Mechanisms affecting the fate of Prosopis flexuosa (Fabaceae, Mimosoideae) seeds during early secondary dispersal in the Monte Desert, Argentina
Abstract
Abstract The fate of seeds during secondary dispersal is largely unknown for most species in most ecosystems. This paper deals with sources of seed output of Prosopis flexuosa D.C. (Fabaceae, Mimosoideae) from the surface soil seed-bank. Prosopis flexuosa is the main tree species in the central Monte Desert, Argentina. In spite of occasional high fruit production, P. flexuosa seeds are not usually found in the soil, suggesting that this species does not form a persistent soil seed-bank. The magnitude of removal by animals and germination of P. flexuosa seeds was experimentally analysed during the first stage of secondary dispersal (early autumn). The proportion of seeds removed by granivores was assessed by offering different types of diaspores: free seeds, seeds inside intact endocarps, pod segments consisting of 2–3 seeds, and seeds from faeces of one herbivorous hystricognath rodent, the mara (Dolichotis patagonum). The proportion of seeds lost through germination was measured for seeds inside intact endocarps, seeds inside artificially broken endocarps, and free seeds. Removal by ants and mammals is the main factor limiting the formation of a persistent soil seed-bank of P. flexuosa: >90% of the offered seeds were removed within 24 h of exposure to granivores in three of four treatments. Seeds from the faeces of maras, on the other hand, were less vulnerable to granivory than were other types of diaspores. These results suggest that herbivory might be an indirect mechanism promoting seed longevity in the soil (and likely germination) by discouraging granivore attack. On the other hand, germination did not seem to have an important postdispersal impact on the persistence of P. flexuosa seeds in the soil. Both direct and indirect interactions between vertebrate herbivores and plants may foster P. flexuosa's seed germination in some South American arid zones.
Introduction
In most ecosystems, a major proportion of newly produced seeds are lost from the surface soil seed-bank during secondary dispersal or redistribution (Chambers & MacMahon 1994). Major mechanisms that account for such losses are germination, consumption by animals, attack from microorganisms and deep burial. Hence, for a plant species to form persistent soil seed-banks, its seeds must not only be capable of some type of dormancy, but also be able to cope with the several postdispersal challenges imposed by mechanisms other than germination.
In the central Monte Desert of Argentina, annual forb seeds appear to form persistent soil seed-banks, whereas perennial grass seeds, shrub seeds and tree seeds form transient soil seed-banks, with most seed output occurring 1–6 months after production (i.e. in autumn and winter; Marone et al. 1998a). The strong decline of grass seeds in the top 2 cm of Monte Desert soil (Marone et al. 1998a) seems to be a consequence of bird and, secondarily, mammal consumption (approximately 50% of grass seeds are lost to granivores), deep burial (30% are lost to burial, including all tiny seeds), germination (<5%), and pathogen attack (Marone et al. 2000b). The postdispersal fate of the larger shrub and tree seeds in the Monte Desert remains largely unknown, despite the strong decline seeds suffer as soon as they land on the ground (Marone et al. 1998a).
Prosopis flexuosa D.C. (Fabaceae, Mimosoideae) is the most abundant tree species in the central Monte Desert, and its seeds are among the biggest in this ecosystem (24–40 mg). Prosopis flexuosa production is highly variable from year to year, with records of 80 000–800 000 seeds per hectare in different areas of the Monte Desert (Ffolliot and Thames 1983; Dalmasso and Anconetani 1993). Despite such a high potential input, only a few P. flexuosa seeds were found in the soil seed-bank of the central Monte Desert during the winters and springs of 1993 through 1998 (Marone et al. 1998a; L. Marone, unpubl. data). This period had at least two events of high fruit production (the summers of 1995 and 1998; L. Marone, pers. obs.). Occasional high seed production along with the absence of a persistent soil seed-bank suggests that at least some mechanisms of P. flexuosa seed loss were effective during that period.
