A common hazard to seed survival is fungal rot. A variety of fungi attack fruits and seeds (e.g., Harman 1983, Cipollini and Stiles 1993b) and have direct effects on seed survival and germination (e.g., Kirkpatrick and Bazzaz 1979, Christensen 1989) as well as indirect effects mediated by decreasing removal by ant and avian dispersers (Cipollini and Stiles 1993a, Crist and Friese 1993). In a Panamanian rain forest, seeds in mammalian defecations and embedded in fruit experienced more fungal infection than seeds cleaned of these substrates (M. Jones, pers. obs.). Both beetles and ants remove seeds from some vertebrate defecations (Byrne and Levey 1993, Estrada and Coates-Estrada 1991, Wicklow, et al.1984). By removing seeds from mammalian defecations and fruits, ants and other insect seed removers may increase seed survivalby decreasing the frequency of fungal rot, thereby providing an"escape" service parallel to vertebrate dispersal of seeds away from high mortality associated with parent plants (Augspurger1983, Schupp 1992).
Although ants have been cited among the principalinsect removal agents for small seeds (Hölldobler and Wilson 1990), beetles may also effectively move small seeds. In particular, dung beetles often remove seeds in the process of forming dung balls from mammalian defecations (Estrada and Coates-Estrada 1986,1991). Since dung beetles are not principally interested in the seeds, there may be different effects of removal by beetles versus ants. Dung beetles usually do not consume seeds from defecations and so more seeds may escape intact when compared to seeds removed by ants (Hanski 1991). Beetles may, however, bury seeds too deeply for effective germination (Estrada and Coates-Estrada 1991). Thus, if dung beetles and ants differentially move seeds from defecations versus other sources (e.g., fruits), there may be different fates for seeds that primary dispersers initially disperse by defecating, by regurgitating, and by dropping.
Even though removal by insects serves a dispersal function for some seeds, many seeds may not benefit from secondary dispersal. Some ant species collect seeds and disperse some of them while transporting them to their nest; however, once at the nest they may consume all of the seeds. Many seeds have external bodies that remain intact after passage through the digestive system of a primary disperser, attract ants, and promote dispersal (Pierce and Cowling 1991, Byrne and Levey 1993, Gibson 1993b). Removal of this elaiosome generally has no effect on seed viability or is beneficial (D. Levey, unpub.data, Lobstein and Rockwood 1993). The exact effect of the plant-insect interaction probably depends on several factors, including plant species, remover species, and method of primary dispersal.
In this experiment I tested the hypothesis that secondary seed removal and subsequent viability vary with species of secondary remover, method of primary dispersal, and plant species. In particular, I examined secondary seed removal and seed viability for two tropical plant species, Ficus maxima and Ossaea quinquenervia. For F. maxima, I presented seeds to a natural litter ant community in several potential substrates and examined the effect of the substrate on removal and seed viability. For O. quinquenervia, I presented the seeds to a natural litter ant community and examined the balance between dispersal and predation, and the effect of ants on seed viability.
