Relationships between frugivory and seed dispersal have been observed in a wide variety of taxa. Many mammals (Howe 1986; Coates-Estrada and Estrada 1988; Herrera 1989; Fragoso 1997; Corlett 1998), birds (Howe 1986; Coates-Estrada and Estrada 1988; Barnea et al. 1992; Corlett 1998; Witmer 2001), reptiles (Rick and Bowman 1961; Van der Pijl 1982; Iverson 1985; Corlett 1998), fish (De Souza-Stevaux et al. 1994; Kubitzki and Ziburski 1994; Horn 1997), and invertebrates (Gervais et al. 1998) aid in seed dispersal, and some may decrease seed germination time (Fenner 1985). Most studies on seed dispersal have involved mammals and birds because they are considered to be more herbivorous than reptiles. However, some birds and mammals may not be the most effective seed dispersers as many may act as seed predators rather than seed dispersers (e.g., Snow 1976; Greeff and Whiting 1999).

Dispersal of seeds by reptiles, or saurochory (Van der Pijl 1982), has been investigated for many species of lizards (e.g., Iverson 1985; Traveset 1990; Valido and Nogales 1994; Corlett 1998; Greeff and Whiting 1999) and turtles (e.g., Rick and Bowman 1961; Hnatiuk 1978; Rust and Roth 1981; Rose and Judd 1982; Braun and Brooks 1987; Cobo and Andreu 1988; Milton 1992; Moll and Jansen 1995; Varela and Bucher 2002; Strong and Fragoso 2006; Guzman and Stevenson 2008; Jerozolimski et al. 2009). These reptiles have been shown to disperse a variety of seeds and play vital roles in seed germination. Examples of mutualism exist between Galapagos tortoises and Galapagos tomatoes (Rick and Bowman 1961) and between Berlandier's tortoise and Opuntia fruit (Rose and Judd 1982). In both cases germination of seeds is enhanced by passage through the tortoise gut.

Turtles, especially terrestrial species, have received more study in seed dispersal because they tend to be more herbivorous reptiles. Only one detailed study to date has been conducted on seed dispersal by an aquatic turtle species, Rhinoclemmys funerea, in Costa Rica by Moll and Jansen (1995), although Vogt and Guzman (1988) suggested that Staurotypus triporcatus may play a role in dispersal and germination of Diospyros digyna seeds. Dietary studies that have been performed on North American turtle species in which seeds have been observed either in their stomachs or fecal contents suggest that these turtles may be possible seed dispersers also (Lagler 1943; Parmenter and Avery 1990; Thomas 1993; Turner 1995).

The objective of this study was to determine seed dispersal capabilities of red-eared sliders (Trachemys scripta elegans) and common snapping turtles (Chelydra serpentina). First, we identified the seed species being consumed in nature, and then we determined seed condition after passage through the turtles' digestive tracts. We investigated the ability of turtle-passed seeds to germinate compared to seeds directly removed from the parent plant. Data were collected to determine whether these turtles could act as seed dispersers, and if so, whether they also enhanced percentage of germination of ingested seeds.

Methods

We collected T. scripta elegans and C. serpentina from May to August 2002 at a farm pond in Christian County, Missouri, using hoop nets baited with sardines. Individual turtles were rinsed to remove any superficial material and housed at Southwest Missouri State University laboratories in 42.5-L plastic tubs (58.2 × 43.2 × 26.4 cm) filled with enough water to cover the turtle's carapace. Turtles remained in these tubs for 48 hours or until a fecal sample was produced. After such time had passed, the water was filtered to obtain all fecal material. Using a low-magnification dissecting scope, we dissected feces and identified and counted all seeds present.

We collected each seed species identified in feces from plants at the farm pond. Wind-dispersed seeds (i.e., barnyardgrass [Echinochloa crus-galli] and curly dock [Rumex crispus]) were collected by gently shaking a seed head over a sieve to ensure that only ripe seeds ready to fall off the plant were obtained. Mulberry (Morus sp.) seeds were obtained directly from fruits with the aid of a dissecting scope. Seeds were stored in manila envelopes at room temperature in the dark until they were fed to turtles or used as germination controls.

