Ruminants are peculiar among mammalian herbivores because they combine a foregut fermentation system with a specific sorting mechanism(Reference Schwarm, Ortmann and Wolf1, Reference Schwarm, Ortmann and Wolf2). This not only facilitates a very efficient reduction in size of ingesta particles(Reference Fritz, Hummel and Kienzle3) but also allows ruminants to consume more feed than other foregut fermenters(Reference Schwarm, Ortmann and Wolf4, Reference Clauss, Hume and Hummel5). In contrast to the historical view that this sorting mechanism operates mainly on the size of ingesta particles in the forestomach, like a simple sieve mechanism, it has more recently been understood that the sorting mechanism in the ruminant forestomach operates in particular on the density of ingesta particles(Reference Lechner-Doll, Kaske, Engelhardt, Tsuda, Sasaki and Kawashima6–Reference Baumont and Deswysen9). Because the size of actual ingesta particles is related to their density(Reference Sutherland, Dobson and Dobson10–Reference Hummel, Südekum and Bayer13), this density-dependent mechanism automatically ensures that particles are sorted according to their size, even if the separation mechanism does not discriminate particles by size itself. In addition to this density-dependent effect, a ‘filter-bed effect’ is assumed to operate in domestic ruminants. The rumen contents of domestic ruminants are usually stratified in different layers, with a ‘fibre mat’ or ‘fibre raft’ above a more fluid phase(Reference Hummel, Südekum and Bayer13). This fibre mat can additionally enhance particle retention, independent of the density-dependent sorting mechanism, by entanglement of particles in the fibre mat that acts as a ‘filter-bed’ that does not release larger particles(Reference Faichney, Milligan, Grovum and Dobson14–Reference Bernard, Chaise and Baumont16). Such a mechanism may represent an additional size- or shape-based sorting mechanism in those ruminants whose rumen contents stratify.
The influence of density on the fate of particles in the ruminant forestomach has repeatedly been investigated with plastic particles of varying size and density in domestic goats and sheep(Reference Pasha, Prigge and Russell17–Reference Katoh, Sato and Yamazaki20), and buffalo and cattle(Reference Murphy, Kennedy and Welch21–Reference Prigge, Fox and Jacquemet30). The general result of these studies is that larger particles are usually retained in the reticulorumen (RR) for a longer time than shorter particles, possibly due to a ‘filter-bed effect’, although this assumption has rarely been stated (but see Prigge et al. (Reference Prigge, Fox and Jacquemet30)). Note that in plastic particles, variations in length are specifically not linked to variations in density. Moreover, as the specific gravity of particles increases from about 0·92 to about 1·44 g/ml, their retention time decreases, or in other words, denser particles are excreted faster. This indicates that low-density particles are retained in the RR by their buoyancy and that denser particles have a higher probability of leaving the RR. However, once a certain density of approximately 1·50 g/ml is surpassed, the retention time again increases, indicating that very high densities make an expulsion from the RR less likely. This fact is recognised in the application of intraruminal devices, which reliably stay in the RR irrespective of their size, if they are at least 1·8 g/ml(Reference Cardinal31, Reference Ghirardi, Caja and Garín32).
Ruminants differ in terms of their forestomach physiology; the two extremes of this range have been termed ‘cattle-type’ (with stratified RR contents) and ‘moose-type’ (without stratification in RR contents)(Reference Clauss, Hume and Hummel5). The adaptive significance of this difference remains hypothetical and might be more related to salivary defences against secondary plant compounds in ‘moose-type’ ruminants and optimisation of microbial harvest from the RR in ‘cattle-type’ ruminants than to mechanisms of particle retention(Reference Clauss, Hume and Hummel5). Nevertheless, an absence of stratification should also translate into a less pronounced ‘filter-bed effect’.
