Pre-dispersal seed predation decreases plant progeny performance: the role of seed position within the fruit

Abstract

Background and Aims

Pre-dispersal seed predation is a critical interaction that may significantly affect species fitness. Most studies have addressed this question by focusing on seed size, while the effect of seed position within the fruit remains unexplored. We tested how seed position within fruits affects pre-dispersal seed predation likelihood and progeny performance in two dominant trees, Vachellia aroma and Gleditsia triacanthos (Fabaceae), that are causing encroachment and invasions in Argentina’s Chaco Forest.

Methods

We quantified pre-dispersal seed predation by bruchid beetles across different seed positions (basal, middle, stylar) within fruits. Seed germination and seedling vigour and survival were compared among seeds from such positions.

Key Results

Pre-dispersal seed predation was highest among seeds from the basal position in both species. In the absence of predators, basal seeds had higher survival and produced more vigorous seedlings than those from the middle and stylar positions.

Conclusions

Pre-dispersal seed predation not only reduced the quantity of seeds but also disproportionately negatively affected plant progeny performance. This selective predation may decrease seedling recruitment and thus impact population dynamics in both species. Our findings suggest that selective seed predation on seeds that would give rise to the most vigorous progeny may help regulate ecological processes such as invasion dynamics, while potentially promoting biodiversity.

INTRODUCTION

Plant population sustainability depends on the relative success of different life history stages. Sexual seed production, germination and seedling establishment are critical stages in the plant life cycle (Fenner and Thompson, 2005). Some seed traits, such as seed size, were commonly associated with germination and seedling performance. In fact, several studies have shown that larger seeds have higher germination percentages (Baloch et al., 2001; Gómez, 2004; van Mölken et al., 2005), produce seedlings with higher establishment rate (Kosiński, 2008), and have higher growth rate and survival, compared with smaller seeds (Cipollini and Stiles, 1991; Leishman et al., 2000; Gómez, 2004; Badano and Sánchez-Montes de Oca, 2022). Notably, seed germination and seedling vigour and establishment may be also affected by the position of the seed into the fruit. The effect of seed position within the fruit has been studied as a trait to assess the likelihood of seed abortion, the order of ovule fertilization, and resource acquisition (e.g. Lee, 1988; Quesada et al., 1993), but its influence on progeny performance has been rarely examined (Rocha and Stephenson, 1990; Quesada et al., 1993; Mena-Ali and Rocha, 2005a). In any case, these two traits (seed size and seed position in the fruit) can also sign the fate of seeds by affecting their likelihood to predation (e.g. Aizen, 1991; Moles et al., 2003; Hokche and Ramírez, 2018; Dylewski et al., 2020).

Seeds represent important food resources for many animal species, and several studies have shown that seed predators preferentially consume larger seeds, at both intra- and interspecific scales (e.g. Moles et al., 2003; Celis-Diez et al., 2004; Gómez, 2004; Lázaro and Traveset, 2009; Dylewski et al., 2020; but see Muschetto et al., 2015). In addition, some studies have tested the consequences to plant fitness of selective seed predation based on seed size. This evidence has allowed a global synthesis on this topic (e.g. Moles et al., 2003; Moles and Westoby, 2004; Dylewski et al., 2020). This form of selective predation is conducted primarily by post-dispersal seed predators, which can choose their resource based on size (Dylewski et al., 2020). Pre-dispersal seed predation takes place while seeds are still on the maternal plant, within the temporal window between seed production and dispersal. Thus, pre-dispersal seed predation is the first ecological filter that may affect the fitness of the progeny available to be dispersed.

Even though seed position might be a relevant trait for pre-dispersal seed predators, far fewer studies have explored the consequences of selective seed predation based on seed position within the fruit (Mena-Ali and Rocha, 2005a, b; Silveira and Fuzessy, 2015). Insects are usually pre-dispersal seed predators and their oviposition might depend on accessibility to the seeds and therefore indirectly select seed position (Boe et al., 1988; Aizen, 1991; Hokche and Ramírez, 2018). If insects access fruit via the pedicel (e.g. walking), the basal seeds of the fruit may have the highest likelihood of predation, whereas if they access via the distal zone of the fruit (e.g. flying) the stylar seeds may have the highest likelihood of predation. By consuming seeds in a non-random manner, pre-dispersal seed predators may have important consequences for plant fitness, affecting both the quantity and overall performance of the progeny available for dispersal and recruitment. Fabaceae species engage in a close interaction with a high diversity of bruchid species (Chrysomelidae, Coleoptera), which are specialist pre-dispersal seed predators (Janzen, 1980; Auld, 1983; Szentesi and Jermy, 1995). Within this family, some species exhibit a higher frequency of pre-dispersal seed predation in the basal sector of the fruit compared to the stylar sector (Boe et al., 1988; Aizen, 1991; Hokche and Ramírez, 2018), but in others species no clear pattern has been observed (Mena-Ali and Rocha, 2005b). Female bruchids lay their eggs externally on fruits, or they glue them to seeds exposed in dehiscent fruits. The larvae then bore into the seed, where they consume its content and complete their development within a seed (Janzen, 1969; Southgate, 1979; Hegazy and Eesa, 1991).

