By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Due to essential maintenance the subscribe/renew pages will be unavailable on Wednesday 26 October between 02:00- 08:00 BST/ 09:00 – 15:00 SGT/ 21:00- 03:00 EDT. Apologies for the inconvenience.
J. Ding, Wuhan Botanical Institute/Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan, Hubei 430074 China. E-mail: firstname.lastname@example.org
Facilitation, both by inter- and intra-specific neighbours, is known to be an important process in structuring plant communities. However, only a small number of experiments have been reported on facilitation in plant invasions, especially between invasive con-specific individuals. Here, we focus on how con-specific neighbours of the invasive alien plant alligator weed affect the tolerance of alligator weed to herbivory by the introduced biological control agent, Agasicles hygrophila. We conducted greenhouse and garden experiments in which invasive plant density and herbivory intensity (artificial clipping and real herbivory) were manipulated. In the greenhouse experiment, artificial clipping significantly reduced plant biomass when plants were grown individually, but when con-specific neighbours were present in the same pot, biomass was not significantly different from control plants. Similarly, when compared to control plants, plants that were subjected to herbivory by A. hygrophila produced more biomass when grown with two con-specific neighbours than when grown alone. Real herbivory also resulted in an increased number of vegetative buds, and again when two con-specific neighbours were present this effect was increased (a 55.3% increase in buds when there was no neighbour, but a 111.6% increase in buds when two con-specific neighbours were present). In the garden experiment, in which plants were grown at high density (6 plants per pot), alligator weed fully recovered from defoliation caused by insects at levels from 20–30% to 100%. Our results indicate that the con-specific association may increase the compensatory ability to cope with intense damage in this invasive plant.
Understanding mechanisms underlying biological invasion is crucial for managing invasive species (Mack et al. 2000). While several hypotheses for invasion success have been proposed (Mitchell et al. 2006), we still have no general conceptual framework (Dietz & Edwards 2006). Recently, an increasing number of researchers (Dietz & Edwards 2006; Theoharides & Dukes 2007) have considered biological invasions as ecological and evolutionary processes in which changing biotic conditions, such as the identity of neighbours, may affect invaders’ performance. In the early stages of an invasion, many exotic plants may occur in small populations (Pyšek & Hulme 2005). As these plants compete with and replace native neighbours, many invaders establish large populations, forming high-density monocultures (Kot et al. 1996; Mack et al. 2000). Therefore, during invasion, the neighbours of invaders will tend to switch from an initial predominance of native species to a majority of con-specific individuals once a high density has been reached. Interactions such as competition and facilitation between invasive and native species have been intensively studied, but interactions between invasive con-specific individuals have received less attention (but see Davis et al. 2004; Reinhart et al. 2006; Jordan et al. 2008).
Facilitation (positive interaction) among plants is known to be an important process in structuring plant communities (Bruno et al. 2003; Reinhart et al. 2006). However, only a small number of studies have focused on facilitation between invasive plants (Bruno et al. 2003), especially between invasive con-specific individuals (Davis et al. 2004; Reinhart et al. 2006; Jordan et al. 2008). Reinhart et al. (2006) reported that canopies of the invasive tree Acer platanoides facilitated recruitment and survival of con-specific seedlings through modification of physical conditions, while the invasive species Bromus inermis and Agropyron cristatum exhibited significant self-facilitation via modification of soil microbes (Jordan et al. 2008). Theoretical models and empirical evidence indicate that the importance of facilitation should increase with increasing stress (Callaway & Walker 1997; Graff et al. 2007; Alberti et al. 2008). However, whether and how invasive con-specific neighbours facilitate each other to cope with increasing herbivory remains unknown. In particular, the number and identity of herbivores attacking plant invaders may change during the invasion process. Native herbivores may accumulate on invasive plants (Siemann et al. 2006), and classical biological control programmes also increase herbivore load on the plants (McFadyen 1998).
