applying the limiting resource model to plant tolerance of apical meristem damage

13
vol. 172, no. 5 the american naturalist november 2008 Applying the Limiting Resource Model to Plant Tolerance of Apical Meristem Damage Michael J. Wise * and Warren G. Abrahamson Department of Biology, Bucknell University, Lewisburg, Pennsylvania 17837 Submitted January 23, 2008; Accepted June 5, 2008; Electronically published October 7, 2008 abstract: Apical meristem damage (AMD) is a common result of herbivory. AMD can have dramatically variable effects on plant ar- chitecture and fitness, ranging from a total loss of reproductive ca- pacity to overcompensation. We explored the influence of environ- mental stresses and meristem limitation on tolerance of AMD by applying the limiting resource model (LRM) of plant tolerance to 17 previously published studies and a new empirical study on Solidago altissima. In the S. altissima experiment, AMD released axillary mer- istems from apical dominance, and fertilizer addition enabled plants to take full advantage of the lateral branches. AMD caused a 58% reduction in seed production in nutrient-stressed plants but only a 6% reduction in seed production in fertilized plants. In 12 of the 18 studies reviewed, tolerance was greater in the high-resource (or low- competition) treatment; in two, tolerance was greater in the low- resource treatment; and in four, resource level did not affect tolerance of AMD. The results of 15 studies (83%) were consistent with LRM predictions. Overcompensation was observed in six studies, and it occurred only in the high-resource treatments in five of these studies, as would be expected from applying the LRM. Keywords: apical dominance, compensatory continuum hypothesis, growth rate model, overcompensation, Solidago altissima, tolerance of herbivory. The apical meristems of plants are susceptible to damage by fire, frost, and wind, and they are particularly vulnerable to damage by vertebrate and invertebrate herbivores. The meristems are sites of rapidly dividing cells, and they are likely to be more tender and nutritious than most other plant parts (Mattson 1980; Benner 1988). They are often * Corresponding author; e-mail: [email protected]. E-mail: [email protected]. Am. Nat. 2008. Vol. 172, pp. 635–647. 2008 by The University of Chicago. 0003-0147/2008/17205-50200$15.00. All rights reserved. DOI: 10.1086/591691 within easy reach of browsers and grazers, and they are particularly susceptible to insect borers and gall inducers. Because of the central importance of the apical meristem to plant growth, apical meristem–damaging herbivores can have profound effects on their host plants. Without the apical meristem, plant growth may be completely cur- tailed, such that even a tiny herbivore that damages a very small amount of tissue can have a disproportionately large impact on plant fitness. At the other extreme, plants may be able to compensate for apical meristem damage (AMD) through the activation of latent axillary meristems. Under certain circumstances, growth and reproduction on axil- lary branches may exceed what the plant would have ac- complished in the absence of AMD (Aarssen 1995). In fact, some of the best evidence of overcompensation for herbivory has come from plants exposed to AMD (Inouye 1982; Paige and Whitham 1987; Paige 1992; Irwin and Aarssen 1996; Lennartsson et al. 1998). Beyond the specific issue of overcompensation, there is still much general debate about why plants exhibit such a wide range in tolerance of damage (Hawkes and Sullivan 2001). The concept of tolerance as a defense trait separate from resistance (traits that affect the amount of damage, including antibiosis and antixenosis) has been around for decades (Painter 1958; Beck 1965). Researchers interested in the evolutionary ecology of tolerance have suggested a variety of ways to define and quantify tolerance (Simms 2000; Stowe et al. 2000; Fornoni et al. 2003). In this article, we use a broad definition of tolerance as simply the effect of herbivore damage on a host plant’s performance. In this sense, tolerance is roughly equivalent to compensa- tion, with compensation indicating the degree of tolerance (i.e., undercompensation, exact compensation, or over- compensation; Strauss and Agrawal 1999). There is widespread agreement that the environmental conditions in which plants are growing, particularly levels of competition and resource availability, can affect plants’ ability to tolerate herbivore damage. However, the question of how differences in environmental conditions translate into differences in tolerance of herbivory is still being ac- tively explored (reviewed in Hawkes and Sullivan 2001;

Upload: bucknell

Post on 19-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

vol. 172, no. 5 the american naturalist november 2008

Applying the Limiting Resource Model to Plant

Tolerance of Apical Meristem Damage

Michael J. Wise* and Warren G. Abrahamson†

Department of Biology, Bucknell University, Lewisburg,Pennsylvania 17837

Submitted January 23, 2008; Accepted June 5, 2008;Electronically published October 7, 2008

abstract: Apical meristem damage (AMD) is a common result ofherbivory. AMD can have dramatically variable effects on plant ar-chitecture and fitness, ranging from a total loss of reproductive ca-pacity to overcompensation. We explored the influence of environ-mental stresses and meristem limitation on tolerance of AMD byapplying the limiting resource model (LRM) of plant tolerance to17 previously published studies and a new empirical study on Solidagoaltissima. In the S. altissima experiment, AMD released axillary mer-istems from apical dominance, and fertilizer addition enabled plantsto take full advantage of the lateral branches. AMD caused a 58%reduction in seed production in nutrient-stressed plants but only a6% reduction in seed production in fertilized plants. In 12 of the 18studies reviewed, tolerance was greater in the high-resource (or low-competition) treatment; in two, tolerance was greater in the low-resource treatment; and in four, resource level did not affect toleranceof AMD. The results of 15 studies (83%) were consistent with LRMpredictions. Overcompensation was observed in six studies, and itoccurred only in the high-resource treatments in five of these studies,as would be expected from applying the LRM.

Keywords: apical dominance, compensatory continuum hypothesis,growth rate model, overcompensation, Solidago altissima, toleranceof herbivory.

The apical meristems of plants are susceptible to damageby fire, frost, and wind, and they are particularly vulnerableto damage by vertebrate and invertebrate herbivores. Themeristems are sites of rapidly dividing cells, and they arelikely to be more tender and nutritious than most otherplant parts (Mattson 1980; Benner 1988). They are often

* Corresponding author; e-mail: [email protected].

† E-mail: [email protected].

Am. Nat. 2008. Vol. 172, pp. 635–647. � 2008 by The University of Chicago.0003-0147/2008/17205-50200$15.00. All rights reserved.DOI: 10.1086/591691

within easy reach of browsers and grazers, and they areparticularly susceptible to insect borers and gall inducers.Because of the central importance of the apical meristemto plant growth, apical meristem–damaging herbivores canhave profound effects on their host plants. Without theapical meristem, plant growth may be completely cur-tailed, such that even a tiny herbivore that damages a verysmall amount of tissue can have a disproportionately largeimpact on plant fitness. At the other extreme, plants maybe able to compensate for apical meristem damage (AMD)through the activation of latent axillary meristems. Undercertain circumstances, growth and reproduction on axil-lary branches may exceed what the plant would have ac-complished in the absence of AMD (Aarssen 1995). Infact, some of the best evidence of overcompensation forherbivory has come from plants exposed to AMD (Inouye1982; Paige and Whitham 1987; Paige 1992; Irwin andAarssen 1996; Lennartsson et al. 1998).

Beyond the specific issue of overcompensation, there isstill much general debate about why plants exhibit such awide range in tolerance of damage (Hawkes and Sullivan2001). The concept of tolerance as a defense trait separatefrom resistance (traits that affect the amount of damage,including antibiosis and antixenosis) has been around fordecades (Painter 1958; Beck 1965). Researchers interestedin the evolutionary ecology of tolerance have suggested avariety of ways to define and quantify tolerance (Simms2000; Stowe et al. 2000; Fornoni et al. 2003). In this article,we use a broad definition of tolerance as simply the effectof herbivore damage on a host plant’s performance. Inthis sense, tolerance is roughly equivalent to compensa-tion, with compensation indicating the degree of tolerance(i.e., undercompensation, exact compensation, or over-compensation; Strauss and Agrawal 1999).

