Is facilitation in arid environments the result of direct or complex interactions?


For several decades, species competition has dominated in ecological theory. However, since the beginning of the 1990s, facilitation has been recognized as a major driving force of species coexistence in plant communities, particularly in harsh environments. A number of authors (e.g. Bertness & Callaway, 1994; Brooker et al., 2005) have proposed that facilitation should become more important when the environmental stress increases; evidence for these models has been provided, in particular from alpine and arctic tundras (e.g. Callaway et al., 2002). However, most recently, other studies conducted in arid environments have obtained contrasting results, which have caused significant controversy among plant ecologists (Lortie & Callaway, 2005; Maestre et al., 2005), and are in some ways reminiscent of the Grime–Tilman debate of the late 1980s and early 1990s (Thompson, 1987; Tilman, 1987). A paper by Cavieres et al. published in this issue of New Phytologist (pp. 59–69) is up to date on this debate because they studied changes in biotic interactions with water stress along an altitudinal gradient in the high Andes.

‘The method used to quantify biotic interactions may strongly affect the outcome of experiments in arid environments.’

Patterns of variation in positive associations and facilitation with elevation in alpine communities

Cavieres et al. used both observational and experimental approaches to test the hypothesis that facilitation should decrease with elevation in Mediterranean mountains because of the prevalence of water stress over temperature stress in dry mountain ranges, as compared with mesic mountain ranges. The authors based their study in the central Andes of Chile (33° S latitude) where the climate is of the Mediterranean type. They selected two positions along the elevation gradient, a dry site at 2800 m and a mesic site at 3200 m, and reported that the cover of the dominant cushion plant Laretia acaulis increased from 22 to 30% with elevation. The authors then went on to assess facilitation in terms of the individual factors that vary along the gradient.

Cavieres et al. first analysed variation, along the elevation gradient, in spatial associations with the cushion plant L. acaulis for a total of 46 species, by comparing vegetation composition within the cushions vs the open areas between cushions. They found that 40% of the whole community was significantly associated with L. acaulis at the low site vs only 7% of the species at the high site. Moreover, four species were restricted to the cushions at the low site, whereas no species were found solely within cushions at the high site. They also monitored seedling survival at both elevations for two species planted within the cushions and in the open areas. Survival was higher within the cushions than in the open for the two species, and these differences in survival were higher at low elevation than at high elevation. Abiotic measurements showed that soil water availability was higher during the growing season within cushions than in the open areas only at low elevation, whereas temperatures registered on the bare ground were higher than those of the cushions, in particular at low elevation. Cavieres et al. concluded that the facilitation observed with decreasing elevation was mainly explained by the provision of water by L. acaulis.

To our knowledge, this is the first study showing that alpine gradients may produce contrasted patterns of variation in positive associations and facilitation with elevation, depending on the climatic context. This result is also consistent with the global study on alpine and tundra communities of Callaway et al. (2002), who found no increase in facilitation with elevation for only one site among their 11 mountain ranges; this site was located in the Sierra Nevada of Spain and hence also has a Mediterranean climate, whereas all other sites were located in temperate, boreal or arctic climates. Choler et al. (2001) also found for the alpine communities of the French Alps that facilitation was more intense in dry exposed sites located on convex slopes than down slope in mesic sheltered sites. However, they concluded that these changes in facilitation with topography were not necessarily driven by higher provision of water by nurse plants in convex slopes vs concave slopes, because other factors may influence the performances of target species along the topographic gradient, and in particular the level of disturbance, which increases with convexity. Furthermore, using gradient analyses, they related the intensity of competitive and facilitative interactions measured in their experiment with the position of the site within the niche of the target species. They showed that when neighbors were removed from around target species at experimental sites that were higher in elevation than the distributional mean of the target species, biomass decreased, whereas this result was not observed when the experimental site was drier (up slope along the topographic gradient) than the distributional mean of the target species. In other words, they showed that facilitation allowed species to move up along the elevation gradient but not along the topographic gradient. This means that neighbors may alleviate a thermic stress but are not likely to alleviate a water stress.

