Introduction

Do forest herbs benefit from fruit consumption by multiple frugivore species? Endozoochory is common in temperate deciduous forest herbs, probably because it is well suited to dispersal of large seeds (Whigham 2004), because seeds move long distances relative to other dispersal modes (Matlack 1994a), and ingestion potentially enhances germination and establishment (Traveset et al. 2007). Most endozoochorous herb species appear to be dispersal generalists consumed by several, or perhaps many, frugivore species in a diffuse mutualistic interaction (e.g., Levey et al. 2006; Loiselle et al. 2007; Brodie et al. 2009). Feeding behaviors and patterns of defecation vary considerably among frugivore species, potentially leading to differences in seed deposition and probability of seedling establishment (Traveset et al. 2007; Schupp et al. 2010). The reproductive success of the parent plant can be viewed as the cumulative result of service provided by each of the individual frugivore species. It is unclear, however, to what extent differences in frugivore life histories and feeding habits actually translate into detectable dispersal effects because multi-vector systems have rarely been examined from the perspective of plant reproduction.

The net result of dispersal by multiple vectors may theoretically be complementary, supplementary, or subtractive relative to a single vector. Two or more frugivore species potentially complement each other by depositing seeds in different microhabitats (Thayer and Vander Wall 2005; Jordano et al. 2007; García-Robledo and Kuprewicz 2009; Amato and Estrada 2010) thereby broadening the habitat diversity experienced by seeds relative to a single frugivore species. Alternatively, frugivore contributions can be additive within the same habitat type, which may be important to plant fitness if vector density is low (e.g., Santos et al. 1999; Loiselle et al. 2007; Garcia and Martinez 2012) or fruit density is high (Herrera and Jordano 1981). It is also possible that a frugivore species providing inferior dispersal service could decrease plant fitness by denying access to a more effective dispersal vector (Jordano 1983, and see Celedon-Neghme et al. 2013). There are few published examples, however, and the relative importance of such interactions is unclear.

Ideally, a rigorous evaluation of dispersal effectiveness would consider the entire process from maturation of seeds to establishment of the vegetative plant (Schupp et al. 2010) but such a broad scope is often logistically difficult. Most studies of multiple dispersal have chosen to focus on the spatial aspects of seed movement. However, fruit removal and digestion are also relevant to dispersal effectiveness and must be considered (e.g., Valenta and Fedigan 2009; Koike et al. 2007). In this paper, we quantify the effect of fruit removal and ingestion on seed survival and germination in a fleshy-fruited forest herb, Podophyllum peltatum (mayapple), by two mammal species. Odocoileus virginianus (white-tailed deer) and Procyon lotor (raccoon) have substantially different feeding behaviors, digestive processes, movement patterns, and range sizes potentially affecting patterns of seed distribution. We ask (a) how each frugivore species affects the reproductive fitness of the forest herb in terms of seed survival and germination, and (b) how their individual patterns of frugivory combine to determine fitness in a real plant community. Specifically, we ask whether they complement or detract from one another as agents of mayapple dispersal.

Methods

Mayapple phenology was monitored in Strouds Run State Park in southeastern Ohio, USA (39°21′6.39′′N, 82°2′5.54′′W). Natural vegetation of the region is a mixed-mesophytic deciduous forest typical of Braun’s (1950) Low Hills belt. Dominant tree species include Quercus spp., Carya spp., Acer spp., and Fagus grandifolia. Liriodendron tulipifera is prominent in disturbed areas. Study sites were situated in long-established second-growth stands on mildly acidic sandy and silty loam soils typical of mature forest (Lucht et al. 1985). The highly dissected ridge-and-ravine topography supports a diverse understory and herb community, and robust populations of raccoons and deer.

Mayapple (Podophyllum peltatum L.) is a shade-tolerant perennial herb common in deciduous forests of eastern North America. In the study area, it is typically found in clusters of 10–100s of shoots in mid-slope positions. In early March, a rhizome section produces either a single peltate leaf or a sexual shoot with a flower bud in the axil of two leaves. The solitary white flowers are pollinated by various bee species in late April (Rust and Roth 1981; Laverty 1992). Developing fruits (unicarpellate berries) are green and odorless with a hard white mesocarp. Upon ripening in mid-summer, the color changes to yellow over the course of 2–3 days with a noticeable odor and a marked softening of the mesocarp. Fruit are globular with a mean diameter of ca. 3 cm and 19 seeds/fruit (±13 SD, n = 74 fruits). Mature seeds weigh 37.5 mg (±8.3 SD, n = 110 seeds). Both raccoons and deer have been observed removing mayapple fruits in the study area (Philhower and Matlack in preparation) and, hence, are potential dispersal agents. Eastern box turtles (Terrapene Carolina) have been reported to disperse mayapple seeds elsewhere in the eastern United States (Rust and Roth 1981; Braun and Brooks 1987) and are common in the study area.