There is widespread agreement that the environmentalconditions in which plants are growing, particularly levelsof competition and resource availability, can affect plants’ability to tolerate herbivore damage. However, the questionof how differences in environmental conditions translateinto differences in tolerance of herbivory is still being ac-tively explored (reviewed in Hawkes and Sullivan 2001;

636 The American Naturalist

Wise and Abrahamson 2005, 2007). The two most oftencited hypotheses regarding how tolerance of herbivory isaffected by environmental conditions make opposite pre-dictions. Simply stated, the compensatory continuum hy-pothesis (CCH) predicts greater tolerance of herbivoredamage in higher-resource, lower-competition environ-ments (Maschinski and Whitham 1989; Whitham et al.1991), while the growth rate model (GRM) predicts lowertolerance in these same conditions (Hilbert et al. 1981).The accumulated mass of empirical evidence over a rangeof plant species, types of herbivory, and resource variationsuggests that neither prediction is correct more than abouthalf of the time (Hawkes and Sullivan 2001; Wise andAbrahamson 2007).

In an attempt to improve tolerance predictions, we in-troduced the limiting resource model (LRM), which in-corporates a simultaneous consideration of the resourcethat varies the most between environments and the typeof herbivore damage, with a focus on which resource isactually limiting plant fitness (Wise and Abrahamson2005). In a subsequent review of 48 published studies oftolerance of leaf herbivory in different nutrient, light, wa-ter, or competition treatments, the LRM predictions wereconsistent with the results of 95% of the 41 studies thatprovided enough information to apply the LRM (Wise andAbrahamson 2007). In this article, we show how the LRMcan be applied to tolerance of AMD. The key is to thinkof meristems as a resource that a plant draws on, muchlike any other abiotic or biotic resource in its environment(Geber 1990; Aarssen 1995; Lehtila and Larsson 2005; Rau-tio et al. 2005). We show that thinking of meristems asresources whose shortage can potentially limit a plant’sfitness enables a substantial advance in the ability to ex-plain the effect of environmental stresses on tolerance ofAMD.

This article has three main parts. First, we provide astep-by-step description of how the LRM can be used ingeneral to make predictions for tolerance of AMD. Second,we provide results from an experiment on Solidago altis-sima growing in nutrient-stressed versus nutrient-richconditions. We use these empirical results to highlight thedetails and assumptions inherent in using the LRM forpredictions on tolerance of AMD. Third, we review pub-lished studies on tolerance of AMD in a variety of plantspecies in different nutrient, light, water, or competitionlevels to see what insight the LRM can bring to the in-terpretation of their results.

The LRM and AMD

The LRM has been discussed in detail in two previousarticles. The first used a flowchart diagram to show a seriesof seven pathways that could lead to three different out-

comes: relatively lower tolerance in resource-poor con-ditions, higher tolerance in resource-poor conditions, orequal tolerance regardless of resource level (Wise andAbrahamson 2005). In this article, we use the dichotomouskey version of the LRM (box 1) from Wise and Abraham-son (2007), which contains the same seven pathways butas a series of question couplets.

The first step in using the LRM is to identify the “focalresource,” which is simply the resource on which the studyfocuses. The focal resource is the resource that is eitheridentified to vary between environments or specificallymanipulated to differ between experimental treatments.For instance, in a study in which different nitrogen fer-tilizer treatments are applied, the focal resource would benitrogen. The first question in the LRM key is whetherthe focal resource is limiting plant fitness in the lower-resource environment (1 vs. 1′ in the LRM key, box 1).For instance, if the addition of nitrogen fertilizer increasesplant fitness, then we would conclude that nitrogen islimiting in the absence of fertilizer. In most publishedstudies, the focal resource will be limiting plant fitness;otherwise, the low-resource environment was not as stress-ful as the investigators probably intended.

If the focal resource was indeed limiting (1 in the LRMkey), the next question is which resource the herbivoredamage primarily affects. Does it mainly affect the ac-quisition of the focal resource, or does it mainly affect adifferent resource (2 vs. 2′ in the LRM key)? For simplicity,we refer to a resource that is not the focal resource as analternate resource. If the focal resource of an experimentis nitrogen, then an alternate resource could be light, water,potassium, and so on. Because damage to the apical mer-istem can have a wide variety of effects on a plant, it isdifficult to predict which abiotic resource will be the mostaffected by AMD in any given case. The most predictableeffect of AMD is that it will change the number of activatedmeristems. It removes the terminal meristem, and it oftenactivates axillary meristems. Following the lead of otherworkers (Geber 1990; Aarssen 1995; Lehtila and Larsson2005; Rautio et al. 2005), we find it useful in applying theLRM to consider the meristems of a plant as a resource.Just as a plant needs nitrogen, water, light, and perhapspollinators, it also needs activated meristems in order togrow and reproduce.

In order to use the LRM key effectively for AMD, thesimplest assumption is that the resource most affected byAMD is the number of active meristems. The focal re-source is an externally varying resource (e.g., nitrogen,light, water, or pollinators). Because the number of activemeristems is not one of these external resources, AMDprimarily affects an alternate resource. Therefore, the an-swer to couplet 2 in the LRM key is 2′, which leads tocouplet 3.

Tolerance of Apical Meristem Damage 637

Box 1: Dichotomous key for predictions of limiting resource model of plant tolerance

The focal resource is the resource that is identified as differing between environments. An alternate resource is the primary resourceaffected by herbivory if it is not the same resource as the focal resource.

1. The focal resource is limiting plant fitness in the low–focal resource (i.e., high-stress) environment.

2. Herbivore damage primarily affects use/acquisition of the focal resource. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Higher tolerance in the high-resource (low-stress) environment.

2′. Herbivore damage primarily affects use/acquisition of an alternate resource.

3. The alternate resource affected by herbivory is not limiting plant fitness in the high–focal resource environment. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Equal tolerance in both environments.

3′. The alternate resource is limiting plant fitness in the high–focal resource environment.

4. Herbivore damage exacerbates the limitation of the alternate resource. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lower tolerance in the high-resource (low-stress) environment.

4′. Herbivore damage ameliorates/removes the limitation of the alternate resource. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Higher tolerance in the high-resource (low-stress) environment.

1′. Focal resource is not limiting plant fitness in the low–focal resource (i.e., high-stress) environment.

5. Herbivore damage primarily affects use/acquisition of the focal resource.

6. Herbivore damage causes the focal resource to become limiting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Higher tolerance in the high-resource (low-stress) environment.

6′. Herbivore damage does not cause the focal resource to become limiting. . . . . . . . . . . . . . . Equal tolerance in both environments.

5′. Herbivore damage primarily affects use/acquisition of the alternate resource. . . . . . . . . . . . . . Equal tolerance in both environments.

Couplet 3 asks whether the alternate resource (the num-ber of meristems) is limiting in the high–focal resourceenvironment. There are at least two plant traits that wouldlead one to predict that the number of meristems wouldnot be limiting in the absence of AMD. The first is weakapical dominance. If apical dominance is weak, then anincrease in resources such as light or nutrients can breakthe dominance and allow activation of axillary meristems(Lortie and Aarssen 1997; Rautio et al. 2005). The secondtrait is indeterminate flowering on the main stem—thatis, there is flexibility in the number of flowers in the apicalinflorescence up to the amount enabled by the increase inresources. If a plant has either of these two traits, then thenumber of active meristems is not as likely to be limitingfitness (3 in the LRM key), and the LRM would predictequal tolerance in high– and low–focal resource con-ditions.

