Highlighting the multiple drivers of change in interactions along stress gradients
The prediction of Bertness & Callaway (1994) that facilitation should increase with increasing physical stress has been recently questioned in arid and semiarid environments, in particular since the publication of the meta-analysis of Maestre et al. (2005), which did not reveal any overall significant trend of variation in interactions in water-limited environments (but see Lortie & Callaway, 2006). Michalet (2006) recently proposed a methodological explanation for these inconsistencies in the literature and now a paper by Sthultz et al. in this issue of New Phytologist (pp. 135–145) adds further to this debate by testing the model of Bertness & Callaway (1994) along a water stress gradient and in accordance with Michalet's recommendations.
‘A number of authors have stressed that facilitation, such as competition, can involve both above- and below-ground effects’
Shift from competition to facilitation along an elevational gradient in a semiarid environment
Sthultz et al. analysed variation in biotic interactions between a foundation tree species, Pinus edulis, and a shrub, Fallugia paradoxa, in a semi-arid environment of southwestern USA. Sites were chosen at two levels of drought stress: four ‘high-stress sites’ at low elevation (approx. 1800 m), and four ‘low-stress sites’ at the ecotone with the P. ponderosa montane forests (approx. 2150 m). They used multiple approaches to test the prediction of Bertness & Callaway (1994) that competition should shift to facilitation with increasing water stress. The first approach was observational, sensu Maestre et al. (2005), in which P. edulis performance was compared below shrubs and in adjacent open areas. The second approach was experimental, whereby they compared P. edulis performance of individuals growing in controlled conditions (below the canopy of the shrub) with individuals growing in shrub patches where either all shrub shoots were severed at ground level or all above-ground parts were tied back. Furthermore, they also analysed the effect of P. edulis age and searched for temporal shifts in interactions during the course of the experiment, in relation to changes in climatic conditions.
Both their observational and experimental approaches supported the model of Bertness & Callaway (1994). The direction and intensity of target responses were very similar with both methods at high-stress sites; P. edulis was as strongly facilitated and there were no differences in target responses between the removal and tied-back treatments, indicating that facilitation was only due to above-ground effects. At low-stress sites, the direction of interactions were also similar with both methods, but total competition for growth was more important in the experiment than in the observational study and both above-ground competition and below-ground competition were observed. There was no significant effect of P. edulis age at the high-stress sites, whereas at the low-stress sites competition increased with seedling age. At the high-stress sites, facilitation occurred during the driest periods, in spring and summer, whereas at the low-stress sites, competition occurred during the mildest periods, in autumn and winter.
Highlighting the importance of above- and below-ground interactions, target age and temporal shifts
Sthultz et al. primarily found that competition shifted to facilitation along a gradient of increasing stress, which definitely supports the model of Bertness & Callaway (1994). This result may seem quite trivial when we know that in the last 10 yr this theoretical prediction has been recurrently supported. However, we want to highlight a number of other results that we consider may improve our knowledge of the mechanisms underlying biotic interactions within plant communities. First, this main result was found using both an observational and an experimental approach (sensu Maestre et al., 2005), and for both growth and survival performances, which renders it less trivial. Maestre et al. (2005) found in a meta-analysis conducted on arid and semiarid environments that growth responses (but not survival responses) were generally positive when authors compared the performance of target species beneath nurse species and in adjacent open areas (observational studies), whereas competition was often detected when authors used removal procedures (experimental studies). Michalet (2006) argued that the time and spatial scales of these interactions were quite different and that the facilitation seen in observational studies was in general a complex process mainly due to differences in soils occurring between open and vegetated areas. In other words, plants may compete at the patch scale, as demonstrated with removal procedures, whereas soil engineering by nurse species may produce positive effects at a much longer temporal scale; these latter effects may only be depicted at a larger spatial scale with observational studies. The similar results observed by Sthultz et al. with both methods demonstrated that F. paradoxa directly facilitated P. edulis at the high-stress sites. However, we suggest that there were likely to be soil differences between the open and shrub patches, at least in the low-stress sites, because competition was twice as high at these sites when measured with the removal procedures than when measured with the observational method.
A number of authors have stressed that facilitation, such as competition, can involve both above- and below-ground effects. However, contrary to competition studies, few experiments focusing on facilitation have separated both effects (but see Holzapfel & Mahall, 1999; Maestre et al., 2003). A second important result of the study of Sthultz et al. was to demonstrate that at the high-stress sites the beneficial effect of the shrub was only above ground, whereas at the low-stress sites both above- and below-ground parts of the shrub negatively affected P. edulis. This result observed with the experimental approach stressed the importance of shade as the main mechanism of direct facilitation in their system. Because there were no differences in the intensity of facilitation measured by the two approaches (observational and experimental), complex facilitation (sensu Michalet, 2006) involving soil differences between open areas and shrub patches were very unlikely to have affected P. edulis in this system, which is in contrast with other semiarid systems (Pugnaire et al., 2004). These differences may be related to the species involved in the interactions and is discussed further later on.
