Extreme stresses, niches, and positive species interactions along stress gradients

ReportsEcology, 95(6), 2014, pp. 1437–1443� 2014 by the Ecological Society of America

Extreme stresses, niches, and positive species interactionsalong stress gradients

QIANG HE1,3

AND MARK D. BERTNESS2

1School of Environment, State Key Laboratory of Water Environment Simulation, Beijing Normal University,Beijing 100875 China

2Department of Ecology and Evolutionary Biology, Brown University, Providence, Rhode Island 02912 USA

Abstract. Since proposed two decades ago, the stress-gradient hypothesis (SGH),suggesting that species interactions shift from competition to facilitation with stress, hasbeen widely examined. Despite broad support across species and ecosystems, ecologists debatewhether the SGH applies to extreme environments, arguing that species interactions switch tocompetition or collapse under extreme stress. We show that facilitation often expandsdistributions on species borders. SGH exceptions occur when weak stress gradients or stressesoutside of species’ niches are examined, multiple stresses co-occur canceling out their effects,temporally dependent effects are involved, or results are improperly analyzed. We suggest thatecologists resolve debates by standardizing key SGH terms, such as fundamental and realizedniche, stress gradients vs. environmental gradients, by quantitatively defining extreme stress,and by critically evaluating the functionality of stress gradients. We also suggest that newresearch examine the breadth and relevance of the SGH. More rigor needs to be applied toSGH tests to identify actual exceptions rather than those due to failures to meet its underlyingassumptions, so that the general principles of the SGH and its exceptions can be incorporatedinto ecological theory, conservation strategies, and environmental change predictions.

Key words: community ecology; environmental gradients; extreme environment; facilitation; generalrule; plant interactions; standardized approach.

INTRODUCTION

An important focus of current ecological studies is on

positive species interactions. After decades of research,

the importance of positive species interactions in

community organization has been generally recognized.

According to the stress-gradient hypothesis (SGH),

positive interactions increase with physical and biolog-

ical stress due to habitat amelioration and associational

defenses, respectively, and physical and biological

stresses are generally negatively correlated (Bertness

and Callaway 1994). The SGH was proposed as a

general didactic model at a time when positive interac-

tions were largely overlooked by ecologists and thought

to rarely be important in natural communities (Connell

and Slayter 1977, Connell 1983). In his 1983 review of

published competitive interaction studies, Connell

(1983) states that ‘‘a few positive interactions were

found, but may be indirect competition mediated by a

third species. . .some may actually be positive.’’ The

SGH pushed ecologists to recognize that positive

interactions are ubiquitous and critical in natural

communities (Callaway 2007, Brooker et al. 2008).

Over the last two decades, hundreds of studies from a

variety of species and ecosystems have tested the SGH.

Despite overwhelming evidence documenting the SGH,

there have been a few apparent exceptions to the SGH.

This has recently led ecologists to think whether

differences in ecological factors, such as stress types

and species traits, can generate variation in how species

interactions change along stress gradients (see He et al.

2013). Several modeling studies and reviews have

addressed these issues, proposing that facilitation should

collapse or shift to competition with extreme stress

(Maestre et al. 2009, Holmgren and Scheffer 2010,

Michalet et al. 2014). While it remains to test the

applicability and generality of these new models with

field experiments, a recent global meta-analysis exam-

ined how a number of ecological factors may affect the

SGH by synthesizing tests in various abiotic and biotic

Manuscript received 4 December 2013; revised 15 January2014; accepted 23 January 2014. Corresponding Editor: B. J.Grewell.

3 E-mail: [email protected]

1437

settings and showed that the SGH held when used as a

general rule of thumb to predict general trends in species

interactions with stress (He et al. 2013). However, it is

still argued that species interactions at the extreme end

of stress gradients (Michalet et al. 2014) or at the

community level were not analyzed. These arguments

are valuable to address the SGH. However, studies

shown to be flawed (Lortie and Callaway 2006, Call-

away 2007, He et al. 2013) were cited as evidence for

these arguments, key terms and concepts central to

current debates were misused or undefined, and a

constructive way forward was not given.