For a plant species for which axillary meristems are atmost only weakly activated by increases in the focal re-source and for which flowering on the main stem is notparticularly flexible, it is more likely that the number ofactivated meristems will be limiting plant fitness in thehigh-resource environment (3′ in the LRM key). Thefourth couplet asks whether damage to the apical meristem

exacerbates or removes this limitation. If damage to theapical meristem does not cause an activation of axillarymeristems, then AMD exacerbates the overall meristemlimitation (by removing the one active meristem), and theLRM predicts a greater relative impact of herbivory (i.e.,lower tolerance) in the high-resource environment (4 inthe LRM key). In contrast, if the damage does break apicaldominance and activate axillary meristems, then the LRMpredicts greater tolerance in the high-resource environ-ment, where the plants can take advantage of the increasedresources to reproduce on the activated meristems (4′ inthe LRM key). This is the scenario in which overcompen-sation for herbivory would not be surprising (Aarssen1995; Irwin and Aarssen 1996).

Finally, consider the case where the focal resource is notlimiting plant fitness in the low–focal resource environ-ment. For instance, if plant fitness were equal in bothtreatments of an experiment with a low- and a high-nitrogen treatment, then we would conclude that nitrogen(the focal resource) was not the limiting resource. Then,the answer to the first couplet in the LRM key is 1′, whichleads to couplet 5. The fifth couplet asks whether the her-bivore affects primarily the focal resource or an alternateresource. We assume that AMD affects primarily the num-

638 The American Naturalist

ber of activated meristems, which we consider to be analternate resource in this context. Thus, the answer to thecouplet is 5′, and the LRM predicts equal tolerance of AMDin high– and low–focal resource environments.

Methods

Experiment on Tolerance of AMD in Solidago altissima

Solidago altissima L. (Asteraceae), or tall goldenrod, is arhizomatous perennial weed common in old fields anddisturbed areas throughout the eastern United States andCanada, and it has been introduced and established inmany other parts of the world (Walck et al. 1999). Withoutdamage to the stem apex, S. altissima ramets (stems) gen-erally remain unbranched (Pilson 1992). In late summer,the apex differentiates into a determinate inflorescence (abranching panicle) with numerous tightly packed capitula(heads), each containing about 10–15 pistillate ray floretsand three to seven hermaphroditic disk florets (Potvin andWerner 1984; Abrahamson and Weis 1997). Each fertilizedfloret matures a single-seeded fruit, or achene.

Goldenrods are susceptible to a wide variety of insectherbivores (Root and Cappuccino 1992). Pilson (1992)noted seven species of insect herbivores that damage theapex of S. altissima, though she specifically identified onlytwo of these: the rosette gall–inducing midge Rhopalomyiasolidaginis (Loew) (Cecidomyiidae) and the spindle gall–inducing moth Gnorimoschema gallaesolidaginis (Riley)(Gelechiidae). To this list of identified apex-damaging her-bivores, we add four more, which may or may not overlapwith the unidentified herbivores of Pilson’s study: the ballgall–inducing fly Eurosta solidaginis (Fitch) (Tephritidae);the rosette gall–inducing fly Procecidochares atra (Loew)(Tephritidae); a new species of rosette gall–inducingmidge, Asphondylia n. sp. (Cecidomyiidae; N. Dorchin,personal communication); and the snap gall–inducingmidge Asphondylia solidaginis Beutenmuller (Cecidomyi-idae). By damaging the stem apices of goldenrod, theseherbivores can cause a release of apical dominance, thusresulting in the potential for branching from axillary mer-istems (Pilson 1992).

The plants for this experiment originated from rhizomesof 26 genets (genetic individuals) excavated in April of2003 from an old-field S. altissima population in UnionCounty, Pennsylvania (40�58�N, 76�57�W). Details onpropagation methods may be found elsewhere (Wise et al.2006, 2008). Briefly, individual ramets (stems) from eachgenet were grown from rhizomes in pots under controlledconditions every year since 2003. In spring of 2006, werandomly chose 15 genets for an AMD experiment. Rhi-zomes of these genets were cut into 2-mL segments (mea-sured by water displacement in a graduated cylinder) and

planted into flats in commercial growing medium (ProMixBX, Premier Horticulture, Dorval, Quebec) in a green-house at Bucknell University. Soon after stem emergence,four ramets from each genet were transplanted individuallyinto 16.5-cm-diameter plastic azalea pots, and the 60 potswere randomized in six rows on a greenhouse bench. Im-mediately after transplantation, each plant was fertilizedwith 59 mL Peters Professional water-soluble 15-16-17(NPK) fertilizer mixed with 3.9 mL/L water (J. R. Peters,Allentown, PA).

On June 15, the ramets were assigned a factorial com-bination of fertilizer and apex damage treatments. Tworamets of each genet were clipped with scissors below theapical meristem, and two were left undamaged. Oneclipped and one undamaged ramet of each genet werefertilized weekly through September at rates describedabove, and the other two ramets received no further fer-tilization. This experiment was conducted on one green-house bench, and nine other benches in the greenhousealso contained S. altissima ramets. (No other species ofplants were in the greenhouse.) In late summer, we placeda NATUPOL class C hive of bumblebees (Bombus impa-tiens) in the greenhouse to enable pollination (KoppertBiological Systems, Romulus, MI). The bees actively visitedall flowering ramets throughout the greenhouse andproved to be very effective pollinators.

We recorded the date at which the first flowers openedon each ramet. After growth finished and each ramet se-nesced, we measured the length of the main stem andcounted and measured the length of the axillary branchesthat were at least 2 cm long and recorded whether theypossessed seeds. We counted the number of capitula oneach ramet and collected 10 capitula from scattered lo-cations on each ramet into plastic vials. In the lab, thesecapitula were dissected and their seeds counted. We mul-tiplied the mean number of seeds per capitulum for aramet by the number of capitula the ramet possessed toestimate the total number of seeds produced by a ramet.Finally, we removed and carefully cleaned the rhizomesfrom each pot, and we counted the number of new rhi-zomes that emerged from the original rhizome segment.We then dried the rhizomes at 60�C to a constant massand weighed the total new rhizome growth for each ramet.

The statistical analyses for this experiment involved aseries of ANOVAs on eight growth and reproduction mea-sures. The basic ANOVA model included three main fac-tors (plant genet, fertilizer treatment, and clipping treat-ment) and all two-way interactions. Genet and theinteractions involving genet were considered random-effects factors. The fertilizer-by-clipping interaction is ofcentral importance because it indicates whether the plants’tolerance of clipping differed between nutrient-stressedand fertilized ramets. The response variables rhizome

Tolerance of Apical Meristem Damage 639

mass, number of capitula, and number of seeds per rametwere natural-log transformed to better meet the distri-butional assumptions of ANOVA. Subsequent to ANOVA,pairwise contrasts were run for total stem length and totalseed production responses using Tukey’s HSD procedure(Sokal and Rohlf 1995). All analyses were performed withJMP-IN 4.0.4 (SAS Institute, Cary, NC).

Review of Studies of Tolerance of AMD

Several criteria were used to decide whether to include astudy in the review. The AMD either had to be caused byan herbivore or else had to be imposed in an effort tosimulate herbivore damage. In addition, the study had tomeasure a performance metric reasonably related to fitnessin at least two environments that differed in an identifiedresource or in competition levels. We used the statisticaltests reported in the articles and the interpretation of re-sults by the original authors to compare tolerance levelsin the different environmental treatments. Again, we useda broad definition of tolerance as the relative impact ofherbivore damage on plant performance.