Feeding trials were conducted from September through November 2002. Ten T. scripta (5 males, 5 females) and 10 C. serpentina (4 males and 6 females) were kept in tubs for at least 2 weeks after an initial fecal sample had been obtained to ensure all seeds consumed in the wild had passed through their digestive tracts. Each turtle was placed in a plastic tub (T. scripta, 42.5 L, 58.2 × 43.2 × 26.4 cm; C. serpentina 55 L, 59 × 43 × 30.5 cm) with enough water to cover its carapace. A 2-tub setup was used for each C. serpentina to prevent escape and for ease in sampling fecal material. The first tub had drainage holes drilled in the bottom and a plastic lid attached with metal wire. Drainage holes (1 cm diameter) were approximately 2 cm apart with each corner containing larger holes (3 cm); the plastic lid had a centrally located hole (4 cm diameter) for feeding and ventilation. This tub, which contained a turtle, was inserted into another 55-L tub containing enough water to cover its carapace.

Turtles were maintained on a daily diet of dog food (Pet Pride®). Turtles were fed seeds from 3 plants, mulberry (Morus sp.; n  =  25 seeds fed to each turtle), barnyardgrass (Echinochloa crus-galli; n  =  50), and curly dock (Rumex crispus; n  =  50) by placing the seeds inside small balls of ground beef (ca. 2 cm in diameter). Transit time was measured as time from initial feeding to the first appearance of seeds in the feces.

We checked turtles 3 times per day. If a bolus was present it was removed and dissected to obtain any seeds. The number of seeds present was recorded, and seeds were packaged in envelopes and stored in the dark at room temperature. Tubs were filtered once a day to obtain all fecal material and seeds and then refilled with water (ca. 25°C). Seeds were later examined under a dissecting scope and ranked as either damaged or undamaged. We defined damage as any break in a seed coat, inferring possible endocarp damage due to entry of digestive enzymes.

We conducted germination experiments from December 2002 through July 2003. The experimental design was adapted from the International Seed Testing Association (1985) guidelines for seed testing. Approximately 25 nondamaged seeds of each plant species were washed in a 20-ml 0.05% hypochlorite solution for 5 minutes. A weak hypochlorite solution was used as a fungicide and not a means to wear away the seed coat. After 5 minutes, seeds were washed twice with 10 ml of sterile water, and placed on sterilized Whatman No. 1 filter paper in sterile glass petri dishes (9 cm diameter) with each turtle's seeds contained within its own dish. Filter paper was moistened with 2 ml of sterile water and placed in an incubator with a 16-hour photoperiod. Seeds were checked daily to prevent desiccation. Germinated seeds were counted and removed every 2 days. Seeds were considered germinated when the radicle or cotyledon ruptured the seed coat.

Due to limited incubator space, not all species of seeds could be tested at the same time; therefore experiments were done in consecutive sets. Mulberry seeds were tested first, followed by barnyardgrass and curly dock seeds.

We used Morisita's index to determine dietary overlap between 3 seed species commonly found in feces from the farm pond. The following equation for Morisita's index of similarity is from Berry (1975):

Chi-square tests were used to analyze number of damaged vs. undamaged seeds passed during each feeding trial. For this test, number of damaged seeds was considered the experimental group, and undamaged seed number was the expected value. This allowed the test to be set up dichotomously where all data collected were mutually exclusive and exhausted. Chi-square was also used to compare percentage of germination of turtle-passed and control seeds. In this test the number of germinated seeds was the experimental group and number of ungerminated seeds was the expected value.

Results

Thirty turtles were captured from the farm pond (18 T. scripta, 12 C. serpentina). A majority of seeds passed were mulberry, barnyardgrass seeds, and curly dock (Table 1). Chelydra serpentina passed large quantities of mulberry seeds, many of which germinated before removal from fecal material. A Morisita's index value of dietary overlap of 0.016 was obtained for the 3 most common seed species mentioned above.

Table 1 Seeds identified in turtle fecal material from a farm pond in southwest Missouri.

View Fullscreen

Ten individuals of each species (T. scripta: 5 males, 5 females; C. serpentina: 4 males, 6 females) were selected for feeding experiments. Individuals were chosen with similar carapace lengths (T. scripta, 198.1 ± 3.54 mm SE; C. serpentina, 215.6 ± 6.20 mm SE) and mass (T. scripta, 1138.0 ± 66.30 g SE; C. serpentine, 2720 ± 224.0 g SE) so that digestive tract length would be similar.

Trachemys scripta passed a significantly larger percentage of damaged mulberry seeds (x¯  =  19.5% ± 0.07% SE) than C. serpentina (x¯  =  8.0% ± 0.03% SE) (χ², p  =  0.013), but a significantly lower percentage of damaged curly dock seeds (x¯  =  4.9% ± 0.02% SE) than C. serpentina (x¯  =  9.7% ± 0.06% SE) (χ², p  =  0.013). For barnyardgrass, T. scripta and C. serpentina passed similar percentages of damaged seeds (T. scripta, x¯  =  26.8% ± 0.03% SE; C. serpentine, x¯  =  29.8% ± 0.04% SE; χ², p  =  0.143). Variations in amount of damage obtained for different seed species did not correlate with transit times (Fig. 1; Table 2).