We suggest that future studies of the SGH explicitly

address issues central to current debates and further

model refinements. We first highlight with examples the

settings where facilitation in extreme environments is

unambiguous and discuss lessons from the exceptions

that have been cited as evidence for alternatives to the

SGH. We then address key terms vaguely used in the

literature and ignored in ongoing debates, propose what

constitutes a rigorous experimental test of the SGH, and

provide future research directions. We conclude by

suggesting that more rigor needs to be applied to tests of

the SGH in order to identify actual exceptions, rather

than those due to failures to meet its underlying

assumptions, so that the general principles of the SGH

and its exceptions can be incorporated into ecological

theory and conservation applications.

SGH ON THE EXTREMES

The best evidence for the SGH has come from

habitats with strong, potentially lethal a priori direc-

tional, physical, and biological stress gradients, e.g.,

intertidal habitats, where organisms are only able to live

at physical and biological extremes in groups, but not as

individuals (Bruno et al. 2003; Fig. 1). Classic compe-

tition-based niche theory predicts that the realized niche

is, by definition, always smaller than the fundamental

niche due to competitive displacement. But, since group

benefits often expand the range of habitats an organism

can live in, including positive interactions in niche

theory leads to the paradox of the realized niche being

FIG. 1. Photographs showing tested examples of positive interactions expanding the fundamental niche and range of plants,seaweeds, and sessile invertebrates. (a) High intertidal distribution of Ascophyllum nodosum and its associated fauna on coastalrocky shores in Maine, (b) high intertidal distribution of the mussel/barnacle zones on rocky shores in southern New England, (c)low intertidal distribution of cordgrass in New England salt marshes, and (d) the shrub Retama sphaerocarpa facilitates anunderstory plant community in Europe. See SGH on the Extremes for details.

QIANG HE AND MARK D. BERTNESS1438 Ecology, Vol. 95, No. 6R

epor

ts

larger than the fundamental niche (Bruno et al. 2003).

Group benefit expansion of the fundamental niche is

common at the physically or biologically stressful limits

of plant and sessile animal distributions (Hay 1986,

Bertness 1989, 1991, Leonard 2000, Stachowicz 2001),

contradicting the criticism that tests of the SGH have

avoided studying environmental extremes (Michalet et

al. 2014). The opposite is true, as documented in these

examples.

Example 1.—Facilitation expands the high intertidal

borders of algae (Fig. 1a). Fucoids as well as canopy-

forming red algae occur at middle to high intertidal

elevations. These algae commonly have high shoreline

borders that are sharply demarcated, with dense

canopies extending as a group to the highest elevation

tolerated by group benefits (Bertness et al. 1999).

Thinning these canopies to low individual densities

results in the death of solitary individuals (also see

Stachowicz 2001), except at higher latitudes and sites

where thermal and desiccation stresses are weak.

Example 2.—Facilitation expands the high intertidal

borders of invertebrates (Fig. 1b). Sessile invertebrates

like mussels live at higher intertidal elevations than they

do outside of groups (Bertness and Grosholz 1985,

Silliman et al. 2011). Experimental thinning of dense

intertidal invertebrate groups leads to the death of not

just habitat-ameliorating high-density invertebrates

(Bertness and Grosholz 1985), but also group-dependent

infaunal organisms (Silliman et al. 2011). Other sessile

invertebrate examples of group benefits extending the

high intertidal distribution are also common (Stacho-

wicz 2001).

Example 3.—Facilitation expands the low intertidal

borders of salt marsh plants (Fig. 1c). Low marsh

sediments in salt marshes are typically anoxic, a

potentially extreme stress lethal to vascular plants with

a terrestrial ancestry. Low marsh grasses occur at high

densities and have oxygen diffusing arrenchyma tissues

that passively diffuse oxygen into the substrate, poten-

tially alleviating anoxia stress (Hacker and Bertness

1995). This typically leads to sharp group-dependent

lower intertidal borders, expansion of the lower inter-

tidal border lower than small or individual plants occur,

and the expansion of the fundamental niche of marsh

plants intra- and interspecifically (Howes et al. 1986,

Bertness 1991).

