BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, researchlibraries, and research funders in the common goal of maximizing access to critical research.

Good Neighbours? Determinants of Aggregation and Segregation among AlpineHerbivoresAuthor(s): Isabel C. Barrio & David S. HikSource: Ecoscience, 20(3):276-282. 2013.Published By: Centre d'études nordiques, Université LavalDOI: http://dx.doi.org/10.2980/20-3-3595URL: http://www.bioone.org/doi/full/10.2980/20-3-3595

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, andenvironmental sciences. BioOne provides a sustainable online platform for over 170 journals and books publishedby nonprofit societies, associations, museums, institutions, and presses.

Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance ofBioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.

Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiriesor rights and permissions requests should be directed to the individual publisher as copyright holder.

Understanding the structure of biotic communities and how different species coexist is one of the main challenges of community ecology (Agrawal et al., 2007). Mechanisms that allow for coexistence depend on the strength and direction of the interactions among the co-occurring

Manipulative experiments are required to infer mechan-

manipulation is not always feasible for some study organ-isms or broader scale studies, so alternative approaches are

needed. Studies of community structure and species inter-actions often involve identifying and quantifying patterns

Darmon et al. -bution of organisms is a prerequisite for the occurrence of interactions among them. In this context, null model analy-ses can provide a means to study mechanisms where experi-

Such analyses are becoming increasingly popular among population ecologists to investigate interactions among ani-mals (Richard et al.the need to jointly study both spatial structures and biotic interactions (Ritchie et al. et al.obtain a comprehensive understanding of the mechan-isms of coexistence. In this sense, determining the relative role of environmental (habitat) variables and interspecific

Good neighbours? Determinants of aggregation and segregation among alpine herbivores1

Isabel C. BARRIO2, and Instituto Pirenaico de Ecología (CSIC), Avda Nuestra Señora de la Victoria s/n, PO Box 64, Jaca, 22700 Spain, e-mail: icbarrio@gmail.com

David S. HIK,

Abstract: Interspecific interactions often determine the structure and stability of biotic communities. In low-productivity and highly seasonal environments such as the alpine tundra, most interactions occur during a short, snow-free period. The strength and direction of these interactions are likely to be determined by the availability of resources, particularly among species of the same ecological guild. Understanding how species interact in such environments can provide insights into the conditions that facilitate their coexistence. We determined the potential for interspecific interactions among 3 resident medium-sized mammalian herbivores inhabiting the alpine tundra and investigated how they share available space and resources. Overlap in their respective activity areas indicated that these species were aggregated at a landscape scale, but other mechanisms allowed their coexistence at a finer scale. Their distributions were primarily associated with shorter distances to heterospecifics and, secondly, with habitat features related to shelter and escape from predation. Our results suggest that these species can (and do) coexist by partitioning their ecological niches. Competition is likely not a major

herbivores in seasonal, low productivity environments.Keywords: alpine tundra, coexistence, facilitation, interspecific interactions, medium-sized mammals.

Résumé : Les interactions interspécifiques déterminent souvent la structure et la stabilité des communautés biotiques. Dans des environnements très saisonniers et peu productifs, comme la toundra alpine, la plupart des interactions se produisent durant la courte période sans neige. L'intensité et la direction de ces interactions sont probablement déterminées par la disponibilité des ressources, particulièrement entre les espèces d'une même guilde écologique. La compréhension de la façon dont les espèces interagissent dans de tels environnements peut fournir des indices sur les facteurs qui facilitent leur coexistence. Nous avons déterminé le potentiel d'interactions interspécifiques chez 3 mammifères herbivores de taille moyenne résidants de la toundra alpine et avons examiné comment ils se partagent l'espace disponible et les ressources. Le chevauchement de leurs aires respectives d'activité indiquait que ces espèces étaient regroupées à l'échelle du paysage, mais que d'autres mécanismes rendaient leur coexistence possible à plus fine échelle. Leurs distributions étaient associées en premier lieu à des distances plus courtes entre individus hétérospécifiques et, en second lieu, à des caractéristiques de l'habitat reliées à l'abri et à l’évitement de la prédation. Nos résultats suggèrent que ces espèces peuvent coexister (et le font) en se divisant les niches écologiques. La compétition n'est probablement pas un important facteur structurant ces

environnements saisonniers de faible productivité. Mots-clés : coexistence, facilitation, interactions interspécifiques, mammifères de taille moyenne, toundra alpine.