View Figure

• Download as Powerpoint
• Download Figure

Table 2 Spearman's Ranked tests showed no correlations between transit time and either amount of damaged seeds or amount of germinated seeds.

View Fullscreen

Turtle-passed seeds did not surpass control seeds in percentage of germination for any seed species tested. There was a significantly higher percentage of germination of control barnyardgrass seeds (n  =  251, 32.7% germination, χ², p < 0.001) and curly dock seeds (n  =  252, 81.4% germination, χ², p < 0.001) than in either T. scripta (barnyardgrass: n  =  250, 4% germination; curly dock: n  =  250, 66.4% germination) or C. serpentina (barnyardgrass: n  =  250, 14.4% germination; curly dock: n  =  247, 52.6% germination) passed seeds (Figs. 2–4).

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

Between turtle species, no dominant trend was observed in percentage of germination. Seeds from T. scripta had a significantly higher percentage of germination than those from C. serpentina with curly dock (χ², p  =  0.002) (Fig. 2), but seeds from C. serpentina had greater percentage of germination with barnyardgrass than those from T. scripta (χ², p < 0.001) (Fig. 3). All 3 treatments showed no significant difference when comparing percentage of germination of mulberry seeds (T. scripta: n  =  191, 19.9% germination; C. serpentina: n  =  229, 19.2% germination; control: n  =  250, 21.6% germination; χ², p  =  0.925) (Fig. 4). Differences observed in seed germination did not correlate with transit time (Fig. 1; Table 2).

Discussion

Both T. scripta and C. serpentina are capable of dispersing viable mulberry, barnyardgrass, and curly dock seeds. Although percentage of germination was not increased by turtle digestion, turtles were passing viable seeds. Therefore seeds could be dispersed along shorelines when turtles emerge to bask or to different areas during a turtle's many overland movements (Gibbons 1970; Moll 1994; Thomas and Parker 2000; Buhlmann and Gibbons 2001). Mulberry, barnyardgrass, and curly dock all produce seeds while T. scripta and C. serpentina are nesting. This allows females to have the greatest probability of dispersing seeds 1000 m or more away from their source (Congdon et al. 1987; Bodie and Semlitsch 2000).

Chemical processes in turtle digestive tracts seemed to have little effect on physical and physiological condition of seeds. Varying transit times during the 3 experiments did not correlate with seed condition or percentage of germination. This suggests that aquatic turtle digestive systems are mild on seeds even when additional time is spent within the digestive tract. Less damage accrued by seeds during digestion does not explain why high frequencies of mulberry seeds were present in the feces of C. serpentina, but not T. scripta during the field survey. Perhaps competitive interactions or diet choice may explain why seeds consumed in other studies, such as mulberries by T. scripta (Thomas 1993), were not consumed in large quantities here.

The effectiveness of a seed dispersal agent depends on the number of seeds being dispersed and the probability that these seeds will become reproductive adults (Schupp 1993). Trachemys scripta and C. serpentina could be considered as effective seed dispersers because they were passing large numbers of seeds (especially mulberry by C. serpentina). Feeding experiments showed that only a small percentage of seeds were damaged as they passed through the digestive tracts. Many of these nondamaged seeds were viable, with only a few treatment groups, such as mulberry seeds passed by T. scripta, having a low percentage of germination.

Consumption and subsequent seed dispersal by T. scripta and C. serpentina offer additional benefits to seeds besides dispersal alone. Seeds located in the water will have a chance of being deposited on land if they are consumed by aquatic turtles. Mulberry seeds were probably being consumed by C. serpentina as they fell into the water from a nearby tree. Trachemys scripta passed more wind-dispersed seeds, which were probably consumed accidentally as turtles fed on duckweed. Large amounts of duckweed were present in T. scripta feces (J.B. Kimmons, pers. obs., 2002), and samples of duckweed from the farm pond contained large amounts of wind-dispersed seeds.

The roles played in seed dispersal by North American aquatic turtles are only now beginning to be discovered. Only 3 seed species were tested in this study, but aquatic turtles could probably disperse a wide variety of other seed species due to the nondamaging effects of their digestive systems. Turtles captured in the field passed many other seed species during our short sampling period. Some seeds appeared in large numbers, such as buttercup (Ranunculus sceleratus), but the germination characteristics of these seeds were unable to be tested because fresh seeds could not be obtained in the field for a comparative analysis of turtle-passed and control seeds.