For each study, in addition to recording the focal re-source and the type of damage, we noted whether theresource level alone affected apical dominance (i.e., thenumber of meristems activated), whether damage affectedapical dominance (by increasing or decreasing the numberof activated meristems), and how the resources and dam-age affected fitness parameters. Given this information, weassigned a prediction outcome from the LRM key (box 1)for each study. Where the results of the study did notmatch the LRM prediction, we attempted to identify atwhich step the prediction failed. Such failures help to il-luminate the limitations of the LRM in predicting toler-ance of AMD. The failures should also highlight the typeof data that will be useful in future studies of AMD tol-erance, and they should serve to inform and refine futuremodels.

Results

Experiment on Tolerance of AMD in Solidago altissima

Fertilization had a large positive effect on nearly everyaspect of plant performance (table 1). For instance, whileramet height did not differ between fertilization treatmentsat the time of clipping, unclipped, fertilized ramets werenearly twice as tall as unclipped, unfertilized ramets (mean� SEM: vs. ) at the end of the experiment.162 � 6 88 � 6Fertilized ramets also initiated more new rhizomes( vs. ) and produced a greater total mass9.5 � 0.7 2.9 � 0.3of rhizomes ( g vs. g) than did6.55 � 0.45 1.25 � 0.15unfertilized ramets. All 30 fertilized ramets produced flow-

ers, while 11 of 30 unfertilized ramets did not, and fer-tilized ramets flowered an average of about 1.5 days earlier.Fertilized ramets produced 29 times more capitula( vs. ) and two more seeds per ca-4,648 � 269 160 � 32pitulum ( vs. ), resulting in 33 times17.5 � 0.6 15.2 � 0.5more total seeds ( vs. ) than79,844 � 4,946 2,401 � 495produced by unfertilized ramets. Importantly, fertilizer ad-dition by itself did not affect apical dominance. No un-clipped ramets, fertilized or not, produced any axillarybranches.

Clipping of the plant apex generally had a detrimentaleffect on plant performance (table 1). Clipping decreasedthe final main stem height by 60% in unfertilized rametsand by 79% in fertilized ramets. Eight of 30 clipped rametsdid not flower, compared with three of 30 unclipped ra-mets, and clipped ramets flowered an average of 5 dayslater (table 1). Clipping reduced total seed number( vs. for unclipped and42,948 � 8,011 39,297 � 7,982clipped ramets, respectively) by reducing the number ofcapitula per ramet but not the number of seeds per ca-pitulum (table 1). However, clipping did not have the samerelative effect on seed production in fertilized and unfer-tilized ramets (significant F # C interaction, table 1).When ramets were nutrient stressed, clipping reduced seedproduction by 58% ( vs. ), but3,383 � 747 1,420 � 566when the plants were fertilized, clipping reduced seed pro-duction by only 6% ( vs. ;82,514 � 6,454 77,174 � 7,660fig. 1). Therefore, nutrient-stressed ramets had lower tol-erance of AMD than did fertilized ramets.

Clipping released ramets from apical dominance, andevery clipped ramet expanded at least two axillarybranches. No unclipped ramet produced even a single ax-illary branch. Clipped, fertilized ramets had nearly twiceas many axillary branches as clipped, unfertilized ramets(table 2). Moreover, axillary branches were three timesmore likely to produce flowers on fertilized ramets thanthey were on unfertilized ramets. In terms of total stemlength (sum of main stem and axillary branches), unfer-tilized ramets were able to just compensate for clipping,while clipped, fertilized ramets produced nearly threetimes more total stem length than unclipped, fertilizedramets (fig. 2). Finally, clipped ramets initiated slightlymore rhizomes than unclipped ramets ( vs.6.8 � 0.9

) and produced a greater total mass of new rhi-5.6 � 0.7zomes ( g vs. g; table 1).4.26 � 0.60 3.54 � 0.58

Review of Studies on Tolerance of AMD

Including this study on S. altissima, we found 18 studiesfrom 13 separate articles that fit the criteria for inclusionin this review (table 3). Only one of these studies usedinsect herbivory (Fay et al. 1996), two studies included amammal grazing treatment (Maschinski and Whitham

Table 1: ANOVAs for ramet growth and reproduction responses to fertilizerand clipping

Source of variation df Mean square F P

Height at clipping:G 14 152.843 14.297 !.0001*F 1 .833 .078 .78Error 14 10.690

Main stem length:G 14 413.067 2.437 .41F 1 20,683.3 57.571 !.0001*C 1 122,402 666.693 !.0001*G # F 14 359.267 .962 .53G # C 14 183.595 .4918 .90F # C 1 21,056.3 56.400 !.0001*Error 14 373.34

No. rhizomes initiated:G 14 16.65 2.465 .12F 1 653.4 87.037 !.0001*C 1 21.6 5.508 .034*G # F 14 7.507 1.606 .19G # C 14 3.92 .839 .63F # C 1 .067 .014 .91Error 14 4.674

Rhizome mass (total):a

G 14 .894 2.114 .18F 1 48.700 86.988 !.0001*C 1 2.702 20.852 .0004*G # F 14 .560 2.101 .089G # C 14 .130 .486 .91F # C 1 1.437 5.392 .036*Error 14 .266

First flowering:b

G 14 103.869 4.537 .0002*F 1 183.469 8.014 .0081*C 1 450.704 19.686 .0001*F # C 1 29.344 1.282 .27Error 31 22.895

No. capitula:a

G 14 3.948 1.420 .39F 1 383.308 121.214 !.0001*C 1 18.587 7.310 .017*G # F 14 3.162 1.082 .44G # C 14 2.543 .870 .60F # C 1 16.381 5.602 .033*Error 14 2.924

No. seeds per capitulum:b

G 14 21.145 9.431 !.0001*F 1 71.232 31.771 !.0001*C 1 1.816 .810 .38F # C 1 .139 .062 .81Error 31 2.242

No. seeds per ramet:a

G 14 2.022 1.717 .31F 1 299.487 224.322 !.0001*C 1 9.058 8.061 .013*G # F 14 1.335 1.042 .47G # C 14 1.124 .877 .60F # C 1 7.003 5.466 .035*Error 14 1.281

Note: G p genet; F p fertilizer; C p clipping.a Analyses were performed on natural-log-transformed data.b Not all interactions could be calculated because 11 ramets produced no capitula.

* Significant at .P ! .05

Tolerance of Apical Meristem Damage 641

Figure 1: Effect of fertilizer and clipping on seed production (in naturallog units). The hatched bars represent the fertilized ramets, and the openbars represent the nutrient-stressed ramets. The error bars representmeans � 1 SEM. Columns with different letters are significantly different( ) using Tukey’s HSD procedure.P ! .05

1989), and the rest used manual clipping to simulate her-bivory. The focal resources included nine studies on in-organic nutrients (including nitrogen), one study on water,one study on water and nutrients combined, two studieson competition for light, and five studies on competitionin which the resource under competition was not specified.In 12 studies, tolerance of AMD was greater in the higher–focal resource treatment (table 3, pt. A); in two studies,tolerance was greater in the lower–focal resource treatment(table 3, pt. B); and in four studies, tolerance was equalregardless of resource level (table 3, pt. C).

In terms of the LRM key (box 1), outcome 4′, which isgreater tolerance in the high-resource (low-stress) envi-ronment, was predicted for 13 studies. In each of these 13studies, the focal resource was limiting in the nominallystressful environment, as increasing resource level led tohigher fitness (1 in the LRM key). With the assumptionthat the primary effect of AMD is on meristem availabilityrather than on any specific abiotic resource, we consideredthe damage to affect an alternate resource (2′ in the LRMkey). Without specific knowledge to the contrary, we alsoassumed that at higher–focal resource levels, the focal re-source is less likely to be the most limiting factor, and thusthe number of active meristems is more likely to be lim-iting in the high–focal resource environment (3′ in theLRM key). The final critical question is whether AMDexacerbates or ameliorates this meristem limitation (4 vs.4′ in the LRM key). If a study reported that AMD releasedthe plants from apical dominance and increased axillarybranching, then the predicted outcome was 4′.