A third important result of the study by Sthultz et al. was to show that the direction of interactions may change during the life history of the target species, at least at their low-stress sites where they observed an increasing negative effect of the shrub as pine seedlings aged. Furthermore, they related the results of another experiment conducted in the same high-stress sites, where the authors found that different shrub species, including F. paradoxa, had strong negative effects on the growth of older P. edulis individuals than those of their own experiment. However, factors other than target age may explain these different results, and in particular differences in climatic conditions (Greenlee & Callaway, 1996), because the two experiments were not conducted during the same year. A number of authors have also found a switch from facilitation to competition increases with target age, particularly in experiments involving a tree species as the target (Kunstler et al., 2006), although contrasting results have been found by Lamaze et al. (2003).
Lastly, another interesting result of this study was to show that temporal changes in interactions during the course of the experiment were consistent with spatial changes along the elevation gradient. This result is in agreement with those of other studies observed either between years and experiments (Greenlee & Callaway, 1996; Gomez-Aparicio et al., 2004) or between seasons in a single experiment (Holzapfel & Mahall, 1999; Kikvidze et al., 2006). However, other authors found contrasting results and either a decreasing importance of facilitation (Kitzberger et al., 2000) or an increasing importance of competition (Tielbörger & Kadmon, 2000) during the driest climatic events of their studies.
Highlighting the importance of species
Sthultz et al. concluded that interactions between spatial and temporal scales that influence the net interactions between plants, such as those seen in this study, may help to explain why several recent studies do not support the prediction of Bertness & Callaway (1994). We suggest that, together with the influence of target age shown by Sthultz et al., factors other than temporal shifts should be included in experiments and theoretical models to explain some inconsistencies found in the literature for dry environments (Maestre et al., 2005). Among them, we would like to highlight the importance of the species involved in the experiments, in particular in dry environments. The study of Sthultz et al. concerned only two interacting species, while there are now several lines of evidence to indicate that changing the nature of either the nurse or the target species may produce, within the same spatial and temporal scales, substantially different results in either intensity or direction.
The effect of the functional type of nurses involved in the interactions is obvious in the literature. Gomez-Aparicio et al. (2004) showed in a meta-analysis conducted on 11 woody species planted within 16 different nurse species that the most important positive effects were found when seedlings were planted within legumes and the most important negative effects within rockroses (Cistus species, known for their allelopathic compounds), with intermediate positive effects for other shrubs. The somewhat exceptional positive effect of legumes is known from a number of other studies (Pugnaire et al., 2004), but the dominant negative effect of some functional groups is certainly not restricted to allelopathic plants, such as Cistus sp. There are increasing lines of experimental evidence (Kunstler et al., 2006; Picon-Cochard et al., 2006) showing that grass species are certainly very poor nurse species, which is probably a result of their strong ability to take up soil resources, as well as their weak light interception, as compared with shrubs (Köchy & Wilson, 2000). Maestre et al. (2003) have shown, in particular, the strong competitive effect of Stipa tenacissima, a widespread dominant species of the North African steppes also present in the same environments of southeast Spain where Pugnaire et al. (2004) and others described the strong facilitative effect of the legume Retama sphaerocarpa.
The effect of the target species seems more complex. In some studies, stress-tolerant light-demanding species have been shown to be less facilitated than stress-intolerant shade-demanding species (Liancourt et al., 2005; Pages & Michalet, 2006). For example, Gomez-Aparicio et al. (2004) showed in their meta-analysis that pines were the least facilitated among all of their ligneous target species. But it has been shown in a number of other studies that, when associated with grass nurses, stress-intolerant tree species are more negatively affected than stress-tolerant tree species (Kunstler et al., 2006); furthermore, the negative effect of the grass generally increases with increasing water stress (Davis et al., 1998; Maestre et al., 2003). This latter result is inconsistent with the model of Bertness & Callaway (1994) but supports the model of Taylor et al. (1990) of increasing competition with decreasing resource availability. These contrasting results may be explained by the balance between positive and negative effects, which is very likely to change depending on the constraint of the system and the functional strategies of both the nurse and target species. When the main constraint is above ground (high vapour pressure deficit, VPD), a stress-intolerant target species is the most likely to be facilitated, in particular with a nurse efficiently intercepting light, and this positive effect generally increases with increasing VPD. In contrast, when the main constraint is below ground (e.g. low water availability), a water-demanding species is the most likely to be negatively affected by neighbours, in particular when neighbours are grasses, and this negative effect generally increases with increasing drought.
The switch from competition to facilitation with increasing water stress found by Sthultz et al. in northern Arizona may have been driven, in part, by the functional characteristics of the interacting species and the limiting factors involved in these interactions (VPD vs soil water stress). For example, Callaway et al. (1996) also found, in a very similar environment in western Nevada, a significant facilitation for another stress-tolerant and closely related pine species (Pinus monophylla) growing in association with Artemisia tridentata, but a significant competition from the sagebrush for a more water-demanding species, Pinus ponderosa. However, further experiments are needed to test the hypothesis that competition should increase for this latter species with increasing water stress, because this study was not designed along a stress gradient.