It has been suggested that stratification and the ‘filter-bed effect’ result in particle sorting in the rumen before sorting in the reticulum in ‘cattle-type’ ruminants, whereas sorting may be limited to the reticulum in ‘moose-type’ ruminants(Reference Clauss, Fritz and Bayer11, Reference Clauss, Fritz and Bayer12, Reference Clauss, Hofmann and Streich33). Lechner et al. (Reference Lechner, Barboza and Collins34) investigated the retention of small v. large particles in muskoxen and moose and did not find a difference between the species using mordanted fibres as particle markers; additionally, faecal particle size – the ultimate measure of the efficiency of the RR sorting mechanism – did not differ between species when fed their natural forages, regardless of whether they were of the ‘moose-type’ or the ‘cattle-type’.
Mordanted fibres have the advantage of closely resembling ingesta in their physical properties; however, they only represent different size classes with similar density. In order to clearly separate the effects of size and density, we conducted additional studies using plastic particles in which the variation of size and density was not inherently linked, using domestic cattle for a pilot trial regarding the method, and muskoxen and moose to test our hypothesis. Because a series of measurements confirmed the fundamental differences between moose v. domestic cattle and muskoxen(Reference Lechner, Barboza and Collins34–Reference Clauss, Adamczewski and Hofmann38), a comparison of moose and muskoxen should be particularly appropriate for investigating the effects of different physiological adaptations on RR particle retention mechanisms. On the basis of previous results, our hypothesis was that muskoxen, with their typically stratified rumen content and thus an expected ‘filter-bed effect’, should not only display a density-dependent particle retention but also a size-dependent particle retention in the RR. In contrast, moose, lacking a rumen content stratification and thus a ‘filter bed’, should have a similar density-dependent but not size-dependent particle retention.
Materials and methods
We used four adult, fistulated domestic oxen (mean 1238 (sd 39) kg) of the Institute of Animal Science of the University of Bonn, Germany, four fistulated, adult castrated male muskoxen (276 (sd 23) kg) of the Robert G. White Large Animal Research Station, Institute of Arctic Biology, University of Alaska, Fairbanks, AK, USA and two adult, fistulated female moose (345 (sd 13) kg) of the Alaska Department of Fish and Game at the Palmer Research Center, AK, USA (Table 1). All animals had received the rumen fistulas for other studies more than a year before the present experiment. All animals were kept individually (wild ruminants in outdoor pens, oxen in a stable) with ad libitum access to water, shade and their respective feed. Adaptation periods to new diets were at least 14 d. The oxen received a diet of grass silage (n 4; trials in autumn 2007). The muskoxen received a diet of either mixed browse (n 4; Salix spp.) or grass hay (n 4; Bromus sp.) in a crossover design (two trials in June/July 2008). The moose received a diet of mixed browse (n 2; mostly Salix spp.) in June 2008 and a diet of grass silage (n 2; Bromus sp.) in October 2008 for ad libitum intake. Browse was harvested on a daily basis for the respective animals. All forages were fed whole (i.e. not chopped). Feed intake and proximate composition of the different diets have already been reported(Reference Lechner, Barboza and Collins34) and are given in Table 2.
Table 1 Animals used in the present study, location and measurements performed in each species
Table 2 Diets used and DM intake (DMI) during the feeding trials in domestic cattle (Bos primigenius f. taurus), muskoxen (Ovibos moschatus) and moose (Alces alces)
(Mean values and standard deviations)
CP, crude protein (in % DM); NDF, neutral-detergent fibre; ADF, acid-detergent fibre (in % DM with residual ash); ADL, acid-detergent lignin (in % DM).