Raccoons (Procyon lotor L.) are medium-sized (5–7 kg) omnivorous mammals with a home range of 40–100 ha, implying a maximum movement distance of 700–1100 m (Lotze and Anderson 1979; Ghert 2003). Consistent with a broadly diversified diet, raccoons have a dental index of 3.1.4.2 with relatively few and narrow molars (Zeveloff 2002). They have a simple stomach, no cecum, and non-complex colons. Ingested material is retained in the gut for ca. 24 h (Clemens and Stevens 1979). Raccoons are known to disperse a variety of tree and shrub species by ingestion and defecation of seeds (Willson 1993; Cypher and Cypher 1999; LoGiudice and Ostfeld 2002).

White-tailed deer (Odocoileus virginianus, Zimmerman) are large cervids (40–100 kg) common in forests across eastern North America. White-tailed deer (hereafter “deer”) are generalist herbivores known to disperse many plant species including several deciduous-forest herbs (Vellend et al. 2003; Myers et al. 2004; Williams et al. 2008; Blyth et al. 2013). As ruminants, they have large well-developed molars and a four-chamber stomach. Ingested food is fermented and re-chewed to reduce structural carbohydrates (Ditchkoff 2011). Multi-stage food processing results in fine shredding of ingested foliage with gut retention times of up to 64 h (Mautz and Petrides 1971; Ditchkoff 2011). Deer are highly mobile with ranges as large as 3.2 km in diameter (Walter et al. 2009).

Phenology

To place mayapple reproduction and fruit removal in a seasonal context, clonal shoots were monitored through the growing season. In March 2010, seven 3 × 3 meter plots were established within mayapple colonies just after the beginning of shoot emergence. Plots were positioned to include ca. 100 stems apiece and situated at least 500 m apart to ensure independence. All plots were at least 50 m from a forest margin to avoid edge effects (Matlack 1994b). Anthesis, fruit initiation, spontaneous abortion, and fruit removal were recorded weekly from early March through senescence of the last shoot in mid-August. Only a small proportion of stems actually produced flowers in 2010. To allow closer examination of reproductive behavior, fifty-six sexual shoots were specifically selected for monitoring in March 2011. Shoots were selected as available in and near the 3 × 3 meter plots used in the previous year.

Gut passage and germination

In July 2010, we fed wild-collected fruits to captive white-tailed deer and raccoons to examine gut passage. All applicable institutional and national guidelines for the care and use of animals were followed. In all cases, the scale of the trials was limited by the number of seeds and fruits available in local populations. A two-year-old doe was fed ten ripe mayapple fruits. Because fully developed fruit sometimes contain no seeds, each fruit was cut beforehand to verify the presence of seeds. Assuming 19 seeds per fruit, the deer was fed ca. 190 seeds. The deer’s enclosure was cleared of fecal pellets prior to feeding and all pellets were collected for 72 h after feeding, a period chosen to ensure complete gut passage. Pellets were gently mashed and washed through a sieve to recover seeds. Because only a small number of seeds were recovered, seeds were not systematically tested for germination.

Each of five captive raccoons was fed two ripe fruits, equivalent to a total of ca. 190 seeds. Feces were collected for 30 h after feeding and washed through a sieve to recover seeds. Seed germination after gut passage was compared to germination of non-ingested ripe seeds and seeds from unripe fruits harvested 3–4 weeks earlier. Seeds from unripe fruit showed the same firmness, moisture, and size as ripe seeds; we assume they were live embryos at an earlier developmental stage. To test germination, seeds were sown in flats of potting soil mixed with sand and a small amount of forest soil. Flats were kept outdoors over winter and covered with 50 % shade cloth to simulate natural light conditions. Seeds were checked weekly for appearance of seedlings and watered as needed to stay imbibed.

The numbers of ingested and non-ingested seeds surviving to spring 2011 were compared using a Chi-square test with the null hypothesis that ingestion and survival were independent. Similarly, non-ingested ripe and unripe seeds were compared on the basis of survival to spring 2011, and ripe and unripe germination was compared in surviving seeds.