Nomenclature

Introduction

Associate Editor: Daniel Fortin.2Author for correspondence.

ÉCOSCIENCE, VOL. 20 (3), 2013

277

interactions in shaping species’ spatial distribution can pro-vide insights into such mechanisms (Azeria, Ibarzabal &

Interactions among co-occurring organisms may show considerable plasticity depending on the local biotic and

-tive interactions prevail over competition in determining the structure of plant communities in harsh environments (Callaway et al., 2002). However, for herbivores inhabiting less productive ecosystems, competition is thought to be the dominant form of interaction (e.g., Mishra et al.Cheng & Ritchie, 2006), although positive interactions may also play an important role in their coexistence (Gross,

-atures, and low net primary production, tundra ecosystems can support a high biomass of herbivores, among which mammalian herbivores are often abundant (Jefferies, Klein

critical period for herbivores to accrue energetic reserves -

petition for resources might be highly localized in time (Xi,

may be crucial in determining the structure and stability of these herbivore communities and how they respond to environmental changes.

We analyzed the potential for (positive or negative) biotic interactions among 3 alpine mammalian herbivores: hoary marmots (Marmota caligata), arctic ground squir-rels (Urocitellus parryii), and collared pikas (Ochotona collaris). Broadly, these species have similar diets, are ter-ritorial, and can be considered central place foragers, but some differences in their life strategies and behaviour allow them to cope with harsh environmental conditions. Marmots

while pikas remain active in winter below the snow cover. In contrast to marmots, pikas and (to a lesser extent) ground squirrels cache food for over-winter survival and emergence from hibernation, respectively. Following the approach pro-posed by Darmon et al. -bution and co-occurrence of the 3 species. Specifically, we used null models to determine if the overlap between their spatial distribution was random (neutral interaction), larger than expected by chance (positive association), or smaller than expected by chance (avoidance). Then, we analyzed the habitat selection of each species in the presence of the other 2, to determine the extent to which the aggregation or segregation of their distribution patterns could be attributed to environmental factors or to biotic interactions with sym-patric herbivores. Given the low productivity of alpine tun-dra ecosystems, we predicted all 3 species to be aggregated in space as a result of similar diet and habitat requirements. However, we also predicted segregation at a finer scale in order to reduce competition for essential resources, particu-larly access to foraging areas and refuges from predation or perceived predation risk.

MethodsSITE AND STUDY SPECIES

The study was conducted in the Ruby Range N W), southwest Yukon, Canada, in a

valley (about 7.2 km2

to 2000 m, where 3 main herbivore species coexist at rela-tively high densities. Landscapes are composed of a mosaic of alpine meadows and tundra vegetation (Hik, McColl

with boulder fields (i.e., talus patches), which represent

–2

David S. Hik, unpubl. data).Three species of medium-sized herbivores, ranging in

and are the dominant herbivores: hoary marmots, arctic ground squirrels, and collared pikas. All 3 species have been the subjects of extensive study at this site over the past

(caribou, Rangifer tarandus Ovis dalli) or more locally (voles, Clethrionomys spp. and Microtus spp.) or exploit other feeding resources (ptarmigan, Lagopus spp.). Predators are relatively low in abundance (Hik, McColl &

SURVEY

Signs of active presence of the 3 herbivores were

active. One observer (ICB) slowly walked the whole study area inspecting for presence signs everywhere. Signs of

were recorded and geo-referenced using a hand-held GPS receiver. Active burrows were considered to be signs of

the presence of active haypiles (over-winter food stores) were considered signs of presence of pikas. Active burrows of marmots and ground squirrels had fresh pellets or recent

pellets. Activity of haypiles was assessed by the presence of

August, as some pikas may only collect vegetation later in the season (Morrison et al.