In this study we tested the effect of seed position within fruits on seed traits (seed mass and germination), seedling performance traits and the likelihood that they experience pre-dispersal seed predation. Fabaceae species with ovules arranged linearly within the ovary serve as an excellent model to test these ideas. This linear distribution may create two potential gradients within the ovary during ovule fertilization, which could influence progeny performance. One of these gradients is related to resource access within the ovary: ovules located at the basal position of the fruit (near the pedicel) are closer to the maternal nutrient source and therefore would be the first to access resources (Lee, 1988). On the other hand, ovules positioned near the entry site of pollen tubes into the ovary (usually the stylar position) are likely to be fertilized by more vigorous microgametophytes (Lee, 1988; Quesada et al., 1993). As a consequence, the relationship between the position of a seed in the fruit and its performance will vary depending on the predominant gradient (Lee, 1988). The performance of the progeny may increase from the basal to the stylar position or vice versa, or from the extremes to the middle position of the ovary, depending on whether the predominant gradient is competition for resources, microgametophytic competition or a combination of the two (Lee, 1988; Quesada et al., 1993; Delph et al., 1998).

In recent years, the Chaco Serrano Forest of central Argentina have undergone drastic changes in its vegetation cover and diversity due to the advance of the livestock farming and urban and agricultural frontiers (Aguilar et al., 2018), fires (Giorgis et al., 2013; Carbone et al., 2025), and particularly the spread of exotic plant species (Giorgis et al., 2016, 2017). In this forest, Fabaceae is one of the richest botanical families (Giorgis et al., 2011), and species of the family play a crucial role in several ecological processes. They contribute to soil stabilization and enrichment through nitrogen fixation. Their flowers supply abundant resources for pollinators, while their leaves and fruits serve as high-quality food for both wildlife and livestock (e.g. Fagg and Stewart 1994; Marquez et al., 2023). Some Fabaceae play a major role as pioneer species, shaping environments with favourable conditions for the development of other species (e.g. Peralta et al., 1992), but some of them also have major impacts on ecosystem processes and biodiversity (Fernandez et al., 2017; Guzmán et al., 2023). For this study, we selected two dominant Fabaceae species that are driving ecological and economic issues in the Chaco Forest of central Argentina: the native Vachellia aroma, which contributes to the encroachment process in areas of livestock production (Guzmán et al., 2023), and the exotic Gleditsia triacanthos (honey locust), which is an invader of several ecosystems worldwide (Richardson and Rejmánek, 2011) and is one of the most successful woody invaders of several ecosystems in Argentina. The latter is displacing native flora with negative impacts on flora and fauna diversity (Fernandez et al., 2017). Both species suffer pre-dispersal seed predation by bruchids (Ferreras and Galetto, 2010), which damage seeds more frequently from the basal position of the fruits (Aizen, 1991; L. Ashworth, pers. obs.). In this study, we aim to determine whether pre-dispersal seed predators of Fabaceae species selectively consume seeds based on their positions within the fruit to understand their consequences on plant progeny fitness (pre-dispersal fitness). Thus, we seek to understand the characteristics of the seeds being consumed: Are there differences in the percentage of seed germination and seedling vigour and survival depending on seed position? Consequently, what types of seeds are being targeted by pre-dispersal predators, and which ones escape predation? Finally, are those patterns similar in both species? Studying these processes in these key species can provide valuable insights into whether pre-dispersal seed predators can function as a biotic barrier to invasion and encroachment processes.