Until now it has been generally assumed that germination and predation by insects of the family Bruchidae are the main fates of seeds of the genus Prosopis (Tschirley & Martin 1960; Smith & Ueckert 1974; Lerner & Peinetti 1996; Ortega Baez et al. 2001). However, germination is probably not an important cause of seed loss during redistribution. For example, Lerner and Peinetti (1996) reported that less than 10% of P. caldenia seeds germinated under natural conditions in central Argentina, whereas Tschirley and Martin (1960) reported that 16% of hulled P. velutina seeds and 45% of seeds planted in pod segments germinated after the first year and a few more during the second and the third years. Regarding seed predation, there is convincing evidence of a high predispersal impact of bruchids, which can decrease Prosopis seed production by 25–70% (Smith & Ueckert 1974; Solbrig & Cantino 1975; Kingsolver et al. 1977; Agrawal 1996). On the other hand, the impact of postdispersal predators on Prosopis seeds is largely unknown. Lerner and Peinetti (1996) found that bruchids infested approximately 35% of the P. caldenia seeds they arranged experimentally on the soil (protected against predation) during the spring following production. Ortega Baez et al. (2001) found that 99% of P. ferox seeds were predated by bruchids after 6 years on the soil in a subtropical mountain desert. Solbrig and Cantino (1975) suggested that ant predation would not be high for P. flexuosa seeds in the northern Monte Desert, although they did not report figures to support their assertion. The great number of animals that are capable of eating Prosopis fruits and seeds (foxes, medium-sized and small mammals, ants) indicates, however, that predation might be an important force in preventing such tree species from forming persistent soil seed-banks in arid areas of southern South America.
Prosopis seeds are frequently dispersed by domestic and wild animals (Peinetti et al. 1993; Campos & Ojeda 1997). Its fruits have morphological and biochemical adaptations to animal dispersal. These adaptations include indehiscent pods with a thin exocarp, a spongy and sweet mesocarp (with a high proportion of sugars and proteins), and a hard and leathery endocarp. At maturity, the seed is free inside a cavity in the endocarp and a hard seed coat prevents water uptake (Kingsolver et al. 1977). The endocarp and seed coat seem to prevent mortality of embryos when seeds pass through the digestive tracts of herbivores (Peinetti et al. 1993; Campos & Ojeda 1997). Although seed dispersal by animals is often interpreted as a classical model of direct animal–plant mutualism (e.g. herbivores obtain energy and foster germination capacity by breaking dormancy in the digestive tract), some indirect consequences of seed consumption and dispersal may be equally important. For example, Lerner and Peinetti (1996), following Janzen (1969), suggested that dispersal of P. caldenia seeds by vertebrates might favour seed longevity in the soil because the passage of seeds through the intestinal tract of vertebrates would eliminate previous infection or discourage new infection by beetles of the family Bruchidae. Likewise, a plausible hypothesis is that seed consumption and dispersal by vertebrate herbivores favours seed longevity in the soil by reducing seed vulnerability to granivores.
We measured the seed loss of P. flexuosa due to germination and predation during the early postdispersal stage (a few weeks after primary dispersal) in the central Monte Desert of Argentina. We investigated which mechanism is more important in preventing this species from forming a persistent soil seed-bank. We also assessed whether seeds coming from the faeces of the hystricognath rodent Dolichotis patagonum (mara) are as vulnerable to granivory as are seeds coming directly from the parent plant, thus testing the hypothesis of an indirect positive interaction between herbivores and plants.
Methods
Study site
The Biosphere Reserve of Ñacuñán (67°58′W, 34°02′S) is located in the central part of the Monte Desert, Argentina (Morello 1958). Grazing has been excluded from the reserve since 1972. One of its most conspicuous communities is an open woodland, with a sparse tree layer dominated by Prosopis flexuosa within a shrub matrix of Larrea divaricata, Condalia microphyla and Capparis atamisquea (tall shrubs), and Lycium spp., Verbena spp. and Accantolippia seriphioides (low shrubs). Grass cover, composed almost exclusively of perennial species, reaches 25–50% (Pappophorum spp., Trichloris crinita, Aristida spp., Digitaria californica, Setaria leucopila, Sporobolus cryptandrus). The climate in the central Monte Desert is dry and temperate, with cold winters. On average, 75% (250 mm, n = 25 years) of the annual rainfall occurs in spring and summer (October–March), when seeds are produced (Marone et al. 1998a). The present study was carried out during the exceptionally wet spring–summer of 1997–1998 (453 mm), associated with a strong El Niño/Southern Oscillation event. Fruit production by P. flexuosa was relatively high in the summer of 1998, with most dispersal occurring during January–February (L. Marone, pers. obs.). Many fruits were available at the site during the course of the experiment.
