Variation in arbuscular mycorrhizal fungi colonization modifies the expression of tolerance to above-ground defoliation

Summary

1. Plant association with arbuscular mycorrhizal fungi (AMF) has been considered a factor increasing plant tolerance to herbivory. However, this positive effect could decrease with colonization density if the benefit : cost ratio of the AMF–plant association changes. We measured plant performance and tolerance to defoliation across a gradient of commercial AMF (Glomus sp.) inoculum concentration.

2. Six genetic families of Datura stramonium were grown under greenhouse conditions and subjected to five increasing levels of AMF inoculum concentration and to defoliation treatments, i.e. the presence/absence of 50% artificial damage, following a full-factorial design.

3. AMF colonization increased linearly with inoculum concentration while foliar area, root mass, flowering phenology and seed production expressed nonlinear functions. Plant genetic variation in the benefit function of AMF colonization was also detected. We show a negative interaction between AMF concentration and plant tolerance to defoliation.

4. Synthesis. The negative correlation between plant tolerance and AMF concentration suggests that defoliation can reduce AMF benefits and that natural variations in AMF can limit the evolution of optimum levels of tolerance. Moreover, genetic variation in the shape of the reaction norms to AMF in the presence/absence of defoliation suggests that plants may evolve in response to variation in densities of AMF and herbivores.

Introduction

Ecologists are gaining an increasing appreciation of how multispecies interactions can act synergistically or antagonistically to alter the ecological and evolutionary outcomes of interactions in ways that differ fundamentally from outcomes predicted by pairwise interactions (Strauss & Irwin 2004). In particular, above-ground–below-ground feedbacks play a fundamental role in controlling plant abundance and distribution and the interactions of plants with other community members (Van der Putten et al. 2001; Wardle et al. 2004). Soil biota contributing to these feedbacks include arbuscular mycorrhizal fungi (AMF), which colonize the roots of a great diversity of vascular plants. AMF obtain photosynthates from plants while enhancing nutrient uptake by the host plant; AMF may also improve plant performance when plants are attacked by pathogens (reviewed in Borowicz 2001) or by insect herbivores (reviewed in Gange 2007).

The AMF–plant relationship, while often considered mutualistic, can also entail costs. The amount of carbon allocated to AMF is estimated to range from 4% to 20% of a plant’s total carbon budget (Smith & Read 1997). Scattered throughout the literature are examples of the conditionality of this relationship exemplified by a continuum of the effects of AMF colonization on hosts from positive, through null to negative (Francis & Read 1995; Johnson, Graham & Smith 1997; Jones & Smith 2004). For any particular host plant−fungus combination, the whole continuum of the effects may be expressed, depending on environmental conditions and genotypes involved (reviewed in Johnson, Graham & Smith 1997). Moreover, it has been suggested that the benefit of a plant associating with fungal symbionts depends not only on the presence of AMF, but also on colonization density (Gange & Ayres 1999). Thus, the plant performance continuum may reflect both the identity and the density of AMF colonization.

Feedbacks between above- and below-ground interactions may occur when other interactors, like herbivores or pollinators, compete for plant resources with AMF. In the presence of herbivores, plants lose not only foliar area and water but also their carbon-fixing capacity through loss of photosynthetic tissue (Gange 2007). Because herbivores and AMF both extract energy from plants, albeit in different forms, they are likely to interact (Gehring & Whitham 1994; Koricheva, Gange & Jones 2009), especially when some resources are limited (e.g. carbon, phosphorus, nitrogen, etc.). If AMF and defoliation create a resource-limiting environment for the plant, an increment in the density of AMF colonization would constrain the plant’s ability to reduce the negative effect of defoliation in terms of fitness (i.e. tolerance). In contrast, whenever AMF provides a surplus of resources for the host plant to decrease the costs of tissue lost by defoliation, a positive relation between the density of AMF colonization and the plant tolerance to defoliation would be expected (Borowicz 1997; Kula, Hartnett & Wilson 2005; Bennett, Alers-Garcia & Bever 2006; Bennett & Bever 2007). Given that the benefit gained by the host plant from the association with AMF can depend on colonization density (Gange & Ayres 1999), we specifically evaluated plant tolerance to above-ground defoliation across a gradient of AMF inoculum concentration.

In this study, we measured foliar area, root mass, flowering phenology, seed production, total colonization and arbuscule percentage across a range of AMF inoculum concentration and across plant genotypes in both the presence and absence of defoliation. In addition, we evaluated whether the expression of tolerance to defoliation (differences in seed production between damaged and undamaged full-sib plants) varied along a gradient of inoculum concentration. Here, we show for the first time a negative interaction between AMF inoculum concentration and the expression of tolerance to above-ground defoliation, suggesting a negative below-ground–above-ground interaction.

