The large brown, round, strongly scented seeds of Ceratocaryum argenteum (Restionaceae) emit many volatiles found to be present in herbivore dung. These seeds attract dung beetles that roll and bury them. As the seeds are hard and offer no reward to the dung beetles, this is a remarkable example of deception in plant seed dispersal.
Mimicry, deception and sensory exploitation of animals by plants are controversial, especially for seed dispersal1. A few plant species produce colourful but hard seeds that are unrewarding to fruit-eating birds1,2. Evidence that this represents visual mimicry or visual sensory exploitation is weak: there is a general absence of co-occurring rewarding similarly coloured fleshy-fruited model plant species, seeds are often toxic and thus the colour is potentially aposematic (warning colouration) and birds largely ignore the seeds resulting in very poor seed dispersal1,2. Chemical deception in seed dispersal also appears to be equally rare, only being suspected in two myrmecochores (ant-dispersed plants)3,4. Plants which produce seeds with low levels of the chemical attractant (oleic acid) do not effectively deceive ants; such seeds are as weakly dispersed as non-myrmecochorous plants without any oleic acid3.
In the Potberg area of the De Hoop Nature Reserve (DHNR), in the southern Cape, we investigated the dispersal of Ceratocaryum argenteum Nees ex Kunth, an endemic Restionaceae plant species that occurs in local fire-prone shrublands on deep sands5. Seeds of this species are unusual; its nuts are the largest in the family (1 cm in length), the nuts have no elaiosome for ant dispersal and instead of having a smooth, black seed coat typical of large nuts in the family, it has a rough, tuberculate and brown outer seed coat (Fig. 1a–e). To the human nose the tuberculate layer has a pungent scent similar to herbivore faeces. Some Cape fynbos plant species with large nuts are buried by scatterhoarding small mammals6. However this does not apply to C. argenteum as neither of the only two known Cape scatterhoarding small mammal species (Acomys subspinosusWaterhouse and Gerbillurus paeba A. Smith) occurs at this site. Here the dominant small mammal is Rhabdomys pumilio Sparrman (striped field mouse)7 (see also Supplementary Information). It is a ubiquitous and widespread Cape omnivore that consumes seeds8 but does not bury them6. By using motion-activated cameras and fluorescent thread markers, we confirmed that despite being granivorous, R. pumilio never consumed or buried intact C. argenteum seeds. It frequently consumed dehusked C. argenteum seeds (Supplementary Movie 1) and intactLeucadendron sessile seeds (Supplementary Table 1).
Figure 1: The similarities between C. argenteum seeds and a dung pellet and dissimilarities with other nut seeds.
a–c, Vertical (a) and side (b) views of a C. argenteum seed as well as one that has been cracked open (c) showing the endosperm and thick woody inner seed-coat layer and the outer tuberculate layer which together form the husk. d,e, Scanning electron microscopy (SEM) of the outer, tuberculate layer and inner seed-coat, with white silicon granules at the boundary between the two layers. f, E. flagellatus. g, Bontebok faeces. h,i, Vertical (h) and side (i) views of an L. sessileseed. j, Cannomois grandis seed with white elaiosome.
A strong scent is unusual among nut seeds as it would facilitate the discovery, even of buried seeds, and thus predation by small mammals9,10. It has been argued that hard-seededness (that is nuts) evolved to reduce scent emission and thus reduce consumption by small mammals10. Therefore we hypothesized that the role of the strong scent of C. argenteum may be to chemically attract dung beetles and induce them to bury these seeds.
After the placement of experimental C. argenteum and L. sessile nuts in the field, we frequently (>20 occasions) observed that the dung beetleEpirinus flagellatus Fabricius11 (Fig. 1f) was attracted to the C. argenteumseeds. The beetles, seven individuals of which were collected for identification purposes, rolled these seeds in typical head-down dung-beetle fashion and then buried them by pulling them underground from below. Although these dung beetles are too small (1 cm in length) to activate motion-activated remote cameras, three dung beetles rolling seeds were incidentally filmed during three of the 214 video incidents that had been captured .
Dung beetles are typically more active during warm and moist periods in most South African ecosystems12. To investigate the consequences of dispersal by the dung beetle E. flagellatus, we placed out five to ten C. argenteum seeds at 31 stations (195 seeds) after a rain event. Of these, 44% (n = 87) were removed from stations within 24 h. Using fluorescent threads as markers (Supplementary Movie 3), we recovered 80% (n = 66) of the removed seeds, and of these 80% (n = 53) were found to be buried. This represents at least 27% of all experimental seeds being buried within 1 day, under these optimum moist conditions. C. argenteum seeds were typically moved 0.21 ± 0.08 m (mean ± s.e.m.) from experimental stations and buried 2.04 ± 0.21 cm (mean ± s.e.m.) deep. Seeds were mostly buried singly (n = 33), but occasionally also in larger caches (two seeds per cache n = 8; four seeds per cache n = 1). C. argenteum seeds are hard and thus provide no reward to small E. flagellatus dung beetles. We excavated all the located buried seeds within 24 h of burial and in no case did we capture any dung beetles or observe any beetle eggs on seeds or any damage to the seeds. This suggests that the deception is ‘discovered’ only when the dung beetle attempts to eat, or oviposit within, the hard seeds. As we were able to discount any seed burial due to small mammals, and frequently observed rolling and burial by dung beetles directly as well as capturing the behaviour on motion-activated videos, we are confident that this plant species is primarily dung beetle dispersed.
