Background and Aims Plants have evolved a variety of seed dispersal mechanisms to overcome lack of mobility. Many species embed seeds in fleshy pulp to elicit endozoochory, i.e. disseminating seed through the animal gut. In contrast to well-studied fleshy fruited plants, dry-fruited plants may exploit this dispersal mutualism by producing fleshy appendages as a nutritional reward to entice animals to swallow their diaspores, but this has been little studied. In this study, it is hypothesized that these accessory fruits represent co-adaptations facilitating the syndrome of mammalian endozoochorous dispersal.
This gets dry and fosters inflammations with the protective order cialis mucus membranes within your intestines… One can experience a total inability to viagra online canadian achieve an erection at any point during the 4 to 6 hours after taking it. There might be other parameters that viagra stores can be used at many occasions, even brought together at family holidays or small trips. Another reason is a misinterpretation of Sigmund Freud’s teachings which led sildenafil rx to the incorrect conclusion that the majority of cases.
Methods Field observations (focal tree watches, faecal surveys and fruiting phenology) with experimental manipulations (examination of seed germination and feeding trials) were conducted over 2 years in a native population of the raisin tree, Hovenia dulcis, which produces enlarged, twisted brown peduncles with external black seeds, in central China.
Key Results Birds were not observed to swallow seeds or carry infructescences away during 190 h of focal tree watches. However, H. dulcis seeds were detected in 247 faecal samples, representative of two herbivore and four carnivore mammalian species. Feeding trials revealed that peduncles attracted mammals to consume the entire infructescence, thereby facilitating effective seed dispersal. The germination rate of egested seeds proved higher than that of unconsumed seeds. It was also noted that this mutualism was most vulnerable in degraded forest.
Conclusions Hovenia dulcis peduncle sets are confirmed to adapt primarily to mammalian endozoochory, a mutualistic association similar in function to fleshy pulp or foliage. This demonstrates that plant organ systems can be adapted to unique mutualisms that utilize animal dispersal agents. Such an ecological role has until now been attributed only to bird epizoochory. Future studies should consider more widely the putative role of peduncle sets and mammalian endozoochory as a dispersal mechanism, particularly for those plants that possess relatively large accessory fruits.
INTRODUCTION
Co-evolved plant–animal interactions can be complex and play a vital role in ecosystem functionality and the persistence of biodiversity (Howe and Smallwood, 1982; Bascompte et al., 2006). Through natural selection, plants have evolved a variety of dispersal units and mechanisms to overcome mobility issues (van der Pijl, 1982; Jordano et al., 2007), the most common of which is to utilize animal agents to facilitate zoochory [dispersal by animals, including endozoochory, epizoochory and hoarding (van der Pijl, 1982; Herrera, 2002)]. For example, up to 90 % of the plant species in tropical forests disperse seeds via animal agents (Howe and Smallwood, 1982; Fleming et al., 1987; Jordano, 2000; Tiffney, 2004). Many dry-fruited plants produce appendages (e.g. hooks and hairs) on their reproductive structures to facilitate epizoochory [i.e. external dispersal by animals (Chambers and Macmahon, 1994)]. Birds (e.g. Loayza and Knight, 2010) and monkeys (e.g. Stevenson, 2011) often act as agents of epizoochory, by carrying infructescences to remote perches to peck or chew on the pulp. The predominant plant dispersal strategy is, however, to produce diaspores that elicit endozoochory [i.e. the consumption and subsequent egestion of viable seed (Robertson et al., 2006; de Vega et al., 2011)] by embedding seeds in fleshy pulp (van der Pijl, 1982; Moore, 2001). Lush foliage can also attract herbivores to small-fruited herbaceous plants, to promote endozoochory (Janzen, 1984).
