Species traits and their non-additive interactions control the water economy of bryophyte cushions

Authors

  • Pascale Michel,

    Corresponding author
    1. Biodiversity and Conservation Team, Manaaki Whenua-Landcare Research, Private Bag 1930, Dunedin 9054, New Zealand
    2. Department of Botany, University of Otago, PO Box 56, Dunedin 9054, New Zealand
    Search for more papers by this author
  • William G. Lee,

    1. Biodiversity and Conservation Team, Manaaki Whenua-Landcare Research, Private Bag 1930, Dunedin 9054, New Zealand
    Search for more papers by this author
  • Heinjo J. During,

    1. Department of Ecology & Biodiversity, Utrecht University, PO Box 800-84, 3508 TB Utrecht, The Netherlands
    Search for more papers by this author
  • Johannes H. C. Cornelissen

    1. Systems Ecology, Department of Ecological Science, Faculty of Earth and Life Sciences, VU University Amsterdam, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands
    Search for more papers by this author

Correspondence author. E-mail: michelp@landcareresearch.co.nz

Summary

1. Ecological processes in mixed-species assemblages are not always an additive function of those in monocultures. In areas with high ground cover of bryophytes, renowned for their considerable water retention capacity, non-additive interactions in mixed-species cushions could play a key role in the ecosystem water economy.

2. We investigated mixture effects on external water loss in natural cushions with different species pairs of mosses and liverworts and the underlying mechanism of any non-additivity through shoot characteristics and canopy traits.

3. Species mixtures in bryophyte cushions had both additive and non-additive effects on the water economy, and these interactions were dependent on the composition of species assemblages and on plant tissue mass. Non-additivity of species mixtures was positive, resulting in the improvement of water retention.

4. Our results suggest advantages for bryophyte species to grow smaller and denser when in mixtures. They appear to alter the surface exchange area to converge in size with their neighbours, thus controlling boundary-layer properties and evaporation to reduce water loss.

5.Synthesis. A shift in bryophyte assemblages thus may influence ecohydrological processes of various ecosystems that cannot be simply predicted from the water economy of the component species when in monospecific cushions. In contrast to vascular plants, bryophytes do not compete for water but share it, and trait plasticity amongst bryophyte mixtures acts as a critical physical strategy in the community water economy.

Introduction

Plant–plant interactions are a key driver of community assembly and ecosystem processes. Plants may do better or worse in coexistence with other plants, and the underlying mechanisms of competition and facilitation are known to be important in shaping vegetation composition and community structure (Brooker et al. 2008). The balance between competition and facilitation can also depend on whether the interactions are intraspecific or interspecific, i.e. whether species grow in monocultures or mixtures (Weigelt et al. 2007; Spitale 2009). For instance, in mixed-species forest stands, plants often alter their root placement in reaction to neighbouring species (Bartelheimer, Steinlein & Beyschlag 2006; da Silva et al. 2009). It has been recognized that monospecific vs. mixed-plant assembly not only matters for plant performance but also for some key ecological processes and the services they control. Indeed, ecological processes in mixed-species assemblages cannot always be predicted simply from the relative contributions of the component species derived from measurements in monoculture. Non-additive effects on processes have been reported particularly for mixed-species litter decomposition, where species interactions alter mass loss, decomposition rate, nutrient concentration and decomposer community and activity (Hättenschwiler, Tiunov & Scheu 2005; Ball et al. 2008; Hoorens, Stroetenga & Aerts 2010). The concept of non-additivity was also implied in explaining patterns of water transport amongst bryophyte communities (Spitale 2009), but bryophyte interaction effects on water economy have not, to our knowledge, been experimentally quantified.

Interactions between bryophyte species are mainly constrained by water and light as the most growth-limiting factors (Bates 1988; Rydin 1997a). The capacity of bryophytes to take up and retain water is often tied to species variation in morphological and life-history traits and to their survival and performance in ecological niches, including ones with prolonged dry periods (Hedderson & Longton 1996). Growth form (i.e. configuration of shoot aggregation) and shoot morphological and anatomical features (e.g. cell papillosity and presence of water-conducting tissues) often dictate water relationships in bryophytes (Proctor 1981, 2009; Bates 1988; Zotz et al. 2000; Rice & Schneider 2004; Rice, Aclander & Hanson 2008; Elumeeva et al. 2011). Size and allometry of plant shoots also contribute to the relative proportion of organs and tissues that control water fluxes and storage. Fine-scale variations in canopy, branch and leaf structure further interact with wind flow, generating turbulences affecting boundary-layer properties and evaporation (Rice, Collins & Anderson 2001; Rice & Schneider 2004). The erect growth form, for example, in combination with high turf density is considered to be an adaptation to water stress by improving uptake and moisture retention and reducing air movement close to the upper leaves in the canopy (Robinson, Vitt & Timoney 1989). Shoot densities in bryophyte cushions are regulated by intrinsic mechanisms (Van der Hoeven & During 1997, 2008) with higher density of shoots improving moisture conditions (Spitale 2009; Elumeeva et al. 2011). Often, the relationships between area for evaporation and water storage volume in moss species are summarized by measures of leaf area-to-volume ratio (Rice & Schneider 2004). However, a mixed-species cushion of morphologically different bryophytes is likely to result in a greater diversity of architecture, thus further modifying the micro-scale water balance (Mulder, Uliassi & Doak 2001). Spitale (2009) suggested a specific pattern of water transport through monospecific colonies, which cannot be cumulative in species assemblages. Thus, water transport amongst colonies, and water retention, might be non-additive as a result of micro-differences in colony architecture. Yet, non-additivity in the local water economy of bryophyte assemblages has not been tested explicitly nor has the role of the associated physical architecture of these cushions.