Materials and methods

Datura stramonium L. (Solanaceae) is a cosmopolitan summer annual species that inhabits disturbed areas and borders of cultivated fields (Weaver & Warwick 1984; Núñez-Farfán & Dirzo 1994). A complete description of this species can be found elsewhere (Fornoni & Núñez-Farfán 2000). Plant material used in this study was gathered from a population of D. stramonium in Central Mexico (18°45′ N, 99°07′ W). Based on a previous study that detected significant additive genetic variation in plant defences (tolerance and resistance) to herbivory under natural conditions (Fornoni, Valverde & Núñez-Farfán 2003, 2004), we specifically choose six genetic families that differ in tolerance but not in resistance (see Espinosa & Fornoni 2006). To minimize environmental maternal effects, each family was self-pollinated under greenhouse conditions for two generations prior to conducting the experiment. In February 2007, seeds from each family were germinated in sterile vermiculite in a greenhouse at UC Davis (California, USA) to obtain 40 plants per family (N = 240). Background soil consisted of a 6 : 4 mixture of commercial potting soil (MetroMix; Sun Gro Horticulture Canada CM Ltd, Vancouver, BC, Canada) and a fine sandy loam collected from Yolo County (California, USA) steam-sterilized twice. Concentrations of nitrogen and phosphorus in the soil were 0.76% and 0.05% respectively. Pots of 1.2 L each were filled with the background soil and inoculated with commercial mycorrhizal inoculum (MycoApply® Endomycorrhizal granular inoculum containing spores of four Glomus species: G. intraradices, G. mosseae, G. aggregatum and G. etunicatum) obtained from Mycorrhizal Applications (Grants Pass, OR, USA) according to five inocula treatments. Inoculum was mixed into the background soil to ensure maximum contact between roots and inoculum.

Inocula treatments consisted of a gradient in the concentration of inoculum within 1.2-L pots: 0, 42, 84, 167 and 333 mL. That is, the proportion of sterile background soil and inoculum was manipulated without changing total soil volume. This gradient produced the maximum levels of root colonization in an earlier study (Bennett & Bever 2009) and provided a range of root colonization from 0% to 31% (see Results). The amount of inoculum added to each pot is not likely to change soil structure significantly, given the highly porous background soil used in the experiment. Immediately after inoculation plants were transplanted. There were eight plants per inoculum concentration per family. All pots were randomized following a complete block design within the greenhouse (25 °C, 60% humidity and 12 : 12 L : D).

Starting 3 weeks after transplanting and continuing until all plants had stopped reproducing after 17 weeks, half the plants (N = 4) in each inoculum treatment were subjected to a ‘defoliation’ treatment on a weekly basis, which consisted of removing 50% of the leaf area of each expanded new leaf with a hole-punch. In the field, plants are attacked by leaf beetles, which feed in a shot-hole pattern and can damage up to 80% of leaf area (Núñez-Farfán & Dirzo 1994). Thus, our defoliation treatment mimics natural damage patterns reasonably well, even though hole-punches lack chemical cues provided by beetle saliva. Besides the apparent absence of alkaloid induction after damage in D. stramonium (Shonle & Bergelson 2000), there is a plethora of defence responses that are likely to be induced after herbivory damage. To reduce this source of variation, we chose artificial damage to ensure that all plants lost the same amount of leaf area. Thus, it is more likely that our artificial damage elicited tolerance responses triggered by alteration of source–sink relations through the loss of leaf area. The relationship between leaf length and leaf area (leaf area = 0.329 × leaf length²; r² = 0.98; Núñez-Farfán & Dirzo 1994) was used to estimate the absolute amount of area to be removed under the defoliation treatment. Plants were not fertilized throughout the course of the experiment.