Among these plants adopting an endozoochorous strategy, a few have evolved an alternative approach, producing specialized accessory fruits with a fleshy peduncle (edible fleshy fruit stalk) and dry drupe (nut/achene). The ecological functions and evolutionary origins of fleshy peduncles in general, however, remain to be fully understood and established. Ridley (1930, p. 419) was the first to comment on the role of peduncles in the cashew (Anacardium occidentale), assuming: ‘ … its diffusion in countries where it has been once introduced is due to birds … .’ Recent field observations of wild cashew [A. excelsum (Smythe, 1970; Beckman and Muller-Landau, 2011)], and related species [A. spruceanum and A. giganteum (van Roosmalen, 1985)] as well as other similar species [Podocarpus latifolius (Geldenhuys, 1993); P. nubigena (Willson et al., 1996); Apodytes spp. (Potgieter and van Wyk, 1994)] have also supported this premise, i.e. fleshy appendages serve as an edible, nutritious reward for bird and bat epizoochory. All these examples apply to fleshy peduncles that are small, bright and colourful (e.g. red), a trait connected to the bird ‘dispersal syndrome’ (Janson, 1983), i.e. co-occurring sets of matched traits between fruit and the behaviour, physiology and morphology of the disperser (Lomascolo et al., 2010; Muñoz and Bonal, 2011; Leroy et al., 2012).
The raisin tree, Hovenia dulcis, has twisted and fleshy raisin-like peduncles, originating from the pedicel (rachis), and supporting the seed pod (dry drupe fruit) developing from the gynoecium (see Fig. 1). Maxwell (1994) and Kopachon et al. (1996) all assumed, from morphology and anecdotal observation, but without empirical corroboration, that, like Anacardium, Podocarpus and Apodytes (Geldenhuys, 1993; Potgieter and van Wyk, 1994; Willson et al., 1996; Beckman and Muller-Landau, 2011), birds and bats would carry infructescences (i.e. ensembles of fruits derived from the ovaries of an inflorescence/fruiting branch) of H. dulcis to a perch in another tree, to peck the fleshy peduncles and achieve epizoochorous dispersal. This assumption is, however, contrary to certain expectations. While avian dispersers are attracted to small, bright and colourful dispersal units (Janson, 1983; Lomascolo et al., 2010), H. dulcis peduncles, in contrast, are large, brown and fleshy with external black fruits (weight: 6·5 g) (see Fig. 1 and Supplementary Data Table S1). Several species of large mammals have been observed to eat H. dulcis peduncles and fruits in previous studies (Zhou et al., 2008a; Hirsch, 2009). Moreover, bats are absent from the Asian sub-tropical and temperature forests where H. dulcis is indigenous (Maxwell, 1994; Kopachon et al., 1996; Smith and Xie, 2008). This leads us to hypothesize that these composite fruits present a different ecological and evolutionary co-adaptation, not evolved primarily to attract bird epizoochory (Janson, 1983), but rather to facilitate the mammal dispersal syndrome.
Raisin tree H. dulcis dispersal units (top) and two key dispersers: the Chinese serow (left, photograph taken by Francisco Palomares) and masked palm civet (right).
We test this by incorporating focal tree observations – to ascertain the relative incidence of bird vs. mammal seed procurement – with evidence from faecal examination. We proceed to a comparison of ingested vs. control seed germination, followed by a series of captive seed disperser feeding trials. We use these data to inform an investigation of how H. dulcis seed dispersal varies with habitat type, thereby deducing how animal dispersal mutualism is likely to be influenced by habitat degradation in Asia.
MATERIALS AND METHODS
Study population
The raisin tree, Hovenia dulcis (Rhamnaceae), has a distribution from Japan, across China to the Himalayas (up to elevations of 2000 m). It produces drupe fruits, consisting anatomically of an outer fibroid exocarp and a mescocap surrounding a shell (pyrena) of hardened endocarp containing the seed; these develop at the end of edible fleshy pedunculate fruit stalks that swell to form a type of accessory fruit (infructescence) (Kopachon et al., 1996) (Fig. 1). This study was conducted in a native population of H. dulcis at Houhe National Nature Reserve (HNNR) (30 °5’N, 111 °42’E), central China, where raisin tree specimens occur commonly in the canopy, at a density of approx. 9 ha−1. The fruiting season extends from October to March, peaking from November to January (Zhou, 2008).