In 11 of the 13 studies for which 4′ was the LRM pre-

diction, tolerance of AMD was indeed greater in thehigher-resource treatment (table 3, pt. A). In the two otherstudies, tolerance was equal regardless of focal resourcelevel (Fay and Throop 2005; Rautio et al. 2005; table 3,pt. C). In Rautio et al.’s (2005) study, fertilizer increasedthe growth and reproduction of tall wormseed mustard(Erysimum strictum). Clipping had a statistically similar(positive) effect on the plants regardless of fertilizer treat-ment (i.e., no interaction between fertilizer and clipping).The authors suggest that because adding fertilizer in-creased branching in the absence of clipping, the plantswere not meristem limited when fertilized, but they weremore likely to be meristem limited when not fertilized.

Rautio et al.’s (2005) study highlights the importanceof knowing which resource is limiting and knowing detailsabout the natural history in order to use the LRM, par-ticularly for AMD predictions. Without our knowing apriori whether increasing resource level would causebranching to such an extent that meristems would not belimiting in any given species, we assumed that meristemswould be limiting plants in the high-resource treatmentsfor all the studies in this review. This assumption led theLRM to make an incorrect prediction regarding the resultsof Rautio et al.’s (2005) study. However, the knowledgeof very weak apical dominance in E. strictum in responseto fertilization could be included in future predictions onthis species. In terms of the LRM, the alternate resource(meristems) would then not be expected to be limitingplant fitness in fertilized E. strictum plants (3 in the LRMkey), and the LRM would then predict equal tolerance offertilized and unfertilized plants, which is consistent withthe results of Rautio et al.’s (2005) study.

Fay and Throop’s (2005) finding that Silphium integri-folium had equal tolerance of AMD in high- and low-competition environments is not readily explained fromthe perspective of the LRM. This result of equal tolerancewas surprising to Fay and Throop, who cited an a prioriprediction of greater tolerance of AMD with less com-petition from neighbors (Aarssen and Irwin 1991). Fayand Throop speculated that differential responses in pro-duction and abscission of primary and lateral leaves in thedifferent treatment combinations may have balanced totalleaf number in a manner such that the plants in differentlight treatments had a similar ability to allocate resourcestoward capitula production. This plasticity in leaf abscis-sion in response to light level differences may have enabledthe plants to have equal tolerance of clipping regardlessof the light competition treatment.

There were three studies (from two articles) in the re-view for which the LRM predicted lower tolerance in thehigh-resource environment (Mutikainen and Walls 1995;Huhta et al. 2000; table 3). In each of these studies, AMDdecreased the amount of branching by the plants. In terms

642 The American Naturalist

Table 2: Axillary shoot production by clipped ramets

No. axillary shoots Percent with seeds Total length (cm)

Unfertilized 3.2 � .2 27 � 8 59 � 10Fertilized 6.1 � .4 82 � 13 424 � 24

Note: Values include branches at least 2 cm in length. Unclipped ramets elongated no

axillary shoots.

Figure 2: Effect of fertilizer and clipping on total stem length (mainstem plus all lateral shoots). The hatched bars represent the fertilizedramets, and the open bars represent the nutrient-stressed ramets. Theerror bars represent means � 1 SEM. Columns with different letters aresignificantly different ( ) using Tukey’s HSD procedure.P ! .05

of the LRM, if the number of axillary meristems was al-ready more likely to be limiting plant fitness in the high-resource environment (3′ in the LRM key), then AMD thatdecreased branching exacerbated this limitation (4 in theLRM key). In two of these studies, tolerance was indeedlower in the high-resource environment (Mutikainen andWalls 1995). In the third study, however, tolerance of AMDwas found to be greater in the high-nutrient treatment,even though clipping decreased branching (Huhta et al.2000).

Huhta et al. (2000) left it an unsolved dilemma as towhy E. strictum plants in a high-nutrient treatment weremore tolerant of AMD, even though AMD caused a re-duction in branching. A possible explanation for theseresults has to do with the strength of apical dominanceand the extent of the damage treatment. In this study, justas in Rautio et al.’s (2005) study on the same species,fertilizer was very effective at breaking apical dominanceeven without clipping, such that reproduction in undam-aged fertilized plants may not have been strongly limitedby the number of activated meristems (3 rather than 3′ inthe LRM key). Recall that in Rautio et al.’s study, tolerancewas equal in high- and low-nutrient conditions. Part ofthe difference in tolerance outcomes between the two stud-ies could be that while Rautio et al.’s damage treatmentinvolved clipping 25% of the main stem, Huhta et al.clipped up to 50% of the stem. It seems likely that themore severe damage might remove not only the apicalmeristem but also a substantial number of axillary mer-istems. When fertilization enabled the activation of mer-istems, the clipped plants simply had fewer latent meri-stems available to expand. Regardless of why clippingreduced the number of branches, the LRM rationale doesnot offer an explanation of why tolerance would have beengreater in the fertilized treatment in this particular study.

For two of the 18 studies in this review (Juenger andBergelson 1997; Rautio et al. 2005), the LRM predictedequal tolerance in the low- and high-resource conditions(table 3, pt. C). In Juenger and Bergelson’s study, nutrientlevel did not affect seed production by scarlet gilia (Ipo-mopsis aggregata). Similarly, Rautio et al. (2005) found thatthe level of competition did not affect seed production intall wormseed mustard (E. strictum). Therefore, in thesestudies, the focal resource was not limiting plant fitnessin the low–focal resource (nominally high-stress) envi-

ronment (1′ in the LRM key). Again, we are assuming thatthe main effect of AMD is on the number of apical mer-istems and, thus, that the damage primarily affects not thefocal resource but an alternate resource (5′ in the LRMkey). Thus, the LRM prediction for these two studies isequal tolerance in high- and low-resource environments,and this is consistent with the findings of these studies(table 3, pt. C).

Discussion

Experiment on Tolerance of AMD in Solidago altissima

In our experiment, fertilized ramets of Solidago altissimahad greater tolerance of AMD than did nutrient-stressedramets. Specifically, AMD reduced relative seed productionmore severely in stressed ramets than it did in fertilizedramets. We had predicted this pattern using the LRM (out-come 4′ in the LRM key) on the basis of our knowledgeof the goldenrod system. Here we show how the resultsof the experiment apply to the model, step-by-step, high-lighting the simplifications and assumptions in applyingthe LRM to studies of tolerance of AMD.

The first question to address in the LRM key is whether

Tolerance of Apical Meristem Damage 643

the focal resource (nutrients) was limiting ramet fitnessin the low-nutrient treatment (box 1). From previous ex-periments, we expected that S. altissima would be limitedby nutrients if not given fertilizer, and, clearly, the greatboost in plant performance when fertilized indicates thatnutrients were limiting (1 in the LRM key). An impliedassumption is that the fertilized ramets were no longernutrient limited. However, we do not know whether thiswas actually the case. Perhaps at twice the fertilizer ratethat we used, the ramets would have grown even largerand produced even more seeds. Nevertheless, the rationalefor the LRM predictions is that nutrients are less limitingin the fertilized treatment than in the unfertilized treat-ment, and thus a different resource is more likely to belimiting in the fertilized treatment.