METHODS

Study species

Vachellia aroma (Gillies ex Hook. & Arn.) Seigler & Ebinger (Fabaceae, Mimosoideae formerly Acacia aroma Gillies ex Hook. & Arn.) is a native species of Argentina, with a South American distribution. It is a small to medium-sized tree found from southern Ecuador to central Argentina (Ebinger et al., 2000). Flowering occurs from October to December; fruits are indehiscent and ripen from February to June (Demaio et al., 2002; Funes et al., 2007). Pre-dispersal seed predation is carried out mainly by the bruchid Pseudopachymerina grata (Coleoptera; Chrysomelidae, Bruchinae; Ferreras and Galetto, 2010). Females of P. grata oviposit on mature fruits of V. aroma, the egg hatches, the larva bores into the fruit wall and the seed coat, and once into the seed it consumes the cotyledons and the embryo. After several larval stages the adult emerges from the seed leaving a typical conspicuous exit hole. Reportedly, a single bruchid emerges per seed (Aizen, 1991).

Gleditsia triacanthos L. (Fabaceae, Caesalpiniaceae) is native to the USA and introduced to Argentina during the first half of the 19th century. It is a large tree, and an extremely important invader in lowland areas near watercourses in different eco-regions of Argentina (Di Iorio, 2005; Ferreras and Galetto, 2010). Flowering occurs between October and December; fruits are indehiscent, they begin to ripen in March and they are dispersed approximately from June to September (Ferreras et al., 2023). Megabruchidius tonkineus (previously identified erroneously as Bruchidius endotubercularis, Coleoptera; Chrysomelidae; Di Iorio, 2015a) is the main pre-dispersal seed predator of G. triacanthos while three species of Lepidoptera: Amyelois transitella, Ectomyelois ceratoniae (both species belonging to Pyralidae, Phycitinae) and an unidentified Tortricidae species were occasional pre-dispersal seed predators that damaged few seeds (Ferreras and Galetto, 2010). In G. triacanthos, seeds with predated cotyledons can germinate (Horvat and Sajna, 2021), but seedlings usually die due to fungus damage (A. E. Ferreras, unpubl. data).

Study site

This study was carried out in the Chaco Serrano Forest in central Argentina (Córdoba province). This is a xerophytic forest located in the slope of hills from 500 to 1350 m a.s.l. (Giorgis et al., 2011). The area belongs to the Chaco phytogeographical region, a seasonal subtropical semi-arid forest (Giorgis et al., 2017). Average annual precipitation is 734 mm, and most rain falls in the warmest months (from the end of September to April). Mean temperature is 24 °C in summer and 10 °C in winter (De Fina, 1992). The vegetation of the region is composed of a mosaic of physiognomies in different successional stages, ranging from close to open woodlands, shrublands and, to a lesser extent, pastures induced by fire and grazing (Giorgis et al., 2017). Furthermore, over 16 % (28 of 171) of tree and shrub species in the Chaco Serrano Forest in Córdoba province are exotics (Giorgis et al., 2011).

We selected three sampling sites where native and exotic invader species co-occur and the forest has similar vegetation composition and structure (Ferreras et al., 2014): ‘La Rancherita’ (31°45′34.58″S, 64°27′34.19″W), ‘La Serranita’ (31°44′18.01″S, 64°26′55.85″W) and ‘Villa Los Aromos’ (31°43′24.7″S, 64°26′30.4″W) (Fig. 1A, B).

Fig. 1.

Location of the study sites and sampling design. (A) Map of Argentina showing the province of Córdoba and the location of the study area. (B) Zoom of Cordoba province showing the spatial distribution of the three study sites. (C) Scheme of sampling design for seedling performance traits of Vachellia aroma and Gleditsia triacanthos. Seed positions: Basal, Middle and Stylar refer to seeds located at the extreme of the fruit near the pedicel, seeds located in the middle portion of the fruit and seeds located at the extreme of the fruit, close to the style, respectively.

Sampling design

During 2008, we selected 42 trees encompassing V. aroma and G. triacanthos (N = 7 mother trees per species per site). To protect developing fruits from seed predators, we covered five fruits of each tree with tulle bags when they were beginning to develop (small and green fruits) and we collected them when they became mature (brown and woody). Subsequently, we collected ten mature fruits from each tree: five fruits exposed to seed predators and five fruits isolated from potential threats (resulting in a total of N = 420 fruits). Fruits exposed to predators were used to quantify seed predation while the isolated fruits were reserved for assessing plant progeny performance (Fig. 1C, scheme of sampling design). Fruits were individually stored in a paper bag and then placed in separate plastic containers according to whether they were exposed to seed predators or isolated. Fruits were maintained in our laboratory at room temperature for a month, allowing seed predators to emerge, and to be easily captured and identified, based on our previous experience. During that period, we checked for the emergence of seed predators once a week and captured them in order to reduce the probability of re-infection of seeds.