Since all 3 species are central place foragers, “high use areas” were defined as those within a certain distance of each species’ respective signs of active presence, accord-ing to the literature available from the study area and

maps of the 3 species yielded a map of herbivore activity

(Table I). To correlate our map to the actual presence of animals we generated a set of random observation points within each intersection category sufficiently represented in

in each category, and the effort was increased in the mar-

Occurrence of the 3 species was assessed by checking points for signs of recent activity (see above) and con-

-ure cannot be considered a true validation of the map,

BARRIO & HIK: SPATIAL DISTRIBUTION OF ALPINE HERBIVORES

because different methods were used, it provides an estimate of the map’s reliability. Overall, the agreement between the activity at observation points and predictions

of direct observations compared to the cumulative activ-ity recorded for the map, we consider this agreement to be very good. To further evaluate potential direct behav-

-

where the mapped distributions of the 3 species overlapped. -

(e.g.marmots and pikas were together. We interpreted these cases as “potential direct interactions”.

STATISTICAL ANALYSES

To test the overlap between the areas of activity of the herbivore species we used a null model (Gotelli &

3 species by generating a number of random points equal to the number of locations recorded in the field (npikas = 206, nmarmots nsquirrels = 746). In the case of pikas, ran-dom points were restricted to talus patches, because the distribution of haypiles is a priori known to be limited to

-domly generated points for pikas lie within a biologically meaningful habitat type. Based on the randomly generated points and the buffer distances considered for each species (see above), we then calculated the randomized high-use areas for each species and their overlap and compared the observed area for each category with the distribution of

P-value for this test.To determine habitat selection of each species we com-

bined our map of species activity with maps of potentially relevant habitat and interspecific variables (distances to locations for the other 2 species). Habitat variables included topographic characteristics (elevation, slope, and aspect at

distances rather than the presence/absence of talus enabled us to account for potential edge effects (Conner, Smith &

FIGURE -

0 500250 1000 m

All 3 herbivores

Marmots and squirrels

Ground squirrel only

Marmot only

No herbivores

N

ÉCOSCIENCE, VOL. 20 (3), 2013

Burger, 2003). Aspect was linearized using a sine trans-

& Hik, 2004). No detailed vegetation map was available at this fine scale, but plant communities are closely deter-mined by local topography in this area (Danby et al.

to low-elevation, southwest-facing gentle slopes.We used variance partitioning to decompose the varia-

tion in the occurrence of each species among 2 blocks of predictors, habitat and interspecific variables, resulting in 3 fractions: the pure effects of habitat alone, those of interspecific variables alone, and the joint effects of both (Figure 2). We used Generalized Additive Models (GAM) with binomial errors and a logit link to calculate the deviance of the general and partial models (Legendre

locations where signs of activity were found and an equal number of random absence points. As the whole area was censused for presence signs, these points can be safely considered true absences. Within each block of predictors

we excluded variables that did not contribute significantly (Pstepwise procedure. We corrected for spatial autocorrelation in all models by including a non-linear spatial term (Bivand,

dimensional smoother of the spatial coordinates. Including the smoother removed spatial trends from the residuals of the model, as assessed by visually inspecting spatial cor-relograms, making the assumption of spatial independence

spatial correlograms.

ResultsThe presence of alpine herbivores was aggregated at a

broader scale, occurring over smaller areas than expected at random (P

the area occupied by herbivores, areas where more than

suggesting some avoidance mechanisms, such as direct avoidance of heterospecifics or segregation among species at a finer scale. Conversely, areas occupied by each species alone were larger than expected at random except for pika-only areas, which did not show a clear trend (PThe main part of the study area was shared by marmots

mainly where the 3 species overlapped (Table I).This spatial pattern was further supported by the habitat

selection of each species. Interspecific variables accounted for the highest proportion of the variance in the occur-rence of the 3 herbivores (Figure 2), while habitat variables

of pure interspecific variables was similar across species

TABLE I. Intersection categories of activity areas of the 3 alpine herbivores. Null model comparison indicates whether the ob-

P-

our map of herbivore activity to the actual presence of herbivores. Absence of values are noted with (.)

Category Area (km2) surface comparison points

arctic ground squirrel

FIGURE 2. Variance partitioning of the factors determining the distribution of 3 mammalian herbivores in the alpine tundra of southwest Yukon. The bars show the contribution of each block to the variance explained for each herbivore, when spatial structures were taken into account.

