at finer scales functional divergence may allow segregation into niches – thus the ultimate character of communities may result from both unifying and diversifying filters operating at different scales. This is equivalent to the α (within-habitat), β (between-habitat) and γ (regional-scale) niches demonstrated by Silvertown et al. (2006), with α niches occupied by phylogenetically diverse species and β and γ niches exhibiting evolutionary conservatism (i.e. broad adaptive convergence). Thus, ‘large-scale convergent’ (β and γ) and ‘small-scale divergent’ (α) filters may operate in concert to determine the character of communities. However, is competition really the main α-scale filter determining functional divergence, or do other filters have principal roles at finer scales?

Grime (2006) suggests that the main β filter is productivity (with the potential for growth determining the extent to which organisms are adapted to compete for, or to conserve, resources) and that disturbance is ‘the more potent driver of trait differentiation and species coexistence at a local scale’. Although large-scale disturbances such as indiscriminate forest clearance may have negative impacts on species richness and community complexity (Mayfield et al. 2005), small-scale disturbances are varied in both character and timing, creating diverse opportunities for establishment, growth and reproduction, potentially promoting biodiversity. Indeed, Ackerly (2004) distinguishes between fire and small-scale disturbances in a chaparral community, which have selected, respectively, for either specific fire adaptations (‘post-fire seeders’, ‘post-fire sprouters’) or more general disturbance adaptation (‘opportunists’). Opportunists are characterized by short leaf life span, high photosynthetic rates and a suite of traits allowing them to ‘complete the life cycle before repeated disturbance’ (equivalent to ruderal species sensu Grime 2001). However, larger-scale disturbances may also be relatively discriminate: Tansley & Adamson's (1925) classic exclosure experiment demonstrated that grazing may target and suppress potential dominants, thereby allowing larger numbers of subordinate species to coexist.

The influence of disturbance on biodiversity should be particularly apparent for communities such as alpine and chaparral communities that are known to exhibit broad β- and γ-scale conservative functional convergence, but which also include ruderal species. Alpine communities may be exposed to semi-discriminate grazing and provide an ideal setting in which to compare the effects of background vs. augmented disturbance intensities. Here we test the hypothesis that subordinate species in an alpine community exhibit primarily opportunistic survival strategies (the dominant species being stress-tolerant), and that grazing may result in: (i) suppression of the dominant species (a comparatively low frequency of the most abundant species in the community), and (ii) relatively high functional diversity and species richness, particularly of opportunistic species.

To this end, we applied three independent tests of plant function [competitor, stress-tolerator, ruderal (CSR) classification, detrended correspondence analysis (DCA), and non-metric multidimensional scaling (NMDS) of key plant functional traits] to determine the influence of disturbance on the function and frequency of species comprising an alpine sedge-dominated community in situ, and compared this with a consistently disturbed pasture in the same alpine vegetation belt. The CSR classification used here (Hodgson et al. 1999) is an applied methodology that uses functional traits of wild plants to assign plant strategies according to CSR theory (Grime 1974, 1977, 1979, 2001). This classification has recently been validated in the field for a similar alpine community at a different site in the western Alps (Caccianiga et al. 2006; note that this study investigated the most frequent species involved in succession, not the coexistence of an entire community).

CSR theory suggests that productive habitats select for the ability to pre-empt resources in the vegetative phase by foraging (competitors; C), chronically unproductive habitats select for more consistent, slow-but-steady growth based on denser tissues with long life spans, in which resources are cached during resource pulses against suboptimal periods (stress-tolerators; S), and disturbance (biomass destruction) selects for regenerative traits, fecundity and rapid completion of the life cycle (ruderalism; R). Thus, in contrast to the denser tissues, low specific leaf area (SLA) and concomitantly slow growth of stress-tolerators (Poorter & Van der Werf 1998; Poorter & De Jong 1999; Weiher et al. 1999), both competitors and ruderals are characterized by high SLA and faster relative growth rates (resulting from greater internal conductivity and lesser investment in structural tissues), although ruderals invest more in the reproductive (cf. vegetative) phase of the life cycle. This CSR scheme also recognizes intermediate strategy spectra, and strategy classes may be either broadly defined (the primary strategies are simply C, S and R) or detailed (e.g. tertiary strategy classes include S/SR and SR/CSR).