Example 4.—Facilitation expands the arid borders of

plants in dry habitats (Fig. 1d). Across the entire

distributional range of Retama sphaerocarpa in Europe,

Armas et al. (2011) found that both the intensity and

frequency of facilitation by this nurse shrub increase

with aridity, being highest in the most arid habitats

where this species occurred. A significant number of the

beneficiary species were only able to survive beneath

Retama. Facilitation thus critically enlarges the niches of

many species in the driest zones. Similarly, facilitation of

pine seedlings by the shrub Artemisia tridentata

increased with aridity across their entire range over the

Great Basin and the Mojave Desert, USA (Ziffer-Berger

et al. 2014). Wilson and Agnew (1992) review many

more examples of terrestrial plant communities where

groups modify the physical environment, expanding

species distributions, calling them positive habitat

switches.

The common denominator of all of these studies

involving vascular plants, sessile invertebrates, and

intra- and interspecific interactions is that they are all

examples of plants and sessile animals living on extreme,

potentially lethal physical stress gradients. These stresses

are lethal to individual organisms, but are alleviated by

group benefits to expand their distributions and size of

their fundamental niche. They illustrate that on the

extreme edge of strong stress gradients, facilitation

functions as a crucial driver of niche expansion,

supporting the SGH (Fig. 2a). This wealth of general

examples contradict the assertion that the SGH has not

been tested at stressful range limits. Quite the contrary,

since in all these common cases, species ranges would

contract from physical stress mortality without neigh-

bors and group benefits. Studies on potentially lethal

extreme stress gradients led to the proposal of the SGH

(Bertness 1989, 1991). The above examples also empha-

size that the SGH was initially suggested by, and equally

applies to, sessile and slow-moving invertebrates, as well

as plants.

EXCEPTIONS AND LESSONS

There are actually only a few apparent exceptions to

the SGH but they have been repeatedly cited in

arguments for alternatives to the SGH (He et al.

2013). Three of the best cited are Tielborger and

Kadmon (2000), Maestre and Cortina (2004), and

Maestre et al. (2005). However, Tielborger and Kadmon

(2000) used permanent shrub plots to examine facilita-

tive effects on understory communities over time (4 yr),

so temporal effects were involved. For example, shrubs

may have grown larger and became competitive over the

time period, which meant that the observed shift from

facilitation to competition with stress may have resulted

from changes in life history rather in stress (Bertness and

Yeh 1994, Armas and Pugnaire 2005). Maestre and

Cortina (2004) did not work on a rainfall gradient that

limited the performance of their study species at all, so

were not working on a functional stress gradient (see

reanalysis in He et al. 2013). The third paper (Maestre et

al. 2005) was widely considered a problematic meta-

analysis. The authors did not even follow the criteria,

defined by themselves, in selecting literature, and stricter

reanalyses of their data showed increasing facilitation

with stress (detailed in Lortie and Callaway 2006,

Callaway 2007, He et al. 2013). These studies are

valuable to improve our understanding of plant

interactions with stress, but should not be considered

robust demonstrations of exceptions to the SGH.

Regrettably, these studies are repeatedly used to justify

arguments for alternatives to the SGH (Maestre et al.

June 2014 1439EXTREME STRESS, NICHES, AND FACILITATIONR

eports

2009, Holmgren and Scheffer 2010, Michalet et al.

2014).

Based on these lessons, we suggest that exceptions to

the SGH (Fig. 2) may occur when any of the following

five factors are present.

Studies involve temporal effects.—See Fig. 2b. For

example, juveniles often cooperate, while adults often

compete as a consequence of juvenile facilitation.

Exceptions to the SGH may occur when the life-history

stage of focal species shift across different stress

conditions. Also, juvenile interactions take precedence

over adult interactions since they establish distribution

patterns (e.g., Bertness and Yeh 1994, Holzapfel and

Mahall 1999).

Weak stress gradients that do not constitute a major

proportion of the species’ niche are examined.—See Fig.

2c. This may lead to random or normal distributions of

interactions with positive interactions randomly distrib-

uted or peaking at intermediate stress levels. Normal

distributions of positive interactions in this case can be a

statistical consequence of nonlethal multiple stresses

distributed independently in space (e.g., Holmgren and

Scheffer 2010).