Identifying the agents of H. dulcis dispersal through diurnal focal tree watches
To investigate and quantify the contribution birds made to H. dulcis seed dispersal (e.g. Maxwell, 1994; Kopachon et al., 1996), we observed bird feeding habits, noting also any visits by mammals, at ten focal raisin trees during the peak H. dulcis fruiting season in HNNR, from 15 December 2005 to 4 January 2006. Each tree was observed over 2 d, during times of peak bird activity (from 08:00 to 11:00 h and again from 15:00 to 18:00 h, providing a cumulative effort of 190 h; observations were suspended during periods of heavy rain). Focal trees were located at least 100 m apart, to increase the likelihood of statistical independence. All focal trees were within an area of 100 ha, in which H. dulcis occurred at typical density. Activity in and under each tree was observed by Y.B.Z. and a trained local assistant, using binoculars from a hide placed at a distance of 20–30 m. For each visit, species, duration and the distance over which fruits were carried were recorded, along with whether or not the animal fed on peduncles and/or consumed seeds. If a group of conspecific animals visited a tree and individual behaviour could not be observed simultaneously, we focused on the individual that was most visible. If the behaviour of individuals from different species could not be observed simultaneously, we focused on the rarer (least often observed) species.
Adaptation to elicit mammal endozoochory: faecal surveys, examination of seed germination and feeding trials
To test whether large mammals (>1 kg; DeMattia et al., 2004; Iob and Vieira, 2008) dispersed seeds and to identify the species responsible, while quantifying the extent of seed dispersal contributed by each, we performed a comprehensive survey and examination of faeces over the study area. Faecal samples were acquired initially during a pilot study from September to November 2004, after which the protocol was standardized and applied consistently from April 2005 to April 2007, in conjunction with a study of the ecology and behaviour of the HNNR’s community of mammals (Zhou et al., 2008a; Zhang et al., 2010). Faeces were collected every 2 weeks along 20 transects (mean length = 2·7 km, range 2·1–3·0 km, total length = 54·3 km). Faeces encountered during concurrent radiotracking studies of the masked palm civet (Paguma larvata) were also checked and/or collected and included in the analysis (Zhou et al., 2008a). Faeces were identified to species reliably by size, shape, texture, nearby tracks/field signs, and characteristic odour (Zhou et al., 2008a). Seeds and/or seed coats were extracted from faecal samples by sieving them through a nylon mesh (0·5 mm). Particular care was taken to detect any remains of seed coats that would indicate the presence of broken seeds. When remains of seed coats from H. dulcis were present, we estimated the number of entire seeds that would account for them. Separate records were kept for intact and broken (estimated from seed coat remains) plus cracked seeds (‘damaged seeds’ hereafter) (Herrera, 1989).
To assess the effectiveness of endozoochorous dispersal, we examined whether seeds passed through the gut intact and the effects of gut passage on seed germination, by comparing the germination success of H. dulcis seeds defecated in the faeces of these mammals with that of seeds obtained directly from wild plants during the peak of fruit consumption (i.e. December). This experimental design accommodated the variable availability of ingested seeds for each disperser species. Ingested seeds were cleaned of faecal material; control seeds were removed from the peduncles (Samuels and Levey, 2005). All seeds in faeces collected from October 2005 until March 2006 were pooled. For the Chinese serow (Capricornis milneedwardsii) and the Asian black bear (Ursus thibetanus), five samples of 400 seeds, chosen at random, were used for germination trials; for the masked palm civet (P. larvata), five samples of 100 seeds were chosen randomly; and for the other three mammalian seed disperser, all available seeds in faeces were pooled per species and split evenly between five samples (Table 1). For control seeds, 20 fruits per individual tree were collected directly from 40 wild plants, and 100 seeds obtained randomly from these fruits were pooled and split evenly between five samples (800 seeds per sample). All seeds were placed on moist filter paper in Petri dishes, which were inspected and watered daily for 6 months. Five replicate groups were placed in incubators (14 h light, 25 °C). Germinated seeds were counted and removed when the radicle reached 2 mm long (Zhou et al., 2008b).