Estimating seed removal
The magnitude of seed losses caused by predation was assessed by using a bait-removal experiment, using a totally randomized design. We measured seed removal from piles of seeds after 24 h exposure to granivores and took a repeated measure of the remaining seeds a week later. This design allowed us to compare both the final amount and the rate of seed removal among treatments.
Treatments consisted of four types of diaspores: (i) free seeds; (ii) seeds inside intact endocarps; (iii) pod segments consisting of 2–3 seeds; and (iv) seeds inside intact endocarps extracted from fresh faeces of maras. For the fourth treatment, we chose seeds that had been eaten by maras because Campos and Ojeda (1997) found that in the central Monte Desert seeds dispersed by this rodent reach maximum germination capacity. Fruits of P. flexuosa were collected in the field immediately before starting the experiments, in March 1998. Free seeds and seeds in endocarps were manually extracted from ripe fruits. The presence of seeds in pod segments was verified by shaking the pods.
Diaspores were offered in plastic trays (12 cm long, 8.5 cm wide, 1 cm deep), which were completely buried in the ground and filled with soil taken from the same site. For every treatment, a set of diaspores containing 20 seeds was superficially placed on each tray, and 40 replicates were arranged over three 0.2-ha plots, approximately 1 km apart, in the open woodland. Within each plot, the trays were placed at least 5 m apart. The trays were arranged over two microhabitats (areas under the canopy of trees or tall shrubs, and exposed areas in between) but seed removal did not differ between microhabitats, so we pooled all trays for subsequent analysis.
We recorded all direct or indirect signals of seed foraging activity around the trays, ensuring that our observations did not disturb granivores while assessing the trays. Our aim was to detect the main animal species that consumed the offered seeds. Although we used seed removal mainly as a surrogate for seed predation or loss from the surface seed-bank, some of the potential predators could have only relocated, rather than consumed or eliminated, a proportion of the seeds removed. For example, ants of several species remove Prosopis seeds and take them to their colonies. At least some of these ants (e.g. Pogonomyrmex and Acromyrmex) occasionally deposit P. flexuosa seeds in granaries, which are buried as deep as 40 cm in the soil (S. Claver and L. Marone, pers. obs.). This activity results in effective loss from the seed-bank, but we do not know the actual proportion of P. flexuosa seeds that were lost from the habitat in this way. Only a few South American rodent species have been shown to be capable of hoarding seeds, always under laboratory conditions (Vásquez 1996). However, no rodent species in the Monte Desert has cheek pouches and, consequently, their hoarding capacity is expected to be lower than that of other mammalian species (e.g. the heteromyids of North America; Marone et al. 2000a). We paid especial attention to interpreting signals of rodent foraging activity around the trays once our experiments had finished. We often found the remains of P. flexuosa seed coats along with rodent tracks, which indicates that some of the seeds were consumed by small mammals directly from the trays (P. Villagra, pers. obs.).
Estimating seed germination
To assess the magnitude of seed germination, we used a totally randomized design. We considered three types of diaspores: (i) seeds inside intact endocarps; (ii) seeds in broken endocarps; and (iii) free seeds.
The seeds used in all treatments were chemically scarified by a 2-min immersion in concentrated sulfuric acid (Ffolliot & Thames 1983). This technique was used because of the large percentage of hard seeds found in this species, and because we consider that this type of chemical scarification is the most similar to natural scarification in the digestive tract of herbivorous animals. Previous studies have found germination percentages reached by acid-scarified seeds to be more than 90%, whereas those of non-scarified seeds reached only 6–25% (Catalán & Balzarini 1992; M. Cony, unpubl. data). To avoid predation, seeds were incubated in 0.5 mm-mesh cotton bags (10 cm × 10 cm). Each bag contained 20 seeds. Bags were randomly distributed over three 0.2-ha plots in the open woodland and buried 1–2 cm deep in the soil. The assay began on 20 March 1998, 4 days after a rainfall of 58.4 mm. No more rainfall occurred during the experiment. Percentages of germinated seeds (those with radical emergence >1 mm) were recorded 7 days after incubation. Twenty bags were proportionally arranged over the same two microhabitats as the removal experiment and, again, given that no differences were found between them, we pooled all the trays for subsequent analysis.