C. argenteum seeds look similar in size, shape and dark brownish colouration to the dung of local small antelope (Fig. 1g). Seeds are remarkably circular (ratio of widest to narrowest seed dimension = 1.02;Fig. 1a,b) whereas scatterhoarded nuts tend to be flattened (L. sessileratio = 1.86; Fig. 1h,i). This circularity would facilitate rolling. Seeds of most of the other Ceratocaryum and closely related Cannomois species are unscented, smaller, smooth, black and have elaiosomes for ant dispersal5 (Fig. 1j). We compared the volatile emissions of C. argenteumwith those of seeds of several other Cape Restionaceae species and with the fresh dung of several African herbivores, using standard headspace sampling and gas chromatography mass spectrometry methods (see Supplementary Information). Currently at DHNR, the main small antelope is the bontebok, an obligate short-grass grazer8 and the most common large antelope is the eland, a mixed feeder8.
The total mass of the volatiles emitted by fresh C. argenteum seeds (<4 weeks after seed release) was almost 300-fold greater than that emitted by seeds of other Restionaceae species and even older C. argenteumseeds (>8 weeks after seed release) emitted a significantly greater amount of volatiles (Supplementary Fig. 1). The total mass of volatiles from fresh bontebok faeces was intermediate between fresh and old C. argenteum seeds (Supplementary Fig. 1). These findings were not altered even after emissions were corrected for mass or the surface area of droppings and seeds (Supplementary Fig. 1).
The volatile composition of emissions from C. argenteum seeds is similar to that of the dung of large mammalian herbivores, particularly eland and bontebok (Fig. 2). Compounds emitted from the seeds that are also emitted by eland and/or bontebok dung include various acids, the benzenoid compounds acetophenone, phenol, p-cresol and 4-ethyl-phenol, as well as the sulphur compound dimethyl sulphone (Supplementary Tables 2 and 3). Most of these compounds are well known as components of the scent of herbivore dung13,14 (see alsoSupplementary Table 3). Similarity in scent is not due to bontebok feeding on C. argenteum as this plant is unpalatable and bontebok feed instead on various grasses (Poaceae)8.
Figure 2: Comparisons of volatile emissions among seeds and large mammal droppings that indicate the similarity of C. argenteum to the dung of some herbivores.
Similarity in the composition of volatile blends of seeds and animal droppings is based on non-metric multidimensional scaling. Symbols for other Restionaceae (Methods, Supplementary Table 3) that overlap are slightly offset for clarity. The composition of scent sampled from Ceratocaryum seeds is very similar (R = 0.75, P = 0.33) to that of dung of local herbivores (eland and bontebok), but differs markedly (R = 1.0, P = 0.028) from that of seeds of other Restionaceae (nested ANOSIM permutation
There are rare examples of dung beetle seed dispersal of dry and fleshy fruits and seeds. Acorns are rarely taken and buried by dung beetles15, but such acorns are predominantly bird-dispersed or scatterhoarded by rodents. Also, dung beetles feed on these buried acorns14, so their positive demographic impact is very weak. Dung beetles roll the fleshy fruit-covered soft seeds of Strychnos madagascariensis and they are attracted to certain of its fruit scent molecules16. However this is primarily a primate-dispersed plant species and this dung beetle is mostly associated with, and attracted to, herbivore dung16. Incidental seed dispersal by dung beetles of small hard seeds that are ingested by large animals and mixed within their dung also occurs17,18. For C. argenteum, however, the dispersal is not incidental via being mixed with dung. This plant species cannot resprout after fire and therefore depends on successful post-fire seedling establishment from an incineration-proof, buried seed-bank5. In conclusion, C. argenteumexploits the visual and olfactory sensory perception of a dung beetle for dispersal and burial of its seeds. The interaction is deceptive with no reward offered for the dung beetles.
Data were collected in January and February 2014 in stands of C. argenteum in the Potberg section of DHNR, South Africa (−34.399907°, 20.554238°). C. argenteum nuts have a mean mass of 0.710 g (n = 20), comprising the tuberculate dry outer pericarp and thick seed-coat (combined mass 0.435 g), as well as a large (0.275 g) nutritious (57.8% protein) inner endosperm and embryo. L. sessile nuts were used as controls as these highly palatable seeds are known to be scatterhoarded6. L. sessile nuts have mean seed mass 0.224 g (n = 20) and a mean endosperm mass 0.056 g that comprises 45% protein. Ltl Acorn 6210M remotely activated camera traps (n = 10) were used to observe the small mammal species interacting with the seeds. Videos were focussed on seed depots from approximately 1 m away and were set at 1 min length videos, preceded by a photograph. We considered an event as an animal having an interaction (consuming, inspecting, removing or ignoring) with a seed, or a change in interaction or a seed, or an animal entering and leaving the field of view. Thus a single event may be spread over several 1 min video clips, for example as an animal chews the same seed. Also, several events may take place within a 1 min video clip, for example if an animal inspected and ignored several seeds. Using these definitions, a total of 214 events were recorded over 3 days, with 202 R. pumilio, nine Otomys irroratus Brants (a foliovore9 which ignored all seeds) and three E. flagellatus events. To determine whether removed seeds were consumed or buried or not we attached a 10 cm length of fluorescent thread to each seed using quick setting glue. Threads from buried seeds could be located with ultraviolet-emitting light-emitting diode torches. We placed five seeds of C. argenteum andL. sessile at ten stations, with each station being ±50 m apart. Video analysis and field observations showed that C. argenteum seeds mostly were ignored by R. pumilio, occasionally moved short distances but never consumed or buried (Supplementary Table 1). All field experiments involving small mammals were approved by the UCT Science Faculty Animal Ethics Committee (2013/V15/GB).