Relationship between seed position and seed predation

To test if seed predation varies among seeds from different positions in the fruit, we manually opened the five fruits per maternal tree and extracted and examined each seed carefully. To quantify the percentage of seed predation per seed position in the fruit we first categorized the seeds according to their position inside the fruit as: (1) Basal (Ba) – seeds located at the extreme of the fruit near the pedicel, (2) Middle (Mid) – seeds located in the middle portion of the fruit and (3) Stylar (Sty) – seeds located at the extreme of the fruit, close to the style (Fig. 1C). It is easy to delimit the position of the seeds in the fruits since both species have seeds ordered lineally. For each tree, we pooled all the seeds from the five fruits from the same position and calculated the proportion of seed predation per tree as: no. of of predated seeds in z position/no. of total seeds in z position.

As seed predators do not completely destroy the seed coat, it is easy to quantify the total number of seeds produced per seed position per tree (i.e. predated and non-predated seeds). Predated seeds can be easily identified by the exit hole left by the bruchids or bruchid parasitoids (Supplementary Data Fig. S1). Additionally, we used a water floating method to identify infested seeds whose predator/parasitoids had not yet emerged: damaged seeds float slightly while healthy ones sink completely. To validate this method, we examined a subsample of 20 seeds per species with X-rays. The results confirmed with 99 % accuracy that floating seeds contained predators or parasitoids, while non-floating seeds were healthy. Moreover, X-rays enabled the identification of the type of insect inside the seeds, distinguishing between bruchids and their parasitoids (Fig. S2).

Relationship between seed position and progeny performance

To analyse whether progeny performance varies with seed position, we quantified seed mass, germination, seedling survival and seedling vigour. For this, we manually opened the five isolated fruits from each mother tree. In each fruit, we randomly selected one healthy seed per position. We used five fruits per tree giving five seeds per position per tree in total (Fig. 1C; total 315 seeds used per species = 7 trees × 3 sites × 3 positions × 5 seeds; replicates per position per species 21 = 7 trees × 3 sites). These seeds were used for the seed germination and seedling experiments. For the latter experiment we grew additional seedlings under the same conditions described below, serving as a backup to ensure five seedlings per seed position per tree for measuring seedling height, relative growth rate (RGR) and leaf number. This precaution was taken in case some seeds failed to germinate or seedlings did not survive during the assay. While most seeds successfully germinated and survived, any shortfall was addressed by randomly selecting extra seedlings from the backup pool to maintain a total of five seedlings for each seed position per tree.

To test if seed mass is related to the position of seeds within the fruit, we weighed the previously selected five seeds per seed position per tree. Each seed was weighed individually on a precision balance (Mettler Toledo PB153 scale). To quantify the percentage of seed germination for each seed position, seeds were previously scarified with sandpaper to overcome physical dormancy (Ferreras et al., 2015). All seeds from an individual mother tree were placed in a Petri dish (i.e. a total of 21 Petri dishes per species; with 15 seeds each). Each Petri dish was divided into three sections to clearly identify the five seeds corresponding to the basal, the middle and the stylar positions. Seeds were placed on filter paper and moistened with distilled water every 5 d. Dishes were kept in germination chambers (12-h photoperiod, light density around 40 µmol m^–2 s^–1, temperature 25 °C, Ferreras and Galetto, 2010). The germination trial lasted 30 d, and seeds were considered germinated upon radicle emergence. The percentage of seed germination per seed position per tree was calculated as: no. of germinated seeds in z position/5 seeds (the replicates for seed position were the trees, N = 21).

For quantifying seedling survival and seedling vigour per seed position, all germinated seeds were transplanted into pots with a mixture of two parts of soil and one part of sand. Transplantation was performed 2 d after germination. Seedlings were grown in a glasshouse with natural photoperiod, sprinkler irrigation for 2 min once a day and a temperature of 25–30 °C. Seedling height, number of leaves and RGR, at 30 and 90 d after seedling transplant, were used as seedling vigour traits. Seedling height was measured with a digital caliper from the cotyledon (or cotyledon scars) to the base of the newest leaf in the apical meristem. We counted the number of completely expanded leaves. Seedling height was used to calculate RGR (cm d^–1) per seedling as: (LnH2 − LnH1/T2 − T1), where LnH1 and LnH2 are the natural logarithm of seedling height measured after 30 (T1) and 90 d (T2), respectively (T2 − T1 = 60; Hunt, 2012). Seedling survival per seed position per tree was quantified after 90 d from transplantation to pots as: no. of survival seedlings in z position/5 seedlings (the replicates for seed position were the trees, N = 21).

