1. Introduction
Tannins (polyphenolics) and essential oils (terpenoids) co-occur in several families well represented, if not dominant, in the Australian flora and elsewhere, especially in the families Lamiaceae, Mimosaceae, Myrtaceae, and Rutaceae [
1,
2,
3,
4]. These secondary plant metabolites are well established as antiherbivore chemicals in many parts of the world [
3,
5,
6]. Tannins are usually non-volatile and inhibit nitrogen assimilation [
7], binding readily to proteins and inhibiting digestive enzyme function and/or reducing absorption of dietary protein [
8]. Tannins are therefore potent feeding deterrents among mammals [
9]. Essential oils are volatile and aromatic. They block the stimulatory effects of sugars on palatability [
10], inhibit the action of bacteria involved in digestion and block the action of digestive enzymes [
11,
12,
13]. Formylated phloroglucinol metabolites (non-volatile combinations of condensed tannins and terpenes) are also potent antifeeding chemicals in eucalypts, the dominant tree in Australian forests and woodlands (Myrtaceae [
14]).
We examined which metabolites—tannin or essential oils—are pivotal in diet selection by the western grey kangaroo (
Macropus fuliginosus (Desmarest, 1817), and therefore their relative importance in deterring herbivory among plants. Western grey kangaroos are facultative browsers, but in the absence of grasslands, they are routinely browsers in heathlands to forests, where both essential-oil- and tannin-bearing plants dominate the ground flora [
2,
15]. The study was conducted with semi-domesticated animals at the Perth Zoo, Western Australia, using their usual diet modified to incorporate tannin, as oenotannin, and two aromatic metabolites: 1,8-cineole (eucalyptol)—common in Australian plants [
3], and pine oil—containing a number of terpenes including α pinene, present in eucalypts [
16]. Both 1,8-cineole and metabolites in pine oil are reported in the foliage of myrtaceous species with essential oils averaging 1–20% dry weight [
3], while phenolics have been noted in the range 2–17% dry weight for shrub species in a wide range of taxa and life stages [
2,
4,
17] and tannins (as a subset) up to 9% in juvenile plants [
18].
Among ringtail and brushtail possums fed a synthetic diet of 1,8-cineole and jensenone, a non-volatile formylated phloroglucinol, Lawler et al. [
19] noted that the jensenone determined how much the animal ate, while the presence of 1,8-cineole provided a learned cue about the presence of jensenone. Hunt et al. [
20] undertook a parallel trial with red kangaroos that learned to use odour cues to avoid ingesting bitter alkaloids in solution. If essential oils have an inhibitory effect in their own right, as noted above, then they should have an additional deterrent effect when combined with tannin rather than the effect remaining unchanged, as expected with a learned cueing (‘Pavlovian’) response. Thus, we tested three hypotheses: (a) the presence of tannin reduces food consumption by western grey kangaroos, (b) the presence of essential oils reduces food consumption by kangaroos, and (c) there is an additive effect when the two groups of metabolites are combined. Since the aromatic essential oils should be detected by smell and the non-volatile tannins by taste, an ancillary hypothesis was that diets lacking either or just tannin would be selected before those containing essential oils, independent of the presence of tannin.
2. Materials and Methods
The western grey kangaroos at Perth Zoo, Western Australia, were selected for use in the trials as they are accustomed to human contact, allowing for their close observation, and our previous work on this population showed that their food choice behaviour was matched with that in the wild [
20]. Presentations were made in the Australian Bushwalk exhibit, covering an area of 0.5 ha, in the feeding area frequented by six western greys, although several red kangaroos and wallabies also used the area. Kangaroo pellets (Glenforest Stockfeeds, Perth, Western Australia), the primary component of macropod diets at Perth Zoo, were modified via the addition of three metabolites: oenotannin (ellagic tannin derived from chestnut (
Castanea sativa) wood, Laffort Oenologie, Bordeaux, France, laffort.com/en/products/tanin-oenologique-2), pine oil (derived from
Pinus pinaster, containing α and β pinene, Range Products, St. Pacoima, CA, USA), and 1,8-cineole (C
10H
16O, eucalyptol BPC, Faulding, Sydney, Australia). These compounds were incorporated into the pellets during manufacture of regular kangaroo pellets (at Glenforest Stockfeeds). Pine oil and 1,8-cineole were added at a concentration of 2% pellet dry weight, while tannin was added at 10% pellet dry weight (three treatments plus control). For treatments incorporating more than one metabolite, the same concentrations were used; thus, concentrations were cumulative (1,8-cineole-tannin, pine-oil-tannin, 1,8-cineole-pine-oil, 1,8-cineole-pine-oil-tannin). Samples were sealed in plastic wrap and placed in jars immediately after manufacture. Post-trial chemical analysis of pellets yielded the same concentrations of metabolites at the beginning of the study (gas chromatography–mass spectrometry screening). The smell of essential oils, as detected by the human nose, was also similar to that noted at the time of manufacture.