I examined secondary seed removal by litter insects and its effect on germination success for Ficus maxima (Moraceae) and Ossaea quinquenervia (Melastomataceae). Ficus maxima (mean seed length = 2.28 mm) and O. quinquenervia (mean seed length = 0.45 mm) fruits are consumed by birds and mammals that disperse the seeds by defecation, by regurgitation, or by dropping fruit parts under the parent plant (Worthington 1982, M. Jones, pers. obs.). Ossaea quinquenervia seeds have a triangularly-shaped, external body (approximately.45 mm height, .12 mm base) that is attached to the seed and is cream colored in contrast to the reddish-brown seed. This external body remains intact after passage through the digestive tract of at least one tropical frugivorous bird, the green honeycreeper(Chlorophanes spiza, M. Jones, pers. obs.). After primary dispersal, seeds may be removed by ants and other insects. On BCI, litter ants are extremely diverse and abundant (> 127 species in 49 genera; Levings 1983) and several are potential seed harvesters. Levings and Franks (1982) reported high densities of generalist myrmicine (0.28 nests/ m2) and ponerine (0.39 nests/m2) ants that are potential seed harvesters on BCI (Hölldobler and Wilson 1990). In particular, Ectatomma ruidum (Ponerinae), Pheidole spp. (Myrmicinae), and Wasmannia spp. (Myrmicinae) were common ants with densities of 0.34, 0.38, and 0.09 nests/m2, respectively (Levings and Franks 1982). Each of these species may be seed harvesters (Horvitz and Schemske 1986b, Roberts and Heithaus 1986, Byrne and Levey 1993). Fifty-nine species of Scarabaeidae dung beetles are found on BCI, including species that tunnel beneath or near the site of dung defecation and species that roll away dung balls (Gill 1991). All of these species are potential seed removers because they inadvertantly remove seeds with the dung that they take, and small seeds are more likely removed than large seeds (Estrada and Coates-Estrada 1991).
For O. quinquenervia, I tested whether insect access (access or control) affected removal. I cleaned seeds of all fruit pulp and presented 600 seeds (mean number seeds per fruit = 659, s.d. = 76, n = 21) in each of five replicates for each ant access and control treatment.
For both plant species, I placed the seeds in 35 mm diameter petri dishes and I excluded vertebrates from the seeds by securing 40 x 25 x 10 cm hardware cloth (1.27 cm mesh) cages over the stations. For the control dishes, ants and other crawling insects were excluded by Tanglefoot® (The Tanglefoot Company, Grand Rapids, Michigan, USA) barriers that surrounded the petri dish containing the seeds. All stations were covered by a 30 x 20 cm plexiglass roof placed 5 cm off the ground to prevent rain from splashing the seeds out of the station.
I placed the stations at the intersections of a grid that was 15 m wide and 20 m long, with 5 m between each axis. This design produced a 4x5 grid, for a total of twenty stations. For F. maxima, one of the three dispersal substrates was randomly chosen for each intersection in the grid, and both an insect access dish and a control dish were placed under the roof at that station. Thus, each of the first 15 stations contained a pair of dishes (access, control) for a randomly chosen dispersal substrate in the Ficus maxima experiment. Each of the last 5 stations contained a pair of dishes (access, control) in the Ossaea quinquenervia experiment.
I monitored seed removal for 2 days, collecting vouchers of insect species found in the petri plates. Previous studies have shown that most seed removal by ants occurs during the first two days (M. Jones, pers. obs., Byrne and Levey 1993). Each station was observed during two 15 min censuses in the morning and two in the afternoon for the two days of the experiment. During each census, I recorded the maximum number of insects of each species present in the petri plate simultaneously, the number present at the midpoint of the census interval, and the number of seeds removed by each insect species. I carefully searched the area within 25 cm of the petri and I scanned the area within 50 cm for seeds that had been dispersed by ants and other insects, and I collected seeds that I found. Also, if ants were seen removing seeds, I followed them back to the nest and searched the area surrounding that nest for dispersed seeds, collecting the seeds. I was unable to relocate seeds that were buried underground by beetles or that were taken into ant nests, so all comparisons in the germination experiments apply only to seeds that I could relocate.
For Ossaea quinquenervia, I analyzed the removal data using insect access (access, control) as the single factor in an ANOVA, with the cumulative proportion of seeds within each station removed as the dependent variable. I analyzed the germination data using external body presence (control, body intact, body removed) as the single factor in an ANOVA. I used the proportion of seeds germinating for the dependent variable.
Figure 2-1. Proportion (mean ± S.E.) of Ficus maxima seeds removed by beetles and ants when presented in three dispersal substrates (feces, fruit, seeds only). Gray bars indicate dishes to which insects has access and hatched bars show dishes from which insects were excluded.