In the context of herbivory, tolerance has been defined as the ability of a plant to relieve negative impacts of herbivory on fitness by compensatory growth and reproduction (Strauss & Agrawal 1999). Recent studies indicate that the identity of neighbouring plants and herbivory may affect the expression of plant tolerance (Eskelinen 2007), and facilitation between con-specific individuals may increase plant compensatory ability (Rand 2004). However, there is little information about the impact of invasive con-specific neighbours on a plant’s compensatory response to herbivory. We propose a hypothesis that facilitation between invasive con-specific individuals may increase plant compensatory capacity in response to increasing herbivory; if so, then herbivores may have a decreasing impact on invaders when plants are at high density.
Here, we test the above hypothesis using the invasive plant alligator weed Alternanthera philoxeroides (Mart.) Griseb as a model species. It is native to South America and has invaded North America, Asia, New Zealand and Australia (Julien et al. 1995). In invasive ranges, alligator weed forms dense monocultures in both aquatic and terrestrial habitats (Sainty et al. 1998; Ma 2001). The host-specific leaf beetle, Agasicles hygrophila (Coleoptera; Chrysomelidae), was introduced from Argentina as a classical biological control agent, first to Florida (USA) in the early 1960s, and from there to China in 1986 (Ma 2001). However, the beetle has only controlled the weed effectively in aquatic habitats, and had no effects in terrestrial habitats (Ma 2001).
In this study, we conducted greenhouse and garden experiments to examine the effects of plant density and herbivory on plant performance in alligator weed. Specifically, we addressed the following question: do con-specific neighbours increase plant compensatory capacity in response to herbivory?
Material and methods
Alligator weed is a herbaceous perennial, with horizontal to ascending stems. Plants root in soil either on land, emerging from shallow water, or as floating mats attached to banks. In the invasive range, the species rarely sets seeds and reproduces by vegetative means from apical stem buds, axillary stems and root buds (Sainty et al. 1998). The adults and larvae of A. hygrophila eat leaves of this plant, often producing feeding ‘holes’ and ‘trenches’. Insects used in all of our experiments were collected from patches of alligator weed in the suburbs of Wuhan, Hubei Province.
To test the interactive effects of con-specific neighbours of differing density and herbivores on plant performance, we conducted an experiment in a greenhouse with natural light in the Wuhan Botanical Garden of the Chinese Academy of Sciences from May to December 2007. In July and August the roof of the greenhouse was covered with shading nets and all windows in the greenhouse walls were open to allow air flow. Temperatures in the greenhouse were between 20–35 °C from May to September and 10–25 °C between October and early December. Utilizing a two-way factorial design, we manipulated plant density (three levels: 1, 3 and 5 plants per pot, hereafter referred to as low, medium and high densities, respectively) and herbivory (three levels of defoliation: 50% leaves removed by artificial clipping, 50% leaves removed by leaf beetle herbivory, and undamaged control, hereafter referred to as artificial clipping, real herbivory and control, respectively) with 7–10 replicates per treatment combination, as available.
In early June 2007, plants were sampled at four sites (∼20 m apart) in an abandoned field in Wuhan. The stems were cut to similar size (length of 4–5 cm), with one node for each stem. Six randomly selected stems were planted vertically in a 16-cm diameter pot filled with a homogenous mixture of peat, topsoil and sand (ratio: 1:1:1). Plants were watered with tap water every 2 days. Twenty days later, pots were randomly assigned to three levels of density, where plants were thinned to one, three or five similar-sized plants per pot.