The second question is whether AMD primarily affectsthe focal resource (nutrients in this experiment) or analternate one. Our assumption was that because AMD hassuch a dramatic effect on ramet growth and architecture,its primary effect is likely to be on the number of meri-stems available for growth and reproduction rather thanon any particular abiotic resource. Thus, the number ofactive meristems is the alternate resource assumed to bemost affected by AMD (2′ in the LRM key). Of course, adrastic change in ramet height and architecture is alsolikely to affect acquisition of resources such as light andnutrients from the environment. In any particular case,this effect might turn out to be of more consequence thanthe effect on the number of meristems. However, this sim-plification of considering the meristems a resource makesapplying the LRM much more practical.

The third question is whether the alternate resource (thenumber of activated meristems) is limiting ramet fitnessin the high-nutrient treatment (particularly withoutAMD). This step requires the most knowledge of the sys-tem and perhaps the most speculation. A plant with acombination of strong apical dominance and determinateflowering is most likely to be meristem limited in a high-resource environment. If apical dominance is relativelyweak, then axillary meristems may be activated by an in-crease in nutrients. If flowering is relatively indeterminate,then the terminal inflorescence may be able to expand tokeep up with an increase in nutrients. Solidago altissimais known to have strong apical dominance (in the absenceof AMD) and a determinate inflorescence (Potvin andWerner 1984; Pilson 1992). Therefore, we decided that thenumber of activated meristems (the alternate resource)was likely to be a limiting factor on plant fitness in thefertilized plants (3′ in the LRM key). Our experiment con-firmed the apical dominance assumption: unclipped fer-tilized ramets never expanded any axillary meristems.However, although the inflorescence may technically bedeterminate, the size of the terminal inflorescence, par-

ticularly the number of inflorescence branches, displayedplasticity in response to nutrient addition.

The fourth and final question is whether AMD exac-erbated or ameliorated meristem limitation. Because ofthe robust growth of axillary branches in response toAMD, particularly in the fertilized treatment, it is reason-able to conclude that AMD tended to remove, or at leastameliorate, the potential meristem limitation (4′ in theLRM key). Therefore, the LRM prediction for our exper-iment was higher tolerance in the fertilized ramets, whichis the outcome we observed. In all, AMD reduced fitnessby 58% in the unfertilized group but only by 6% in thefertilized group.

Review of Studies on Tolerance of AMD

Although our main goal in this article was to show howthe LRM can be applied to tolerance of AMD, it is worthcomparing the patterns found in the literature review withthe predictions of the two most cited models regardingthe effect of environmental conditions on tolerance: thecompensatory continuum hypothesis (CCH) and thegrowth rate model (GRM; Hawkes and Sullivan 2001; Wiseand Abrahamson 2007). Though these models are moremultifaceted than often portrayed, at the simplest level,the CCH predicts that plants will be relatively more tol-erant of herbivory when growing in resource-rich, low-competition, or otherwise benign environments (Ma-schinski and Whitham 1989; Whitham et al. 1991). Incontrast, the GRM predicts that plants will be more tol-erant of herbivory in resource-deprived, high-competition,or otherwise stressful environments (Hilbert et al. 1981;Alward and Joern 1993).

Of the 18 studies in this review, 12 (67%) were con-sistent with the CCH prediction, two (11%) were consis-tent with the GRM prediction, and four (22%) showedno significant effect of resource level on tolerance. In ourprevious review of 48 studies of tolerance of leaf herbivory,the relative successes of the CCH and the GRM were re-versed (31% and 48%, respectively). In contrast, the LRMsuccess was more consistent across damage types. The re-sults of 83% of the AMD studies and 95% of the leafherbivory studies were consistent with LRM predictions(Wise and Abrahamson 2007). From both reviews, it isclear that with the minor additional considerations in-volved in the LRM comes substantially better explanatorypower.

Several authors, following the lead of Hawkes and Sul-livan (2001), have suggested that dicots are more likely tofollow the pattern predicted by the GRM; that is, dicotsshould have lower tolerance in high-resource conditions(De Deyn et al. 2004; MacDonald and Bach 2005; Throop2005). The 18 studies in the current review on AMD in-

644

Tabl

e3:

Stu

dies

ofth

eef

fect

sof

reso

urc

ele

vels

onto

lera

nce

ofap

ical

mer

iste

mda

mag

e

Pla

nt

spec

ies

(fam

ily)

Foca

lre

sou

rce

Dam

age

Fitn

ess

met

ric

Eff

ect

ofh

ighe

r–fo

cal

reso

urc

ele

vel

onE

ffec

tof

dam

age

on

bran

chin

g

LRM

path

way

pred

icti

onR

efer

ence

Fitn

ess

Bra

nch

ing

A.

Tole

ran

cegr

eate

rat

hig

h–f

ocal

reso

urc

ele

vel:

Thl

aspi

arve

nse

(Bra

ssic

acea

e)N

utr

ien

tsa

Clip

pin

gN

o.se

eds

Incr

ease

Incr

ease

Incr

ease

4′Y

Ben

ner

1988

Ipom

opsi

sar

izon

ica

(Pol

emon

iace

ae)

Nu

trie

nts

Gra

zin

g/

clip

pin

gN

o.fr

uit

sIn

crea

seN

otre

port

edIn

crea

se4′

Y

Mas

chin

ski

and

Wh

itha

m

1989

I.ar

izon

ica

(Pol

emon

iace

ae)

Not

spec

ified

(com

peti

tion

)

Gra

zin

g/

clip

pin

g

No.

fru

its

Incr

ease

Not

repo

rted

Incr

ease

4′Y

Mas

chin

ski

and

Wh

itha

m

1989

Bet

ula

pube

scen

s(B

etu

lace

ae)

Not

spec

ified

(den

sity

)

Clip

pin

gB

iom

ass,

heig

htIn

crea

seN

one

Incr

ease

b4′

YH

jalt

enet

al.

1993

Ant

hris

cus

sylv

estr

is(A

piac

eae)

Nit

roge

nC

lippi

ng

Rep

rodu

ctiv

em

ass

Incr

ease

Incr

ease

cIn

crea

se4′

YH

anss

on19

94

Epi

lobi

umci

liatu

m(O

nag

race

ae)

Nu

trie

nts

Clip

pin

gN

o.se

edca

psu

les

Incr

ease

Incr

ease

Incr

ease

4′Y

Irw

inan

dA

arss

en19

96

E.

cilia

tum

(On

agra

ceae

)Li

ght

(com

peti

tion

)C

lippi

ng

No.

seed

caps

ule

sIn

crea

seIn

crea

seIn

crea

se4′

YIr

win

and

Aar

ssen

1996

Silp

hium

inte

grif

oliu

m(A

ster

acea

e)N

utr

ien

tsan

dw

ater

dC

ynip

idga

llsR

epro

duct

ive

mas

sIn

crea

seIn

crea

seIn

crea

se4′

YFa

yet

al.

1996

Ery

sim

umst

rict

um(B

rass

icac

eae)

Nu

trie

nts

Clip

pin

g

Rep

rodu

ctio

nan

d

grow

the

Incr

ease

Incr

ease

Dec

reas

e4

NH

uht

aet

al.

2000

E.

stri

ctum

(Bra

ssic

acea

e)N

otsp

ecifi

ed

(com

peti

tion

)

Clip

pin

gR

epro

duct

ion

and

grow

the

Incr

ease

Incr

ease

Incr

ease

4′Y

Hu

hta

etal

.20

00

Mad

iasa

tiva

(Ast

erac

eae)

Wat

erC

lippi

ng

No.

seed

head

sIn

crea

seIn

crea

seIn

crea

se4′

YG

onza

lez

etal

.20

08

Solid

ago

alti

ssim

a(A

ster

acea

e)N

utr

ien

tsC

lippi

ng

No.

seed

sIn

crea

seN

one

Incr

ease

4′Y

Th

isst

udy

645

B.