Materials and methods

study sites

The study was conducted at two sites in the eastern European Alps. The first is a community dominated by the sedge Carex curvula (specifically, Hygrocaricetum curvulae Braun 1913; Grabherr & Mucina 1993), representing the late-successional stage of a primary succession sere on a glacier foreland in Val Cedec (Cedec Valley, Valfurva, Lombardy, Italy), between 46°27.164′N, 10°34.506′E and 46°27.146′N, 10°34.418′E, at an altitude of 2744–2762 m. This site, being a glacier foreland, is characterized by low temperatures (see supplementary Appendix S1 for details of the cold, continental alpine climate at the nearest weather station, 1004 m below the study site).

The second site is characterized by traditional cattle grazing practices, and thus additional disturbance – apparent as the presence of cattle and characteristic cattle-herder's stone or brick huts, known locally as Malghe (singular, Malga). Thus, grazing has undoubtedly been a long-term, continuing, selection pressure for local vegetation. This site is in the vicinity of the Santa Maria/Umbrail pass in the Valle del Braulio (Braulio Valley, Lombardy, Italy), between 46°32.526′N, 10°25.922′E and 46°32.742′N, 10°25.376′E, at an altitude between 2484 and 2598 m.

The principal difference between the study sites is that Valle del Braulio includes an important alpine pass, sufficient to support extensive cattle movement. In contrast, Val Cedec is a blind-ended hanging valley lacking roads. Although a single active Malga is present at the mouth of Val Cedec, it is located 520 m below and 3.3 km south of the study site. In other respects the two sites are similar: both belong to the alpine bioclimatic belt, overlie the same geology (metamorphic phyllites and micaschists), share the same acid soil type (‘alpine rasenbraunerde’; Giacomini & Pignatti 1955), and have similar aspect. Both communities exhibit a high degree of vegetation cover (a mean of 93% at Valle del Braulio and 83% at Val Cedec, with the gaps between vegetation consisting of exposed rock; n = 6, data not shown). Although detailed edaphic data for our study sites were not available, Caccianiga et al. (2006) determined that leaf nutrient contents in a C. curvula-dominated community were on average lower than neighbouring communities, suggesting that nutrient availability is a significant selection pressure for this type of community. Fieldwork at Val Cedec was conducted mostly during July 2005 (an interval with snow cover excluded) and at Valle del Braulio during July 2006.

analysis of plant functional types

The calculation of CSR plant strategies for each species followed Hodgson et al. (1999), with CSR coordinates adapted for ternary plots as described by Caccianiga et al. (2006). The spreadsheet used to calculate CSR strategies is available as supplementary Appendix S2. To summarize, seven traits were either measured in situ (canopy height, lateral spread), from material collected in situ and transported immediately to the laboratory (leaf dry weight, leaf dry matter content, SLA: from the same individuals as the canopy height/lateral spread measurements), or based on field observations (flowering period, flowering start). These traits were determined for six replicate individuals of all species [nomenclature follows Pignatti's (1982) flora of Italy], previously determined to be present in the two communities by floristic surveys.

Laboratory measurements followed the standardized methodologies detailed by Cornelissen et al. (2003): for the determination of leaf fresh weight (LFW) and leaf area (LA, i.e. the mean surface area of fully expanded leaves) leaf material was stored at 4 °C overnight to obtain full turgidity. LA was determined using a digital leaf area meter (Delta-T Image Analysis System; Delta-T Devices Co. Ltd, Burwell, Cambridgeshire, UK). Leaf dry weight (LDW) was then determined following drying for 24 h at 105 °C, and parameters such as SLA (i.e. LA divided by LDW) were calculated.

The frequency of CSR strategy groups in each community was used to provide a measure of functional diversity: CSR classification uses the squared Euclidean distance (or ecological distance; ED) to delimit strategy groups, with 19 primary, secondary and tertiary CSR strategies recognized (Hodgson et al. 1999). Thus, the number of CSR strategy groups present divided by 19 provides a simple functional diversity index (f.d.i.) where 1 denotes that all possible CSR strategy groups are present and 0 indicates that no strategies are present.