Studies examine a nonlimiting stress to the study

species.—See Fig. 2c. Not all stresses are equally

important or are species limiting, an idea known as

Liebig’s law of the minimum. Stresses examined that are

not ameliorated by neighbors and/or are not a critical

FIG. 2. Five graphical models of predicted species interactions along stress gradients, including (a) change in speciesinteractions with stress (either physical or biotic) formulated in the original stress-gradient hypothesis (SGH), (b) switch fromfacilitation to competition with stress due to changes in life-history stage of the focal species, (c) random or normal distributions ofinteractions resulting from examining a weak or a nonlimiting stress gradient, (d) collapse of facilitation due to high stress outsideof the realized niche of the focal species, and (e) random or normal distributions of interactions resulting from spatial correlation/canceling-out effects of other stresses (either physical or biotic) along the examined stress gradient. Filled circles indicatehypothesized data points.


epor

ts

limiting factor in a community will not follow the

predictions of the SGH. For example, on rocky shores,

thermal and desiccation stress are dominant, communi-

ty-shaping stresses (Bertness 1989, Bertness et al. 1999),

while nutrient availability is not. Thus, studies examin-

ing interactions on thermal and desiccation stress

gradients will conform to the SGH, while studies

examining nutrient stresses may not. Similarly, while

water and nutrients are limiting factors in sand dune

communities, sand mobility and its concomitant distur-

bances are the overwhelming stresses that shape sand

dune communities (Forey et al. 2008). Sand dune studies

of disturbance and sand mobility are predicted to adhere

to the SGH (Franks and Peterson 2003), while

secondary stresses like water and nutrient limitation

should not (Donovan and Richards 2000). SGH studies

of nonlimiting stresses will lead to random or normal

distributions of positive interactions along stress gradi-

ents depending on the distribution of other nonlimiting

or limiting stresses in space.

Studies examine high stress habitats that are outside of

the realized niche of the species.—See Fig. 2d. This will

lead, by default, to positive interactions peaking at

intermediate stress levels (e.g., He et al. 2011). An

implicit assumption of the SGH is that it deals with

interaction strength within the natural range or niche of

a species (Bruno et al. 2003).

Multiple potentially lethal stress gradients are not

spatially correlated, so cancel out effects.—See Fig. 2e;

see also Michalet (2006), Mod et al. (2014). This can

lead to random or normally distributed interactions

along stress gradients, depending on the spatial corre-

lation of stresses.

ABANDON THE SGH OR MORE RIGOROUS TESTS?

Disparities in the SGH literature force rethinking what

a robust test of the SGH should and should not do. We

suggest these guidelines for rigorously testing the SGH.

Define or clarify terms vaguely used in the literature

Debates on the SGH are often related to several key

terms that have not been well defined. A first step to

resolve debates and move forward should be to

unambiguously define the terms on which the SGH

was formulated.

Fundamental and realized niche.—Fundamental niche

is the set of abiotic conditions that enable a species to

maintain self-sustaining populations, without negative

or positive species interactions, while the realized niche

is the set of abiotic and biotic conditions in which the

species maintains self-sustaining populations (Sax et al.

2013). As described in SGH on the Extremes, facilitation

expands the fundamental niche of species leading to a

realized niche larger than their fundamental niche

(Bruno et al. 2003). The SGH should be tested within

a species’ niche.

Stress gradients vs. environmental gradients.—These

terms need to be differentiated and used rigorously.

Stress is any environmental factor or a sum of

environmental factors limiting species fitness such as

survival, growth, or reproduction (Grime 1979, Menge

and Sutherland 1987; detailed in Lortie 2010). Environ-

mental factors that do not limit species fitness should

not be assumed as stress. Similarly, plant or animal

success without neighboring interactions (or productiv-

ity at the community level) is always directly related to

the stress gradient, declining with increasing stress, while

it may have different ways of varying along an

environmental gradient, either linearly decreasing,

reaching a peak, or reaching an asymptote along the

gradient. For example, environmental gradients of

rainfall, temperature, salinity, and nutrients are not

necessarily stress gradients.

Extreme stress.—The term extreme stress (and low,

medium, and high stresses) should be quantitatively

defined. For the study species with phytometers, we

suggest that extreme stress occurs where .80% of

species performance (e.g., survival, growth, productivi-

ty) without intra- or interspecific interactions is limited

within a specific time span (e.g., a growing season or life

stage) in the study system. Low, medium, and high

stresses occur where ,20%, 30–50%, and 50–80% of

species performance is limited, respectively. The severity

of the examined stress should be always quantified, as

thoroughly as possible (Lortie 2010).