Tole

ran

cegr

eate

rat

low

–foc

al

reso

urc

ele

vel:

Urt

ica

uren

s,U

rtic

adi

oica

dioi

ca,

Urt

ica

dioi

caso

nden

ii

(Urt

icac

eae)

fN

utr

ien

tsC

lippi

ng

Infl

ores

cen

cem

ass

Incr

ease

Incr

ease

Dec

reas

e4

YM

uti

kain

enan

dW

alls

1995

U.

uren

s,U

.di

oica

dioi

ca,

U.

dioi

ca

sond

enii

(Urt

icac

eae)

f

Not

spec

ified

(com

peti

tion

)C

lippi

ng

Infl

ores

cen

cem

ass

Incr

ease

Incr

ease

Dec

reas

e4

YM

uti

kain

enan

dW

alls

1995

C.

Tole

ran

ceeq

ual

atdi

ffer

ent

foca

l-

reso

urc

ele

vels

:

Ipom

opsi

sag

greg

ata

cand

ida

(Pol

emon

iace

ae)

Nu

trie

nts

Clip

pin

gN

o.se

eds

Non

eIn

crea

seIn

crea

se5′

YJu

enge

ran

dB

erge

lson

1997

E.

stri

ctum

(Bra

ssic

acea

e)N

otsp

ecifi

ed

(com

peti

tion

)

Clip

pin

gN

o.se

eds

Non

eIn

crea

seIn

crea

se5′

YR

auti

oet

al.

2005

E.

stri

ctum

(Bra

ssic

acea

e)N

utr

ien

tsC

lippi

ng

No.

seed

sIn

crea

seIn

crea

seIn

crea

se4′

NR

auti

oet

al.

2005

S.in

tegr

ifol

ium

(Ast

erac

eae)

Ligh

t(c

ompe

titi

on)g

Clip

pin

gN

o.ca

pitu

laIn

crea

seIn

crea

seIn

crea

se4′

NFa

yan

dT

hroo

p20

05

Not

e:T

he

foca

lre

sou

rce

isth

ere

sou

rce

that

vari

edbe

twee

nen

viro

nm

ents

orw

asm

anip

ula

ted

inth

eex

peri

men

tal

trea

tmen

ts.

Th

elim

itin

gre

sou

rce

mod

el(L

RM

)pr

edic

tion

for

each

stu

dyis

show

n

byth

eou

tcom

en

um

ber

from

the

cou

plet

sin

box

1.Y

indi

cate

sth

atth

ere

sult

sof

the

stu

dym

atch

edth

eLR

M’s

pred

icti

on;

Nin

dica

tes

that

the

resu

lts

did

not

mat

chth

eLR

M’s

pred

icti

on.

aR

esu

lts

are

for

earl

yn

utr

ien

tad

diti

onan

dea

rly

apex

-clip

pin

gtr

eatm

ents

.b

Cu

ttin

gth

eto

psof

fof

the

plan

tsre

leas

edap

ical

dom

inan

ce,

and

the

upp

ersh

oots

grew

mor

eu

nti

lon

eu

pper

shoo

tre

gain

eddo

min

ance

.c

Fert

ilize

rca

use

dan

incr

ease

inso

me

mea

sure

sof

side

rose

tte

grow

thin

un

cut

plan

ts.

dIn

aga

rden

expe

rim

ent,

fert

ilize

ran

dw

ater

wer

efa

ctor

ial

trea

tmen

ts,

but

tole

ran

cere

sult

sw

ere

disc

uss

edm

ain

lyfo

rpl

ants

rece

ivin

gbo

thre

sou

rces

vers

us

nei

ther

.e

Res

ult

sar

efo

rM

AN

OV

Aof

aco

mbi

nat

ion

ofgr

owth

and

repr

odu

ctio

nm

easu

res.

fM

ult

ivar

iate

AN

OV

Are

sult

sar

epr

esen

ted.

The

over

all

patt

ern

sar

esi

mila

rfo

ral

lth

ree

spec

ies.

gR

esu

lts

are

for

the

“mow

ing

vers

us

clip

pin

gex

peri

men

t,”

whe

rem

owin

gis

assu

med

topr

imar

ilyaf

fect

com

peti

tion

for

light

.

646 The American Naturalist

volved 13 different species of dicots. (Monocots were notrelevant to this review because they have basal rather thanapical meristems.) In only two of the 18 studies did tol-erance of the dicots follow the GRM prediction. Combinedwith the results of our previous review on leaf herbivory,it seems clear that the type of damage (combined with thetype of resource) has much more to do with the effect ofresources on tolerance than whether the plant is a dicotor a monocot.

Finally, we noted above that the conditions that led toscenario 4′ in the LRM key were those in which overcom-pensation was most likely to occur. In particular, if a plant’sapical dominance is so strong that its reproduction is se-verely limited by the availability of meristems in a high-resource environment, then the plant might do better ifan herbivore damaged its apex and allowed axillary mer-istems to expand. Of the 13 articles on AMD in our review,the authors provide evidence of overcompensation in six.In five of these, overcompensation occurred in the morebenign treatments (i.e., high resources, low competition;Benner 1988; Maschinski and Whitham 1989; Hjalten etal. 1993; Irwin and Aarssen 1996; Huhta et al. 2000). Inthe sixth (Rautio et al. 2005), there was a trend for over-compensation at all four combinations of nutrient andcompetition treatments, but the trend was greatest in theunfertilized, high-competition treatment (i.e., the higheststress treatment). However, the stress-by-clipping inter-actions for seed number were not close to statistical sig-nificance in this study; thus, it is questionable whethertrue overcompensation occurred (Rautio et al. 2005). Thepreponderance of data, therefore, indicates that overcom-pensation for AMD is most likely to occur under high-resource, low-stress conditions, as is consistent with theLRM predicted outcome 4′.

Acknowledgments

We thank C. P. Blair, N. Dorchin, P. J. March, and twoanonymous reviewers for constructive criticism on pre-vious versions of the manuscript. We are also grateful toM. Jones for assistance in counting and dissecting gold-enrod capitula and counting and weighing rhizomes andto R. and A. Helbig for permission to collect goldenrodrhizomes on their property. This work was supported fi-nancially by the David Burpee Endowment of BucknellUniversity and by a National Science Foundation grant(DEB-0515483) to W.G.A. and M.J.W. Any opinions, find-ings, and conclusions expressed in this material are thoseof the authors and do not necessarily reflect the views ofthe National Science Foundation.

Literature Cited

Aarssen, L. W. 1995. Hypotheses for the evolution of apical domi-nance in plants: implications for the interpretation of overcom-pensation. Oikos 74:149–156.

Aarssen, L. W., and D. L. Irwin. 1991. What selection: herbivory orcompetition? Oikos 60:261–262.

Abrahamson, W. G., and A. E. Weis. 1997. Evolutionary ecologyacross three trophic levels: goldenrods, gallmakers, and naturalenemies. Princeton University Press, Princeton, NJ.

Alward, R. D., and A. Joern. 1993. Plasticity and overcompensationin grass responses to herbivory. Oecologia (Berlin) 95:358–364.

Beck, S. D. 1965. Resistance of plants to insects. Annual Review ofEntomology 10:207–232.

Benner, B. L. 1988. Effects of apex removal and nutrient supple-mentation on branching and seed production in Thlaspi arvense(Brassicaceae). American Journal of Botany 75:645–651.