Significantly more seeds were removed from access dishes than from control dishes (F1,24 = 25.00, p = .0001). Beetles removed a few seeds from the control dishes in the feces and fruit substrates because beetles were attracted to these substrates (but not to seeds without a substrate) and were able to fly or crawl over the Tanglefoot® barriers. Significantly more seeds were removed from feces than from fruit substrate and no substrate dishes (F2,24 = 3.87, p = .0348). This is most likely because dung beetles (Canthidium sp.) were attracted to the feces but not to the fruit substrate nor to seeds without a substrate. Beetles often arrived within 30 s of seed placement and made dung balls that contained most of the embedded seeds. Beetles then removed the balls from the station and buried them in tunnels that they had dug. Seeds removed by these beetles were easily relocated because the tunnels were generally within 10 cm of the station. On average, I recovered 23.8 (53.4%) seeds of those dispersed from the defecation stations. Most of these seeds had been separated from the dung and were found within 1 to 2 mm of the soil surface in loose soil turned over by beetles. Because all dung was removed by beetles from stations to which they had access, seeds left behind in these dishes were cleaned of dung.
Seed removal from the fruit substrate and no substrate dishes was principally due to ants. Ants, mainly Trachymyrmex sp., Wasmannia sp., and Ectatomma ruidum, often recruited to the seeds within fifteen minutes of seed placement (Table 2-1), but only Trachymyrmex sp. and Ectatomma
| F. maxima | O. quinquenervia | ||||||
|---|---|---|---|---|---|---|---|
| Mean | Max. | N | Mean | Max. | N | ||
| All ants | 0.50 | 0.92 | 48 | 0.53 | 1.23 | 43 | |
| Ectatomma ruidum | 1.33 | 2.67 | 3 | 1.00 | 1.00 | 1 | |
| Trachymyrmex sp. | 2.33 | 3.83 | 6 | 0.33 | 1.44 | 9 | |
| Wasmannia sp. | 0.75 | 2.00 | 4 | 1.58 | 3.17 | 12 | |
| Cyphomyrmex sp. | 2.00 | 4.00 | 1 | -- | -- | -- | |
| All beetles | 1.50 | 2.00 | 8 | -- | -- | -- | |
| Canthidium sp. | 2.00 | 2.00 | 1 | -- | -- | -- | |
| Other | 1.43 | 2.00 | 7 | -- | -- | -- | |
ruidum took seeds out of the stations. Some of these seeds were dropped outside of the station and others were returned to ant nests, generally within 150 cm of the station. Many seeds that were brought into ant nests were found distributed in a ring around the nests, apparently undamaged, on subsequent days. I recovered on average 2.6 (12.4%) and 3.8 (17.0%) seeds that had been dispersed from the fruit and seeds only substrates, respectively. Seeds left behind in the fruit dishes generally remained in fruit when collected at the end of two days.
Seeds that remained in controls, particularly those in feces and fruit substrates, began to show fungal growth by Rhizopus sp. during the two days of the removal experiment (G.Gilbert, pers. comm.).
Fungal growth continued for the week-long incubation period, but only remained on seeds in feces in control dishes from which insects were excluded. Three of these five stations showed substantial fungal growth, while no other treatments did.
Seed germination followed a sigmoidal trajectory for all treatments, with most of the germination occurring within an 8 day period. In the first ANOVA, significantly fewer seeds germinated from control dishes than from insect access dishes (Figure 2-2a; F1,20 = 6.85, p = .0165). In addition, the interaction between insect access and dispersal substrate was significant (F2,20 = 3.67, p = .0439); that is, seeds in feces had significantly lower germination when they were from control dishes than when they were exposed to insects, whereas seeds presented in fruit or without a substrate showed no such effect.