Frequency of H. dulcis seeds and percentages of intact seeds in faeces from six mammals collected in October–March 2005–2006 and 2006–2007, including the body mass of these mammals
We performed a series of feeding trials to evaluate experimentally whether fleshy peduncles attract mammalian seed dispersers to devour seeds. Ten captive masked palm civets and six Chinese ferret-badgers (Melogale moschata) were acclimatized to experimental conditions on a diet of cultivated fruits (persimmons, apples and bananas), supplemented with chicken and earthworms. Drinking water was provided ad libitum. Three intact dispersal units, three peduncles only and three drupes only were offered to these animals for 3 d. Remains of intact dispersal units, peduncles only and drupes only were checked after 15 min of feeding. After each trial, all of the remains were removed, and familiar foods were re-provisioned. Each trial was carried out once per day and three replicate samples were consequently obtained for the comparison of individual variation. The passage rate of seeds through the gut was established by examination the faeces for the presence of seeds, accounting for the proportion of uneaten drupes and seeds. All animal husbandry and handling procedures in these experiments adhered to the guidelines of the American Society of Mammalogists for the use of wild mammals in research (Gannon et al., 2007).
Assessing the effects of habitat degradation on the H. dulcis–mammal mutualism
Mutualisms involving the dispersal of seeds by animals can be particularly disrupted in degraded habitats (Cordeiro and Howe, 2003), especially where larger species serve as dispersal agents and large-seeded trees are involved (Wotton and Kelly, 2011; but see Matias et al., 2010). We gained insight into the effects of forest degradation on H. dulcis seed dispersal by comparing the monthly number of faecal samples per transect and the frequency of occurrence of H. dulcis seed remains in faeces per transect kilometre between primary and logged forests; combining selectively logged forest and logged forest as ‘degraded forests’, and excluding forest plantation and farmland where H. dulcis were absent, only 1·3 % of faeces were collected from these habitats combined.
Statistical analysis
All analyses were performed in R version 2·6·1, with measured parameters presented as the mean ± s.d. All data were tested for normality prior to analysis. Spearman’s rank correlations (ρ) were employed to examine the relationship of the frequency of occurrence and number of H. dulcis seeds per faecal sample to the body mass of mammalian dispersers. Differences in germination percentage between ingested and control seeds were analysed using the non-parametric Wilcoxon signed ranks tests (WSR tests). For civets and ferret-badgers, analysis of variance (ANOVA) tests were used to analyse individual variation in seed passage rate through the gut, and t-tests were used to analyse the feeding preference trials. Paired t-tests were used to compare the differences in the monthly number of faecal samples and the density of dispersed seeds per transect kilometre between primary and degraded forests, and WSR tests were subsequently used to compare the frequency of H. dulcis seed in these faeces.
RESULTS
Identifying the agents of H. dulcis dispersal through focal trees watches
We recorded 381 visits by 16 bird species and 256 visits by seven small species of mammal during 190 h of diurnal watches. While peduncle pecking was observed for 82 visits by seven bird species, this never involved birds actually swallowing seeds or carrying infructescences away; thus, we conclude that birds are not major contributors to H. dulcis seed dispersal. Three squirrel species were also seen to eat peduncles (116 visits), but only in situ – chewing and destroying seeds and thus failing to disperse them effectively. A Pallas’ squirrel (Callosciurus erythraeus) removed two dispersal units that had fallen on the ground, but to distances of only 0·28 and 0·45 m, respectively, and then they chewed the seeds and ate the peduncles completely. The red-billed blue magpie (Urocissa erythrorhyncha) and Pallas’ squirrel accounted for 46 and 37 % of visits where in situ peduncle consumption was observed, with the Pallas’ squirrel and pheasant (Tragopan temminckii) accounting for 44 and 31 % of visits to fallen fruits on the ground (details are given in Supplementary Data Table S2). Visits were infrequent (0·4 and 0·3 visit events h−1 for magpies and Pallas’ squirrels, and 0·7 and 0·4 to fallen fruits for Pallas’ squirrels and pheasants). In addition, the red and white giant flying squirrel (Petaurista alborufus), the complex-toothed flying squirrel (Trogopterus xanthipes), the Chinese bamboo rat (Rhizomys sinensis) and the Siberian weasel (Mustela sibirica) made single visits to trees, without dispersing any seeds.