Progeny fitness as a function of seed position

To estimate progeny fitness per seed position at the pre-dispersal stage, we calculated two multiplicative indices for each mother tree. The pre-dispersal fitness index was calculated as the proportion of seed germination × proportion of seedling survival at 90 d × proportion of non-predated seeds (Nelson and Johnson, 1983). As a comparative measure, we also calculated a potential pre-dispersal fitness index, which reflects progeny performance without considering pre-dispersal seed predation. This index was calculated as the proportion of seed germination × proportion of seedling survival at 90 d. The last index allows us to determine what the progeny’s fitness would have been in the absence of pre-dispersal seed predation.

Statistical analysis

Statistical analyses were performed with R software (R Core Team, 2022). For each species, to evaluate the effect of seed position within fruits on seed predation, seed traits (seed mass, germination) and progeny performance traits (seedling height, number of leaves, RGR, survival and fitness indexes), we used generalized linear mixed models (GLMMs) with seed position within fruit as the fixed effect (three levels) and site (i.e. three sites) and mother tree (i.e. 21 individuals) as random factors. For variables with a Gaussian distribution of errors (seed mass, seedling height, RGR, and pre-dispersal and potential pre-dispersal fitness), we used the lmer function from the lme4 package. For variables with a binomial distribution of errors and a logit link function (e.g. proportion of predated seeds, proportion of seed germination and proportion of seedling survival) and for those with a Poisson distribution (number of leaves), we used the glmer function from the lme4 package (Bates et al., 2015). The significance of the fixed effect was tested by comparing the global model (including the fixed effect) with nested models excluding the fixed effect, using a likelihood ratio test (LRT); the same methodology was used to test the significance of random effects. The significance of random effects (site and mother tree) is given in Supplementary Data Table S1. Model fit was assessed using residual diagnostics in the DHARMa package (Hartig, 2022) (Fig. S3). When the fixed factor was found to be significant, a post hoc Tukey’s test was performed to evaluate differences between levels of the fixed factor (using the glht function from the multcomp v.1.4-7 package, Hothorn et al., 2016). All statistical analyses were conducted separately for each species.

RESULTS

Relationship between seed position and seed predation

The number of seeds per fruit varied from six to 16 in V. aroma and from ten to 31 in G. triacanthos and the average number of seeds per fruit (±s.d.) was 8.52 (±1.82) and 21.4 (±3.86) respectively. For V. aroma the average number of seeds predated per fruit was 2.30 ± 1.80 (27 ± 11 %) while for G. triacanthos the average number of predated seeds per fruit was 6.79 ± 5.15 (31 ± 13 %). For both species, the percentage of seed predation increased from the stylar to the basal section of the fruit and differed among positions, being higher in the basal position, and similar between the middle and the stylar position (Fig. 2). Regarding random factors, significant variation in seed predation was observed only among mother trees of G. triacanthos (Supplementary Data Table S1).

Fig. 2.

Effect of seed position within fruits (Ba: basal, Mid: middle, Sty: stylar) on the percentage of seed predation (mean ± s.d.) of Vachellia aroma (A) and Gleditsia triacanthos (B). Different letters above bars indicate significant differences among seed positions according to Tukey’s posteriori tests, P < 0.05.

The main seed predators of V. aroma and G. triacanthos were the bruchids Pseudopachymerina grata and Megabruchidius tonkineus, respectively. Moreover, parasitoids of P. grata (Horismenus sp. Hymenoptera; Eulophidae) and M. tonkineus (probably Allorghas sp.) also emerged from some seeds (Supplementary Data Figs S1 and S4). The exit hole made by parasitoids was much smaller than the hole made by bruchids (Fig. S1).