The following variables involved in the interaction with AMF and damage were measured or estimated: foliar area, root mass, days to first flowering, seed production, total colonization, arbuscule percentage and tolerance to defoliation. Foliar area was estimated at week 12, just before plants started to flower. The same relationship between leaf length and area mentioned above was used to estimate total foliar area per plant. Following harvest, plant roots were washed to free them from soil, dried at 35 °C for 3 days and weighed using an electronic balance (OHAUS Corporation, Pine Brook, NJ, USA). To estimate total colonization, a 2-g sample of dried roots from each plant was hydrated in water overnight and stained with Trypan Blue. Colonization levels were assessed using the gridline intersection method (McGonigle et al. 1990) and c. 120 intersections per slide were recorded to give a measure of percentage root length colonized. A site was considered colonized if hyphae, vesicles, arbuscules or spores were present. Arbuscule percentage was then estimated as the number of arbuscules between total AMF colonization, thus this variable represents a fraction of the total AMF infection. Non-AMF were detected in roots at levels below 3% (A. E. Bennett, unpublished data) corresponding to expected airborne greenhouse contamination levels. Finally, to estimate tolerance, seed production was first standardized per plant genotype to control for differences in vigour among families. Then, given that defoliation was experimentally imposed, we estimated tolerance as the difference in standardized seed production between damaged and undamaged replicates for each plant genotype (Strauss & Agrawal 1999). This way of estimating tolerance avoids possible sources of variation that are usually incurred when damage is inflicted by natural herbivores (Tiffin & Inouye 2000; Lehtilä 2003). For instance, herbivore preference for particular host genotypes can generate differences in damage and thus in tolerance.

Statistical analysis

The effect of a gradient in AMF inoculum concentration on the plant response variables was analysed by anova. The model included the following predictors of performance: family, defoliation, inoculum and all interactions between these factors. Inoculum concentration was considered as a continuous effect; thus, quadratic effects were included in the analyses. All other effects were considered fixed. Root mass was root-square-transformed to improve normality. Days to flowering were analysed as a survival analysis following the Cox regression model (Cox 1972). Because the distribution of the variables of seed production, total colonization and arbuscule percentage best fit a Poisson distribution, we used a generalized linear model with the log-link function option. The analyses for these three variables were corrected for overdispersion. Because all these measurements were taken on the same replicates, a Bonferroni correction to maintain the overall experiment-wise error rate was performed. Finally, a regression analysis between inoculum concentration and plant tolerance to defoliation was performed. All analyses were carried out in jmp 7.0 (SAS Institute Inc. 2007).

Results

There was genetic variation in foliar area, flowering day, seed production and arbuscule percentage (Table 1). The defoliation treatment had a significant effect on all the variables measured except AMF colonization (Table 1). In particular, plants under the defoliation treatment had 30% more foliar area (98.7 ± 2.7 cm²) than undamaged plants (68.4 ± 2.4 cm²). This increase in foliar area was, at least in part, the result of more leaves being produced by defoliated plants relative to non-defoliated ones (F_1,238 = 13.64; P = 0.0003), suggesting compensation for foliar area lost. On the other hand, plants under the defoliation treatment had 24% less root mass (0.71 ± 0.02 g) than undamaged plants (0.93 ± 0.06 g). Damaged plants flowered on average 4 days later (88.3 ± 1.4 days after transplant) than undamaged plants (84.6 ± 1.3 days after transplant). Finally, damaged plants produced 30% fewer seeds (75.2 ± 5.5 seeds) than undamaged plants (106.8 ± 7.1 seeds).

Table 1.   Results of the anovas for the effects of genetic family, defoliation and inoculum gradient concentration on plant performance. Values followed by asterisks were significant after a Bonferroni correction (*P < 0.01, **P < 0.001, ***P < 0.0001). A significant quadratic effect of inoculum indicates a nonlinear relationship between the response variable and the gradient of inoculum concentration

Source of variation	d.f.	Plant response variables				Interaction response variables
		Foliar area	Root mass	Flowering day	Seed production	Total colonization	Arbuscules
		F	F	χ^2	χ^2	χ^2	χ^2
†Values significant at a P < 0.05.
Family	5	4.14**	2.76†	60.85***	32.96***	6.31	16.7*
Defoliation	1	42.11***	18.37***	18.39***	25.26***	5.35†	4.97
Defoliation × Family	5	1.03	1.1	7.05	14.59†	3.84	16.49*
Inoculum	1	51.84***	39.34***	27.21***	35.51***	61.17***	43.89***
Inoculum^2	1	71.29***	22.17***	47.18***	30.22***	0.34	8.71
Inoculum × Family	5	10.08***	8.53***	34.9***	31.13***	2.6	5.36
Inoculum^2×Family	5	13.15***	3.42*	56.28***	39.68***	1.42	10.49
Inoculum × Defoliation	1	2.19	16.03***	11.55**	15.49***	0.62	2.54
Inoculum^2×Defoliation	1	5.33†	7.85*	9.52†	12.28**	7.02*	4.89
Inoculum × Defoliation × Family	5	4.44***	4.91**	11.42†	12.52†	4.0	2.5
Inoculum^2×Defoliation × Family	5	5.55***	1.56	15.14*	33.67***	2.54	0.47
Block	3	14.82*	4.21*	22.93***	207.49***	2.17	0.68