These findings demonstrate that birds were peduncle feeders only, not ingesting seeds and not playing a substantial role as seed dispersers, while squirrels were seed and peduncle predators (i.e. they destroyed the seeds by mastication). Consequently, neither birds nor squirrels were capable of achieving substantial and effective H. dulcis seed dispersal. No legitimate seed disperser species, as established from examination of faeces (described below), were seen feeding on seed/peduncles (or fallen seed/peduncles) directly by day. This was, however, unsurprising because these species forage mostly at night. Equivalent nocturnal surveys were impractical as, even with night vision equipment, entire tree canopies could not be observed reliably in the dark.
Faecal surveys, examination of seed germination and feeding trials
A total of 4051 faecal samples were examined from April 2005 to April 2007, from 13 species of mammal and various unidentifiable birds. Of these, H. dulcis seed remains were detected in 247 mammalian faecal samples (118 in the first winter and 129 in the second winter), from two herbivores and four carnivores, collected during the October to March fruiting season (in total, 1447 faecal samples were examined for these six primary dispersal agent species) (Fig. 2). The frequency of occurrence of all plant seeds combined differed among these six mammals, being highest for the Asian black bear (82 %) and lowest for the Chinese goral Naemorhedus griseus (10 %) (Table 1). For H. dulcis seeds, however, the frequency of occurrence in these faecal samples was the highest for the yellow-throated martens (Martes flavigula) and lowest for the Chinese goral (see Table 1). This frequency of occurrence peaked in November (25 %) and December (42 %) of the first study winter and in December (37 %) and January (56 %) of the second winter (Fig. 2). Four species accounted for 96 % of seed-bearing faeces, where the relative contribution made by each, measured as the number of seeds found per faecal sample, was 49 % for Chinese serow, 17 % for masked palm civets, 15 % for yellow-throated martens and 15 % for black bears; note, however, that bears hibernate and so no samples of their faeces were found during the peak H. dulcis fruiting period (Table 1). For these mammals, the frequency of occurrence of H. dulcis seeds did not relate to body mass (ρ = – 0·257, P = 0·623), but the number of H. dulcis seeds per faecal sample increased with body mass (ρ = 0·943, P = 0·005) (Table 1). These findings demonstrate that large mammals (>1 kg) were frequent and substantial dispersers of H. dulcis seed.
Proportion of H. dulcis seeds recovered from faecal samples of six mammalian species (colour-coded columns: a, serow; b, goral; c, bear; d, ferret-badger; e, marten; f, civet) and relationship to the relative abundance of H. dulcis fruits (as number of accessory fruits per sample branch; means ± s.d.): October–March 2005–2006 (A), and 2006–2007 (B). The sample size is shown above each column. In 2005–2006, one sample was not included because it was not identified as belonging to a specific species (i.e. from civet or marten).
The percentages of intact seeds passing through the gut were high (>97 % for all carnivorous species, 85 and 79 % for herbivorous serow and goral, respectively), indicating that these mammals seldom destroyed seeds during mastication and passage through the gut (Table 1).
Seeds planted after ingestion by serow, goral and martens germinated in significantly higher proportions than did control seeds (WSR test: Z = 2·02–2·03, P < 0·05), while seeds consumed by ferret-badgers, civets and bears all germinated in proportions that were similar to control seeds obtained directly from wild plants (WSR test: Z < 1·75, P > 0·05) (Fig. 3). That is, seeds ingested by certain mammals exhibited a germination advantage over undigested seeds.
Cumulative seed germination percentages for each seed manipulation treatment. The dotted line represents the control seeds; solid lines represented ingested seeds.
In feeding trials, all individuals consumed the whole dispersal unit (100 % of the peduncle and most of the drupe) and the peduncle in isolation (100 %), readily rejecting entirely drupes when presented alone. Gut passage rates were similar for individual civets and ferret-badgers (ANOVA: F = 0·64, P = 0·751; and F = 1·42, P = 0·286). Civets, however, devoured a higher proportion of seeds (94 %, n = 30) than did ferret-badgers (88 %, n = 18; t = 7·7, P = 0·008). Drupes are accessible without the necessity of consuming the peduncle mass, so the fact that these frugivores devoured dry drupes only when they were accompanied by the twisted peduncle demonstrates that peduncles provide the incentive for large mammal endozoochory.