Relationship between seed position and progeny performance

The mass of the seeds in V. aroma was 0.049 ± 0.012 g (mean ± s.d.), and it did not significantly differ among positions. Conversely, in G. triacanthos the mass of seeds from the basal and middle positions did not differ, but seeds from the stylar position were the lightest (Table 1). Seed mass varied significantly among mother trees in both species (Supplementary Data Table S1). The percentage of seed germination was high for both species [V. aroma: 85.85 % (±7.57), mean (±s.d.); G. triacanthos: 73.33 % (±3.81)], and it did not differ significantly among seed positions in either of the species (Table 1). In V. aroma the survival of seedlings differed significantly among the three positions, increasing from the stylar to the basal position (Table 1). In G. triacanthos seedling survival also differed significantly among positions: seedlings originated in the basal position had higher survival rates than those from the remaining two positions, and survival of seedlings from the middle position did not differ from those from the stylar position (Table 1). For both species, seed germination and seedling survival showed no significant variation among mother trees and sites (Table S1).

Table 1.

Effect of seed position within fruits (Basal, Middle, Stylar) on seed traits and seedling performance traits (mean ± s.d.) of Vachellia aroma and Gleditsia triacanthos: seed mass (g), percentage of seed germination, percentage of seedling survival, seedling height (cm), number of leaves and relative growth rate (RGR). Different letters indicate significant differences among seed positions according to Tukey’s posteriori tests, P < 0.05.

Traits	Vachellia aroma					Gleditsia triacanthos
	Basal	Middle	Stylar	Stat	P	Basal	Middle	Stylar	Stat	P
Seed mass (g)	0.053 ± 0.011a	0.048 ± 0.013a	0.047 ± 0.012a	χ2 = 4.107	0.128	0.193 ± 0.011a	0.191 ± 0.013a	0.168 ± 0.012b	χ2 = 16.072	<0.0001
Seed germination (%)	91.428 ± 16.213a	88.889 ± 17.111a	77.222 ± 17.758a	χ2 = 3.654	0.161	77.143 ± 17.071a	73.333 ± 20.331a	69.522 ± 26.547a	χ2 = 1.581	0.454
Seedling survival at 90 d (%)	82 ± 27.453a	73 ± 35.703b	72 ± 19.894c	χ2 = 8.324	0.016	87.619 ± 13.381a	55.238 ± 29.601b	49.524 ± 24.182b	χ2 = 33.935	<0.0001
Seedling height at 30 d (cm)	4.900 ± 1.294a	3.910 ± 0.718b	3.600 ± 0.995b	χ2 = 16.539	<0.0001	11.600 ± 1.899a	9.520 ± 2.479 b	9.38 ± 2.103b	χ2 = 20.472	<0.0001
Seedling height at 90 d (cm)	16.696 ± 1.913a	13.207 ± 2.219b	13.183 ± 3.086b	χ2 = 23.201	<0.0001	20.589 ± 2.446a	16.640 ± 3.275b	16.341 ± 2.767b	χ2 = 32.291	<0.0001
No. of leaves at 30 d	9.65 ± 1.268a	8.15 ± 1.387b	7.7 ± 1.380b	χ2 = 20.346	<0.0001	8.952 ± 1.117a	7.667 ± 1.110b	7.667 ± 1.064b	χ2 = 22.608	<0.0001
No. of leaves at 90 d	25.370 ± 2.387a	22.210 ± 2.791b	21.860 ± 3.980b	χ2 = 14.934	<0.0001	13.290 ± 1.533a	11.95 ± 1.796b	11.240 ± 1.389b	χ2 = 15.299	<0.0001
RGR	0.012 ± 0.002a	0.012 ± 0.001a	0.012 ± 0.002a	χ2 = 1.576	0.455	0.006 ± 0.001a	0.006 ± 0.001a	0.006 ± 0.002a	χ2 = 0.102	0.95

Seedling height and mean number of leaves at 30 d differed significantly among seed positions following the same pattern in both species. Seedlings from the basal position were taller and had more leaves than those from the middle and stylar positions, and seedlings from the middle and stylar position did not differ in these traits. This pattern persisted for both species at 90 d (Table 1). Seedling RGR did not differ significantly among seed positions for V. aroma or for G. triacanthos (Table 1). For both species there were significant variations in seedling height (at 30 d but not at 90 d), number of leaves (at 30 and 90 d) and RGR among mother trees (Supplementary Data Table S1).

Progeny fitness as a function of seed position

For V. aroma, pre-dispersal fitness did not differ among the basal, middle and stylar positions (Fig. 3). However, potential pre-dispersal fitness (in the absence of seed predators) was significantly higher at the basal position, intermediate at the middle and lower at the stylar position (Fig. 3). For G. triacanthos, pre-dispersal fitness was significantly higher in the basal position, intermediate in the middle and lower in the stylar position (Fig. 3). Potential pre-dispersal fitness was also significantly higher in the basal position, but no differences were found between the middle and stylar positions (Fig. 3). Regarding random factors, significant variability in pre-dispersal and potential pre-dispersal fitness were observed among mother trees in both species (Table S1).