De Deyn, G. B., C. E. Raaijmakers, and W. H. van der Putten. 2004.Plant community development is affected by nutrients and soilbiota. Journal of Ecology 92:824–834.

Fay, P. A., and H. L. Throop. 2005. Branching responses in Silphiumintegrifolium (Asteraceae) following mechanical or gall damage toapical meristems and neighbor removal. American Journal of Bot-any 92:954–959.

Fay, P. A., D. C. Hartnett, and A. K. Knapp. 1996. Plant toleranceof gall-insect attack and gall-insect performance. Ecology 77:521–534.

Fornoni, J., J. Nunez-Farfan, and P. L. Valverde. 2003. Evolutionaryecology of tolerance to herbivory: advances and perspectives. Com-ments on Theoretical Biology 8:643–663.

Geber, M. A. 1990. The cost of meristem limitation in Polygonumarenastrum: negative genetic correlations between fecundity andgrowth. Evolution 44:799–819.

Gonzalez, W. L., L. H. Suarez, M. A. Molina-Montenegro, and E.Gianoli. 2008. Water availability limits tolerance of apical damagein the Chilean tarweed Madia sativa. Acta Oecologica 34:104–110.

Hansson, M. L. 1994. Response of Anthriscus sylvestris (L.) Hoffm.to defoliation and different nitrogen supply levels. Swedish Journalof Agricultural Research 24:21–29.

Hawkes, C. V., and J. J. Sullivan. 2001. The impact of herbivory onplants in different resource conditions: a meta-analysis. Ecology82:2045–2058.

Hilbert, D. W., D. M. Swift, J. K. Detling, and M. I. Dyer. 1981.Relative growth rates and the grazing optimization hypothesis.Oecologia (Berlin) 51:14–18.

Hjalten, J., K. Dannell, and L. Ericson. 1993. Effects of simulatedherbivory and intraspecific competition on the compensatory abil-ity of birches. Ecology 74:1136–1142.

Huhta, A.-P., K. Hellstrom, P. Rautio, and J. Tuomi. 2000. A test ofthe compensatory continuum: fertilization increases and below-ground competition decreases the grazing tolerance of tall worm-seed mustard (Erysimum strictum). Evolutionary Ecology 14:353–372.

Inouye, D. W. 1982. The consequences of herbivory: a mixed blessingfor Jurinea mollis (Asteraceae). Oikos 39:269–272.

Irwin, D. L., and L. W. Aarssen. 1996. Effects of nutrient level oncost and benefit of apical dominance in Epilobium ciliatum. Amer-ican Midland Naturalist 136:14–28.

Juenger, T., and J. Bergelson. 1997. Pollen and resource limitation

Tolerance of Apical Meristem Damage 647

of compensation to herbivory in scarlet gilia, Ipomopsis aggregata.Ecology 78:1684–1695.

Lehtila, K., and A. S. Larsson. 2005. Meristem allocation as a meansof assessing reproductive allocation. Pages 51–75 in E. G. Reekieand F. A. Bazzaz, eds. Reproductive allocation in plants: physio-logical ecology. Elsevier Academic, Burlington, MA.

Lennartsson, T., P. Nilsson, and J. Tuomi. 1998. Induction of over-compensation in the field gentian, Gentianella campestris (Gen-tianaceae). Ecology 79:1061–1072.

Lortie, C. J., and L. W. Aarssen. 1997. Apical dominance as an ad-aptation in Verbascum thapsus: effects of water and nutrients onbranching. International Journal of Plant Sciences 158:461–464.

MacDonald, K. P., and C. E. Bach. 2005. Resistance and toleranceto herbivory in Salix cordata are affected by different environ-mental factors. Ecological Entomology 30:581–589.

Maschinski, J., and T. G. Whitham. 1989. The continuum of plantresponses to herbivory: the influence of plant association, nutrientavailability, and timing. American Naturalist 134:1–19.

Mattson, W. J., Jr. 1980. Herbivory in relation to plant nitrogencontent. Annual Review of Ecology and Systematics 11:119–161.

Mutikainen, P., and M. Walls. 1995. Growth, reproduction and de-fence in nettles: responses to herbivory modified by competitionand fertilization. Oecologia (Berlin) 104:487–495.

Paige, K. N. 1992. Overcompensation in response to mammalianherbivory: from mutualistic to antagonistic interactions. Ecology73:2076–2085.

Paige, K. N., and T. G. Whitham. 1987. Overcompensation in re-sponse to mammalian herbivory: the advantage of being eaten.American Naturalist 129:407–416.

Painter, R. H. 1958. Resistance of plants to insects. Annual Reviewof Entomology 3:267–290.

Pilson, D. 1992. Aphid distribution and the evolution of goldenrodresistance. Evolution 46:1358–1372.

Potvin, M. A., and P. A. Werner. 1984. Seasonal patterns in waterrelations of two species of goldenrods (Solidago) grown on anexperimental soil moisture gradient. Bulletin of the Torrey Botan-ical Club 111:171–178.

Rautio, P., A.-P. Huhta, S. Piippo, J. Tuomi, T. Juenger, M. Saari, andJ. Aspi. 2005. Overcompensation and adaptive plasticity of apicaldominance in Erysimum strictum (Brassicaceae) in response tosimulated browsing and resource availability. Oikos 111:179–191.

Root, R. B., and N. Cappuccino. 1992. Patterns in population changeand the organization of the insect community associated withgoldenrod. Ecological Monographs 62:393–420.

Simms, E. L. 2000. Defining tolerance as a norm of reaction. Evo-lutionary Ecology 14:563–570.

Sokal, R. R., and F. J. Rohlf. 1995. Biometry. W. H. Freeman, NewYork.

Stowe, K. A., R. J. Marquis, C. G. Hochwender, and E. L. Simms.2000. The evolutionary ecology of tolerance to consumer damage.Annual Review of Ecology and Systematics 31:565–595.

Strauss, S. Y., and A. A. Agrawal. 1999. The ecology and evolutionof plant tolerance to herbivory. Trends in Ecology & Evolution 14:179–185.

Throop, H. L. 2005. Nitrogen deposition and herbivory affect bio-mass production and allocation in an annual plant. Oikos 111:91–100.

Walck, J. L., J. M. Baskin, and C. C. Baskin. 1999. Relative competitiveabilities and growth characteristics of a narrowly endemic and ageographically widespread Solidago species (Asteraceae). AmericanJournal of Botany 86:820–828.

Whitham, T. G., J. Maschinski, K. C. Larson, and K. W. Paige. 1991.Plant responses to herbivory: the continuum from negative topositive and underlying physiological mechanisms. Pages 227–256in P. W. Price, T. M. Lewinsohn, G. W. Fernandes, and W. W.Benson, eds. Plant-animal interactions: evolutionary ecology intropical and temperate regions. Wiley, New York.

Wise, M. J., and W. G. Abrahamson. 2005. Beyond the compensatorycontinuum: environmental resource levels and plant tolerance ofherbivory. Oikos 109:417–428.

———. 2007. Effects of resource availability on tolerance of her-bivory: a review and assessment of three opposing models. Amer-ican Naturalist 169:443–454.

Wise, M. J., R. J. Fox, and W. G. Abrahamson. 2006. Disarming theparadox of sublethal plant defense against insects: Trirhabda virgatalarval development time and leaf tissue loss on Solidago altissima.Entomologia Experimentalis et Applicata 120:77–87.

Wise, M. J., J. M. Partelow, K. J. Everson, M. K. Anselmo, and W.G. Abrahamson. 2008. Good mothers, bad mothers, and resistanceto herbivory in Solidago altissima. Oecologia (Berlin) 155:257–266.

Associate Editor: Susan KaliszEditor: Monica A. Geber