Study area

The experimental study area is located at the Barranco del Espartal, a seasonal watercourse in the arid Guadix-Baza Basin (Granada Province, southeastern Spain). The climate at the study area is continental Mediterranean with strong temperature fluctuations (ranging from -14°C to up to 45°C) and high seasonality (hot summers, cold winters). Annual precipitation rarely exceeds 300 mm and potential evapotranspiration is 3–4 times higher than precipitation, severely conditioning vegetation diversity and cover [41]. The soil is characterized by a sandy–loam texture, high pH, low water-retention capacity and high salinity. The vegetation is an arid open shrub-steppe dominated by Artemisia herba-alba Asso, A. barrelieri Bess. (Asteraceae), Salsola oppositifolia Desf. (Amaranthaceae), Stipa tenacissima L. (Poaceae), Retama sphaerocarpa L. (Fabaceae), Ononis tridentata L. (Fabaceae) and Lygeum spartum L. (Poaceae) [41].

The study area comprises private and governmental land and suffered different human pressure: some of the properties are used exclusively for hunting (mainly rabbits, hares and partridges) and grazing by domestic ungulates has not been permitted for at least the last 25 years. The rest of the area is open to be used by domestic ungulates (mainly sheep) and numbers has varied from 500 to 700 sheep (50 or more years ago) to around 100 sheep (since 2011 to today). In the areas to which they have access, ungulates cause a pronounced impact on annual and short-lived plant populations [42].

Model system

The short-lived Brassicaceae species Moricandia moricandioides (Boiss.) Heywood is highly abundant in the study area and was used as a model system. Moricandia moricandioides plants grow as a vegetative rosette during winter, and produce reproductive stalks during late spring [42]. Most individuals reproduce only once, but few individuals are able to resprout the next season and reproduce more than one year (less than 10% of the population). As a Brassicales, M. moricandioides is provided with potent chemical defenses, with the glucosinolate-myrosinase system being probably the most important defense [43]. In particular, seeds are expected to be one of the most well defended tissues, due to their value in terms of fitness [44].

The pre-dispersal seed predator Crossobela trinotella (Herrich-Schaffer, 1856; Lepidoptera, Gelechiidae) is among the most abundant herbivores of M. moricandioides. The attack rate of C. trinotella fluctuates between years and among populations [González-Megías et al. in prep.]. This Brassicaceae specialist oviposits on flowers or immature fruits, and the caterpillar develops inside the fruits feeding on seeds [45]. There can be more than one caterpillar per fruit, each of them eating 8–12 seeds (fruits have usually 20–60 seeds).

Experimental set-up and data collection

We selected 12 populations of M. moricandioides in the study area (S1 Fig); six populations located in areas with ungulate presence (UNG+), and six populations in areas inaccessible for ungulates (UNG-). As populations, we refer to patches of M. moricandioides isolated from other M. moricandioides patches or individuals for at least 100 m.

During July-August 2013, we selected 15 random plants from each of the 12 experimental populations. We counted the total number of fruits per plant in the field and collected five fruits of each of the 15 plants per population (900 fruits in total). Once in the lab and during autumn, we recorded the presence (PSP+) or absence (PSP-) of the pre-dispersal seed predator at both plant and fruit level (see S2 Fig flow diagram). The percentage of fruits with seed predators found during the study year was 13% for the surveyed populations.

C. trinotella was the main seed predator found in the fruits (84% of the PSPs), followed by an unidentified curculionid. Only in three fruits (0.33% of fruits) both PSP species were present together. Due to the predominance of C. trinotella, seed predator identity was not considered in the following analyses.

Seed quality

Seed quality was measured as seed mass, C and N contents and the concentrations of chemical defenses (glucosinolates) in the seed. Each of these attributes was estimated in a subset of plants (S2 Fig). We weighed groups of 10 seeds from each fruit to estimate the mean weight of a seed for each fruit (with a 0.01 mg accuracy Sartorius Cubis MSE-125P precision balance). Mean seed weight from 209 fruits from 62 plants that belonged to populations with and without ungulates, and plants with and without PSPs was estimated. Fruit sample size across treatments varied between plants (see sample sizes in S2 Fig) and at within-plants (fruits with and without PSPs from the same plant; S2 Fig).