Fig. 3.

Potential pre-dispersal fitness (proportion of seed germination × proportion of seedling survival at 90 d, solid line) and pre-dispersal fitness (proportion of seed germination × proportion of seedling survival at 90 d × proportion of non-predated seeds, dashed line) of progeny of Vachellia aroma (A) and Gleditsia triacanthos (B) according to seed position within fruits (Ba: basal, Mid: middle, Sty: stylar). Values are mean ± s.d., and different letters above bars indicate significant differences among seed positions according to Tukey’s posteriori tests, P < 0.05.

DISCUSSION

Pre-dispersal seed predation plays a crucial ecological role in naturally regulating plant population growth. This interaction can be harnessed as a biological control mechanism for problematic plant species (Busch et al., 2012; Fernandez et al., 2017). We observed a consistent pattern in the two studied Fabaceae species: the likelihood of seed predation increased from the stylar to the basal positions within fruits. This pattern has been previously observed in V. aroma and G. triacanthos (Aizen, 1991; L. Ashworth, pers. obs.), as well as in other Fabaceae species (Hokche and Ramírez, 2018). Our main finding is that progeny performance mirrored the seed predation pattern, i.e. the vigour and survival of the progeny increased from the stylar to the basal position, showing that bruchids are targeting those seeds that would otherwise produce the highest performance seedlings compared to those from the other positions. Bruchids probably access the fruit by the pedicel and prey more frequently on the first seeds they reach, i.e. those in the basal position (Aizen, 1991). To the best of our knowledge, this is the first study showing that pre-dispersal seed predators may modify the performance of the progeny by predominantly damaging seeds from a specific section of the fruit. Indeed, previous studies have found similar effects of seed predators on progeny performance, but they were focused on seed size not on seed position (Nelson and Johnson, 1983; Cipollini and Stiles, 1991; Moegenburg, 1996). Finally, since the same pattern was observed in both the native and non-native species, propagule pressure and post-dispersal events may play a critical role in the encroachment of V. aroma and the invasive potential of G. triacanthos (Ferreras and Galetto, 2010; Guzmán et al., 2023).

Differences in the performance of the progeny produced in different positions within the fruits may result from a differential maternal resource allocation to the developing embryos, microgametophytic competition or a combination of the two (Rocha and Stephenson, 1991; Delph et al., 1998). In V. aroma, where the pollen dispersal unit is the polyad, resource competition is more likely to explain the observed differences rather than pollen competition. One polyad contains 16 pollen grains and the ovary has 11–17 ovules (Ashworth, 2004; Casiva et al., 2004). Given that the pollen ovule ratio is near 1:1, pollen competition is unlikely. Furthermore, genetic studies have shown that only one polyad fertilizes all the ovules in the ovary and therefore within a fruit all seeds are full-sibs (Casiva et al., 2004). This fact further rules out the likelihood of competition among different pollen donors. Gleditsia triacanthos belongs to the Caesalpinieae subfamiliy, where the pollen grains are released in monads (Banks et al., 2010), thus in this species pollen competition is possible and competition for resources cannot be ruled out. Our results on seed mass suggest a resource provisioning gradient of the developing seeds, with provisioning increasing from the stylar to the basal fruit position in both species. In addition, to disentangle the effect of pollen and resource competition on progeny performance, specific studies controlling pollination time, pollen quantity and number of pollen donors are needed (see Delph et al., 1998).

In both species, in the absence of predators, the proportion of seeds potentially capable of producing seedlings (potential pre-dispersal fitness) and the vigour of these seedlings (height and number of leaves) were higher at the basal position compared to the middle and stylar positions. After considering seed predation (pre-dispersal fitness), the differences in fitness between seed positions disappeared, with this pattern being more evident in V. aroma than in G. triacanthos. Thus, by consuming in particular the progeny with the highest potential survival, bruchids homogenized the fitness of the progeny across seed positions. Through this predation pattern, the bruchids decreased the overall performance of the progeny available to be dispersed. Studies on pre-dispersal seed predation have shown that this ecological interaction can reduce seedling recruitment and plant population growth rate (reviewed by Kolb et al., 2007). Interestingly, our results suggest that it is not just a reduction in progeny quantity, but rather a selective elimination that specifically compromises the progeny capable of producing the highest-performing seedlings.