Statistical analyses
Removal percentages were subjected to a one-way anova for repeated measures, and germination percentages to a one-way anova. As data did not meet the assumption of homogeneity of variance, predation percentages were transformed using ranks (Conover & Iman 1981), whereas germination percentages were reciprocally transformed (y = 1/(x + 0.05)). Tukey tests were used for a posteriori mean comparisons. Mean differences were considered to be significant at P < 0.05.
Results
Seed loss due to removal by granivores
Almost 95 and 100% of the offered seeds were removed in three of the four treatments after 24 h and 8 days, respectively (Table 1). The seeds from mara faeces, however, were removed less and at a lower rate than any other diaspore (Tables 1,2). The removal rate of free seeds was also lower than that of seeds inside endocarps and pod segments, as indicated by their lower removal values on the first day of the experiment (Tables 1,2).
| Removal after 24 h (%) | TT | Removal after 8 days (%) | TT | |
|---|---|---|---|---|
| Pod segments | 92.1 (3.9) | a | 100 (0) | a |
| Inside endocarp | 99.1 (25.9) | a | 100 (0) | a |
| From Dolichotis faeces | 76.5 (34.6) | c | 93.5 (17.5) | b |
| Free seeds | 91.0 (16.8) | b | 99.4 (2.3) | a |
- Values represent the mean (standard deviation). ‘TT’ indicates the results of Tukey's test for detecting significant differences in removal percentages between types of diaspores on each date (different letters indicate significant differences at P < 0.05).
| Sources | d.f. | MS | F | P |
|---|---|---|---|---|
| (A) Type of diaspore | 3 | 73273 | 16.99 | <0.0001 |
| (C) Date | 1 | 79380 | 44.14 | <0.0001 |
| A × C interaction | 3 | 9430 | 5.24 | <0.0017 |
| Error | 152 |
Several species of ants (e.g. Acromyrmex lobicornis, Pheidole bergi and Forelius sp.) were observed removing all types of diaspores. Some of these ants were seen to carry the diaspores to their nests and then discard the endocarps onto refuse piles around the nests. We also observed abundant evidence of seed predation by mammals which, unlike ants, appeared to consume at least a proportion of the seeds near the trays, discarding both endocarps and seed coats. We found no evidence of seed consumption by birds.
Seed loss due to germination
Despite the fact that seeds were scarified, almost no germination was observed in the treatment involving seeds inside intact endocarps (Tables 3,4). Germination percentages increased as seed protection decreased due to scarification, reaching the maximum values in free seeds (Tables 3,4).
| Germination after 7 days (%) | TT | |
|---|---|---|
| Complete endocarp | 1 (2.6) | a |
| Broken endocarp | 4.6 (5.9) | b |
| Free seeds | 42.4 (30.9) | c |
- Values represent the mean (standard deviation). ‘TT’ indicates the results of Tukey's test for detecting differences between types of diaspore. Different letters indicate significant differences (P < 0.05).
| Sources | d.f. | MS | F | P |
|---|---|---|---|---|
| (A) Type of diaspore | 2 | 968 | 31.7 | <0.00001 |
| Error | 57 |
Discussion
Our results coincide with those of other recent studies, which propose that seed consumption is an important ecological force in some warm South American deserts (Marone et al. 2000a). In the Monte Desert, Marone et al. (1998b, 2000b) indicated that autumn–winter granivory may be the main mechanism governing grass seed loss from seed-banks. Granivory might also be a major mechanism affecting the fate of P. flexuosa seeds during redistribution.