Conduct experimental tests of the SGH

using the following recommendations

1) Quantify the fundamental niche (without neighbors)

and realized niche (with neighbors) of a species along

a study stress gradient, and operationally evaluate

the functionality of the stress gradient.

2) Have multiple stress levels. We recommend four:

low, medium, high, and extreme levels of stress.

Where studies with four levels of stress are imprac-

tical, the severity of the examined stress should be

evaluated and defined as low, medium, high, or

extreme.

3) Hold the ontogenetic stage and identity of neighbors

and targets constant across stress levels.

These recommendations are best applicable to testing

how pairwise species interactions change with stress.

Nevertheless, community-level tests of the SGH should

also follow these points. In particular, community-level

studies should differentiate stress gradients vs. environ-

mental gradients and quantitatively define low, medium,

high, and extreme stresses.

Develop new research to examine the breadth

and relevance of the SGH

Examine how the frequency, intensity, and importance

of competition and facilitation may change independently

along stress gradients.—Despite previous suggestions

(Brooker et al. 2008), the importance of facilitation

along stress gradients has been rarely tested (but see le


eports

Roux and McGeoch 2010). The original SGH suggested

that the frequency of positive species interactions

increases with stress, and by far, most studies examine

the intensity of interactions. The concept of importance

has been proposed for a decade, a term used to

distinguish the most crucial drivers of community

organization from others (Brooker et al. 2008).

Test the separate and interactive roles of stresses in

strong multiple stress habitats.—Past studies of the SGH

often test how species interactions change along

gradients of either a physical or biotic stress, but it is

common in nature that multiple stresses co-occur across

a single habitat. Multiple co-occurring stresses (such as

co-occurring herbivory and aridity stresses) have been

suggested to confound the SGH (Smit et al. 2009), but

empirical studies are rare.

Utilize species with contrasting functional traits to

examine the SGH across species.—The study of func-

tional traits is an emerging research area. Functional

traits (e.g., stress tolerance) have been suggested to

determine the nature of species interactions (e.g., He et

al. 2012).

Test the SGH in less studied species and ecosystems,

such as in higher animals, microbes, and tropical rain-

forests.—Although the SGH was proposed as a general

ecological model, it has been widely examined only in

plant communities. Quite few animal (except sessile

invertebrates in coastal marine habitats) and microbial

studies on the SGH have been conducted (but see Fugere

et al. 2012, Dangles et al. 2013). Also, most tests of the

SGH have been conducted in herb-, shrub-, or small tree-

dominated ecosystems (He et al. 2013). The SGH needs

to be tested in forests, particularly tropical forests.

CONCLUSIONS

The SGH has guided the study of positive plant, sessile

animal, and algal interactions and distributions for two

decades and holds as a general rule of thumb to predict

general trends in species interactions with increasing

stress. In contrast to assertions that the SGH has not

been tested in extreme habitats, there is a vast body of

literature showing the critical role of facilitation in

expanding the niche of both plants and animals at their

habitat edge. Exceptions to the SGH often occur when

(1) weak stress gradients, nonlimiting stresses, or stresses

outside of species’ niche are examined, (2) multiple

stresses co-occur and cancel out their effects, (3)

temporally dependent effects are involved, or (4)

improper statistical analysis is encountered.

Moving forward, ecologists should use identical

definitions and protocols to more rigorously test the

SGH and to develop new research to examine the

breadth and relevance of the SGH. It is still too early to

say no to the SGH. Rather, more rigor needs to be

applied to tests of the SGH to identify actual exceptions

to the SGH rather than those due to failures to meet its

underlying assumptions, so that the general principles of

the SGH and its exceptions can be incorporated into

ecological theory, conservation strategies, and environ-

mental change predictions.

ACKNOWLEDGMENTS

Q. He was supported by National Key Basic ResearchProgram of China (2013CB430406). M. D. Bertness wassupported by a Fulbright Fellowship and the National ScienceFoundation.

LITERATURE CITED

Armas, C., and F. I. Pugnaire. 2005. Plant interactions governpopulation dynamics in a semiarid plant community. Journalof Ecology 93:978–989.