Ten days after assigning the density treatment, herbivory treatments were initiated. For each density treatment, one third of the pots received insects (real herbivory treatment); one third received clipping (artificial herbivory treatment); and the last third were control pots. Two days before the real herbivory treatment, half of each plant was caged with nylon gauze. Each cage covered nearly half of the stems and leaves of each plant. In the real herbivory treatment, seven to nine adults of A. hygrophila were released into each cage. Prior tests indicated that this number of insects could defoliate half of the plant in 1 week. In the artificial clipping treatment, all caged leaves in each pot were clipped on the seventh day of the real herbivory treatment. To keep exposure to sunlight at the same level between all treatments, plants in the control treatment were caged, as were the real herbivory and artificial herbivory treatments. When the required levels of defoliation were achieved (in 7 days), all the cages and insects were manually removed. Plants were then allowed to grow for an additional 80 days. The plants were watered with tap water every 2 days. All pots were randomly distributed and their positions were switched every 2 weeks.
We counted the number of shoots and measured the primary stem length on 6 October and again on 5 December, when the experiment ended. We counted the numbers of leaf and root buds and measured root length, then separated and dried the above (shoot and leaf, hereafter referred to as shoot biomass) and belowground (root) parts at 80 °C for 48 h before weighing.
To test the effects of levels of defoliation on plant re-growth under relatively natural conditions, we conducted a garden experiment in a field near the Wuhan Botanical Garden from early June to December 2007. Air temperatures in the field were between 15–38 °C from May to early September and 8–20 °C between November and early December. In early June 2007, we sampled plants from the same population used in the greenhouse experiment and established similar treatments. Ten stems (length of 4–5 cm) were planted vertically into a 25 cm diameter pot filled with topsoil. All plants were caged with nylon gauze immediately to exclude insects. All pots were randomly distributed in four rows in the field. Plants were watered every 2 days and pots weeded as necessary. Twenty days later, every pot was thinned to six similar-sized plants.
Potted alligator weed plants were randomly assigned to the control treatment (no herbivores) or one of three levels of defoliation by A. hygrophila. To obtain three levels of defoliation (20–30%, 40–50% and 100% defoliation), we manipulated the number of insects released into each cage to achieve the necessary level of defoliation. The number of insects per cage varied from 4 to 16, depending on the required levels of defoliation for each treatment. After insect release, the plants were visually inspected for damage. Depending on the level of defoliation observed at intervals of 1 or 2 days, more insects were added into some cages to increase herbivory intensity when necessary. About 7–8 days after insect release, all three levels of defoliation were achieved and the insects were removed. Plants were allowed to grow for an additional 80 days. Each treatment was replicated 7 to 10 times and the positions of all pots were randomised. We then harvested the plants and made the same measurements as in the greenhouse experiment.
Prior to analysis, biomass and the ratio of root to shoot biomass (R/S) were log-transformed to improve normality and reduce heterogeneity of variances. To examine facilitation between con-specific neighbors, effects of herbivory, plant density and their interactions on plant vegetative and reproductive characteristics were assessed. A general linear model (GLM) repeated measures analysis of variance (anova) was used with the length of primary stems and number of shoots as response varaibles, each variable having been recorded in October and December. When significant interactions occurred, further tests for differences were made using Bonferroni post hoc multiple comparisons. The impacts on total plant biomass, total number of buds and root length at the end of the experiment were evaluated by a two-way anova. A mancova was carried out on the root mass, shoot mass, R/S, and numbers of leaf and root buds. For root and shoot masses and R/S, the mancova was performed using the total biomass as a covariate, and with the root and shoot masses and R/S as response variables, whereas for the leaf and root buds, the total number of buds was used as a covariate and the numbers of root and leaf buds as response variables. When a significant interaction between herbivory and plant density occurred, further tests for differences among treatments were made using Bonferroni post hoc multiple comparisons.