In this study, we hypothesize that bryophyte species mixtures have non-additive effects on the water economy of cushions as dictated by cushion architecture. To test this hypothesis, we investigated eight subarctic species of mosses and liverworts by comparing the water economies of monocultures and two-species mixtures that naturally occur in the field. Specifically, we experimentally assessed the time for individuals in single- and mixed-species cushions to lose 50% of their external water. In controlling for wind effect in greenhouse conditions, we only tested for the interactions between desiccation rate and water retention potential in a cushion. We also measured canopy characteristics in cushions of different compositions that could underpin differences in their water economy.

Materials and methods

Field sampling and experimental work were carried out at Abisko Scientific Research Station (68º22′N, 18º46′E, 403 m a.s.l.), North Sweden, from 3 to 16 July 2009. The study included seven species of mosses (for nomenclature see Hill et al. 2006): Aulacomnium palustre (abbreviation Aulpal), Cinclidium stygium (Cinsty), Dicranum scoparium (Dicsco), Hylocomium splendens (Hylspl), Pleurozium schreberi (Plesch), Polytrichum commune (Polcom) and Tomentypnum nitens (Tomnit); and one species of liverwort: Lophozia lycopodioides (Wallr.) Cogn. (Loplyc). Based on their natural occurrence locally in cushions in monoculture and species mixtures, eight different monocultures and eight different species-mixture combinations were selected. Only mixed-species communities of two species, containing c. 50% by volume of each species, were sampled. Samples for typical monocultures and mixtures of a given species were taken in close proximity to one another, from apparently similar environments, avoiding spatial clustering of monocultures vs. mixtures. Each cushion was first harvested for the water relationships experiment and for measurement of canopy traits (n = 96). Three subsamples (averaging a total of 20 g wet mass) of each cushion were randomly collected for measurements of shoot characteristics, and mean values of shoot traits were used in the final statistical analysis (n = 16).

Water Relationships Experiment

For each cushion type, six core samples of 5 cm in diameter were collected using a soil corer, resulting in six replications of 16 bryophyte ‘communities’. Each core was taken immediately from the field into the laboratory, cut to only keep the upper live green plant material while limiting disturbance of the cushion structure and retaining field shoot density. Sporophytes and non-target materials (e.g. soil, dead leaves, live vascular plant parts or other bryophyte species, invertebrates) were removed using forceps because if included, they are likely to overestimate water content and alter boundary layer and thus evaporation rate of a cushion. Each sample was then placed into a numbered pot of 5 cm in diameter and 5 cm in height, with a mesh at the bottom to provide good drainage. Three point measurements of cushion height were taken to the nearest 0.1 mm to derive the mean cushion height of each sample. The experimental design comprised six blocks, each represented by a tray on which the 16 pots (one replicate for each cushion type) were randomly arranged in a partially (c. 50%) shaded, well-ventilated glasshouse.

All pots were watered to saturation and left for 12 h allowing plants to rehydrate. They were then watered a second time and left to drip for 30 min. Finally, cushions were watered using a spray bottle 2 min prior to first weighing. Bryophyte cushions were weighed at time 0, 2, 4, 8, 12 h and then every 12 h until all samples had reached 50% water loss. After each weighing, pots were randomly repositioned within an experimental block (tray). Final dry mass was measured at the end of the experiment, as described below. To account for possible effects of variability in environmental conditions, the experiment was repeated in exactly the same way, with the same samples, 12 h after the first run. Ambient temperature and relative air humidity in the glasshouse were recorded for the whole experimentation period. Air temperature and relative humidity for the area were obtained from the Abisko Scientific Research Station weather station, located 25 m from the glasshouse.