A second germination trial was begun in 2011 because many seeds sown in 2010 appeared to have been removed by small animals. Six raccoons were fed 16 ripe fruits equivalent to ca. 304 seeds. Recovered seeds were sown outdoors in flats caged with wire mesh to prevent seed removal. To examine the effects of ingestion, non-ingested ripe seeds were sown in the same flats. As before, the interaction of ingestion and survival over winter was assessed by a Chi-square test. Among surviving seeds, the interaction of ingestion condition (ingested or non-ingested) and germination was similarly tested.

Results

Mayapple shoots emerged rapidly in late March and early April (Fig. 1). Most vegetative shoots were present for 6–8 weeks and then senesced over a period of ca. 4 weeks in June. Sexual shoots and a small number of vegetative shoots remained green until early August. In 2010, only fifteen (2.6 %) of the 593 shoots produced flowers in our plots. Flowers opened in early May and fruit was initiated in all sexual shoots 1–2 weeks later (Fig. 1). However, seven fruits (47 %) were aborted (indicated by drying, shriveling, and browning) within 2 weeks of anthesis (Fig. 2a). Five developing fruits were removed before June 17 while still hard and green. The three remaining fruit were removed upon ripening at the end of July and early August leaving a 10- to 20-cm stem.

Fig. 1
figure 1

Phenology of mayapple (Podophyllum peltatum) at seven forested sites in southeastern Ohio in 2010. A maximum of 593 total shoots were recorded in mid-May. The numbers of sexual shoots are shown on the right-hand axis for clarity

Full size image
Fig. 2
figure 2

Fate of mayapple fruit through the fruiting season at plots in southeast Ohio (N = 15 fruit) and in 2011 (56 fruit). Chart begins on the date of peak fruit number. In 2011, seven fruit aborted before the peak at May 17 and, so, are not shown here. Arrows indicate the approximate beginning of fruit ripening

Full size image

In 2011, fruit abortion was again concentrated in the first two weeks but continued at low levels until July 19, nine weeks after anthesis (Fig. 2b). Only 14 fruits (25 %) continued development and 13 of these were eaten unripe in June or the first week of July. In June, fruit removal was accompanied by loss of leaves, clipping of the stem at 10–20 cm, and removal of many surrounding vegetative shoots, suggesting deer grazing. The single remaining fruit ripened and was quickly removed at the end of July, leaving an erect 10- to 20-cm stem. Of all sexual shoots encountered in 2010 and 2011, 65.6 % failed to produce fruit due to abortion and only 5.5 % of fruit eventually ripened.

Ingested seeds

Only two seeds were recovered from the 10 fruits fed to the deer, representing ca. 1 % of ingested seeds. Both seeds were excreted between 24 and 48 h after consumption. The two seeds recovered both appeared to be viable, remaining firm, moist, light-brown, and free from fungi (indeed, one of the seeds germinated in the lab). However, the small sample size prevented systematic testing of germination.

In the 2010 feeding trial, 53 seeds (ca. 28 % of seeds ingested) were recovered from the five raccoons, all within 30 h of consumption. In the germination trial many seeds decayed, were eaten by slugs or millipedes, or simply disappeared within the first few months (Table 1a). No raccoon-ingested seeds survived to the spring of 2011. In contrast to ingested seeds, 61 % of non-ingested ripe seeds survived, showing a strongly significant ingestion × survival interaction (χ 2 = 10.75, 1 df, P = 0.001). Failure to survive was largely caused by the high rate of disappearance of ingested seeds (63 %) relative to non-ingested ripe seeds (22 %). In contrast to disappearance, similar proportions of raccoon-ingested seeds (12.5 %) and ripe non-ingested seeds (10.2 %) decayed.

Table 1 Fate of mayapple seeds ingested by raccoon and non-ingested seeds collected from ripe and unripe fruit
Full size table

Sufficient non-ingested seeds survived to compare seeds from ripe and unripe fruit (Table 1a). Ripe seeds were more likely to survive until spring than unripe seeds (χ 2 = 11.64, 1 df, P = 0.001), primarily due to the higher rate of decay of unripe seeds (36.5 and 10.2 % decayed; χ 2 = 9.55, 1 df, P = 0.002). Germination of surviving seeds was independent of fruit ripeness (χ 2 = 1.19, 1 df, P = 0.28). Six ripe seeds survived an additional year and germinated in 2012, bringing the total germination of hand-harvested ripe seeds to 40.7 %. In contrast, only 13.5 % of unripe seeds eventually germinated and none germinated in 2012.