For the dataset in the garden experiment, a one-way anova was performed to test the effects of levels of defoliation on total plant biomass, total number of buds, number of shoots, stem and root lengths. An ancova was carried out on root and shoot masses, R/S, numbers of leaf and root buds. For root and shoot masses and R/S, an ancova was performed using total biomass as a covariate and with the root and shoot masses and R/S as response variables, whereas for leaf and root buds, the total number of buds was used as a covariate and numbers of root and leaf buds as response variables. All data were analysed using spss 13.0 for Windows (spss Inc., Chicago, IL, USA)
Effects on stem and shoot growth
The only factor that significantly affected stem length was plant density (stems were shorter at higher plant densities, Table 1, Fig. 1A). However, the number of shoots was significantly affected by plant density, herbivory treatment, time, the interaction of plant density and herbivory treatment, and the interaction of plant density and time (Table 1).
Table 1. GLM repeated measures anova for effects of plant density, herbivory and their interaction on the length of primary stems and number of shoots in the greenhouse experiment.
source of variation
no. of shoots
F-values and their level of significance are given for density, herbivory, time and their interactions. ***P < 0.001; **P < 0.01; n.s. = not significant.
D × H
T × D
T × H
T × D × H
Plants grown at low density produced significantly more shoots when defoliated by real herbivory than if they were clipped or kept as control plants (P = 0.023) (Fig. 1B). However, neither type of herbivory significantly affected shoot growth at medium or high plant densities (Fig. 1B).
From 6 October to 5 December, the total number of shoots averaged across treatments increased by 22.6% (P < 0.001) at the low plant density, while at medium and high densities the number of shoots did not change significantly (P > 0.05). On 6 October, there was no significant difference in the number of shoots between the medium and high densities (P > 0.05), whereas by the end of the experiment (5 December) the number of shoots at medium density was higher than that at high density (P < 0.05).
Effects on reproductive buds, biomass and root growth
Plant density, herbivory treatment and their interaction influenced reproductive buds (Table 2). The total number of buds decreased as plant density increased (Fig. 2E). At low and medium densities, the total number of buds increased after real herbivory by 55.3% (P < 0.001) and 111.6% (P = 0.0027) respectively, while no such differences were found in the clipping treatments (Fig. 2E). At high density, the total number of buds was not affected by real herbivory or clipping (Fig. 2E).
Table 2. Effects of plant density, herbivory and their interaction on plant vegetative and reproductive traits in the greenhouse experiment.
source of variation
total no. of buds
D × H
source of variation
no. of leaf buds
no. of root buds
Tests carried out were two-way anova for total biomass, total number of buds and root length, and mancova for biomass (total biomass as covariate, root and shoot masses and R/S as dependent variables) and reproductive buds (total number of buds as covariate, numbers of leaf and root buds as dependent variables) allocation to above- and belowground. F-values and their level of significance are given for density, herbivory and their interaction. ***P < 0.001; **P < 0.01; *P < 0.05; n.s. = not significant. For mancova, interaction terms between plant density, herbivory and covariates were first introduced in the model, then non-significant interaction terms, including covariate, were removed to obtain the final model.
D × H
Total biomass was significantly affected by plant density, herbivory treatment and their interaction (Table 2). Con-specific neighbours reduced total biomass as plant density increased (Fig. 2A). At low and high densities, there were no significant differences between real herbivory and control plants (Fig. 2A), but at medium density real herbivory increased biomass (P = 0.0005) compared to undamaged control plants (Fig. 2A). The ratio of root to shoot biomass (R/S) was influenced by plant density (Table 2): R/S was greater at high and medium densities than at low density (P = 0.0162 and P = 0.0004 for high and medium densities, respectively).
The shoot mass (Fig. 2B), number of leaf and root buds (Fig. 2F, G) and root length (Fig. 2H) at low density were significantly higher than at medium (all P < 0.05) and high densities (all P < 0.05). However, there was not a signficant effect of herbivory treatment and its interaction with plant density on these plant traits (Table 2).
Total biomass and root mass did not differ significantly between defoliation treatments (Fig. 3A, C); however, biomass allocation to shoots was affected (Fig. 3B). Defoliation at medium and high levels (40–50% and 100%) deceased plant shoot mass, compared to undamaged control plants (Fig. 3B).