Retention times of fluid and forage particle markers have been described previously for these animals(Reference Lechner, Barboza and Collins34, Reference Lechner, Barboza and Collins39). In addition, we applied a set of plastic particle markers similar to those described by Kaske & von Engelhardt(Reference Kaske and von Engelhardt18) and Kaske et al. (Reference Kaske, Hatiboglu and von Engelhardt19). These particles were of three different densities (1·03, 1·22 and 1·44 g/cm3) and three different lengths (1, 10 and 20 mm), with a common diameter of 0·7 mm. Polyethylene (high-pressure polyethylene 1840H; Basell, Frankfurt am Main, Germany) and barium sulphate (Honeywell Specialty Chemicals Seelze GmbH, Seelze, Germany) were mixed in different proportions (1·03 g/cm3: 86:14, w/w; 1·22 g/cm3: 69:31, w/w; 1·44 g/cm3: 55:45, w/w).
For each density, three batches with different pigments (yellow UN1750, orange UN2255, red UN3927, white UN0005, beige UN8016, black UN0055, green UN66003, violet UN5046, blue UN5001; COLOR-Service, Hainburg, Germany; at 0·5–1 % of the total mix) were produced. The material was first mixed in a tumbling mixer (and, in the case of high barium sulphate proportions, additionally by hand) and then melted at < 200°C and homogenised in a co-rotating twin-screw extruder (Teach-Line® ZK 25 T; Dr Collin, Ebersberg, Germany). The material was extruded as a long string, cooled in water and cut into small pieces using a string granulator (CSG 171/1; Dr Collin). The resulting material was homogenised in the tumbling mixer and then extruded by a single-screw extruder (Teach-Line® E 20 T; Dr Collin) at < 150°C using a nozzle with twenty-four openings of 0·7 mm each. To avoid adhesion of the individual strings, they were allowed to cool in 1·8 m vertical descent in air at ambient temperatures, fixed with adhesive tape at each 0·5 m, and coiled by hand.
Plastic strings were cut to specified lengths using paper cutters. We verified densities of the resultant particles with an Ultrapycnometer 1000 (Quantachrome Instruments, Boynton Beach, FL, USA). Low-density particles ranged from 1·02 to 1·03 g/cm3, intermediate-density particles were 1·20 g/cm3 and high-density particles were 1·37–1·44 g/cm3. The 10 and 20 mm particles were flexible.
We applied markers as a pulse dose. In the domestic oxen, the particles were placed by hand on top of the fibre mat in the middle of the rumen. In the wild ruminants, the smaller cannulae did not allow direct placement of the dose by hand. For these animals, we mixed and packed the particles into plastic tubes of the same diameter as the cannula. The mixtures of markers in the tubes were then saturated with water and frozen. The frozen mixture was then removed from the tubes and dosed through the cannula into the upper to middle layer of the rumen contents in the central (neither cranial nor caudal) region. A thawing test with a frozen marker in a 38°C water-bath resulted in complete thawing after 80 s.
All animals received the markers in the morning between 08.00 and 10.00 hours and received their morning feed directly afterwards. Domestic cattle received a dose of 20 g of the marker of each of the nine coloured markers. Although yellow and white particles were easy to tell apart in the raw state, these two colours could not be differentiated in the faeces of domestic cattle (that is, we could not differentiate yellow particles of 1 mm and 1·03 g/cm3 from white particles of 1 mm and 1·20 g/cm3). Therefore, in the trials with wild ruminants, the white marker was not used; 1·20 g/cm3 particles were thus only represented by 1 mm (black) and 20 mm (beige) particles. Also, after analysing the cattle faeces, we decided to increase the marker dose for the wild ruminants (relative to the body mass) to enhance the marker signal, so that the moose received 25 g and the muskoxen received 16 g of each of the eight coloured markers.
Three faecal samples taken from the animals before marker dosage were used for baseline values. After marker dosing, faeces were sampled at progressively increasing intervals: 4 h (day 1–2), 6 h (day 3), 8 h (day 4–5), 12 h (day 6–9) and 24 h (day 11–13); in doing so, all faeces defaecated during the time period were collected, mixed, and a representative subsample (approximately 10 % of the total sample) was taken. All samples were stored frozen at − 20°C until analysis.