The weather was mild during the trial, which was undertaken throughout a single day until all replicates were presented. Treatments were randomly allocated in groups of four, such that the eight treatments were offered with every two successive presentations (
Figure 1). This was to ensure that kangaroos were not overwhelmed by the presentation of all treatments at once. The eight treatments were numbered 1 to 8; then, their order was repeatedly randomised in fours for each pair of presentations. This was continued until (4 × 2) × 10 presentations were offered with ten random replicates per treatment. A black plastic tray (30 × 40 cm) with small plastic dishes with pellets in each of the corners (diameter 12 cm) was used per treatment. These were spaced sufficiently far apart (2 m) to allow kangaroos to consciously select between the four treatments and restrict the aroma of neighbouring pellets interfering in the selection process. Normal feeding regimes were maintained, and care was taken to keep human disturbance to a minimum.
Preliminary test trials involving the same animals were conducted to provide an indication of the level of their interest. A sufficient volume of each treatment type was required to ensure preferred pellets did not run out, leading animals to eat less-preferred types and possibly confound the results. It was concluded from the trials run during the day before the study that 100 g (25 g per dish) of each treatment per presentation was ideal, with trays able to be visited by kangaroos for a total of five minutes. Presentation time was halved if two kangaroos visited together, and so on, to quantify feeding frequencies between trays relative to kangaroo participation. The timing of presentations was paused with the departure of the feeding kangaroos and restarted when the same or another individual approached. Since other macropods reside within the enclosure, it was necessary to cancel presentations when interrupted by other marsupials. At the end of each presentation, the remaining pellets per treatment were bulked, bagged, and weighed later.
Kangaroos were observed throughout pellet presentations, and behavioural activity was noted. Order of pellet selection was recorded, and a tally was made of the first treatment selected in each presentation. A total of six kangaroos visited the trays to feed, four females and two juveniles, although in which order they visited was not monitored as we were only interested in the role of food type on feeding. While all animals participated in the trial, the four adults fed more often and apparently equally in number of visits, although this was not quantified (the distinctive markings on each animal made it easy to identify individuals). For presentations where more than one kangaroo participated, only the first treatment eaten was recorded, as selection by the second or third kangaroo was often determined by what was remaining to feed upon. String was attached to each tray, and if under visit by a kangaroo when the time was up, the tray was pulled away by the observer.
Statistical Analyses
Two-way analyses of variance were performed on consumption and feeding times for the ± tannin ± 1,8-cineole/pine oil treatments (SPSS for Mac OS X, 2002, SPSS Inc., Chicago, IL, USA). Order of pellet selection was regressed against amount consumed to indicate if selection order was indicative of overall feeding preferences. Fisher’s Exact test was used on the order of diet selection. First preference among the aromatic treatments was compared with the nonaromatic by contingency table analysis. Time spent feeding was regressed against food ingested to determine if the former could be used as a surrogate for the latter.
4. Discussion
The individual addition of three sources of secondary metabolites—tannin, 1,8-cineole, and pine oil—greatly reduces the acceptability of pellets to western grey kangaroos (
Table 1). In Australia, only specialist mammals are observed to consume foliage containing essential oils, despite its abundance in key elements of the flora over many millions of years [
21]. The kangaroos tended to avoid the 1,8-cineole- and pine-oil-rich pellets, with their aromatic properties providing an olfactory cue [
19,
20] and their bitterness apparently providing a later taste cue to their unpalatability. This was supported by the kangaroos choosing pellets lacking essential oils to taste first. In the absence of other cues as to their presence, this confirms that the large ellagitannin molecules are non-volatile, and so the pellets containing them lacked a detectable aroma. Jones et al. [
1] concluded that the absence of any herbivory by kangaroos in five of seven myrtaceous genera/species (in contrast to morphologically matched nonaromatic control species) must have been due to their emission of aromatic essential oils. Not until 1,8-cineole and pine oil were combined (4% air dry weight) was consumption reduced to negligible levels (1%) compared with their effects alone. This further supports that the intensity of the aroma was the key to feeding rather than the actual chemical composition of the essential oils, noting that the pine oil was a mixture of three compounds. While the concentration and range of essential oils are within acceptable bounds [
1,
3], this does not negate the possibility that the tannins present in aromatic species might still have an antiherbivore function as their role is different (see Introduction).