Shoot Characteristics and Canopy Traits of Bryophyte Communities

Size (vertical length, horizontal width and surface area) and aspect ratio were considered a critical set of traits in the regulation of water-holding capacity in bryophyte species. Furthermore, size in plants is often dictated by coexisting species in a community (Økland & Økland 1996) and thus is likely to vary in bryophyte cushions of mixed species. Canopy traits of bryophyte cushions were expressed in terms of shoot density, total dry mass, canopy height, shoot surface area-to-volume ratio and similarity in terms of shoot traits between the species making up the canopy.

Each bryophyte species and foreign material (twigs, litter and other plant materials) in a sample were separated, dried at 60 °C for 48 h and weighed for dry mass. Shoot density of each bryophyte species in a cushion was also determined prior to drying and weighing. Mean dry mass of a shoot was then calculated to derive species contribution to cushion total biomass and to adjust predicted community values for mass. The cylinder formula was used to calculate the volume of a cushion from mean height measurements (Balci & Kennedy 2003; Rice & Schneider 2004).

Cushion subsamples were thoroughly washed to remove soil and invertebrates, air-dried, sealed into packets with silica gel and sent to be processed in PC1 Lab facilities at the Department of Botany, University of Otago, New Zealand, for shoot measurements (Ministry of Agriculture & Forestry, Biosecurity, New Zealand, Permit No: 2009036513). After rehydration for 24 h, shoot morphometry was quantified following Bond-Lamberty & Gower (2007), who used a flatbed scanner to measure shoot morphometry to derive bryophyte-specific leaf area (SLA) and leaf area index (LAI) from the projected area. Bond-Lamberty & Gower (2007) concluded that although the method may be inaccurate in the estimates of stand-level LAI, the precision was high in measurements of projected area. In our study, six shoots of each species in each of the 16 cushion types were scanned using an Epson Expression 10000 XL scanner, and digitized (350 dpi), to obtain vertical length (L), horizontal length (W), aspect ratio (W/L) and horizontally projected shoot area using the image analysis function in the software package WinFolia. Total surface area of a cushion was estimated by multiplying surface area of a shoot by the numbers of shoots.

Statistical Analysis

The water economy in bryophytes is adequately expressed on the basis of water loss rate (Zotz et al. 2000). We thus interpolated time (h) to 50% water loss (T50) for each monoculture and mixed-species cushion. Variation in time to 50% water loss between cushion types observed during the first experimental run strongly correlated with values observed during the second run ( inline image = 0.864, P < 0.01). Climatic conditions were similar during the two experimental runs. Temperatures averaged 15.9 ± 3.5 °C (min. = 10.6 °C, max. = 26.5 °C) during the first run and 16.1 °C ± 4.1 °C (min. = 9.3 °C, max.25.7 °C) during the second. Relative humidity averaged 73%± 15% (min. = 41.4%, max. = 99.5%) and 68.9%± 13.7% (min. = 37%, max. = 94.2%), respectively. Thus, values for T50 used in all analyses were the averages of observed values from both runs. Variance in T50 between experimental blocks and cushion type was first tested using one-way anova with Tukey post hoc comparison. Homogeneity of variance was checked with Levene’s test, and if significant, data were square-root transformed. If Levene’s test remained significant after transformation, the nonparametric Kruskal–Wallis test was used, with the Mann–Whitney U-test used for post hoc comparisons.

From the measurements (observed T50 values) in monocultures, we also derived expected (predicted) T50 values for species mixtures in two ways: (i) averaging of observed values in monocultures (non-weighted values) and (ii) adjusting observed values from monoculture by percentage contribution of each species to the total cushion biomass (weighted values). We considered an effect of species mixture on external water loss to be non-additive when observed values for T50 were either significantly higher or lower than predicted values (values significantly different from zero). Thus, non-additivity was expressed by calculating an interaction strength value as: 1 − [predicted/observed] (Hoorens, Stroetenga & Aerts 2010). We tested whether mean interaction strength of a mixed-species cushion differed from zero using one-sample t-tests (value = 0). When observed T50 was significantly higher than expected, the interaction was positive (facilitation with respect to water retention), and when lower, the interaction was negative (competition) (Hoorens, Stroetenga & Aerts 2010).

The construction of functional dendrograms from information about species’ traits is increasingly used in ecology to determine how functional trait composition varies between species assemblages (Petchey & Gaston 2007; Mouchet et al. 2008). Greater dissimilarity in shoot structure between bryophyte species is likely to increase the complexity of cushion architecture. We thus used a cluster analysis to calculate shoot trait similarity (%) between two bryophyte species in a mixture, expressed by Bray–Curtis similarity coefficients (calculated in the computer package primer V5), to measure structural complexity of mixed-species cushions. Data were first log +1 transformed and standardized. Species clusters were then plotted against canopy traits and evaporation rate. To examine whether cushion architecture explained water loss in bryophytes, we tested nonparametrically for relationships between shoot or canopy traits and: (i) observed external water loss, (ii) non-weighted interaction strength (with respect to water loss) and (iii) weighted interaction strength using the Spearman rank correlation coefficient.