Effects of habitat degradation on H. dulcis–mammal mutualism
While we detected no statistically significant differences between the frequency of occurrence of H. dulcis seeds in faeces collected in primary and degraded forests (Fig. 4; WSR test, P > 0·05), we did note significant differences in the extent to which seed-dispersing species utilized forest types: martens, bears, serow and goral deposited more faeces in primary forests than they did in degraded forests (Fig. 5; paired t-test = 5·46–17·16, all P < 0·001). The majority of faecal samples (>92 %) in degraded forests were, however, within selectively logged forest, or within 2 km of primary forest, indicating that dispersers were sensitive to habitat degradation. In contrast, seed dispersal by civets and ferret-badgers exhibited no significant habitat bias (Fig. 5; paired t-test = 1·94 and 0·71, P > 0·05). Furthermore, the absolute density of dispersed seed showed a similar pattern (Fig. 6): martens, bears, serow and goral deposited a higher density of seeds in primary forest than in degraded forests (paired t-test = 2·28–3·21, all P < 0·05), while civets and ferret-badgers exhibited no significant habitat bias (paired t-test = 1·92 and 0·43, P > 0·05). This demonstrates that forest degradation reduces the occurrence of dispersal agents and thus limits H. dulcis seed dispersal.
Frequency of occurrence of H. dulcis seed remains in the faeces from six mammalian species collected between October and March of 2005–2006 and 2006–2007 in primary and degraded forests (as indicated) in HNNR, China. Statistical results for effect of forest type are from Wilcoxon signed ranks test.
Monthly number of faecal samples per kilometre from six mammalian seed dispersers (a, serow; b, goral; c, bear; d, ferret-badger; e, marten; f, civet) in primary and degraded forest, collected between April 2005 and April 2007 in HNNR, China.
Number of H. dulcis seeds dispersed per kilometre by six mammalian seed dispersers (a, serow; b, goral; c, bear; d, ferret-badger; e, marten; f, civet) in primary and degraded forest, collected between October and March of 2005–2006 and 2006–2007 in HNNR, China.
DISCUSSION
Given the substantial differences in how animal taxa perceive fruits and disperse seeds, in combination with similarly substantial differences in fruit traits among plant taxa, ‘dispersal syndromes’ have been identified (Howe and Smallwood, 1982; Janson, 1983; Jordano et al., 2007). These are defined according to co-occurring combinations of fruit traits matched to the behaviour, physiology and morphology of different types of dispersers (Lomascolo et al., 2010; Leroy et al., 2012). While researchers such as Maxwell (1994) and Kopachon et al. (1996) have posited that H. dulcis fleshy peduncles, like those of other peduncle-fruited species (Ridley, 1930; Geldenhuys, 1993; Potgieter and van Wyk, 1994; Willson et al., 1996), would function to attract birds, our observations reveal a different reality: (1) birds pecked at H. dulcis peduncles exclusively in situ, and did not consume any seeds – thus drupe (seed) transportation was incidental and minimal; (2) squirrels facilitated short-distance epizoochory, but they masticated seeds, destroying their viability (Muñoz and Bonal, 2011); conversely (3) seed dispersal units proved to be adapted primarily to large mammal endozoochory in several respects.