Hoary marmot

Arctic ground squirrel

Collared pika

Proportion of explained variance

Unexplained variance Habitat Interspecific Shared

Variance partitioning

0 0.20 0.40 0.60 0.80 1.00

BARRIO & HIK: SPATIAL DISTRIBUTION OF ALPINE HERBIVORES

The occurrence of all species was related to shortest dis-tances to heterospecifics and talus patches (Table II). In the case of arctic ground squirrels, within the interspecific

-tion of arctic ground squirrels was also positively linked to southwest-facing slopes.

DiscussionThe 3 alpine mammalian herbivores were aggregated at

a broad scale but segregated at a finer scale, and their dis-tribution was mainly determined by interspecific variables. This scale dependence is not surprising, since processes determining co-occurrence depend on the spatial scale at which species associations are analyzed (Redfern, Ryan & Getz, 2006). Co-occurrence at a coarse habitat scale may result from abiotic constraints on herbivore distribu-tion (Redfern, Ryan & Getz, 2006) and does not neces-sarily imply sympatry at the smaller patch scale, because patches can be exploited differently by each species (Martin

-tion of marmots, pikas, and squirrels may suggest similar ecological requirements, and might be related to environ-mental variables not accounted for in the present study. The distribution of these species overlaps in different parts of their ranges, and similar associations have been described

American pikas (Ochotona princeps), hoary marmots, and Columbian ground squirrels (Urocitellus columbianus) in talus slopes of Glacier National Park (Montana, USA). He hypothesized that, despite their similarities, these species were able to coexist due to differences in their spatial and

although the foraging areas of marmots and squirrels over-lapped, marmots tended to forage closer to boulder fields

-iods (Morrison et al.timing of their energetic requirements suggests differences in the intensity and timing of use of certain areas. Being obligate hibernators, ground squirrels and hoary marmots need to maximize fat gain, and their activity peaks in

increase their foraging behaviour later in the season, as they can rely on dry vegetation to store in their haypiles (Morrison et al.herbivores at a local scale seems then to be a result of eco-logical differences among species, rather than territorial competition for space, as has been described for other taxa

According to niche theory, coexistence requires some form of differentiation or partitioning between species, which might be allowed by neutral relationships, facilita-tion processes, or segregation at finer scales of the species’ ecological niche (e.g.relationships may result from the exploitation of different resources, when species share non-limiting resources or when other mechanisms (e.g., disturbance) prevent com-

-ate the habitat for other co-occurring species (Stachowicz,

3 species has been suggested, it has not been clearly dem-onstrated and no (direct) behavioural interactions have been described (Morrison et al.contrary, it has been hypothesized that these species may

For example, pikas are known to respond to sympatric heterospecific calls from hoary marmots and arctic ground

which may help them to escape from shared predators. Ground squirrels seem to respond as well to such calls, but

Marmota flaviventris). Other mechanisms by which these species may facilitate each other are habitat modification, through sharing burrows (Karels, Koppel & Hik, 2004), and feeding facilitation (Arsenault & Owen-Smith, 2002). For example, pika herbivory dramatically alters plant community composition along the edges of

-ability to the other herbivore species, as has been shown for other grazers (Mysterud et al.strength and direction of the interactions among these spe-cies does not need to be symmetrical, especially among

For example, in our study pikas were positively associated to shorter distances to arctic ground squirrels, but ground squirrels were not affected by the presence of pikas. Further studies identifying the mechanisms by which these species interact will help understand how biotic interactions struc-ture alpine herbivore communities.

Habitat selection by the 3 herbivore species was mainly determined by interspecific variables and to a lesser extent by abiotic habitat variables. In all cases, spatial distribution

TABLE II. Variables explaining the distribution of hoary marmot (a), arctic ground squirrel (b), and collared pika (c) in southwest Yukon, as assessed with Generalized Additive Models. The non-linear terms (s[x,y]) account for the spatial trends in the data.

a) Hoary marmot Estimate (± SE) Z P-value

edf a 2 P-values(x,y b) Arctic ground squirrel Estimate (± SE) Z P-value

2 P-values(x,y c) Collared pika Estimate (± SE) Z P-value

2 P-values(x,y

a edf: estimated degrees of freedom.