For analysis, the samples were dried at 60°C and subsequently ground in a regular coffee grinder. Kaske & von Engelhardt(Reference Kaske and von Engelhardt18) had observed that this procedure did not change the size of plastic particles in the faeces, and we made the same observation. Applying a coffee grinder to dried ruminant faeces apparently only disrupts the cohesion of the dried particles but does not change their size. The dry mass of the sample was determined by weighing; the sample was washed (5–15 min) over a sieve with a pore size of 0·5 mm and dried again. The plastic particles were then sorted out by hand under bright light conditions and a degree of magnification preferred by the person doing the sorting. Plastic particles of each colour were weighed, and the concentration of marker was expressed as g particles/g faecal DM for each density and initial particle size.
Separation of plastic particles from faecal material was much more labour-intensive than we assumed from reports on this method in the literature. Neither washing with detergent solution nor treatment with 72 % H2SO4 made sorting in the remaining material easier, and preliminary attempts at separation by density were not successful due to the varying density of digested plant material. The faeces from grass-fed animals were distinctively easier to sort than the faeces from browse-fed animals. Sorting required approximately 3 h per 30–85 g of sample, in addition to the 10 min required for grinding and up to 25 min of washing before sorting. Although we originally intended that all samples were sorted by one investigator, helpers had to be employed. All 127 cattle samples were sorted by the second author, with some support from the first author. Approximately, 50 % of all muskox (n 212) and moose (n 88) samples were sorted by the second author and the rest by five additional helpers, including the first author. No distinction was made whether particles had been ruminated upon or not, but subjectively, it appeared that the majority of the 10 and 20 mm particles had been ruminated.
For the domestic oxen, the results for yellow, white and black particles were not used, due to the difficulties described earlier and due to putative difficulties in retrieving black particles. The same markers had also been given to the fistulated reindeer used in this set of trials(Reference Lechner, Barboza and Collins34, Reference Lechner, Barboza and Collins39), but the plastic markers were chewed to such an extreme fineness that manual sorting was considered too laborious.
The mean retention times (MRT) for the whole gastrointestinal tract (GIT) (MRTGIT) was calculated according to Thielemans et al. (Reference Thielemans, Francois and Bodart40) as:
where C i is the marker concentration in the faecal samples from the interval represented by time t i (hours after marker administration) and dt i is the interval (h) of the respective sample,
Faeces were sampled up to 240 h after marker dosage in oxen, 264–288 h in muskoxen and 264–278 h in moose. Apart from individual cases of small, intermediate and heavy particles in the muskoxen on a grass diet, all cases of small, intermediate and heavy particles in the muskoxen on browse and all intermediate and heavy particles in the moose, particle excretion had not terminated at the end of the sampling period. Therefore, the MRT values reported for these particles are truncated, similar to the results from Kaske & von Engelhardt(Reference Kaske and von Engelhardt18).
Several authors confirmed that fluids and particles move more or less in parallel in the distal GIT of ruminants(Reference Grovum and Williams41–Reference Wylie, Ellis and Matis44). In contrast, Siciliano-Jones & Murphy(Reference Siciliano-Jones and Murphy45) found differences in the passage of plastic particles of various density and size through the distal GIT of cattle, which might have been due to the inclusion of very high-density particles (1·77 g/ml) in their study. We followed Kaske & von Engelhardt(Reference Kaske and von Engelhardt18) in calculating the MRT for the RR (MRTRR) by subtracting the fluid MRT for the distal digestive tract(Reference Lechner, Barboza and Collins34) from the particle MRTGIT; for this procedure, the fluid MRT for the distal digestive tract was calculated as the difference between the fluid MRTGIT, calculated as described earlier, and the fluid MRTRR, as calculated by the decrease the faecal liquid marker concentration C i with time according to the equation 
or ln C i = − kt i+b (fluid MRTRR is then k− 1)(Reference Grovum and Williams46).