Large accessory fruits are difficult for birds to carry or swallow (Janson, 1983), because of birds’ relatively small body size, narrow gape and lack of teeth (Levey, 1987). Hovenia dulcis peduncles are heavier (3·9 g vs. <1 g for Podocarpus spp. and Apodytes spp.; Geldenhuys, 1993; Potgieter and van Wyk, 1994; Willson et al., 1996) and infructescences are larger (6·5 g vs. <2 g for Podocarpus spp. and Apodytes spp.) than in other peduncle-bearing tree species, where bird epizoochory predominates. Hovenia dulcis drupes thus appear not to have undergone selection for easy transportation, but rather for having an enticing pulp volume. The integrity of the connection between the seed and the pulp (or fleshy appendage) is an important factor determining the chance of legitimate seed dispersal by frugivores (Levey, 1987; Loayza and Knight, 2010). With greater seed–pulp attachment integrity, frugivores have thus been found to swallow seeds along with pulp more readily (Stevenson, 2011). In the case of H. dulcis, the twisted structure of peduncles, and the associated difficulty in removing seeds, leads to a greater likelihood that seeds will be ingested, and hence dispersed. We consequently propose that because the animals involved in H. dulcis peduncle carrying (epizoochory) eat only the peduncle and/or predate seeds and do not defecate undamaged seeds, this is a sub-ordinate strategy for this tree species, making a very limited contribution to seed dispersal.
The large and brown peduncles of H. dulcis, with small black and dry fruits, also do not appear to be the product of natural selection for ease of detection, to appeal to birds, which have tetrachromatic vision (Janson, 1983; Schaefer and Schmidt, 2007); although they can see black fruits. Mammals, in contrast, have a much greater reliance upon smell/taste, rather than vision, for finding food (Corlett, 1996; Lomascolo et al., 2010), and often consume fruits that are brown, yellow or orange (Willson and Whelan, 1990). We note that the main mammalian H. dulcis peduncle consumers were not observed during our diurnal focal tree watches, as these species feed nocturnally – in darkness (Zhou, 2008). This provides further evidence that vivid colour has not been a selective criterion in the evolution of this dispersal mechanism. To human perception, mature peduncle has a sweet smell and taste – traits that seem to fit the classical mammal dispersal syndrome (Janson, 1983; Herrera, 1989; Hodgkison et al., 2007; Lomascolo et al., 2010). In contrast, birds have a poor sense of smell and taste (Bennett and Théry, 2007).
Generally, fruits with high carbohydrate contents (specifically high sucrose and low glucose and fructose; Ko et al., 1998) and low, or no, lipid content are an adaptation to attract mammalian dispersers, whereas high lipid contents attract avian dispersal agents (Corlett, 1996). Fitting the mammal dispersal profile (Corlett, 1996; Ko et al., 1998), H. dulcis peduncles contain 48·8 % carbohydrate, 8·6 % protein and only 1·2 % lipid (Lin, 2009), with relatively high sucrose (24 %) and low glucose (7·9 %) and fructose (9·5 %) levels (Li et al., 1996). Furthermore, H. dulcis exhibits a fruiting syndrome synchronized specifically (peaking from November to January) to coincide with fruit paucity among other sympatric trees (Zhou et al., 2008a). Under these circumstances, the high protein and carbohydrate content of peduncles (details are given in Supplementary Data Table S1) would prove important for sustaining consumers during exposure to cold stress (King and Murphy, 1985).
Most species of fleshy fruits have morphological structures – fleshy pulp or epimatium enclosing the seed – that enhance endozoochory and thus elicit long-distance seed displacement (Jordano, 1995). For species possessing accessory fruits, however, the division of dry fruits by fleshy pulp within these structures can function to facilitate seed scattering beneath, or near to, parent plants during avian epizoochory (Levey, 1987), when peduncles borne of transported infructescences are pecked (Ridley, 1930; Geldenhuys, 1993). The Janzen–Connell model, one of the widely accepted explanations for the maintenance of tree species biodiversity (Janzen, 1970; Connell, 1971), however, predicts lower seed and seedling survival beneath, or near to, parent plants. Interestingly, our feeding trial showed that peduncles actually attracted mammals to devour seeds; an ecological function paralogous to having attractive fleshy pulp or epimatium. A similar trait has been observed with the lush foliage of small-fruited herbaceous plants, for which fleshy leaves may function to attract large herbivores, facilitating endozoochorous seed dispersal (Janzen, 1984).