Figure 2-2. Cumulative proportion (mean ± S.E.) of Ficus maxima seeds germinated when presented in three dispersal substrates (feces, fruit, seeds only), A) in dishes to which insects had access (gray bars) and from which insects were excluded (hatched bars); and, B) for dispersed (gray bars) and non-dispersed (hatched bars) seeds when insects had access to the seeds.
This difference was associated with fungal growth only on feces, which presumably damaged the seeds in the feces. When beetles took fecal material from the insect access stations they left the remaining seeds free of feces, and hence of fungus.
In the second ANOVA, which compared germination between dispersed and non-dispersed seeds from stations with insect access, no factors or interactions were significant (F1,19 = 1.55, p = .228). Mean germination for dispersed seeds was consistently lower than for non-dispersed seeds (Figure 2-2b) and lack of significance may have been due to a lack of power, since dispersed seeds were difficult to relocate.
Ossaea quinquenervia seeds were removed significantly more from insect access dishes than from control dishes (Figure 2-3; F1,19 = 7.60, p = .0130). Several ant species discovered seeds in the stations, including Trachmyrmex sp., Wasmannia sp., and Ectatomma ruidum (Table 2-1), but only Trachymyrmex sp. were observed removing seeds.
Figure 2-3. Proportion (mean ± S.E.) of Ossaea quinquenervia seeds removed from dishes to which insects had access and from which they were excluded.
I was unable to relocate seeds after removal by ants due to the extremely small size of the seeds. Some ants, principally Wasmannia sp., removed the external bodies from seeds in the seed dishes, leaving behind both intact seeds and seeds missing the external body. Germination of intact seeds and control seeds (immediately taken from fruit) was significantly greater than germination of seeds from which the external body had been removed by ants (Figure 2-4; F2,18 = 504, p = .0001).
Figure 2-4. Cumulative proportion (mean ± S.E.) of Ossaea quinquenervia seeds germinated for intact and control seeds and for seeds from which ants removed the external body.
Secondary seed removal by ants and beetles was high and variable for F. maxima and O. quinquenervia. My experiments demonstrated that seed removal patterns differ for ants and beetles, that germination patterns differ for seeds removed by ants and beetles, that ants in the same microhabitat can be both seed predators and dispersers, depending on plant species, and that some insects do provide escape services analogous to primary dispersal.
These results emphasize the importance of fungi as causes of seed mortality. Although previous studies fully recognized the role of fungi in fruit rot (Cipolliniand Stiles 1993a, 1993b) and seedling mortality (Augspurger 1983,Augspurger and Kelly 1984), the ecological relevance of fungal attack on seeds has received less attention (Crist and Friese1993). In this study, fungal infection was more pronounced in feces, and fewer seeds that remained in feces were viable than seeds that had been separated from feces.
The importance of seed mortality because of fungi also demonstrates that secondary seed removers can indeed provide important dispersal services similar to primary dispersers (Howe and Smallwood 1982, Howe 1989). For F. maxima, beetle removal of the dung is effectively dispersal, regardless of whether the seed is moved from the dish, because the seed escapes from the harmful environment of the feces and subsequent fungal attack. Therefore, beetles provided an escape service which is analogous to escape provided by primary dispersers, such as escape from pre-dispersal predation (Dirzo and Domínguez 1986, Howe and Smallwood 1982, Willson 1992). Neither ants nor beetles moved seeds over long distances in this study (10 - 150 cm), so it is unlikely that these secondary dispersers help seeds to colonize new habitats as some primary dispersers do (Brokaw 1986, Schupp, et al. 1989). Finally, both ants and beetles may move seeds to beneficial microsites, but I did not test this hypothesis. Both the seeds I recovered from the surface and seeds brought into ant nests and dung beetle tunnels may have had higher germination rates than those that were not moved by secondary dispersers. Further studies should examine the effect of dispersal and mode of primary dispersal on germination and seedling establishment in situ.