Although the total number of buds was not influenced (Fig. 3E) by herbivory, the numbers of leaf and root buds differed significantly between defoliation treatments (Fig. 3F, G). The number of leaf buds significantly increased with 100% defoliation, compared to undamaged control plants, whereas the number of root buds decreased after 50% and 100% defoliation.
Herbivory did not affect shoot number (Fig. 3H). The primary stems of plants at the low level of defoliation (20–30%) were significantly longer than those of plants defoliated at the high level (100%) (Fig. 3I). Low levels of defoliation (20–30%) significantly reduced the length of primary roots, while higher levels defoliation had no effect (Fig. 3J).
Our results indicate significant interactive effects of the type of herbivory and plant density on the compensatory capacity of alligator weed. When plants were grown individually, artificial herbivory (clipping) significantly reduced plant biomass (Fig. 2A). But plants were able to fully recover from artificial herbivory when two or four con-specific neighbours were present in the same pot. When grown individually, real herbivore-defoliated plants fully recovered, while they overcompensated in biomass production when growing together with two con-specific neighbours. Plants responded to real herbivory by increased production of leaf and root buds (and ergo plant reproductive ability). This equated to a 55.3% increase in total number of vegetative buds when there was no neighbour, but 111.6% when two con-specific neighbours were present. These results suggest that facilitation between con-specific individuals increased the compensatory ability of alligator weed, as hypothesized.
Facilitation among neighbouring individuals resulting in increased plant compensatory capacity has been reported in other systems. For example, Rand (2004) illustrated that neighbouring rushes increased the ability of the annual forb, Atriplex patula var. hasata, to compensate for herbivory through soil shading. Callaway et al. (2001) suggested that the presence of soil fungi and the co-occurring native species Nassella pulcbra promoted a compensatory response of the invasive species Centaurea melitensis to herbivory. Parmesan (2000) reported that herbivory by the butterfly Euphydryas editha reduced fitness of the native montane annual Collinsia torreyi at low density, but not at high density. However, intra-specific facilitation between monoculture-forming invasive species has rarely been studied and, to our knowledge, the present study is the first to demonstrate that con-specific neighbours can increase invasive plant compensatory capacity.
There are several possible explanations for facilitation between individuals under the stress of herbivory. First, neighbours may relieve plants from abiotic stresses, i.e. by reducing salinity or desiccation stress, which may otherwise limit the expression of plant compensatory response, as confirmed by the findings of Rand (2004). She found that removal of neighbouring plants significantly increased salinity and reduced soil oxygen availability, which resulted in decreased plant compensatory capacity. Second, neighbours may modify the activity or composition of soil microbes and thus benefit plant compensation. Kuba et al. (2005) reported that soil microbes, i.e. mycorrhizal symbiosis, could increase plant regrowth following defoliation. Both of the above two mechanisms may help explain the results of this study, although further work on soil nutrients and microbes is needed to identify the mechanisms involved. Our study also found that the total number of buds increased when neighbours were present, suggesting increased reproductive ability of the plant in response to defoliation. In addition, improved resource uptake ability may lead to increased compensatory ability. Schooler et al. (2006) demonstrated that improved resource uptake ability through increased root growth is crucial for compensation in alligator weed. Therefore, plant neighbors may relieve negative impacts of herbivory and provide an associational defence through improved resource uptake and reproductive ability. This possible mechanism warrants further investigation.
Facilitation and competition may operate simultaneously, and the net outcome of interactions among neighbours is determined by the relative strength of each of these components, which are regulated by abiotic and biotic conditions (Callaway & Walker 1997; Alberti et al. 2008). We manipulated density gradients in our experiments to reflect the range of densities observed in the field. Competition dominated the interaction between con-specific individuals without herbivory, however, facilitation and competition operated simultaneously when herbivory was present. Herbivory significantly reduced the length of primary roots at low levels of defoliation (20–30%), while it had neutral effects at medium (40–50%) and high levels (100%) of defoliation. Also, herbivory stimulated the formation of leaf buds at high level of defoliation, but had no effects at low and medium levels of defoliation when neighbours were present. Thus, the results of this study indicate that herbivory shifts the interaction among con-specific individuals from competition to facilitation. These results are in line with the stress–gradient hypothesis (Callaway & Walker 1997), which states that the relative importance of facilitation increases, and the relative importance of competition decreases, with increasing stress.