For most seed and seedling traits we found significant variation among mother trees but not across sites, suggesting early maternal effects on these traits during early development. We also found variability in seed predation among mother trees of G. triacanthos. This suggests that trees may vary in their chemical and physical defences against seed predators or their attractiveness to bruchids (e.g. fruit quantity), among other traits (Jones and Comita, 2010; Lima et al., 2016).

It is necessary to consider that other ecological interactions that occur during the seed dispersal or post-dispersal stages may modify the final progeny performance. The most common seed dispersers for the studied species are mammals, cattle for both and foxes for V. aroma (Ferreras and Galetto, 2010; Fernandez et al., 2017). It has been observed in other Fabaceae species that bruchids could be killed by ingestion by herbivorous mammals and seeds might escape from predation if they are eaten in early stages of bruchid infestation (Coe and Coe, 1987; Or and Ward, 2003). Thus, bruchids from both studied plant species are probably killed after passing through the digestive tract of cattle, mitigating their impact on seeds. In the case of foxes, a preliminary study showed that bruchids were not killed after passing through their digestive tract, allowing beetles to survive and continue their life cycle (Di Iorio, 2015b). Thus, foxes serve as effective dispersers for seeds (Varela and Bucher, 2006), and bruchid beetles for V. aroma (Di Iorio, 2015b). Additionally, a low level of post-dispersal seed predation by rodents has been observed in G. triacanthos compared to other native species in Pampas grasslands (Busch et al., 2012; Muschetto et al., 2015, 2022). Such differences may be due to rodents in these grasslands being generally small in size, while G. triacanthos seeds are larger than those of native species. It has been suggested that the large size or the harder seed coat of G. triancanthos may acts as a barrier for these animals (Muschetto et al., 2015). Thus, by preying more heavily on native seeds than on exotic seend, rodents may be less effective as biological control agents than insects, which damage both native and exotic seeds at a similar rate (this study). However, studies testing pre- and post-dispersal seed predation in both native and non-native species within the same ecosystem are essential to fully understand their combined impact on plant fitness and biological control.

Finally, previous studies have shown that seed predators may drive ecological processes with important impacts at the community level (e.g. Agrawal et al., 2012). More vigorous progeny may have higher competitive ability and greater survival in harsh environments compared with less vigorous ones (Westoby et al., 1996; Moles and Westoby, 2006). Thus, by specifically consuming the most vigorous progeny, bruchids may model the competitive ability of plant populations (Agrawal et al., 2012).

Ecological and management implications

Pre-dispersal seed predation has been proposed as a biological control for reducing the invasiveness of plant species (Busch et al., 2012; Fernandez et al., 2017). In the Chaco Forest G. triacanthos is an exotic, highly invasive species (Ferreras and Galetto, 2010). It is generally thought that invasive species might escape seed predation (Keane and Crawley, 2002), particularly that caused by specific seed predators. However, as shown previously, G. triacanthos establishment seems to be affected by seed predators, even after the fruits are released from the plant (Ferreras and Galetto, 2010; Fernandez et al., 2017). Thus, even though fruit production and seed predation rate might vary from year to year (Kolb et al., 2007; Ferreras et al., 2014), the predation of seeds that might produce the most vigorous seedlings may mitigate the severity of G. triacanthos invasion and the encroachment processes of V. aroma. Manipulative experiments that exclude pre-dispersal seed predators and assess the effect of cattle on bruchids will be crucial to elucidate the role of bruchids in controlling plant population dynamics and invasion processes and the role of cattle in controlling bruchids and indirectly plant population dynamics.

FUNDING

This work was supported by Fondo para la Investigación Científica y Tecnológica (BID-PICT-2019-01897). A.L.C. is a fellowship holder of Fondo para la Investigación Científica y Tecnológica; A.F. and L.A. are researchers from CONICET.

SUPPLEMENTARY DATA

Supplementary data are available at Annals of Botany online and consist of the following.

Figure S1: Photographs of healthy and predated seeds from Vachellia aroma and Gleditsia triacanthos, with visible exit holes made by bruchid beetles and their parasitoids. Figure S2: X-ray photographs comparing healthy and infested seeds. Figure S3. Goodness-of-fit plots and diagnostic tests for all analysed models and response variables. Figure S4. Photographs of seed predators and parasitoids recorded for both Fabaceae species. Table S1. The significance of random effects (site and mother tree) based on likelihood ratio tests (LRTs).