Data are presented as means and standard deviations. The effects on the retention time were analysed with the General Linear Models module of Statistica version 8.0 (StatSoft (Europe) GmbH, Hamburg, Germany)(47), using particle size and particle density as continuous predictor variables, and species and diet (browse or grass) as categorical factors.We controlled for intake introducing it as a covariate. The models included two-way interaction terms; when these were NS (α-level 0·05), the models were reanalysed without the interactions. The cattle were analysed separately from the muskoxen and moose, because of the differences in the marker sets described earlier. For the cattle results, interaction terms were not used because of the unbalanced experimental design, with only one particle size at the intermediate particle density.
Results
General remarks
All animals appeared to be in good health during the trials. Two muskoxen were unusually reluctant to accept the grass hay, leading to a high standard deviation in feed intake (Table 2), and consequently in the retention parameters. Interestingly, the summer/autumn feed intake of the moose hardly varied between the browse and the grass silage diet.
Cattle
In the cattle, there was a significant effect of particle size (P < 0·0001 for both the GIT and the RR; this was best seen in 1·44 g/ml particles, in which all size classes were used) and of density (P < 0·0001 for both the GIT and the RR) on the MRT, with a longer retention of larger and lower-density particles (Table 3 and Fig. 1).
Table 3 Retention time (MRT) in the gastrointestinal tract (GIT) of cattle (Bos primigenius f. taurus), muskoxen (Ovibos moschatus) and moose (Alces alces) for plastic particles of varying density and size
(Mean values and standard deviations)
Fig. 1 Retention time (MRT) of plastic particles of varying length and density in the ruminoreticulum (RR) of cattle on grass silage. □, 1 mm; 
, 10 mm; ■, 20 mm. Values are means, with standard deviations represented by vertical bars.
Muskoxen and moose
In the muskoxen and moose, a similar pattern was evident as seen in the cattle (Table 3 and Fig. 2). However, the moose had generally shorter retention times than the muskoxen. The differences between small (1 mm) and larger particles (10 and 20 mm) were distinct. Retention of the low-density particles (1·03 mg/ml) increased with particle size in the muskoxen on browse, but the differences between the two large particle classes were not distinct for other densities or diets. Retention of the low-density particles (1·03 mg/ml) was notably longer on the grass diet than on the browse diet. In the comparison of muskoxen and moose, species, particle density, particle size, diet and level of intake all had significant effects on the MRT (Table 4).
Fig. 2 Retention time (MRT) of plastic particles of varying length and density in the reticulorumen (RR) of muskoxen (Ovibos moschatus) on (a) grass hay, (b) browse and of moose (Alces alces) on (c) grass silage and (d) on browse. □, 1 mm; , 10 mm; ■, 20 mm. Values are means, with standard deviations represented by vertical bars.
Table 4 Significant effects on the mean retention time (MRT) in the gastrointestinal tract (GIT) and the reticulorumen (RR) of muskoxen (Ovibos moschatus) and moose (Alces alces) for plastic particles of varying density and size
When comparing the differences in retention in the RR between the small and large particles of any given density, the muskoxen always retained the larger particles longer compared with the smaller ones than the moose (Fig. 3(a), (c) and (e)), which is reflected in the significant interaction term (species × particle size) in Table 4. In contrast, when comparing the differences in retention in the RR between high- and low-density particles of any given size, there was no systematic difference between the muskoxen and the moose. The data for cattle (from the present study) also matched this pattern (Fig. 3(b), (d) and (f)); accordingly, the interaction term (species × particle density) was not significant in the muskoxen–moose comparison.
Fig. 3 Relationship of mean retention times (MRT) in the ruminoreticulum (RR) of muskoxen (Ovibos moschatus) and moose (Alces alces) between small (1 mm) and large (20 mm) particles of ascending density: (a) 1·03 mg/ml (
, muskoxen; ○, moose); (c) 1·20 mg/ml (, muskoxen; ○, moose); (e) 1·44 mg/ml (●, cattle; , muskoxen; ○, moose); between high-density (1·44 mg/ml) and low-density (1·03 mg/ml) particles of ascending size: (b) 1 mm (, muskoxen; ○, moose); (d) 10 mm (●, cattle; , muskoxen; ○, moose); (f) 20 mm (●, cattle; , muskoxen; ○, moose). The data for domestic cattle (from the present study) were added where available. The line represents y = x.