Tannin alone was highly effective at reducing consumption (and equally effective as the essential oils alone), confirming earlier feeding trials with other semi-wild kangaroos that used plant species and synthetic diets with different levels of tannin content [
22,
23]. The requirement to taste before rejection may account for part-ingestion of foliage containing relatively high levels of tannins, despite their overall negative correlation with total consumption [
22]. Time spent feeding mimicked consumption amounts, although kangaroos took longer to consume pellets if they contained tannin and/or essential oils (
Figure 2). Thus, to ingest 10 g of pellets containing these metabolites took, on average, 25% longer than to consume 10 g of the controls. Deterrence was greatest when tannin and individual essential oils were combined such that total pellet consumption approached zero. Thus, essential oils must have an inhibitory role in their own right (see Introduction), rather than just as a cue to the presence of high tannin levels, especially as tannin/polyphenolic concentrations are just as high in many nonaromatic species [
2,
4].
We accept that a sample of six animals is small. However, one advantage was that it meant that the design could be kept compact without having to prepare large quantities of the special feeds, and dishes and trays were easily handled and replaced. Second, the lack of crowding around the trays meant that the two observers could easily monitor which treatment was visited first, one of the key objectives of the trial. Third, kangaroos could move unimpeded toward the preferred treatment without other animals interfering (getting in the way physically or being aggressive) to alter the actual treatment visited. Besides, this same group of animals had been shown to match the feeding behaviour of wild kangaroos in an earlier study that was based on living plant material [
22].
5. Conclusions
Based on a synthetic diet that removes other confounding factors present in a previous study of western grey kangaroos at Perth Zoo using actual species foliage [
22], our study highlights the cumulative effect of different secondary metabolites in reducing the acceptability of potential plant food. While the detection of essential oils via smell and tannins via taste is well known [
24], this is the first time that a clear temporal aspect has been demonstrated: a diet high in aromatic essential oils is avoided first, while a diet high in tannins is only avoided later after initial tasting [
1]. Since all aromatic species contain tannins, it seems likely that they act additively to deter herbivory, independently of the target animal’s sense of smell. A negative correlation between tannin/phenolic levels and levels of herbivory is recorded frequently in multivariate trials that include aromatic species [
16,
17,
20,
25,
26,
27]. We conclude that plant consumption by western grey kangaroos is deterred by both their essential oil and tannin contents, that the effect is additive, and that essential oils are detected initially by smell (and apparently later by taste) while tannins are detected (later) by taste. This is consistent with aromatic species experiencing much lower levels of herbivory than (matched) nonaromatic species [
1].
We note that many secondary metabolites invariably have alternative/additional functions that are more related to physiological adaptations to heat and drought, and so it does not follow that they necessarily are involved in plant-animal relations. However, our work shows that the most widespread and abundant secondary metabolites in the Australian flora—tannins and essential oils—also have a vital role in reducing herbivory by the indigenous fauna (see reference list). Among western grey kangaroos, it may be significant that aromatic species are not even given a preliminary nibble before rejection, unlike those high in tannin [
1], although our results indicate that this may be concentration-dependent. Western greys are facultative browsers and readily graze grasslands in the wild and in captivity, where aromatic species are conspicuously absent. Given a choice, grass-like species (Poaceae, Restionaceae, Cyperaceae) are always preferred in the diet of this kangaroo, and these tend to occur bunched in woody vegetation [
17,
28]. If this preference has a phylogenetic basis, it might explain their unexpected aversion to the many shrubs in forests that possess essential oils that would normally be expected to be included in the diet of obligate browsers [
9,
29].
An understanding of the factors driving dietary selection among kangaroos should provide much-needed information on plant species that are suitable for parks, reserves, zoo exhibits, and semirural properties (with gardens), and rehabilitation of post-disturbed lands [
18]. Herbivore-resistant species may be identified according to the presence of particular secondary metabolites without the need for trial-and-error plantings. The formulation of suitable carrier agents and application methods for such metabolites may provide species highly sensitive to herbivory a greater chance of survival (especially at the seedling stage), reducing the need for physical protection methods such as plant guards and fencing.