Armas, C., S. Rodrıguez-Echeverrıa, and F. I. Pugnaire. 2011.A field test of the stress-gradient hypothesis along an ariditygradient. Journal of Vegetation Science 22:818–827.

Bertness, M. D. 1989. Intraspecific competition and facilitationin a northern acorn barnacle population. Ecology 70:257–268.

Bertness, M. D. 1991. Interspecific interactions among highmarsh perennials in a New England salt marsh. Ecology 72:125–137.

Bertness, M. D., and R. Callaway. 1994. Positive interactionsin communities. Trends in Ecology and Evolution 9:191–193.

Bertness, M. D., and T. Grosholz. 1985. Population dynamicsof the ribbed mussel, Geukensia demissa: the costs andbenefits of a clumped distribution. Oecologia 67:192–204.

Bertness, M. D., G. Leonard, J. M. Levine, P. Schmidt, andA. O. Ingraham. 1999. Habitat modification by algalcanopies: testing the relative contribution of positive andnegative interactions in rocky intertidal communities. Ecol-ogy 80:2711–2726.

Bertness, M. D., and S. M. Yeh. 1994. Cooperative andcompetitive interactions in the recruitment of marsh elders.Ecology 75:2416–2429.

Brooker, R. W., et al. 2008. Facilitation in plant communities:the past, the present, and the future. Journal of Ecology 96:18–34.

Bruno, J. F., J. J. Stachowicz, and M. D. Bertness. 2003.Inclusion of facilitation into ecological theory. Trends inEcology and Evolution 18:119–125.

Callaway, R. M. 2007. Positive interactions and interdepen-dence in plant communities. Springer, Dordrecht, TheNetherlands.

Connell, J. H. 1983. On the prevalence and relative importanceof interspecific competition: evidence from field experiments.American Naturalist 122:661–696.

Connell, J. H., and R. O. Slatyer. 1977. Mechanisms ofsuccession in natural communities and their role in commu-nity stability and organization. American Naturalist 111:1119–1144.

Dangles, O., M. Herrera, and F. Anthelme. 2013. Experimentalsupport of the stress-gradient hypothesis in herbivore–herbivore interactions. New Phytologist 197:405–408.

Donovan, L. A., and J. H. Richards. 2000. Juvenile shrubsshow differences in stress tolerance, but no competition orfacilitation, along a stress gradient. Journal of Ecology 88:1–16.

Forey, E., B. Chapelet, Y. Vitasse, M. Tilquin, B. Touzard, andR. Michalet. 2008. The relative importance of disturbanceand environmental stress at local and regional scales inFrench coastal sand dunes. Journal of Vegetation Science 19:493–502.

Franks, S. J., and C. J. Peterson. 2003. Burial disturbance leadsto facilitation among coastal dune plants. Plant Ecology 168:13–21.

Fugere, V., P. Andino, R. Espinosa, F. Anthelme, D. Jacobsen,and O. Dangles. 2012. Testing the stress-gradient hypothesiswith aquatic detritivorous invertebrates: insights for biodi-


epor

ts

versity-ecosystem functioning research. Journal of AnimalEcology 81:1259–1267.

Grime, J. P. 1979. Plant strategies and vegetation processes.Wiley, Chichester, UK.

Hacker, S. D., and M. D. Bertness. 1995. Morphological andphysiological consequences of a positive plant interaction.Ecology 76:2165–2175.

Hay, M. E. 1986. Associational plant defenses and themaintenance of species diversity: turning competitors intoaccomplices. American Naturalist 128:617–641.

He, Q., M. D. Bertness, and A. H. Altieri. 2013. Global shiftstowards positive species interactions with increasing envi-ronmental stress. Ecology Letters 16:695–706.

He, Q., B. Cui, and Y. An. 2011. The importance of facilitationin the zonation of shrubs along a coastal salinity gradient.Journal of Vegetation Science 22:828–836.

He, Q., B. Cui, M. D. Bertness, and Y. An. 2012. Testing theimportance of plant strategies on facilitation using congenersin a coastal community. Ecology 93:2023–2029.

Holmgren, M., and M. Scheffer. 2010. Strong facilitation inmild environments: the stress gradient hypothesis revisited.Journal of Ecology 98:1269–1275.