Carbon and nitrogen contents in seeds were measured at fruit level for another set of seeds in five UNG+ and five UNG- populations of M. moricandioides. A pool of ~15 seeds from each of 197 fruits from 64 plants were analyzed (range 1–5 fruits per plant), most of the plants being the same than those used for assessing seedling emergence (see below). Fruit selection was done to ensure a minimum of 10 fruits per treatment (see between-plant and within-plant sample sizes in S2 Fig). Carbon and nitrogen contents in seeds were measured using a CHN Elemental Analyzer (Thermo Scientific Flash 2000) and the C/N ratio calculated.

Glucosinolate concentrations were quantified for another a set of seeds from five UNG+ and six UNG- populations of M. moricandioides. For glucosinolate concentrations at between-plant level, a pool of 20–25 seeds from all collected fruits of 99 plants was analyzed, many of the plants being the same than those used for assessing seedling emergence rate, and with a varying sample size across treatments (S2 Fig). For glucosinolate concentrations in fruits within the same plant, a pool of 20–25 seeds from all fruits without PSPs was compared to a pool of 20–25 seeds from all fruits with PSPs from the same plant. Due to limitations in seed number, this analysis was restricted to 19 plants from 6 UNG- populations. Seeds were ground and extracted three times in 80% methanol, adding p-hydroxybenzyl glucosinolate (sinalbin) as an internal standard at the first extraction. Glucosinolate extraction and conversion to desulfoglucosinolates as well as measurement by high performance liquid chromatography coupled to a diode-array detector (1200 Series, Agilent Technologies) were done as previously described [46]. Desulfoglucosinolates were identified by comparison of UV-spectra and retention times to those identified in earlier studies [46]. Peaks were integrated at 229 nm and response factors of 1 for aliphatic glucosinolates and 0.26 for indolic glucosinolates were considered and related to the internal standard (response factor 0.5) and sample dry weight for calculation of concentrations.

Seedling emergence

To determine the effects of ungulates and PSPs on seedling emergence rate, ten M. moricandioides populations were selected (five UNG+ and five UNG-). In each population, up to eight plants were selected according to the presence of PSPs (four PSP+ and four PSP-). For each plant, we selected up to 25 intact seeds (range 8–25) from each of the five fruits collected from the field (S2 Fig). A total of 6317 seeds were planted in black peat moss between 10 and 12-Dec-2013, in 34 seedbeds at a greenhouse with natural temperature and photoperiod conditions. Seedbeds were rotated in the same direction every other day to avoid possible location effects.

Seedling emergence was supervised every other day from the planting day until the end of January 2014 (last emergence was on 24-Jan-2014), when emergence of M. moricandioides stopped. Additionally, we calculated emergence time as the time lapse from planted to emerge.

Transgenerational effects of ungulates and pre-dispersal seed predators on seedlings: Field experiments

We carried out two experiments in the field to test transgenerational effects on seedling survival, performance and herbivory on M. moricandioides:

Statistical analyses

A) Effects of herbivores on seed quality and seedling emergence rate

There was no effect of herbivores on seed mass at neither between-plant level nor within-plant level (Table 1). Carbon content in seeds was negatively affected by ungulates at between-plant level and within-plant level (Fig 1; Table 1). There was no effect of herbivores on nitrogen content and C/N ratio at between-plant level or at within-plant level (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Effects of ungulates on seed carbon content and seedling emergence rate.

Ungulate effect on seed carbon content of Moricandia moricandioides at between-plant (a) and at within-plant level (b). Ungulate effect on emergence rate at between-plant (c) and at within-plant level (d). Mean ± SE.

https://doi.org/10.1371/journal.pone.0207553.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Ungulate and PSP effects on seed quality and glucosinolate concentrations at between-plant and within-plant levels of Moricandia moricandioides.

https://doi.org/10.1371/journal.pone.0207553.t001

There was no effect of ungulates or PSPs on the concentration of total glucosinolates, aliphatic glucosinolates, or indolic glucosinolates in seeds at between-plant level or within-plant level (Table 1).

Ungulates negatively affected emergence rate at between-plant level (Fig 1C) and within-plant level (Fig 1D; Table 2). Emergence time was not affected by ungulates or PSPs at any level (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Ungulate and PSP effects on seedling emergence and emergence time at between-plant and within-plant levels of Moricandia moricandioides.

https://doi.org/10.1371/journal.pone.0207553.t002

B) Transgenerational effects of ungulates and pre-dispersal seed predators on seedlings

Experiment 1: Between-plant level.

There was no transgenerational effect of ungulates or PSPs on the number of leaves produced per seedling or on seedling survival rate (Table 3).