Ants and beetles were attracted to different dispersal substrates and removed F. maxima seeds in different patterns. Seed removal by beetles from feces was clearly an indirect effect of their harvesting dung. Nevertheless, they dispersed many more seeds than ants, and at least some dispersed seeds were separated from the dung balls after removal from the seed dishes (not all beetle-dispersed seeds were recovered). Some seeds dispersed by beetles and the seeds left behind in the dishes by beetles were free of dung. As a consequence, fungal attack, which persisted on feces in control dishes from which beetles were excluded, was largely absent from both dispersed and non-dispersed seeds from the insect access dishes. Access by dung beetles thus reduced fungal attack on seeds regardless of whether the seeds were dispersed because dung was removed from the seeds.
Ants removed F. maxima seeds from both fruit substrate and no substrate dishes approximately equally, but avoided the howler monkey defecations. In contrast, ants readily removed seeds from mammal defecations (Roberts and Heithaus 1986) and from bird defecations (Byrne and Levey 1993) in previous studies. Although Roberts and Heithaus (1986) proposed that Ficus seeds lack any special attractant for ants, Kaufmann et al. (1991) showed that F. microcarpa has a fleshy, lipid filled exocarp that may be an attractant for ants. Such an attractant may also be present on F. maxima. Seeds removed by ants were recovered in areas surrounding ant nests and in the paths between the station and the nests, as in previous studies (Horvitz and Schemske 1986b, Roberts and Heithaus 1986). In general, ants dispersed seeds over larger distances than beetles (10 - 150 cm vs. 0-20 cm) and dropped seeds in less of a clumped pattern. Thus, effective seed densities of the seeds I recovered after ant dispersal were lower than after beetle dispersal.
These differences in F. maxima removal by beetles and ants resulted in differences in seed viability. Germination was high ( > 90%) for seeds in fruit and without substrate, regardless whether insects had access to seeds. In contrast, seeds that remained in feces in control dishes had much lower viability than those that were cleaned of dung by beetles that had access to seeds. Thus, ants had no effect on F. maxima viability and beetles substantially improved F. maxima viability by reducing the effect of fungi.
Although ants removed a large proportion of O. quinquenervia seeds, they also consumed the external bodies from many of the seeds in situ. Seed viability was reduced to almost zero for those seeds from which external bodies were removed. This is inconsistent with studies of other melastomes that have external bodies (D. Levey, unpub. data). With such effective mortality of non-dispersed seeds, it is difficult to assess whether dispersal by ants would provide any benefit. If seeds are transported to the nest, they are likely to die because ants are likely to remove the external bodies before discarding the seed. If they are dropped along the way to a nest, however, the seed may have an improved chance of survival by being moved away from the dense clump of its siblings in a bird defecation or fallen fruit (Howe 1989, Loiselle 1990). Indeed, O. quinquenervia produces many fruits with many small seeds (~ 650 per fruit), resulting in a large fruit crop each year. The external body may attract ants to the seeds and promote removal from an intensely competitive environment (see Loiselle 1990). The parent plant may benefit from the rare event in which a seed is removed but dropped along the way rather than brought all the way back to the nest. Although many studies of myrmecochory have shown that elaiosome removal has no effect on seed viability and that seeds are discarded after the elaiosome is removed (e.g., Auld 1986, Lobstein and Rockwood 1993), others have shown that seeds bearing elaiosomes are harvested as much for seed consumption as for elaiosome consumption (Pierce and Cowling 1991).
Removal rates and seed viability are affected by harvester species, mode of primary dispersal, and plant species. Secondary dispersal can indeed be analogous to primary dispersal in that beetles provide escape from the extreme hazards of fungal attack for F. maxima. For O. quinquenervia, in contrast, seed removal by ants and removal of the external body by ants may not have served any dispersal function at all. Because the effect of ants on seed viability strongly depended on the plant species being tested, it may be difficult to generalize the significance of plant-insect interactions across plant species.
[Abstract] [Chapter 1] [Chapter 3] [References] [My Home Page]