In the 2011 feeding trial, 83 seeds were recovered from raccoon feces, representing 27 % of seeds ingested. Recovered ingested seeds were sown beside ripe non-ingested seeds in the wire cages. As in 2010, many seeds appeared to have been eaten or decayed, but no seeds disappeared (Table 1b). Ingestion of seeds and survival to spring 2012 were independent (χ 2 = 0.18, 1 df, P = 0.67) with little difference in proportional survival. However, germination of surviving raccoon-ingested seeds was significantly higher (100 %) than surviving non-ingested seeds (61 %; χ 2 = 11.3, 1 df, P = 0.0007).

Discussion

Development of mayapple in our plots was similar to other early-spring perennials such as Hydrastis canadensis L. (Eichenberger and Parker 1976) and Trillium spp. (Lubbers and Lechowicz 1989; Sage et al. 2001) in which stored reserves are used to promote rapid leaf expansion, and flowers open shortly thereafter. The peak in mayapple shoot number and the beginning of fruit expansion roughly corresponded to canopy closure in mid-May. After canopy closure, fruit development continued through the summer supported by stored resources and a portion of current assimilate (Kudo et al. 2008).

Infrequent flowering and low fruit set observed here may reflect the expense of reproduction in the low-light forest environment (Rust and Roth 1981; Sohn and Policansky 1977). Alternatively, reproduction is often pollen-limited in mayapple (Swanson and Sohmer 1976; Crants 2008), potentially causing the high level of abortion observed in our plots. A similar survey in southeastern Kentucky found 63 % of sexual shoots produced no fruit (Krochmal et al. 1974) possibly due to inadequate pollination (Stephenson 1981). Consistent with our observations, studies in Wisconsin and Indiana reported only ca. 3 % of mayapple shoots flowering and fewer than 10 % of flowers producing fruit (Swanson and Sohmer 1976; Geber et al. 1997). The infrequency and inferred expense of seed set suggests that the fitness cost of inappropriate dispersal is quite high.

Frugivory

Of those fruits not aborted, most (82 %) were removed in June while still unripe. Intensive grazing associated with June fruit removal suggests that unripe fruits were removed by deer, consistent with a previous photo-monitoring study which showed deer grazing of mayapple narrowly concentrated in June (Philhower and Matlack, unpublished). Unripe seeds were less likely to survive over winter than ripe seeds and less likely to germinate in the third year. Thus, early removal by deer is probably of little dispersal value to the plant because it results in distribution of low-viability seeds.

Ripe mayapple seeds were destroyed in the digestive process by both raccoons and deer, but deer caused a much greater reduction in seed numbers. Deer also consume a co-occurring forest herb Panax quinquefolius L. (American ginseng), apparently ingesting the relatively large seeds as they graze the foliage. No viable seeds were recovered from 16,800 deer-consumed fruits (Furedi and McGraw 2004), consistent with our observation of seed destruction in mayapple. Indeed, the captive deer used in this study was also fed corn to supplement herbaceous forage available during the feeding trial. Only a few recognizable corn kernels were identified in 3 days’ worth of feces, indicating the food processing efficiency of the deer. Paradoxically, many species have been observed to survive deer digestion at sites in western New York, southern Connecticut, and central Ohio (Myers et al. 2004; Williams et al. 2008; Blyth et al. 2013). However, these are almost entirely small, hard-coated species typical of open habitats. The degree of seed destruction appears to depend on seed size and exocarp hardness (Vellend et al. 2003) suggesting that many forest herb species cannot be dispersed by deer because their seeds are too large and/or soft to survive gut passage. It is worth noting that two viable seeds were recovered from fecal pellets in our feeding trial so occasional dispersal cannot be ruled out. With long gut-retention times and movement distances measured in kilometers, deer may potentially move seeds long distances (Vellend et al. 2003) with demographic consequences out of proportion to the low dispersal frequency (Neubert and Caswell 2000).