There was no effect of previous generation herbivores on the number of leaves attacked in total or by chrysomelid beetles (Table 3). However, there was a negative transgenerational effect of ungulates on the number of leaves attacked by leaf miners (Fig 2, Table 3). The number of leaves attacked by grasshoppers was lower in seedlings derived from maternal plants that suffered from both herbivores, as shown by the significant interaction term (Fig 2B, Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Transgenerational effects of ungulates and PSPs on next generation seedling herbivory.

a) Transgenerational ungulate and PSP effects on attacked leaves of Moricandia moricandioides seedlings by leaf miners at between-plant level (Field Experiment 1). b) Transgenerational ungulate and PSP effects on total number of attacked leaves in seedlings at between-plant level (Field Experiment 1). c) Transgenerational ungulate and PSP effects on total number of attacked leaves in seedlings at within-plant level (Field Experiment 2). d) Transgenerational ungulate and PSP effects on total number of attacked leaves in seedlings by grasshoppers at within-plant level (Field Experiment 2). Mean ± SE.

https://doi.org/10.1371/journal.pone.0207553.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Ungulate and PSP effects on seedling performance and herbivory (seedling resistance) at between-plant and within-plant levels of Moricandia moricandioides.

https://doi.org/10.1371/journal.pone.0207553.t003

Effects of herbivores on seed quality and seedling emergence rate

The present study supports the idea that the maternal biotic environment has important consequences for the offspring. In particular, ungulates had strong and significant negative effects on M. moricandioides offspring. Ungulates effects on the mother plants significantly reduced the emergence rate of their seedlings in greenhouse conditions. This transgenerational effect of ungulates on emergence rate has been previously reported for other herbs and shrubs [22,52]. We presume that this effect may even be underestimated, as any effects of herbivores on seedling emergence are expected to be much higher under field conditions [40]. Furthermore, damage by ungulates to mother plants also significantly lessened the survival of seedlings in field conditions. Interestingly, this effect was mostly evident in plants attacked also by PSPs, suggesting that PSPs may also impinge some, albeit weak, transgenerational effects. Therefore, our results suggest that ungulates may have much larger long-term effects on M. moricandioides population than inferred from their solely effects on seed production. The current study, thus, indicates that an accurate and precise estimate of ungulate effects on plant population dynamics requires not just quantifying their intra-generational impacts but also the transgenerational consequences of the damage. When these two processes act synergistically, the long-term negative effects on plant populations are greatly amplified and cannot be deduced from studies focusing exclusively on intra-generational effects.

Many theoretical and empirical studies have found that the effects of herbivores on plant performance are stronger in short-lived herbs than in perennials [17,53]. This is logical given that reductions in fecundity for short-lived plants directly translate into a reduction in lifetime fitness, whereas the effects of herbivores on perennial plant fitness are more difficult to determine because perennials can compensate for herbivore damage across years [17]. Recruitment is crucial for the fitness of short-lived plants, as the majority of plant mortality occurs at this stage [54]. Our results on M. moricandioides, in which more than 90% of the individuals reproduce only once, support the evidence that herbivores can strongly influence the performance of short-lived plants.

Ungulates affected seed viability by decreasing seed mass in some arid and semiarid shrubs [23,25]. In our system, ungulates affected seed quality by reducing the content of carbon in the seeds, but there was no effect in seed mass or nitrogen content. This reduction in carbon content in seeds might explain the lower seed viability, and even the higher mortality of seedlings. Net carbon balance is necessary for plant growth and survival, and modest changes in carbon allocation patterns may have large consequences for seedling emergence and seedling survival [55]. Indeed, recent studies in Arabidopsis reveal that carbon-dependent signaling pathways could be ubiquitous regulators of seed germination and as important as nitrogen content and seed mass in determining germination success [56,57]. Carbon limitation in seeds could be a cost of resprouting after being grazed or trampled by ungulates, which immediately reduce the photosynthetic capacity and affect the rate of accumulation of water-soluble carbohydrates in the plant [58]. Ungulate damage does not only lead to tissue loss and thus losses of carbon, but also affects the stored reserves, which mostly consist of carbon resources [22]. Thus, ungulates limit the amount of carbon that could be allocated to seeds, because the resources used for regrowth may translate to fewer resources allocated to reproduction [59]. Our previous work also showed that floral herbivores alter carbon and nitrogen content in seeds of M. moricandioides resulting in a reduction in seedling emergence and establishment [40].