Results

Observed Time to 50% Water Loss

The time bryophyte communities took to reach 50% water loss differed amongst cushion types (F15,80 = 9.3, P < 0.0001), and these results were not affected by a block effect within the greenhouse (F1,46 = 0.47, P = 0.88). Three species, i.e. A. palustre, D. scoparium and L. lycopodioides, lost external water more slowly than the average, and P. commune was the fastest to lose external water (Table 1). Mixtures did not provide greater efficiency in water economy than monocultures overall (F1,94 = 0.017, P = 0.898), instead water retention depended greatly on species composition. Only three species (A. palustre, D. scoparium and P. schreberi) showed significant differences in desiccation rate when growing in mixtures in comparison with monocultures (Fig. 1), with mixtures evaporating faster in most cases. D. scoparium was the only species that benefitted from growing in a mixture with A. palustre. These two species together reached the maximum T50 of all communities.

Table 1.   Canopy traits and water retention capacity [observed time to 50% water loss (T50 ± SE)] for single- and mixed-species cushions of bryophytes in a greenhouse experiment (n = 6)
CommunityCanopy traitsObserved
Shoot densityShoot area to volume (cm cm−3)Canopy height (cm)Dry mass (g)T50 (h) (Mean ± SE)
  1. Abbreviations for bryophyte species can be found in Fig. 1. None of the canopy trait measures differed significantly between monocultures and mixtures (P > 0.05), and all differed between cushion type (P < 0.05) (n = 96, one-way anova).

Monocultures
 Aulpal133.1 ± 13.619.76 ± 1.553.43 ± 0.231.00 ± 0.0640.2 ± 4.7
 Dicsco89.0 ± 10.014.67 ± 1.642.16 ± 0.101.26 ± 0.0536.0 ± 3.3
 Loplyc432.5 ± 83.28.41 ± 1.591.75 ± 0.291.15 ± 0.1329.3 ± 6.1
 Tomnit33.5 ± 5.93.35 ± 0.304.03 ± 0.420.93 ± 0.1723.7 ± 5.1
 Cinsty76.0 ± 7.24.56 ± 0.725.28 ± 0.400.67 ± 0.1121.3 ± 1.8
 Plesch73.6 ± 2.718.41 ± 2.124.15 ± 0.581.15 ± 0.0921.2 ± 0.7
 Hylspl17.3 ± 0.66.90 ± 0.454.63 ± 0.291.19 ± 0.8018.1 ± 2.2
 Polcom42.8 ± 5.67.38 ± 0.847.38 ± 0.141.56 ± 0.2212.2 ± 1.2
 Mean monocultures112.7 ± 20.810.43 ± 0.964.10 ± 0.271.09 ± 0.0625.3 ± 1.7
Mixtures
 Aulpal + Cinsty91.8 ± 14.49.52 ± 2.212.66 ± 0.280.74 ± 0.0722.6 ± 2.7
 Aulpal + Dicsco105.3 ± 4.411.51 ± 2.772.83 ± 0.331.33 ± 0.0856.6 ± 7.3
 Aulpal + Hylspl50.5 ± 3.711.93 ± 0.683.36 ± 0.170.95 ± 0.0925.6 ± 4.7
 Hylspl + Loplyc216.6 ± 30.37.29 ± 1.832.13 ± 0.191.08 ± 0.0923.6 ± 2.1
 Hylspl + Plesch38.0 ± 3.87.35 ± 0.454.70 ± 0.481.03 ± 0.1218.17 ± 1.9
 Hylspl + Polcom52.0 ± 19.15.60 ± 3.095.65 ± 0.391.31 ± 0.1717.6 ± 2.6
 Polcom + Plesch69.5 ± 27.56.69 ± 3.034.53 ± 0.281.11 ± 0.1316.0 ± 1.3
 Tomnit + Cinsty48.1 ± 4.94.59 ± 0.344.40 ± 0.190.73 ± 0.7418.8 ± 2.1
Mean mixtures84.0 ± 9.78.06 ± 0.693.78 ± 0.191.03 ± 0.0524.9 ± 2.2
Figure 1.

 Water retention capacity of eight bryophyte species in monoculture and mixture (mean T50 ± SE). Aulpal = Aulacomnium palustre, Cinsty = Cinclidium stygium, Dicsco = Dicranum scoparium, Hylspl = Hylocomium splendens, Loplyc = Lophozia lycopodioides, Plesch = Pleurozium schreberi, Polcom = Polytrichum commune and Tomnit = Tomentypnum nitens. T50 differed significantly amongst species (one-way anova, F = 8.736, P < 0.001). Only three species (Aulpal, Plesch and Dicsco) showed significant differences in their ability to retain water when growing in mixtures in comparison with monocultures [one-way anova, different letters indicate significance at P < 0.05 (Tukey), n = 6]. Species showed differences in response. For example: growing in mixture with Cinsty and Hylspl is detrimental to Aulpal. In contrast, Dicsco benefits by cohabiting with Aulpal.