Adaptive trends at each site were confirmed by DCA (using MSVP 3.13o software; Kovach Computing services, Pentraeth, Anglesey, UK) and Kruskal's NMDS (Systat 9, SPSS Inc., Chicago, IL, USA) of species × traits matrices for each community. (Note that the two communities were analysed separately because the aim was to characterize the variation in each.) All data were checked for normality and, if not normally distributed, were either log- or square-root-transformed and range-standardized prior to multivariate analysis. Statistically significant correlations between DCA or NMDS axes and plant traits/strategies were identified using Spearman's correlation coefficient, following ranking of data comprising the species × traits matrix (using Systat 9).

CSR strategies were represented by colour on the DCA and NMDS projections, by converting CSR ternary triplets (e.g. C: S: R = 10 : 70 : 20%) into red, green, blue (RGB) triplet colour combinations – the system used by the human eye, electronic displays and graphics software packages, in which red, green and blue wavelengths are blended to produce a final colour. In this system colour triplets sum to 255 rather than 100%: for example, pure blue is denoted by 0,0,255 (e.g. RGB = 0,127,128 describes a green/blue colour). Thus, a CSR strategy of 10 : 70 : 20% (stress-tolerant, with ruderal tendencies) is represented by a RGB triplet of 26,178,51 (bright green with a hint of blue).

determination of species abundance

The frequency of species was determined by point analysis within six 25-m2 quadrats (5 × 5 m) at each site. Quadrats were divided into a grid of 1-m2 resolution (i.e. with 36 sampling points within each quadrat, including the corners of each quadrat), with 3-mm-diameter aluminium rods driven into the ground at each sampling point. The presence of a species was denoted by contact between one or more leaves and the rod, ultimately used to calculate the relative frequency of each species within the plot. Note that extremely infrequent species were sometimes not detected using this method, despite being observed in the floristic survey: these species were recorded qualitatively as ‘rare’. This approach had the advantage of investigating large areas of terrain and thus including rarer components of the community that would otherwise have been excluded from the functional analysis.

Results

The most abundant species in the relatively undisturbed community at Val Cedec was Carex curvula (frequency = 52.0%; Fig. 1a; supplementary Appendix S3), which exhibited a plant strategy incorporating substantial stress-tolerance and, to a much lesser extent, competitive ability (C : S : R = 17.2 : 72.9 : 9.9%): a life form that included an extremely low SLA (8.9 ± 2.28 mm2 mg−1; Appendix S2). None of the 32 other species present exhibited a similar frequency (i.e. all were subordinate to a single dominant species), and typically accounted for less than 5% of the community (Fig. 1a). The second most abundant species (Alchemilla pentaphyllea; 16.3%) exhibited a relatively generalist strategy (28.8 : 36.9 : 34.3), and the third most abundant species (Primula glutinosa; 7.8%) a broadly ruderal strategy (14.3 : 22.1 : 63.6) (Fig. 1a). The strategies of subordinate species in general ranged from extreme ruderalism (e.g. Euphrasia minima; 4.9 : 17.3 : 77.8) to strong stress-tolerance [e.g. the grasses Avenula versicolor (17.5 : 75.5 : 7.0) and Festuca halleri (16.3 : 70.6 : 13.1)]. The rarest elements of the community exhibited a general functional divergence, ranging from ruderalism (e.g. Primula daonensis; 27.4 : 7.3 : 65.3), through SR strategies (e.g. Cerastium alpinum; 9.9 : 41.8 : 48.3) to stress tolerance (e.g. Gnaphalium supinum; 17.9 : 65.4 : 16.7). No highly competitive species were present, the most competitive species being primarily stress-tolerant (Salix herbacea; 35.5 : 49.1 : 15.4), with SR strategies the most frequent in the absence of grazing (Fig. 2).

Figure 1.

 The frequency of plant strategies in an entire alpine community in the eastern Alps, with: (a) a background level of disturbance, dominated by a single species (Carex curvula), and (b) extensive artificial disturbance (cattle grazing), exhibiting loose co-dominance by five species (Avenula versicolor, Carex curvula, C. sempervirens, Nardus stricta and Trifolium alpinum) and with a greater range of plant functional types, particularly ruderals and some competitive ruderals adapted to disturbed conditions.

Figure 2.

 The diversity of primary, secondary and tertiary CSR plant strategies in an entire alpine community in the eastern Alps, with either a background level of disturbance or extensive artificial disturbance (cattle grazing). CSR classification after Hodgson et al. (1999).