Interestingly, there was variation among families in the effect defoliation had on arbuscule percentage (Defoliation × Family interaction; Table 1). There was a positive linear relationship between inoculum concentration and AMF total colonization (r = 0.50; P < 0.0001). In general, mycorrhizal fungi increased plant performance. Specifically, there was a positive relationship between inoculum concentration and foliar area, root mass, seed production and colonization but a negative one with flowering day (Table 1; Fig. 1). We also detected a significant quadratic effect of inoculum upon all variables measured except colonization (Table 1; Fig. 1). The quadratic decelerating function between inoculum concentration and plant performance indicates an optimum level of AMF concentration (167 mL of inoculum) that maximizes AMF benefit (Fig. 1). Importantly, significant genetic variation in plasticity among families in the linear and nonlinear component of the relationship between inoculum concentration and plant performance was detected for all the variables measured except colonization (Inoculum and Inoculum² × Family interactions; Table 1). Moreover, a significant interaction between AMF and defoliation was detected for all variables except foliar area and arbuscule percentage (Inoculum and Inoculum² × Defoliation interactions; Table 1), indicating that plant performance was affected by both environmental factors. In general, defoliation reduced the effect the fungi had on root mass, flowering day, seed production and total colonization (Fig. 2). In particular, plants under the defoliation treatment had, on average, less seeds across the whole gradient of inoculum in comparison with undamaged plants (Fig. 2c), suggesting that herbivore damage could diminish the beneficial effects of AMF on plant seed production. In addition, we found evidence of genetic variation in the linear and quadratic components of the norm of reaction for all variables except colonization (Inoculum and Inoculum² × Defoliation × Family interactions; Table 1; Fig. 3). Specifically, there were differences among genotypes in their performance response that are conditional on AMF inoculum concentration.

Figure 1.  Linear and quadratic effect of arbuscular mycorrhizal fungi inoculum concentration on (a) foliar area, (b) root mass, (c) flowering day, (d) seed production, (e) total colonization percentage and (f) arbuscule percentage of total colonization. In general, plants achieved their maximal performance at a concentration of 167 mL of inoculum.

Figure 2.  Differences between undamaged (continuous line) and damaged (dashed line) plants in (a) root mass, (b) flowering day, (c) seed production and (d) total colonization percentage along a fungal inoculum concentration gradient. In general, the positive relationship between inoculum and performance disappeared with defoliation.

Figure 3.  Genetic variation among plant families in plasticity for foliar area, root mass, flowering day and seed production along the arbuscular mycorrhizal fungi inoculum concentration gradient in the presence and absence of damage. Each line represents one family. Mean values and error bars were omitted.

Because there were no differences in tolerance among families when averaged across AMF inoculum concentration levels (F_4,24 = 0.72; P = 0.6149), we estimated tolerance for each level of inoculum concentration pooling the data of all plant families at the end of the experiment. These estimates were used in a regression analysis including only the linear and quadratic components of the inoculum factor. We found that the AMF concentration gradient had a significant negative linear effect (r² = −0.40; F_1,27 = 5.89; P = 0.0222; Fig. 4) and a marginally significant quadratic effect (F_1,27 = 3.48; P = 0.0728) on tolerance level expressed by D. stramonium. This result indicates a negative interaction between AMF colonization and tolerance to defoliation (i.e. below-ground–above-ground interaction).

Figure 4.  Tolerance to defoliation decreased linearly with inoculum concentration. Tolerance was calculated as the difference in standardized seed production between related damaged (D) and undamaged (U) plant families (D–U).

Discussion

The results presented here show a significant multi-species interaction between the plant response to AMF and the plant response to foliar damage. Specifically, defoliated plants experienced reduced benefits from the association with AMF. This below-ground–above-ground interaction was expressed as a negative correlation between tolerance to defoliation and concentration of AMF inoculum. It is likely that the negative interaction arises because both AMF and herbivores consume resources from the host plant. In turn, natural variations in AMF densities may also limit the evolution of optimum levels of tolerance to defoliation and may account for the maintenance of genetic variation and the presence of intermediate levels of tolerance found within plant populations (Núñez-Farfán, Fornoni & Valverde 2007). At the same time, reduction in the benefits of the mutualism between AMF and the host plant through defoliation may also condition the evolutionary outcome of this interaction.