Discussion
The present study shows that there are not only general differences in the magnitude of the MRT between similar-sized ruminant species(Reference Hummel, Südekum and Streich48) and differences in the retention of fluid and the ratio of fluid v. small particles(Reference Clauss, Hummel and Streich49) but also differences in the mechanics of particle retention. While the influence of particle density appears to be relatively similar across species (Fig. 3(b), (d) and (f)), the difference in retention may occur in relation to particle size (Fig. 3(a), (c) and (e)).
Evidently, caution should be applied when interpreting the results from a comparison of two different individual species(Reference Garland and Adolph50). Extrapolation to general rules about ‘cattle-type’ v. ‘moose-type’ ruminants, or even further interpretation in the sense of generalisations about the digestive physiology of grazing and browsing ruminants, will require evaluation of more species. However, this may prove very difficult in practice, as it will require fistulation of more browsing and grazing wild ruminant species and the relatively laborious sorting of faecal samples as described in the ‘Methods’ section. The present results only demonstrate two different physiological strategies, which might be linked to other findings in these and other ruminant species. Additionally, the present study was limited by the low sample size (n 2) of available, fistulated moose and the unusual hesitance of some of our muskoxen to readily ingest the grass hay offered; these factors made the intake level a significant contributor to the differences in the MRT and led to a significant interaction term, diet × intake level (Table 4).
Forages ingested by ruminants, regardless of their source, exhibit common characteristics in flotation experiments: the floating fraction being consistently comprised of larger particles and the sedimenting fraction of smaller particles(Reference Hristov, Ahvenjärvi and McAllister7, Reference Sutherland, Dobson and Dobson10–Reference Hummel, Südekum and Bayer13, Reference Dardillat and Baumont51, Reference Evans, Pearce and Burnett52), most probably due to differences in adhering fermentation gas bubbles(Reference Wattiaux, Satter and Mertens53). Thus, the process of particle separation by their flotation behaviour, i.e. their density, automatically assures a sorting by particle size and leads to the uniformly small particles that escape the RR of any ruminant species(Reference Fritz, Hummel and Kienzle3, Reference Lechner, Barboza and Collins34). The main location of sorting according to density is the reticulum(Reference Lechner-Doll, Kaske, Engelhardt, Tsuda, Sasaki and Kawashima6, Reference Kaske and Midasch54), where a relatively high fluid content enables the separation by flotation and sedimentation(Reference Clauss, Fritz and Bayer11–Reference Hummel, Südekum and Bayer13). In domestic cattle, a separation of particles according to density (and hence size) has also been described for the rumen itself, where particles become segregated between the dorsal and the ventral rumen or between the fibre mat and the more liquid phase beneath(Reference Hummel, Südekum and Bayer13, Reference Evans, Pearce and Burnett52, Reference Balch and Kelly55–Reference Kovács, Südekum and Stangassinger59).
Separation due to density and size in the rumen itself has not been demonstrated in rumen contents of domestic sheep(Reference Sutherland, Dobson and Dobson10) or wild ruminants(Reference Clauss, Fritz and Bayer11, Reference Clauss, Fritz and Bayer12). Reasons for this might be that either their rumen contents are inherently homogeneous, or because particles forming the fibre mat are of a heterogeneous nature: Sutherland(Reference Sutherland, Dobson and Dobson10) and Hummel et al. (Reference Hummel, Südekum and Bayer13) suggested that low-density particles in the lower part of the fibre mat can support less-buoyant particles in the upper mat that would not stay in that position by their own buoyancy. This effect, combined with simple physical entrapment especially of elongated particles, creates the ‘filter-bed effect’, which retains particles in the fibre mat for a longer time than determined by their own disposition alone. Since the proportion of low-density particles in the lower rumen is characteristically higher on grass diets(Reference Hummel, Südekum and Bayer13), one would expect a pronounced filter-bed effect on those diets, especially in particles of lower density that are more susceptible to this effect. The present results are in accord with this expectation, with low-density particles being particularly affected by the difference in the diet (Fig. 2). Similarly, desBordes & Welch(Reference desBordes and Welch24) concluded that low-density plastic particles are especially subject to rumination, given their propensity to being retained in the fibre mat.