Carex curvula was amongst the most frequent species in the grazed community at Valle del Braulio, but was much less prevalent than at the relatively undisturbed site (6.4% cf. 52%). Indeed, C. curvula shared weak co-dominance with four other stress-tolerators: Avenula versicolor (18.6 : 70.4 : 11.0; frequency = 9.1%), Carex sempervirens (25.2 : 67.8 : 7.0; 15.0%), Nardus stricta (22.8 : 65.2 : 12.0; 15.7%) and Trifolium alpinum (exhibiting an SR/CSR strategy: 27.1 : 45.6 : 27.3; 12.1%) (Fig. 1b; supplementary Appendix S4). All other species each accounted for less than 5% of the community (Fig. 1b). Biodiversity was much greater in the grazed community, with 76 species present in total. However, five of these were so rare that sufficient replicates could not be found for the functional analysis (Agrostis alpina, Androsace obtusifolia, Koeleria hirsuta, Luzula sudetica and Poa laxa). Greater species richness was mirrored by greater functional diversity, apparent as the presence of larger numbers of ruderal species (particularly R/CR, R/SR and SR/CSR strategies), and strategies not evident in the relatively undisturbed community (CR, CR/CSR and S/CSR; Figs 1 & 2). Indeed, the f.d.i. was 0.68 in the grazed community, in comparison with 0.53 at Val Cedec. Highly ruderal species in the grazed community included Galium pumilum (12.9 : 21.3 : 65.8), Gentianella germanica (18.0 : 0.0 : 82.0), Primula daonensis (27.4 : 2.3 : 70.3), P. farinosa (13.9 : 21.9 : 64.2), and the orchids Coeloglossum viride (28.2 : 0.0 : 71.8) and Nigritella nigra (22.6 : 1.5 : 75.9) (Appendix 4). Competitive ruderals included Cirsium spinosissimum (38.8 : 0.0 : 61.2) and Pulsatilla alpina (56.0 : 7.4 : 36.6) (Appendix 4). Strategies intermediate between S and R were also apparent, such as the stress-tolerant ruderals Daphne striata (26.3 : 36.5 : 37.2) and Gentiana kochiana (23.6 : 35.9 : 40.5).

DCA axis 1 and NMDS axis 1 were strongly and positive correlated with ruderality (R; P≤ 0.001), and negatively with stress-tolerance (S) at both sites (Table 1; Fig. 3). This reflected correlations between DCA1 and functional traits. DCA1 was positively correlated with SLA and flowering period, and negatively with leaf dry matter content, leaf dry weight and canopy height (Table 1). Indeed, DCA and NMDS projections of both communities demonstrated a general shift in plant strategies from stress-tolerant (represented by greener colours) to ruderal strategies (bluer colours), with larger numbers of extreme ruderal strategies present in the disturbed community (Fig. 3), i.e. a functional trait spectrum was present that directly reflected the CSR strategy spectrum. Competitivity (C) was significantly correlated with DCA1 in both communities (Table 1), and was strongly correlated with NMDS1 in the grazed community. However, C had a consistently weaker influence than R (Table 1) and no truly competitive survival strategies were found.

Table 1.   Spearman's correlation coefficients between plant traits and DCA and NMDS scores.
(a) Background disturbance, n = 33
TraitDCA1 (39.2%)DCA2 (13.3%)NMDS1NMDS2
Canopy height–0.521***0.113–0.452***–0.403*
Leaf dry matter content–0.547***–0.742***–0.682***0.371*
Flowering period0.730***0.403*0.786***–0.415**
Lateral spread–0.187–0.248–0.250–0.131
Leaf dry weight–0.637***0.500***–0.443**–0.662***
Specific leaf area0.713***0.2490.732***–0.062
Flowering start–0.386*–0.183–0.387*–0.237
C–0.354*0.301–0.226–0.420**
S–0.454***–0.728***–0.596***0.398*
R0.572***0.628***0.672***–0.253
(b) Additional artificial disturbance (grazing), n = 71
TraitDCA1 (28.5%)DCA2 (21.8%)NMDS1NMDS2
  1. Significantly different at *P≤ 0.05, **P≤ 0.01, ***P≤ 0.001. Numbers in parentheses denote Eigenvalues of each DCA axis.