Interaction Strength

Predicted values for T50 calculated from c. 1 : 1 volume mixtures without accounting for species contribution to cushion total biomass were, on average, not significantly different from the mean for observed values (t = 0.12, d.f. = 67, P = 0.90). However, when predicted values were adjusted to weight, they became significantly different from observed values (t = 3.01, d.f. = 65, P = 0.004). Interaction strength (difference ratio between predicted and observed) was on average significantly different from zero (non-weighted values: t = −2.42, d.f. = 47, P = 0.019; weighted values: t = 3.71, d.f. = 47, P = 0.0005) and greatly different between communities (non-weighted values: F7,40 = 3.2, P = 0.009; weighted values: F7,40 = 10.8, P < 0.001). Interaction strength for T50 calculated for c. 1 : 1 volume mixtures, without accounting for species contribution to cushion total biomass, was significantly different from zero for only two communities (A. palustre D. scoparium and A. palustre C. stygium) (Table 2). When weighting was implemented, four pairs of species (A. palustre D. scoparium, H. splendens L. lycopodioides, H. splendens P. commune and T. nitens C. stygium) produced a positive deviation from ‘neutrality’ (additivity), suggesting that in these communities, species-mixing plays an important role in retaining external water.

Table 2.   Predicted water retention capacity [weighted and non-weighted predicted time to 50% water loss (T50 ± SE) derived from observed values of monocultures] and additive effect of mixture [mean interaction strength (±SE)] for mixed-species cushions of bryophytes in a greenhouse experiment (n = 6). Abbreviations for bryophyte species can be found in Fig. 1
CommunityShoot trait similarity (%)†Non-weightedWeighted
T50 (h) Mean ± SEInteraction strength Mean ± SET50 (h) Mean ± SEInteraction strength Mean ± SE
  1. *Interaction strength values differing from zero at P < 0.05 (one-sample t-test, test value = 0).

  2. **Interaction strength values differing from zero at P < 0.01 (one-sample t-test, test value = 0).

  3. †Values of Bray–Curtis similarity coefficient based on shoot traits of each individual bryophyte species (Appendix 1).

Aulpal + Cinsty91.9130.8 ± 2.6−0.45 ± 0.20**27.0 ± 1.1−0.27 ± 0.14
Aulpal + Dicsco92.4938.1 ± 2.80.28 ± 0.08*26.0 ± 1.20.51 ± 0.04**
Aulpal + Hylspl75.3729.2 ± 3.0−0.39 ± 0.3622.1 ± 0.70.13 ± 0.33
Hylspl + Loplyc43.5323.7 ± 3.1−0.07 ± 0.2013.5 ± 1.80.41 ± 0.07**
Hylspl + Plesch90.7619.7 ± 1.1−0.11 ± 0.0816.6 ± 0.80.05 ± 0.09
Hylspl + Polcom93.2216.7 ± 0.50.06 ± 0.149.0 ± 1.10.48 ± 0.03**
Polcom + Plesch91.8122.5 ± 2.2−0.11 ± 0.1216.6 ± 0.3−0.09 ± 0.09
Tomnit + Cinsty88.3524.5 ± 1.3−0.28 ± 0.3111.3 ± 0.80.39 ± 0.03*

Bryophytes Shoot and Canopy Traits

The eight bryophyte species formed three main clusters based on shoot measurements of vertical length, horizontal length, aspect ratio and surface area (Fig. 2). Shoot characteristics were 70% similar (Bray–Curtis similarity coefficients) amongst all moss species, but only 45% similar to the only liverwort species in this study, L. lycopodioides. Mosses were further separated at 90% similarity into two groups: (i) P. commune, H. splendens and P. schreberi and (ii) A. palustre, T. nitens, D. scoparium and C. stygium.

Figure 2.

 Time to 50% water loss (T50) for single- and mixed-species bryophyte cushions plotted against canopy traits: (panel a) shoot density; (panel b) shoot area to volume; (panel c) canopy height and (panel d) dry mass. Clusters of bryophyte species according to similarity in shoot characteristics are circled as follows: blue = group 1, green = group 2, red = group 3. Abbreviations for bryophyte species can be found in Fig. 1. Graphic representations showed clearly the three defined clusters, with communities of species from group 2 generally evaporating the fastest. Cushions formed by the liverwort Lophozia lycopodioides (Loplyc) (group 3) distinctively had the densest and shortest canopy and showed average desiccation rates.