One factor related to the buoyancy of particles is their shape. Lirette et al. (Reference Lirette, Milligan and Cyr60) found that elongated, fibre-like particles are more likely to float than particles of a more cubic nature. Those authors found that these particles also differed in lignin content, which could not only be an indication for advanced digestion but also for an influence of fibre composition on fractionation patterns. Browse, in general, has a higher proportion of lignin in its fibre fraction than grass(Reference Hummel, Südekum and Streich48), and a different arrangement of vascular bundles (parallel in grass but branched in browse)(Reference Sanson, Grigg, Jarman and Hume61, Reference Nultsch62). Among others, these properties could be responsible for different fracture properties that lead to more elongated, fibre-like particles in comminuted grass and more cubic or polygonal particles in comminuted browse(Reference Troelsen and Campbell63–Reference Van Wieren and Van Wieren65). This pattern was also observed in the present study (Fig. 4). Others have suggested that the more cubic shapes of comminuted browse particles are less apt to form fibre mats with intertwined particles(Reference Spalinger, Robbins and Hanley64–Reference Clauss, Lechner-Doll and Streich66), thereby resulting in less retention of low-density particles in browse v. grass diets (Fig. 2).
Fig. 4 Faecal particles in muskoxen (Ovibos moschatus) fed (a) grass hay and (b) browse, and moose (Alces alces) fed (c) grass silage and (d) browse. Note the general difference in shape between grass and browse particles. The scaling is 1 mm.
Durkwa(Reference Durkwa67), as presented in Murphy et al. (Reference Murphy, Kennedy and Welch21), found little difference in retention or rumination among 1–5 mm particles in cattle. Similarly, differences between 2 and 5 mm plastic particles were not consistent between density classes in cattle(Reference Kennedy, McSweeney and Ffoulkes26). Prigge et al. (Reference Prigge, Stuthers and Jacquemet29) did not find a difference between 1 and 3 mm particles in cattle, but did find a longer RR retention for 5 mm nylon particles, and later reported results from a similar trial, wherein RR retention increased continuously from 1–3 to 5 mm nylon particles(Reference Prigge, Fox and Jacquemet30). Stetter Neel et al. (Reference Stetter Neel, Prigge and Townsend25) also found a shorter RR retention for 1 mm than for 3 mm nylon particles in cattle, and Kaske et al. (Reference Kaske, Hatiboglu and von Engelhardt19) described an increased RR retention of 1 v. 5 mm particles in sheep. In contrast to reported differences between lengths < 5 mm, differences between 10 mm and smaller particles were of a larger magnitude(Reference Durkwa67) (as presented in Kaske et al. (Reference Kaske, Hatiboglu and von Engelhardt19), Murphy et al. (Reference Murphy, Kennedy and Welch21) and in the present study), suggesting that the RR retention of 1–10 mm particles might be a continuous function of particle size, although at times difficult to demonstrate between similar-sized particles.