Canopy height–0.616***–0.082–0.547***0.413***
Leaf dry matter content–0.655***0.139–0.675***0.495***
Flowering period0.232*0.758***–0.134–0.629***
Lateral spread0.0910.501***–0.375**–0.035
Leaf dry weight–0.445***0.014–0.499***0.154
Specific leaf area0.799***–0.1590.697***–0.666***
Flowering start–0.085–0.725***0.0280.429***
C–0.531**0.267*–0.641***0.188
S–0.540***0.039–0.504***0.480***
R0.694***–0.1340.711***–0.501***
Figure 3.

 Detrended correspondence analysis (DCA) and non-metric multidimensional scaling (NMDS) biplot axes 1 and 2 for species × traits matrices of an alpine sedge-dominated community (background disturbance) and a grazed pasture in the same alpine vegetation belt (additional artificial disturbance – grazing). CSR strategies (i.e. CSR ternary triplet coordinates) are represented by RGB (red, green, blue) ternary triplet values, i.e. pure colours represent pure strategies, with mixed colours representing intermediate strategies (e.g. green/blue = SR strategy; see text for details). CH = canopy height, LS = lateral spread, SLA = specific leaf area, LDW = leaf dry weight, LDMC = leaf dry matter content, FS = flowering start, FP = flowering period.

Discussion

filters governing biodiversity

We demonstrated α-scale functional divergence principally in response to disturbance. This was evident as a spectrum of ruderal to stress-tolerant survival strategies within the CSR footprint of the community as a whole. Increased disturbance intensity encouraged both functional and species diversity (particularly of ruderals) and weaker co-dominance, as potentially dominant stress-tolerating graminoids were suppressed. At the β-scale, functional convergence in response to stress is suggested from the stress-tolerance required for dominance and by the constraint of survival strategies to a subset of potential strategies (i.e. with competitor strategies filtered out). These empirical data, measured in situ, support Grime's (2006) supposition that conservative vs. competitive strategies are imperative at β-scales, with disturbance a principal filter governing α-scale functional divergence, and thus coexistence and biodiversity.

The presence of ruderals, and a smaller proportion of CR strategists in the more disturbed community, not only confirms that disturbance strongly influences niche segregation, but also suggests that the edaphic environment could be a deciding factor: faster-growing ruderals and competitive ruderals necessarily require a local abundance of nutrients and, due to rapid inherent development and completion of the life cycle, are well adapted to propagate genes before ephemeral nutrient patches become exhausted. Consolidation of nutrients can occur during litter decomposition (particularly from the relatively rich litter of hemiparasites; Quested et al. 2003, 2005), nitrogen fixation or nutrient redistribution by animals (Grime 2001). Hemiparasites, considered keystone species by Press & Phoenix (2005), are unlikely to play a leading role in niche differentiation, being extremely infrequent in both communities (i.e. Euphrasia minima, Pedicularis kerneri and P. tuberosa; Appendices S3 & S4). The nitrogen-fixing symbiosis of legumes (Fabaceae) is potentially more influential (Trifolium alpinum was the only non-graminoid co-dominant in the grazed community). Additionally, although grazing is primarily a disturbance, the association of competitively inclined ruderals with grazing suggests that nutrient consolidation during defecation could produce niches with less stringent abiotic limitations – an additional form of facilitation.

Contrasting resource management strategies are more important than foraging and resource competition in these communities, and the facilitation invoked in the present study is indeed more evident than suppressive competition as a factor encouraging biodiversity in alpine habitats in general (Callaway et al. 2002). Niche differentiation is unlikely to be based on simple resource availability gradients (sensu Tilman 1988): a model in which resource availability in neighbouring niches may be either relatively stable (and limited for a substantial proportion of the year) or comparatively dynamic (and frequently abundant) would be more realistic. Species are adapted to the former situation by ‘resistance and persistence’ (S) and to the latter by ‘fast fecundity and fatality’ (R), and combinations thereof. The hypothesis that plants have further adapted to consistently luxurious niches in productive habitats with ‘pre-emptive plasticity’ (C) is compatible with this view.