ÉCOSCIENCE, VOL. 20 (3), 2013

was associated with closer distances to other herbivores, suggesting a positive effect of the presence of heterospecif-ics. For example, when sharing predators, less preferred prey might select habitats supporting high densities of com-petitors that are a preferred prey because of reduced preda-

mere presence of heterospecifics can also provide valuable information to animals, providing them with cues to habitat

presence of heterospecifics could reflect other habitat vari-ables not accounted for in the present study. In our study, habitat variables mostly reflected areas maximizing shelter from predators and weather extremes, i.e.in the case of ground squirrels, distribution was also related to more productive southwest-facing slopes. Predation risk is a main constraint for the foraging activities of these herbivores (Morrison et al., 2004) even if predator pressure might not be as high as in other habitats (Hik, McColl &

Describing the spatial associations and ecological requirements of coexisting species is a first step to under-standing the potential mechanisms involved (Darmon et al.,

studying interactions when experimental studies are not an option (Richard et al. -gests that these 3 alpine herbivores can (and do) coexist, by partitioning their ecological niches at a finer scale. Their aggregated pattern, together with other lines of evidence, may imply that competition is not playing a major role in

interactions may play an important role in allowing the coexistence of herbivores inhabiting stressful, less product-ive ecosystems (Barrio et al.

year-round) restrict their direct interactions to the summer, when resources are most abundant in the tundra and compe-tition might be less important.

Our work represents a snapshot of a dynamic picture, and changing environmental conditions can reverse the strength and direction of interspecific interactions (Dunson

-able to ongoing climate change (Post et al. -standing these interactions and how they may change will help anticipate the potential responses of alpine herbivore guilds and their cascading effects on ecosystem functioning (Jefferies et al.

AcknowledgementsSpecial thanks are due to F. C. Hik, A. Kolar, and R. Mitten

for their invaluable help in the field, to S. Williamson for assist-ance with GIS, C. Calenge for statistical advice, and C. G. Bueno for helping with the modeling. T. Bao, E. Cameron, S. Nyanumba, K. Peck, A. Shaw, and H. Wheeler provided useful comments on an earlier draft. Funding was provided by the Natural Sciences and Engineering Research Council (Canada). I. C. Barrio was sup-ported by a postdoctoral fellowship provided by the Consejería de Educación, Ciencia y Cultura (JCCM, Spain) and the European Social Fund. We thank Kluane First Nation for permission to con-duct this research on their traditional lands.

Literature cited

Agrawal, A. A., D. D. Ackerly, F. Adler, A. E. Arnold, C. Cáceres, D. F. Doak, E. Post, P. Hudson, J. Maron, K. A. Mooney, M. Power, D. Schemske, J. Stachowica, S. Strauss, M. G. Turner & E. Werner, 2007. Filling key gaps in popula-tion and community ecology. Frontiers in Ecology and the

Arsenault, R. & N. Owen-Smith, 2002. Facilitation versus compe-

characteristics and interspecific interactions on co-occurrence patterns of saproxylic beetles breeding in tree boles after forest

Toronto, Ontario, Canada: University of Toronto Press.-

Extending the stress-gradient hypothesis: Is competition among animals less common in harsh environments? Oikos,

occupy unique nutritional feeding niches. Proceedings of the

Spatial Data Analysis with R. Springer, New York, New York.

C R A N . R - p r o j e c t . o rg / p a c k a g e = n c f ( A c c e s s e d o n st

of Washington and Alaska. American Midland Naturalist,

-position and hibernation in free-living arctic ground squirrels.

Callaway, R. M., R. W. Brooker, P. Choler, Z. Kikvidze, C. J. Lortie, R. Michalet, L. Paolini, F. I. Pugnaire, B. Newingham, E. T. Aschehoug, C. Armas, D. Kikodze & B. J. Cook, 2002. Positive interactions among alpine plants

Cheng, E. & M. E. Ritchie, 2006. Impacts of simulated livestock grazing on Utah prairie dogs (Cynomys parvidens) in a low

Conner, L. M., M. D. Smith & L. W. Burger,. 2003. A comparison of distance-based and classification-based analyses of habitat

of plant community change in the alpine tundra of southwest

Darmon, G., C. Calenge, A. Loison, J.-M. Jullien, D. Maillard -

tion in coexisting species of mountain ungulates. Ecography,

BARRIO & HIK: SPATIAL DISTRIBUTION OF ALPINE HERBIVORES

-tors in community organization. American Naturalist,

-ability: Small-scale segregation underpins large-scale coexist-

Forsman, J. T., J.-T. Seppänen & M. Mönkkönen, 2002. Positive fitness consequences of interspecific interaction with a poten-tial competitor. Proceedings of the Royal Society of London B,

Franken, R. J., 2002. Demography and metapopulation dynamics of collared pikas (Ochotona collaris) in the Southwest Yukon. PhD thesis. University of Alberta, Edmonton, Alberta.