Species from group 1 tended to have the longest and largest stems, with the greatest shoot surface area and the shortest T50 of all the studied species (Appendix 1, Fig. 2). These species also showed the greatest variability in shoot characteristics (vertical length, horizontal length and surface area) depending on community composition, and in general, they were much smaller in size when in mixture (Appendix 1). Species from group 2, with the exception of T. nitens, differed significantly only in vertical length when in mixture. Cinclidium stygium in mixture with T. nitens was the only case of an increasing dimension (height) to adjust to the coexisting species in the cushion. In general, adaptation to coexisting species affected primarily the vertical length of a species and consequently the surface area. Three species (P. commune, H. splendens and A. palustre) reduced their height when in mixture by up to half and more than when in monoculture. In these cases, species did not appear to adjust the height to that of the coexisting species, but instead both species in the mixture reduced their height equally to form shorter and denser cushions.

Canopy traits differed between cushion types (shoot density: χ15,80 = 76.563, P < 0.001; shoot area to volume: χ15,80 = 63.644, P < 0.001; canopy height: F15,80 = 20.656, P < 0.001 and dry mass: F15,80 = 4.246, P < 0.001) and did not between monocultures and mixtures (shoot density: F1,94 = 0.127, P = 0.722; shoot area to volume: F1,94 = 3.003, P = 0.086; canopy height: F1,94 = 0.270, P = 0.604 and dry mass: F1,94 = 0.69, P = 0.408) (Table 1). Monoculture cushions of P. commune were the tallest and heaviest of all studied communities and showed the lowest T50. The only liverwort, L. lycopodioides, was much smaller and had greater shoot density and dry mass to volume than any of the moss species (Table 1, Appendix 1).

Cushion Architecture and External Water Retention Capacity

The importance of shoot traits to water retention capacity was manifested in the water loss responses of different species clusters based on similarities in shoot traits (Fig. 2). Species from group 1 in monoculture or in mixtures of the same cluster consistently lost external water the fastest (12–25 h). However, T50 did not relate to individual shoot traits of bryophyte species and instead strongly related to canopy traits of the cushions (Table 3). T50 increased with increasing shoot density, dry mass and shoot area to volume and with decreasing cushion height.

Table 3.   Spearman correlation coefficients between shoot and canopy traits of bryophyte cushions and water relations in a greenhouse experiment. Mean values of shoot traits for the six measured shoots collected from three randomly chosen subsamples of each community (n = 16) and observed values of canopy traits for the 96 collected cushions in the experiment were entered in the correlation analysis
 Time to 50% water lossInteraction strength
Non-weightedWeighted
  1. *P < 0.05, **P < 0.01; †n = 16, ‡n = 8.

Shoot traitsn = 16n = 8n = 8
Vertical length−0.411−0.145−0.092
Horizontal width−0.470−0.303−0.185
Aspect ratio0.038−0.445−0.259
Surface area−0.380−0.331−0.123
Canopy traitn = 96n = 48n = 48
Shoot density0.595**0.299*0.303*
Dry mass0.379**0.694**0.486**
Shoot area to volume0.499**0.276−0.016
Cushion height−0.466**−0.0270.005
Shoot traits similarity0.010†0.242‡0.142‡

The correspondence of water retention in mixed-species bryophyte communities with values estimated from observations of monocultures was strongly related to the shoot density and biomass of a cushion (Table 3). The weighting of species contribution to cushion total biomass in the calculation of predicted water loss values did not affect these interactions. Correlation results suggested that mean interaction strength values increased with increasing biomass (and to a much lesser extent shoot density) (Fig. 3). Graphic representations further suggested that cushions of extreme dry mass (the highest and the lowest) were more likely to produce deviations from the ‘neutral additivity’. Our measure of canopy complexity (shoot traits similarity) did not correlate with water loss T50 or interaction strength, possibly failing to capture the intertwined branching patterns of canopy structure.

Figure 3.

 Correlation plots for the mean interaction strength ± SE [1 – (predicted/observed)] in T50 of water loss against canopy traits of bryophyte cushions that show significance. Larger black dots indicate mean interaction strength values significantly differing from zero (P < 0.05, one-sample t-test). Positive values indicate a facilitation effect and negative values a competition effect of species mixture on cushion water economy. (Panel a) Non-weighted values; (Panel b) dry-mass-weighted values. Strong correlations are observed between mean interaction strength and dry mass (RSpearman = 0.694, P < 0.001), and only cushions of extreme plant tissue mass and shoot density (the highest and the lowest) displayed non-additivity of species mixture on water economy (non-weighted values). When species contribution to cushion biomass was considered, these relationships weakened.

Discussion

This study presents the first explicit demonstration of non-additive effects of species mixtures on water economy. External water loss in two-species bryophyte cushions was partially a function of neighbouring species, and plant tissue mass of the species assemblages may contribute to the additive or non-additive effect of species mixture. External water loss of c. 1 : 1 volume mixtures could successfully be predicted for 75% of the studied species assemblages by averaging observed values without accounting for species contribution to total biomass. Adjusting for species biomass reduced the success of our predictions to only 50%, and predictions always resulted in an underestimation of desiccation rate in comparison with the actual data. For these communities, species mixture was beneficial and water loss could not be predicted from monocultures. In general, non-additivity acted in cushions of dense plant tissue mass, independently of shoot characteristics and canopy traits, possibly because bryophyte species showed the ability to converge with their neighbours in shoot morphometry to form denser cushions.