Larger mammals often ingest larger quantities and more varieties of fruit (Herrera, 1989; Malo and Suárez, 1995). Due to their more extensive home ranges [e.g. 43·5 km2 for black bears (Trent, 2010); 7·2 km2 for yellow-throated martens (Grassman et al., 2005)], these larger mammals are able to deposit undamaged seeds over relatively greater distances, and are thus regarded as particularly effective vectors of endozoochory (Malo and Suárez, 1995; Jordano et al., 2007). This wider dispersal of seed by larger dispersal agents consequently has the potential to play a substantial role in determining large-scale ecological processes, such as population diffusion, gene flow between populations and the colonization of unoccupied habitats (Herrera, 1987; Hamilton, 1999; Jordano et al., 2007). Additionally, because frugivorous carnivores and herbivorous mammals tend to defecate at particular sites (e.g. latrines), a proportion of the seeds they deposit may arrive at advantageous microsites (Nakashima et al., 2010), and thus contribute to directed seed dispersal (Wenny and Levey, 1998), which may influence plant recruitment patterns and species diversity.
In terms of habitat integrity and forest management, we found that degraded forest was utilized less by dispersal agents, and this could thus limit H. dulcis seed dispersal and future population viability. Declines in frugivore distribution and abundance have been noted to disrupt plant–animal mutualisms and inhibit plant recruitment (Palmer et al., 2008; Wotton and Kelly, 2011), which appears to be particularly detrimental for plants which depend on large animal dispersers (Hansen and Galetti, 2009). Deficiencies in seed dispersal, caused by the fragmentation of high quality habitat by degraded zones (Cordeiro and Howe, 2003), as well as other human activity [e.g. hunting (Wright et al., 2000), conditions promoting colonization by new plant and disperser species (Christian, 2001) or the introduction of invasive species (Traveset et al., 2012)], may expose fruiting plants to risk. This is a particular issue for those plants that produce large fruits and seeds and thus depend predominantly on megafauna (e.g. large frugivores) for dispersal function (Hansen and Galetti, 2009). This can have direct and indirect ecosystem cascade effects for pollinators, dispersal agents and the plants that they serve and, in small forest remnants, could drive species to local extinction (Cordeiro and Howe, 2003; Estes et al., 2011). Forest degradation (Fortuna and Bascompte, 2006) and megafauna extirpation (Hansen and Galetti, 2009) in combination can therefore exacerbate loss of forest integrity.
In conclusion, with this first reported example (of which we are aware), we expand on knowledge of the evolved ecological function of peduncles, elaborating on how peduncle sets in dry-fruited plants function to elicit endozoochorous seed dispersal. Hovenia dulcis peduncles appear to have evolved primarily to facilitate large animal endozoochory, a mutualistic association similar in function to fleshy pulp or foliage (Van der Pijl, 1982; Janzen, 1984; Moore, 2001). The fleshy peduncle with an outer dry drupe, a relatively infrequent diaspore form in angiosperms, occurs commonly in several families (e.g. Podocarpaceae, Anacardiaceae, Icacinaceae and Rhamnaceae), in which a dry drupe develops at the end of a fleshy pedunculate fruit stalk. Generalizing the evidence of our study across these families implies that it is oversimplistic and inconsistent with broader evidence to presume that the function of the fleshy peduncle is solely to attract epizoochory (Maxwell, 1994; Kopachon et al., 1996).
This study provides an elegant example of how functional sets of plant organs can undergo natural selection to facilitate seed dispersal, highlighting the importance of endozoochory in generating diaspore divergence for dry-fruited plants, which may have been underestimated in the past. Consequently, future studies should consider how peduncle sets act as a dispersal mechanism, particularly for those plants that possess relatively large accessory fruits, in order to assess the global significance of this plant–animal interaction, and explore the origins of fleshy peduncles and related evolutionary mechanisms.
These findings enhance our understanding of the broad paradigm of the evolution of seed dispersal mechanisms, with particular regard to plant–mammal co-adaptation (Jordano et al., 2007; Lomascolo et al., 2010). Unfortunately, many large mammals in Asia are threatened currently with local extirpation due to human overexploration and habitat destruction (Corlett, 2007), which could result in broad consequences for forest ecosystem functionality (Hansen and Galetti, 2009).