Holzapfel, C., and B. E. Mahall. 1999. Bidirectional facilitationand interference between shrubs and annuals in the MojaveDesert. Ecology 80:1747–1761.

Howes, B. L., J. W. H. Dacey, and D. D. Goehringer. 1986.Factors controlling the growth form of Spartina alterniflora:feedbacks between above-ground production, sedimentoxidation, nitrogen and salinity. Journal of Ecology 74:881–898.

le Roux, P. C., and M. A. McGeoch. 2010. Interaction intensityand importance along two stress gradients: adding shape tothe stress-gradient hypothesis. Oecologia 62:733–745.

Leonard, G. H. 2000. Latitudinal variation in species interac-tions: a test in the New England rocky intertidal zone.Ecology 81:1015–1030.

Lortie, C. J. 2010. Synthetic analysis of the stress-gradienthypothesis. Pages 125–148 in F. I. Pugnaire, editor. Positiveplant interactions and community dynamics. CRC, Boca,Raton, Florida, USA.

Lortie, C. J., and R. M. Callaway. 2006. Re-analysis of meta-analysis: support for the stress-gradient hypothesis. Journalof Ecology 94:7–16.

Maestre, F. T., R. M. Callaway, F. Valladares, and C. J. Lortie.2009. Refining the stress-gradient hypothesis for competition

and facilitation in plant communities. Journal of Ecology 97:199–205.

Maestre, F. T., and J. Cortina. 2004. Do positive interactionsincrease with abiotic stress? A test from a semi-arid steppe.Proceedings of the Royal Society B 271:S331–S333.

Maestre, F. T., F. Valladares, and J. F. Reynolds. 2005. Is thechange of plant–plant interactions with abiotic stresspredictable? A meta-analysis of field results in arid environ-ments. Journal of Ecology 93:748–757.

Menge, B. A., and J. P. Sutherland. 1987. Communityregulation: variation in disturbance, competition, and preda-tion in relation to gradients of environmental stress andrecruitment. American Naturalist 130:730–757.

Michalet, R. 2006. Is facilitation in arid environments the resultof direct or complex interactions? New Phytologist 169:1–2.

Michalet, R., Y. Le Bagousse-Pinguet, J. P. Maalouf, and C. J.Lortie. 2014. Two alternatives to the stress-gradient hypoth-esis at the edge of life: the collapse of facilitation and theswitch from facilitation to competition. Journal of Vegeta-tion Science 25:609–613.

Mod, H. K., P. C. le Roux, and M. Luoto. 2014. Outcomes ofbiotic interactions are dependent on multiple environmentalvariables. Journal of Vegetation Science. http://dx.doi.org/10.1111/jvs. 12148

Sax, D. F., R. Early, and J. Bellemare. 2013. Niche syndromes,species extinction risks, and management under climatechange. Trends Ecology and Evolution 28:517–523.

Silliman, B. R., M. D. Bertness, A. H. Altieri, J. N. Griffin,M. C. Bazterrica, F. J. Hidalgo, C. M. Crain, and M. V.Reyna. 2011. Whole-community facilitation regulates biodi-versity on Patagonian rocky shores. PLoS ONE 6:e24502.

Smit, C., M. Rietkerk, and M. J. Wassen. 2009. Inclusion ofbiotic stress (consumer pressure) alters predictions from thestress gradient hypothesis. Journal of Ecology 97:1215–1219.

Stachowicz, J. J. 2001. Mutualism, facilitation, and thestructure of ecological communities. Bioscience 51:235–246.

Tielborger, K., and R. Kadmon. 2000. Temporal environmentalvariation tips the balance between facilitation and interfer-ence in desert plants. Ecology 81:1544–1553.

Wilson, J. B., and A. D. Q. Agnew. 1992. Positive-feedbackswitches in plant-communities. Advances in EcologicalResearch 23:263–336.

Ziffer-Berger, J., P. J. Weisberg, M. E. Cablk, and Y. Osem.2014. Spatial patterns provide support for the stress-gradienthypothesis over a range-wide aridity gradient. Journal ofArid Environments 102:27–33.


eports

Documents

Extreme stresses, niches, and positive species interactions along stress gradients