Explaining Non-Additive Effects on Water Retention from Species’ (Plasticity in) Morphology

Bryophyte cushions of mixed species, especially of mixed growth form, are expected to be architecturally more complex than single-species cushions, resulting in greater water-holding capacity and lower evaporation rate (Mägdefrau 1982; Scandrett & Grimingham 1989; Rixen & Mulder 2005). Our results do not support this argument, and instead, the strongest mutually beneficial association in our study was observed between the structurally similar acrocarpous mosses (A. palustre and D. scoparium), both species displaying reduced evaporation rate when growing together. Despite their horizontal growth form adapted to facilitate water absorption (water content = 485–625% dry mass), P. schreberi and H. splendens showed little interaction on evaporation rate when growing together. Reduced water retention was indeed observed when P. schreberi grew in mixture with the tall and large endohydric moss P. commune or H. splendens with the acrocarpous moss A. palustre, i.e. when coexisting species varied in conducting structures, for example internal transport through a central cylinder (P. commune) and external along the surface of the leafy plants (P. schreberi).

The relationship between plant characteristics and desiccation time often acts at two levels: water storage capacity of cushions and boundary-layer properties of the canopy (Hedderson & Longton 1996), both components strongly affecting loss rates by evaporation. In a similar study by Elumeeva et al. (2011), there was little correlation between colony structure or shoot characteristics to water retention capacity in monocultures of the five moss species also discussed in our study (although Elumeeva et al. did find strong correspondence of colony structure and water retention in a broader species set). Instead, water storage capacity in monocultures was greatly related to shoot capacity and cushion dry mass, strongly influencing desiccation rates. In our study, non-additivity of different species assemblages could also to some degree be predicted from biomass measurements. Species contribution to cushion biomass and shoot density thus can regulate water dynamics in a cushion by modifying water availability gained from individual shoots. In general, plant tissue mass, rather than the structural complexity, plays a role in the ability of bryophyte species and communities to hold water externally (Elumeeva et al. 2011). Furthermore, growth forms in bryophytes often are critical to water storage capacity and water transport strategies (Glime 2007; Proctor 2009). For example, with the lowest water storage capacity (95–125% water content to dry mass) and fastest evaporation rate (T50 = 12 ± 1 h), Polytrichum species rely on internal transport for up to 67% of total water conduction (Glime 2007). Yet, boundary-layer fluctuations were likely to be reduced in greenhouse conditions and may play a bigger role under field conditions. The lack of relationships between structural complexity and loss rate by evaporation in our study also may result from measures failing to capture the intertwined branching patterns of mixed-species canopies.

The decreasing effect of canopy height on water retention could be an artefact of our experimental design with airflow being reduced and water-holding potential increased for species lying low inside the pot (<5 cm). Lophozia lycopodioides in monoculture formed the smallest cushions and lost external water at a rate close to the average. Only one monoculture, P. commune, was significantly taller than the pot and displayed the fastest rate of evaporation. However, airflow was likely to play a minor role in our greenhouse experiment, and variations in water loss can be better explained by species’ ability to converge with their neighbours. Changes in cushion height instead may lead to more compact cushions with species growing smaller and denser in an adaptive strategy to reduce water loss. Thus, interactions between canopy traits and water economy appeared to be driven by underlying factors including species composition and shoot characteristics.

Growth in a mixture thus contributed to an increased cushion water storage and retention capacity through individual species converging in shoot morphometry. In our study, most species maintained their rate of evaporation when growing in mixtures by forming more compact cushions (smaller and denser cushions). Bryophytes are known to be highly plastic (Vitt 1990; Glime 2007), changing physical characteristics with environmental gradients and performance of coexisting species. For example, size at the shoot and population levels is often determined by the total density (cover) of the mixed-species bryophyte carpet, suggesting that the interactions in multispecies mixtures are mostly between neighbouring shoots whether or not these belong to the same species (McAlister 1995; Økland & Økland 1996). Growth form of coexisting species further dictates the nature and outcome of these interactions, because growth form generally affects the vertical dynamics of individual shoots (Økland 2000). Although larger individuals may have advantages over smaller conspecifics by storing more water and having higher boundary-layer resistance (Zotz et al. 2000; see also Rixen & Mulder 2005), it appeared advantageous for some species to grow smaller and denser when in mixed-species cushions. By altering the surface exchange area, plants can modify boundary-layer properties and water flux to reduce water loss (Rice & Schneider 2004).