Franken, R. J. & D. S. Hik. 2004. Influence of habitat quality, patch size and connectivity on colonization and extinction dynamics of collared pikas Ochotona collaris. Journal of

Gillis, E. A. 2003. Breeding dispersal, male mating tactics, and population dynamics of arctic ground squirrels. PhD thesis. University of British Columbia, Vancouver, British Columbia.

Smithsonian Institution Press, Washington, DC.-

ground squirrels more stressed in the boreal forest than in alp-

dynamics of occupancy and survival of a population of collared pikas (Ochotona collaris) in the Ruby Range, Yukon Territory. MSc thesis. University of Alberta, Edmonton, Alberta.

herbivores and northern plant communities: Reciprocal influ-

Jefferies, R. L., J. Svoboda, G. Henry, M. Raillard & R. W. Ruess,

G. R. Shaver & J. Svoboda (eds). Arctic Ecosystems in a Changing Climate: An Ecophysiological Perspective. Academic Press, New York, New York.

Karels, T. J., L. Koppel & D. S. Hik, 2004. Fecal pellet counts as a technique for monitoring an alpine-dwelling social rodent, the hoary marmot (Marmota caligata). Arctic, Antarctic, and

Amsterdam.MacKenzie, D. I., L. L. Bailey & J. D. Nichols, 2004.

Investigating species co-occurrence patterns when species are

warblers: Ecological differences or interspecific territoriality?

current season grazing by collared pikas on above-ground biomass and species richness in subarctic alpine meadows.

Mishra, C., S. E. Wieren, I. M. A. Heitkönig & H. H. T. Prins, 2002. A theoretical analysis of competitive exclusion in a Trans-Himalayan large-herbivore assemblage. Animal

Morrison, S., L. Barton, P. Caputa & D. S. Hik, 2004. Forage selection by collared pikas, Ochotona collaris, under vary-ing degrees of predation risk. Canadian Journal of Zoology,

Influence of food hoarding behavior on the over-winter sur-vival of pikas in strongly seasonal environments. Oecologia,

Mysterud, A., D. O. Hessen, R. Mobæk, V. Martinsen, J. Mulder

grazing in an alpine ecosystem. Basic and Applied Ecology,

-ces to quantify the strength of species interactions. Oikos,

Post, E., M. C. Forchhammer, M. S. Bret-Harte, T. V. Callaghan, T. R. Christensen, B. Elberling, A. D. Fox, O. Gilg, D. S. Hik, T. T. Hoye, R. A. Ims, E. Jeppesen, D. R. Klein, J. Madsen, A. D. McGuire, S. Rysgaard, D. E. Schindler, I. Stirling, M. P. Tamstorf, N. J. C. Tyler, R. van der Wal, J. Welker,

dynamics across the Arctic associated with recent climate

Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna.

Redfern, J. V., S. J. Ryan & W. M. Getz, 2006. Defining herbivore assemblages in the Kruger National Park: A correlative coher-

Richard, E., C. Calenge, S. Saïd, J.-L. Hamann & J.-M. Gaillard, -

tions of large herbivores: A null model approach. Ecography,

Separating the influences of environment and species inter-actions on patterns of distribution and abundance: Competition between large herbivores. Journal of Animal Ecology,

ground squirrel responses to heterospecific alarm calls. Animal

-

-hood: Collared pika (Ochotona collaris) responses to playback

in a community of talus slope mammals. American Midland

selection under shared predation: Tradeoffs between risk and

and marginal likelihood estimation of semiparametric general-ized linear models. Journal of the Royal Statistical Society B, 73: 3–36.

-ally suppress caterpillar performance and enhance plant bio-

-tors of insect herbivores: An asymmetrical interaction mediated