We assume the late-removed fruits were consumed by raccoons. Camera-trap data show that raccoons ignore unripe mayapple fruits early in the summer but quickly consume fruits when they ripen in late July (Philower and Matlack, unpublished) possibly cued by odor. Removal of fruits 10–20 cm above the ground is inconsistent with turtle activity, but easily achieved by raccoons. Although raccoons destroyed most seeds in the feeding trials, a substantial number survived ingestion potentially allowing dispersal. Ingestion may actually improve fitness of surviving seeds. Those defecated by raccoons germinated at a higher rate than un-ingested seeds in the 2011 trial, perhaps due to the removal of a germination inhibitor (Krochmal et al. 1974; Cypher and Cypher 1999). Loss of raccoon-ingested seeds from uncaged sowing sites was higher than non-ingested seeds, apparently due to removal of seeds by small vertebrates. If raccoon ingestion makes seeds more attractive to cache-forming seed predators, the benefits of secondary dispersal may also increase fitness of ingested seeds (LoGiudice and Ostfeld 2002; Niederhauser and Matlack, unpublished). In contrast to deer, consumption by raccoons appears to have several benefits for the mayapple notwithstanding the destruction of a portion of seeds in digestion.

Other studies report that box turtles commonly consume mayapple fruits, passing a higher proportion of viable seeds than raccoons and leading to greater germination (Rust and Roth 1981; Braun and Brooks 1987). We found no direct evidence of turtle frugivory (nor did Philhower and Matlack, unpublished), but the abundance of turtles in the study area suggests that they contribute at least occasionally to dispersal.

Cumulative fitness effects

Multiple frugivory appears to have a negative effect on the dispersal component of mayapple fitness in our study area. Raccoons discovered 100 % of ripe fruits but only 28 % of seeds survived gut passage implying a 72 % reduction in reproductive fitness at the dispersal stage. Assuming a hypothetical clone produced ten ripe fruits, raccoon frugivory would result in ca. 53 seedlings. In a second scenario, only deer consume mayapple fruits. In contrast to raccoons only ca. 1 % of deer-ingested seeds survived digestion (these are likely to be low-quality seeds collected before maturity) but they failed to find 18 % of fruits, with a net result of 82 % reduction in reproductive fitness. Thus, in the case of deer frugivory the hypothetical clone could potentially produce ca. 34 seedlings. However, the surviving 18 % of fruits would not be dispersed and seedlings would potentially experience some form of compensatory mortality at high density around the parent plants (Giladi 2006). Thus, the realized number of seedlings would probably be much less than the 34 seeds which escaped deer predation.

In a third scenario, more realistically corresponding to observations in the field, the fruits surviving deer predation are subsequently discovered and consumed by raccoons, compounding the fitness reduction. Of the 18 % escaping deer predation, seed number would be reduced a further 72 % by raccoons resulting in a cumulative 5 % survival of seeds, equivalent to ca. 10 seedlings in the hypothetical clone. Comparison of the hypothetical seedling numbers suggests that deer predation reduces mayapple reproductive fitness by preventing raccoon (and possibly turtle) dispersal. Complementary or supplementary interactions are precluded by the near-complete destruction of seeds by deer.

Preemption of raccoon frugivory in late summer is a form of exploitation competition in which deer dominate by virtue of earlier fruit consumption. The situation is analogous to the classic study of Ficus-consuming parrots in lowland Costa Rica (Jordano 1983), in which competitive dominance was expressed as greater numerical abundance of the inferior dispersal agent (see also Amato and Estrada 2010). The contrast between deer and raccoon dispersal is particularly important to mayapple fitness because successful fruit set is relatively uncommon (only 2.6 % of shoots reproduced in 2010 and 66 % of these were lost to spontaneous abortion). If reproduction is limited by slow accumulation of resources at the forest floor (Whigham 2004) loss of even a small number of seeds to inappropriate frugivores may substantially reduce reproductive success on a scale of years or decades.

Conclusion

The negative effects demonstrated here are probably common in the deciduous forest community because multiple frugivory is common among forest herbs, and individual frugivore species differ in the character of seed and fruit processing (e.g., Bonaccorso et al. 2006; Brodie et al. 2009). On an evolutionary time scale, variation in dispersal-vector quality potentially exerts selective pressure for fruit traits that specifically attract beneficial frugivores and avoid less helpful species. In the short term, white-tailed deer populations are currently at historically high levels within the range of mayapple (Iverson and Iverson 1999; Côté et al. 2004), suggesting unprecedented pressure on mayapple reproduction and perhaps reproduction of other species as well. The potential damage to forest herb populations provides an additional reason for controlling white-tailed deer.