There was no prominent transgenerational effect of PSPs on M. moricandioides. In other systems, pre-dispersal seed predation reduces seedling emergence and recruitment in plant species differing in life cycle [37]. On the contrary, M. moricandioides plants attacked by PSPs overcompensated increasing seed production [Aguirrebengoa et al. under review, González-Megías et al. in prep.] Therefore, the lack of any transgenerational effect of PSPs on seeds support the idea that there is a positive net effect of PSPs on M. moricandioides and opens an interesting debate about how this presumably antagonistic interaction has evolved to become a "mutualistic" one. This type of interaction is even more difficult to explain in resource-limited environments where both the plant and the insects undergo extreme and unpredictable abiotic conditions.

Effects of ungulates and seed predators on seedling herbivory and performance

Our study demonstrates significant transgenerational and interacting effects of ungulates and PSPs on plants by affecting the herbivory experienced by the offspring of the same host plant. Moreover, the transgenerational effects varied according to the herbivore feeding mode and their specialization degree, which suggests that multiple mechanisms are involved.

There are very few examples in the literature of maternal herbivory affecting the influences of other herbivore guilds on the offspring than the inducer herbivore (see [6]). We found that when the parental generation was affected by ungulates, the offspring was less susceptible to leaf miner attack. Host specificity and adaptations to plant defenses tend to be very high in this herbivore guild [60,61], and leaf miners are also quite sensitive to leaf nutrient content and quality [62–64]. Hence, the ungulate transgenerational effect on leaf miners may be the result of the lower carbon content in seeds that could be inherited in seedlings. Analyses of seedling leaf nutrient content and defenses will be necessary to elucidate the specific mechanism behind our results.

One of the most surprising results of our study is that PSPs and ungulates had an interacting transgenerational effect on M. moricandioides seedlings. When maternal plants suffered the pressure of both herbivores (ungulates and PSPs), seedlings increased resistance against generalist herbivores (grasshoppers). The fact that only generalist herbivores but not specialists ones (leaf miners and chrysomelids) were negatively affected by this interaction suggest that the combined impact of ungulate and PSP herbivory on maternal plants triggers a defense response in seedlings, because these are usually more effective against generalist than specialist herbivores [65,66]. Several studies show that maternal herbivory can induce physical (trichomes) and chemical defenses [9,11,13].

Transgenerational defense induction can depend on the degree of predictability of future attack [67]. According to [10], transgenerational induced resistance to herbivores could be expected in those plants subjected to a non-predictable attack rate (i.e. PSPs). On the contrary, if the herbivore attack rate remains constant over time (i.e. ungulates) a constitutive resistance could have evolved. Transgenerational defense induction to reduce herbivory on offspring would therefore be expected for PSPs but not so clearly for ungulates that remain constant for decades and mainly affect plants by trampling. Indeed, ungulate constant impact could have also favored the development of tolerance that allows survival and reproduction [22,68]. Other authors suggested that seedlings of stressed plants are able to rapidly induce defenses in response to a stress similar to the one suffered by the parental plants [69,70]. Seedlings have limited structures required for resource acquisition, and thus, they might rely more on induced rather than constitutive resistance [71,72]. By which features herbivores were deterred in offspring is difficult to elucidate and needs further studies.

Biotic transgenerational effects on seeds and seedlings at within-plant level

Contrary to our prediction, we found no effect in seed traits and seedling performance of PSPs at within-plant level. Within-plant pre-dispersal seed predation effects have been observed in some trees [73,74]. A higher emergence rate of un-attacked fruit seeds due to nitrogen allocation to these fruits has been reported in Mimosa bimucronata [73]. Similarly, higher seed mass and emergence rate of un-attacked cone seeds have been observed in Juniperus thurifera [74]. All examples in the literature of within-plant variation on seed quality due to PSP pressure are from long-lived plants. Plant capacity for localized response within the plant may thus depend on life cycle and predictability of seed predator attack rates.

The most novel result arising from our study is that there was a within-plant transgenerational response to ungulates and PSPs. While some plant performance traits such as seedling emergence or survival were only affected at plant level, transgenerationally induced responses to herbivores occurred at both plant and within-plant levels, which may be related to within-plant variation in plant defense, i.e. glucosinolate induction in M. moricandioides. Defense induction in plants often occurs in the specific tissue damaged, and locally in the specific damaged part [66,75,76]. Regarding within-plant variation, a recent study shows that the patterns of DNA cytosine methylation in leaves are highly variable within individuals, and within-individual variance even surpasses the variance between individuals [77]. Through within-individual variation, plants probably better cope with the heterogeneous environment and optimize the exploitation of resources [39]. Therefore, within-plant variation in transcriptional responses may be caused by within-plant transgenerational differences in gene expression regulation or defence-inducing hormones [77].