Both ants and mammals seemed to be major removers and consumers of P. flexuosa seeds. Claver (2000) reported that P. flexuosa fruits and seeds are, in autumn months, among the main food items of Acromyrmex lobicornis in the Ñacuñán Biosphere Reserve. The workers of Acromyrmex lobicornis, Pheidole bergi and Forelius sp. actively transported the offered pods and seeds during the present experiment. Among vertebrate consumers, mammals took many of the offered seeds and consumed some of them in the vicinity of our trays. Birds were not observed to consume the seeds. This was not surprising, because avian granivores, most of them in the family Emberizidae, eat seeds weighing 0.06–0.60 mg, whereas a seed of P. flexuosa weighs 24–40 mg (Marone et al. 1998b). Only two bird species have been recorded while consuming P. flexuosa seeds, the Burrowing Parrot (Cyanoliseus patagonus) and the Monk Parakeet (Myiopsitta monacha), although they appear to consume unripe seeds on trees more frequently; these species would therefore be regarded as mainly predispersal predators (F. Roig, M. Cony, F. Milesi & P. Villagra, pers. obs.).
Unlike removal by predation, seed loss because of germination was very low, even after chemical scarification. Only free seeds achieved moderate germination, whereas those in endocarps or portions of fruits, the way seeds are more frequently found in the soil under natural conditions, remained dormant. These results are consistent with the low seed loss due to germination reported for other Fabaceae trees of semi-arid regions such as P. caldenia (Lerner & Peinetti 1996) and Acacia albida (Hauser 1994).
Postdispersal seed loss caused by other agents, such as bruchids or other pathogens, should also be low in most years. This is because the high rate of predation by ants and small mammals as soon as seeds reach the soil could exhaust seed reserves before the attack of such organisms. For example, Lerner and Peinetti (1996) observed that infestation of P. caldenia seeds by bruchids in the soil seed-bank, when experimentally protected against predators, occurred only 6 months after primary dispersal, during the following spring. Given the high rate of seed removal reported here, postdispersal bruchid infestation on seeds in the soil might be an extremely infrequent event in P. flexuosa in the central Monte.
Our results reveal a subtle mechanism that may be involved in the mutualistic relationship between Prosopis trees and some herbivorous animals in arid regions. The consumption of fruits by maras appeared to directly and indirectly increase the likelihood of germination of P. flexuosa seeds. Directly, this occurred through the breaking of dormancy when seeds pass through the digestive tract of herbivores (Peinetti et al. 1993; Campos & Ojeda 1997). Campos (1997) found that P. flexuosa seeds coming from the faeces of maras and cows had higher germination capacities than did seeds from faeces of other domestic or wild herbivores of the Monte Desert. Coincidentally, mara and cow faeces contained the highest proportion of free seeds (Campos 1997) and, as shown in Tables 3 and 4, free P. flexuosa seeds were more likely to reach germination than any other of the diaspores studied. Thus, the most efficient dispersers of P. flexuosa seeds might be those herbivores such as maras and cows, whose digestive tracts release some proportion of the seeds from endocarps without injuring the embryos.
However, herbivory by maras might also act indirectly to foster the germination success of P. flexuosa seeds by reducing seed vulnerability to granivorous animals (Tables 1,2). Longer times in the soil should provide more opportunities for germination in arid zones. If, as our results suggest, seed predation is an important selective pressure for Prosopis trees (Tables 1,2), the digestive process of some herbivores might indirectly contribute to the restoration of Prosopis populations by increasing seed longevity in the soil due to a reduction in palatability to, or even detectability by, granivorous animals (Miller 1993).
In summary, P. flexuosa seeds suffered a very high granivore postdispersal pressure in Ñacuñán, such that most seeds are removed only a few days after reaching the soil. Germination of seeds within endocarps, the type of diaspore that prevails during secondary dispersal, appeared to be very low under natural conditions. These results suggest that the lack of a persistent P. flexuosa soil seed-bank may be largely a consequence of seed removal and consumption by granivorous animals. However, there was evidence for an indirect positive interaction between some herbivores and P. flexuosa seeds, which enhances germination. Seed persistence in the soil would be increased through reduced granivore attack on those seeds that had been previously consumed and dispersed by herbivores.
Acknowledgements
We thank J. Lemes, B. E. Rossi, G. Zalazar and N. Horak for their kind collaboration. This paper was partially supported by CONICET and Agencia Nacional de Promoción Científica y Tecnológica of Argentina. This is contribution number 20 of the Desert Community Ecology Research Team (Ecodes), UF & EV, IADIZA Institute.