beyond purely resource-based niche theories

A lack of mechanistic or evolutionary insight from resource-based niche theories has nurtured languor in niche research (Silvertown et al. 1999) and encouraged neutral theories of biodiversity, which assume that functional differences between organisms are so slight as to have negligible effects on community structure (Hubbell 2001). By disregarding the functional differences between organisms, neutral theories are founded on self-admittedly unrealistic assumptions – but are potentially more useful than inadequately expressed niche theories (Hubbell 2005). Inertia in niche research stems from the belief that the limited types of resources required for growth limits the number of niches that can exist (Silvertown 2004). Limited classes of resources do exist for plants, including light, water, nutrients, CO2, O2, physical space, pollinators, dispersal vectors and other symbionts. Not all are necessary for growth, but all may be required for survival: a resource can be defined as any external factor required to complete a developmental stage. Additionally, resource availability is not the only survival risk to which plants must be adapted (e.g. temperature extremes), and combinations of environmental factors fluctuate over varied spatio-temporal scales. Just as all music is composed of only eight basic tones, habitats include characteristic repetitive rhythms, sequences of events and combinations of factors co-occurring over different scales and within and between heterogeneous microhabitats. For example, light exhibits variation in intensity and quality depending on season, time of day, local topography, vegetation type (and position within the vegetation) and cloud cover. Direct sunlight may also act as an injurious stressor, particularly in concert with temperature extremes (e.g. Manuel et al. 1999). Thus, light alone defines innumerable niches to which plants are adapted, to the extent that different members of the same genus may exhibit diverse adaptive traits that enhance or diminish light interception (Pierce 2007). A dynamic annual regime of selection pressures governs adaptive radiation, with a symphony of biotic and abiotic factors creating an immeasurably vast number of spatio-temporal niches, and therein lies the problem: attempting to define all of the realized niches present in a particular habitat, or the full extent of the fundamental niche of each species, would be impractical. Indeed, credible attempts to delimit realized niches have been restricted to the identification of a single overarching environmental factor (Silvertown et al. 1999; McKane et al. 2002). A complementary approach, demonstrated by the present report and studies such as Caccianiga et al. (2006), is the detailed quantitative comparison of the survival strategies (eco-physiological/functional types) of component organisms in situ, and the functional trait space occupied by the community as a whole.

In the search for unifying principles in ecology it is easily forgotten that biology already has a ‘grand unified theory’ (Dawkins 2004) in which the key processes of natural and sexual selection rely on functional adaptation and survival, not simply growth. Plant strategies represent contrasting general approaches to survival, and specific functions such as growth and competitive ability are survival tools that contribute to the overall functional capability of the organism. Thus, strategies may be characterized not only by different relative growth rates, but also by differential investment between phases of the life cycle (sensu Grime 2001). Survival does not rely simply on differences in growth and competitive ability (sensu Tilman 1988). This view is of central importance to the coexistence debate: even recent contributions assume that resource competition universally dominates the evolution of plant traits and thus niche segregation and coexistence (Craine 2005, 2007; Tilman 2007; cf. Grime 2007). However, resource competition theories cannot be applied to habitats where competition is not the main selection pressure or where other phases of the life cycle are decisive for survival. Furthermore, productive habitats and grasslands are relatively modern, with stress dominating early terrestrial plant life (competition is unlikely to have been of importance prior to the Devonian explosion, 409–363 Myr bp; Bateman et al. 1998; reviewed by Pierce et al. 2005). Although grasslands have become widespread and important since the early/mid Miocene (20–10 Myr bp; Jacobs et al. 1999), and are accessible and easy to manipulate experimentally, it is unreasonable to assume that the selection pressures structuring these communities have been equally influential throughout all biomes for 490 million years of plant terrestrial adaptive radiation. A growing body of empirical data demonstrates the importance of facilitation, injurious stress and disturbance in relatively unproductive habitats (e.g. Callaway et al. 2002; Pierce et al. 2005; this study), suggesting that the integrity and general applicability of resource-competition theories should be queried in depth. Indeed, omission of these factors could explain Tilman's (2007) conclusion that ‘our theory and experiments are simplifications that miss important constraints and trade-offs that lead to coexistence in nature’.

Acknowledgements

We thank Elena Castelli, Mattia Castiglioni, Alessio Martinoli, Alessandro Ossola, Francesca Pilotto, Lorenzo Tarenghi and Debora Tollardo for assistance in the field and laboratory, the workers at the Pizzini-Frattola refuge for making the mountain more hospitable, and a number of anonymous reviewers for constructive criticism of an earlier version of this manuscript. S.P. and A.L. were supported by the Centro Flora Autoctona della Regione Lombardia (the Native Flora Centre of the Lombardy Region), via the University of Insubria.

Ancillary