Mycorrhizal fungi, plant performance and defoliation

Gange & Ayres (1999) proposed a simple model predicting a general curvilinear relation between colonization density and plant benefit, where benefit is maximized at some intermediate value of colonization. Accordingly, we found a curvilinear relationship between a proxy of colonization density (AMF inoculum concentration) and plant performance. All plant characteristics except flowering day increased significantly, reached a plateau (at an intermediate point) and then declined. An increase in plant performance due to AMF is the most common result reported in the literature whenever the presence/absence of AMF is manipulated (Smith & Read 1997). However, our study shows that positive associations between AMF and plant fitness may not be proportional and that at high colonization densities, mycorrhizas may also compete with plants for nutrients, immobilize nitrogen and affect root exudation and the rhizosphere microflora, all of which could lead to reduced benefits (Johnson, Graham & Smith 1997; Gange & Ayres 1999; Jones & Smith 2004). Our results indicate that benefits for the host plant from associating with AMF appear to be maximal at intermediate levels of inoculum (i.e. at an optimum density of mycorrhizal colonization). Moreover, we found genetic variation in the shape of the function response to AMF and defoliation (Fig. 3). Specifically, there was variation in the optimum level at which plant families achieved higher performance, probably because the benefit : cost ratio of the association with AMF changes nonlinearly; thus, this optimum point can evolve according to variation in the density of AMF and herbivores.

Unlike previous expectations, we found no evidence that mycorrhizal colonization provided an advantage for defoliated plants. In fact, AMF had little effect on plant performance in the presence of damage, suggesting that the presence of herbivores could limit the plant’s ability to benefit from AMF. Our results are similar to those found by Gange, Bower & Brown (2002) where AMF had no effect on plant biomass when insects were abundant, but a positive one when insects were reduced. In other words, mycorrhizal infection could be beneficial to host plants only when the herbivore load is reduced. Nevertheless, it is possible that the use of commercial AMF and manual defoliation may not accurately reflect true relationships between AMF and damage in the field. The commercial mix used in this study includes commonly found mycorrhizas and previous analyses (not shown) suggest that plants achieve similar fitness when growing with commercial and natural inoculum collected from the field (χ² = 0.03; P = 0.8691). On the other hand, because plants were subjected to artificial damage, we were not able to evaluate the full set of induced responses that would have occurred after natural herbivory. Thus, the question remains if the interaction found between AMF, plants and artificial damage is similar to that expected under natural conditions. So far, we have only shown that evaluating the effect of a gradient in mycorrhizal colonization can give us novel insights about the AMF–plant–herbivore interaction. Given that tolerance responses differ among plant species and also depend on the intensity of damage (Fornoni & Núñez-Farfán 2000; Huhta et al. 2003), future studies should address this multispecies interaction in other plant systems and under different densities of herbivores.

Mycorrhizal fungi and tolerance to defoliation

Here, we found that even high levels of AMF colonization did not reduce the negative effects of damage on plant fitness. Furthermore, it is likely that both AMF and defoliation reduced the levels of photosynthates in the host plant, decreasing not only the benefits of AMF colonization but also the capacity for compensating damage. Because the experiment was conducted outside the growing season, we cannot rule out the possibility that due to reduced light availability the stress caused by the loss of photosynthates due to AMF and defoliation could have intensified. Additionally, tolerance to defoliation decreased linearly with increasing AMF inoculum concentration, suggesting that mycorrhizas join with herbivores to further limit the plant’s access to resources. This result is in accordance with the expectation that tolerance would be greater in resource-rich environments (Compensatory Continuum Hypothesis, see Wise & Abrahamson 2005). Thus, at low density of AMF colonization plants would be able to allocate a greater amount of limiting resources towards tolerance to above-ground defoliation.

It has recently been proposed that to fully understand the evolution of plant defences, it is necessary to consider the role played by below-ground biota (Van der Putten 2003). Our study provides evidence that the adaptive value of tolerance to defoliation could be the result of interactions between a plant’s genetic background and variation in AMF colonization experienced by host plants. The findings of this study show that not only did AMF decrease tolerance to herbivory, but also that this association can potentially change the long-term dynamic of plant–herbivore interactions as a result of reduced tolerance levels in the presence of AMF. Whenever tolerance represents the only profitable strategy to cope with an increasing amount of damage (Jokela, Schmid-Hempel & Rigby 2000), AMF can reduce the adaptive value of tolerance through an ecological cost.

Acknowledgements

This study was financed by PAPIIT IN 200807 to JF and the California Agricultural Experiment Station and the College of Biological Sciences at UC Davis to SYS. We are grateful to Irene Sánchez and Julio Martínez-Romero who made the slides for assessing mycorrhizal colonization, to Pamela Riley for invaluable help in the greenhouse, and to Rubén Pérez for technical support. We also appreciate the critical and constructive comments of W. Van der Putten and two anonymous referees. E.G. is grateful to the Programa de Movilidad Internacional de Estudiantes (DGEP, UNAM), Posgrado en Ciencias Biológicas (UNAM) and CONACyT.