Ecological Significance of Non-Additive Effects on Water Retention

Bryophyte species often occupy distinct ecological niches with respect to water relationships and their growth form, and physiological tolerance to hydrological extremes (desiccation/submergence) dictates their distribution (Robinson, Vitt & Timoney 1989; Wasley et al. 2006). Additive and non-additive interactions with neighbouring species could thus influence species’ ability to persist in otherwise unsuitable habitat types (Kooijman & Kanne 1993; Rydin 1993, 1997a; Kooijman & Bakker 1995; McAlister 1995). Experimental data have challenged the notion of interspecific competition in bryophytes, a concept often derived from descriptive data (Rydin 1997b). After experimental vegetation removal, bryophyte re-establishment showed alleged stronger competitors to be unable to out-compete the other species and instead to coexist with them (Lloret 1994). Competition occurs mostly in the juvenile stage, and positive interactions are common in structuring bryophyte communities (Økland 1994). Our results also support previous findings showing no indication of either competition or facilitation between the dominant pleurocarpous species P. schreberi and H. splendens (Lloret 1994; Rydin 1997a). Furthermore, the non-additive effect of species mixture on water economy in natural community structure generally was positive, resulting in the improvement of water retention in bryophyte cushions. Community structure in bryophytes thus may fit the traditional notion of equilibrium coexistence (Rydin 1997a; Rydin 1997b), and non-additivity of species mixture may act as a facilitative mechanism in bryophyte communities by improving water retention (Rydin 2009; Bowker, Soliveres & Maestre 2010). This highlights once again the ecological consequences of the fact that the water economy of bryophytes is fundamentally different from that of vascular plants, which have evolved mechanisms to regulate water uptake, transport and retention internally and thereby tend to compete for water rather than share it to support community water economy.

Non-additive interactions between bryophyte species also can play a key role in water economy in ecosystems, more specifically in habitats where a dense cover of bryophytes contributes extensively to ecosystem water storage. For example, epiphytic bryophytes are known to prolong the time required for the forest canopy to saturate and dry, and to alter the transfer of water through the forest canopy (Pypker, Unsworth & Bond 2006). The species studied here frequently experience periods of desiccation in the field, e.g. in summer; during such periods, they can presumably not be photosynthetically active and will also lose their buffering function with respect to soil moisture (N.A. Soudzilovskaia, unpubl. data). Terrestrial as well as epiphytic bryophytes further contribute to water dynamics in ecosystems by intercepting rainfall, dew and fog (Beringer et al. 2001; Chang, Lai & Wu 2002; Hölscher et al. 2004). Pypker, Unsworth & Bond (2006) reported that bryophytes in a typical old-growth Douglas-fir forest cover 25–50% of the ground area, providing 485 kg ha−1 of biomass and absorbing 8–14 times their dry mass in water (an additional 0.60 mm of potential water storage). In arctic ecosystems, bryophyte cover influences soil thermal and hydrological regimes, consequently acting on a wide range of ecological processes including energy transfer, litter decomposition, nutrient cycling and vascular plant growth (Gornall et al. 2007).

Bryophyte richness in cushions can further enhance these processes, for instance through increased biomass production and moisture retention (Mulder, Uliassi & Doak 2001), although this effect varies greatly between ecosystems and abiotic stress conditions (e.g. duration of drought) (Rixen & Mulder 2005). A shift in species assemblages (e.g. changes in the density of a species within a community) of terricolous and epiphytic bryophytes thus can influence these processes through cushion composition controlling the local water balance through additive and non-additive interactions. Global environmental changes are more likely to impact on ecosystems at the level of species assemblage dynamics than at the individual species level (Brooker 2006; Suding et al. 2008). Non-additive effects of mixed-species bryophyte cushions on water economy thus potentially play a key role in controlling the hydrology and carbon and nutrient dynamics of an ecosystem. As such, the hydrological dynamics of bryophyte-dominated vegetation in the subarctic invites in-depth examination of the occurrence, mechanisms and ecological consequences of such bryophyte species interactions also in other biomes and ecosystems (Lindo & Gonzalez 2010). It may also pave the way for ecological investigations into non-additive species effects on hydrology in vascular plant systems, for instance through species litter mixture effects on water retention capacity as determined by their differences in morphology and surface properties (Proctor 2000).

Acknowledgements

We acknowledge the Abisko Research Station, Jurgen van Hal, Vickey Tomlinson, Gretchen Brownstein and Bastow J. Wilson for their technical assistance. Fernando T. Maestre, Matthew Bowker and two anonymous reviewers gave particularly constructive suggestions on the draft manuscript. This study was funded by a postdoctoral fellowship from the New Zealand Foundation for Research, Science and Technology and grant 047.018.003 of the Netherlands Organisation for Scientific Research (NWO) to JHCC.

Ancillary