The relative importance of facilitation and competition is also determined by the fitness components selected in our experiments. In terms of biomass, competition is the main interaction among neighbours. In terms of reproductive ability, facilitation is more important under moderate stress (i.e. at medium density), while competition overshadows facilitation under more stressful conditions (i.e. at high density). Thus, detection of the relative importance of facilitation and competition can depend on the fitness parameter examined (Rand 2004).
The results of this study also indicate that increased compensatory ability resulting from facilitation between con-specific individuals may enable invaders to cope with intense damage caused by accumulation of released biological control agents. In the garden experiment, alligator weed fully recovered from defoliation at levels from 20–30% to 100% caused by the introduced biocontrol insect A. hygrophila when plants were grown at high plant density, as predicted. In another experiment, we found that 100% defoliation by A. hygrophila significantly reduced performance of plants from four populations (including the population used in this study) when they were grown individually (Lu & Ding, unpublished results). Thus, we assume that the efficient compensatory capacity described in this study resulted from con-specific association.
Our results have broad implications for management of invasive plants. Although biological, mechanical and chemical methods have been used for decades, alligator weed is still a growing threat to ecosystems in many countries. Increased compensatory ability might help alligator weed to tolerate defoliation caused by introduced biocontrol agents, such as A. hygrophila. It might also allow the plant to tolerate other stress factors in the introduced range, such as herbivory by native insects, vertebrates, fire, drought or harsh climate, because similar compensatory signals might be activated after defoliation by these events (Fujita et al. 2006). Since con-specific neighbours may enhance plant compensation capacity and help plants to tolerate stressful conditions, control efforts in the early stage of invasion, when plants grow at low density, would be more effective than that in the later stages of invasion, when invaders form monocultures at high density. Indeed, the local density of invasive species has been shown to have significant effects on population dynamics and efficacy of other aspects of invasive plant management (Taylor & Hastings 2004).
Success in biological control programmes relies on the ability of natural enemies to suppress target invasive plant populations. Halpern & Underwood (2006) proposed that ignorance of the importance of density-dependence in invasive population regulation might explain a high failure rate of biological control. Müller-Schärer & Schaffner (2008) also pointed out that insufficient understanding of the impact of herbivory on plant fitness might lead to failure of biological control. Our study addressed the impact of the interaction between herbivory and plant density on plant invader fitness. We found that plant density may regulate the magnitude of herbivory impact if the invader possesses the ability to effectively compensate for herbivory. The result may help explain why some biological control practices have failed, even though biological control agents successfully established and formed large populations in target sites (McFadyen 1998). For example, a leaf-feeding beetle, Galerucella calmariensis was introduced to British Columbia, Canada, to control the exotic species Lythrum salicaria. Although the agent established at all surveyed non-tidal sites and caused intense defoliation, the stem density at some sites increased in a 4-year survey (Denoth & Myers 2005). Given that efficient compensatory capacity is common in plant invaders (Müller-Schärer et al. 2004), it is important to predict biological control efficacy before agent release through understanding the plant response to herbivory under differing biotic and abiotic conditions.
We thank Wei Huang for field and lab assistance. The manuscript was improved by comments from John Wilson, Evan Siemann, Ashley Baldridge, Matthew Barnes, Diane Byers and three anonymous reviewers. This work was funded by the Knowledge Innovation Programme of the Chinese Academy of Sciences and by the National Science Foundation of China (30871650) while preparing the manuscript.