Changes in species interactions along water gradients in arid environments

The way species interactions may change along water gradients is becoming a much-debated issue in plant ecology. Most experimental studies have shown that facilitation increases with aridity (e.g. Greenlee & Callaway, 1996; Pugnaire & Luque, 2001; Bertness & Ewanchuk, 2002; Maestre et al., 2003; Liancourt et al., 2005), but others have observed conversely that competition increases with aridity (Tielbörger & Kadmon, 2000; Bellot et al., 2004; Maestre & Cortina, 2004). Gomez-Aparicio et al. (2004) analysed the results of multiexperiments conducted in the Sierra Nevada of Spain on 18 000 seedlings of 11 woody species planted with or without 16 different nurse shrubs. They showed that facilitation increases with aridity, both spatially and temporally, as proposed by the Bertness & Callaway model. In contrast, Maestre et al. (2005) showed in another meta-analysis that facilitation does not increase with stress in arid environment. However, a detailed statistical reanalysis of this data by Lortie & Callaway (2005) has shown that more rigorous data selection criteria, in particular changing gradients lengths between studies, do not support the original conclusions. Although we agree with Lortie & Callaway (2005) that the study of Maestre et al. (2005) included some bias and that therefore the overall conclusions must be treated carefully, some results may shed light on the mechanisms of facilitation in arid environments. In particular, they found that in the case of growth performances, the outcome of experimental studies were opposite for observational and experimental studies. Facilitative interactions were generally observed when the authors (such as Cavieres et al.) only compared the performance of target species beneath nurse species vs in adjacent open areas (Pugnaire & Luque, 2001; Bertness & Ewanchuk, 2002; Maestre et al., 2003; Castro et al., 2004; Gomez-Aparicio et al., 2004). In contrast, most of the experiments showing competitive interactions between dominants and subordinated species used removal procedures (Burger & Louda, 1995; Davis et al., 1998; Pugnaire & Luque, 2001; Maestre et al., 2003), although competitive interactions were also observed in comparisons with open areas (Flores-Martinez et al., 1998; Tielbörger & Kadmon, 2000; Bellot et al., 2004; Maestre & Cortina, 2004).

Direct or complex facilitation in arid environments?

The method used to quantify biotic interactions may strongly affect the outcome of experiments in arid environments; this is because the high levels of natural or anthropogenically induced disturbances produce strong variation in soil depth related to the occurrence of dominants (Aguiar & Sala, 1999). This may explain the contrasting results observed in the recent literature (Fig. 1). In the arid ecosystems of south-east Spain, Maestre et al. (2003) have shown that the direct water provision due to shading by the nurse plant Stipa tenacissima was outweighed by its own water uptake, leading to direct competition with the subordinated shrub Pistacia lentiscus, when measured with removal procedures (right-hand side of Fig. 1). Furthermore, Pugnaire & Luque (2001) and Davis et al. (1998) have shown that this competition for water increases with aridity (see Fig. 1). Additionally, Maestre et al. (2003) found that water availability and the performance of Pistacia lentiscus were the lowest in open areas between the tussocks. They concluded that the net effect of Stipa tenacissima was positive and increased with aridity (left-hand side of Fig. 1), but that this effect was complex and driven by differences in physical soil properties and water infiltration rate. Pugnaire & Luque (2001) also found that facilitation increased with aridity in a similar comparison of species performances in the open vs beneath the shrub canopies. Following the results of Pugnaire & Luque (2001), the increase in facilitation, as measured by a comparison with the open areas, overrides the increase in direct competition, as measured with removal procedures (Fig. 1). The facilitation observed by Cavieres et al. in the dry central Andes may not be caused by a decrease in evapotranspiration. This complex interaction may involve differences in soil depth and properties between open areas and the cushions, which are likely to increase with aridity because of increasing disturbances. These effects may have produced a higher facilitation when increasing aridity. Other studies, including both removal procedures and comparisons with open areas, are needed to understand better the relative parts of direct and complex interactions in arid ecosystems. However, even if facilitation is not a direct interaction occurring at the patch scale in arid environments, as suggested by an increasing amount of experimental evidence, the complex positive effect demonstrated by Cavieres et al. and others at the community scale may be considered as an insurance for biodiversity conservation and ecosystem functioning in arid landscapes.

Figure 1.

Predicted changes in direct (right-hand side) and complex (left-hand side) interactions between a dominant species (in gray) and a subordinated species (in black) under decreasing stress with elevation in an arid mountain range. Direct interactions are quantified by comparing the performances of the subordinated species with and without the dominant species, using removal procedures, whereas complex interactions are quantified by comparing the performances of the subordinated species within the canopy of the dominant species and in naturally occurring open areas.