The differences between larger particles (the 10 and 20 mm particles of the present study) may be less distinct. Similar to the observation of the present study, that the differences in retention between the 10 and 20 mm particles were often small (Fig. 2), Kaske et al. (Reference Kaske, Hatiboglu and von Engelhardt19) observed differences in the retention of 10 and 20 mm plastic particles in the RR of sheep that only tended towards significance, and Schwarm et al. (Reference Schwarm, Ortmann and Wolf68) and Lechner et al. (Reference Lechner, Barboza and Collins34) did not find differences in the retention of 10 and 20 mm mordanted fibre particles in wild ruminants. Thus, it appears that at a particle size above 10 mm, little further contribution to retention due to increases in size should be expected. The main potential difference between the 10 and 20 mm particles might be that, at about 20 mm length, passage through the Ostium reticuloomasale is actually physically prevented by particle size(Reference Kaske, Hatiboglu and von Engelhardt19); this is in accord with McBride et al. (Reference McBride, Milligan and Turner69), who observed the passage of 10 mm particles through the Ostium reticuloomasale (but did not assess 20 mm particles). Therefore, in future studies, when the number of particle sizes that can be investigated is limited, it may be more informative to investigate a combination of 1, 5 and 10 mm particles than 1, 10 and 20 mm particles. For even larger plastic particles, Welch(Reference Welch70) demonstrated that flexible 35 and 70 mm particles could be ruminated and cleared from the rumen in sheep, whereas 300 mm particles could not.
The effect of density on particle retention in the RR is similar in cattle, muskoxen and moose, and we therefore hypothesise that particle separation based on density is a mechanism common to all ruminants. In contrast, there are differences in the effects of particle size. The effect of size may be related to a general difference between species in RR content stratification and formation of a fibre mat. The occurrence of RR content stratification is influenced by the type of forage ingested, grass material tending more towards the formation of a mat and also by the physiology of the animal. Compared with cattle and ‘cattle-type’ ruminants, moose characteristically have higher RR fluid viscosity, lower RR fluid throughput, less distinct separation of the RR retention of small particles and fluids, more uniform ruminal papillation, smaller differences in the DM content between the dorsal and the ventral rumen, weaker ruminal pillars and absence of an intraruminal gas dome(Reference Clauss, Hume and Hummel5, Reference Clauss, Fritz and Bayer12, Reference Lechner, Barboza and Collins34, Reference Tschuor and Clauss36, Reference Clauss, Lechner-Doll and Streich66). All of these characteristics are associated with limited RR content stratification and a less distinct ‘filter-bed effect’, and may contribute to the less pronounced effect of particle size on retention in the RR observed in the present study. Moose are additionally characterised by comparatively small omasa(Reference Clauss, Hofmann and Hummel35) and shallow reticular crests(Reference Clauss, Hofmann and Streich33), both linked to a low RR fluid throughput. Many of these characteristics show some degree of convergence among wild ruminant species having similar natural diets, indicating that species having digestive physiology similar to moose are browsers(Reference Codron and Clauss71).
To conclude, we propose that a more pronounced ‘filter-bed effect’ as demonstrated in muskoxen in the present study is one of several(Reference Clauss, Hume and Hummel5) advantages the ‘cattle-type’ ruminants derive from physiological adaptations that enhance RR fluid throughput and rumen content stratification. Due to differences in fermentation characteristics between browse and grass forages, grazers particularly benefit from extended particle retention produced by the ‘filter-bed effect’(Reference Hummel, Südekum and Streich48).
Acknowledgements
The present project was supported by SNF grant 3100A0-115958/1 to M. C. We thank Bill Hauer, Peter Reynoldson, Sandy Garbowski, Tabitha Hughes, Betsy Wagner and Patrick Steuer for support in animal husbandry. Experiments and handling procedures for animals were approved by the Institutional Animal Care and Use Committee, University of Alaska Fairbanks under protocol no. 07-23 (muskoxen and reindeer) and no. 07-21 (moose). I. L. thanks the Robert G. White Large Animal Research Station and the Institute of Animal Science of the University of Bonn, and M. C. the Robert G. White Large Animal Research Station and the University of Alaska Fairbanks – Matanuska Experiment Farm Palmer for their hospitality and use of their facilities. We thank Professor Michael Kreuzer, Group of Animal Nutrition at the ETH Zurich, for his logistic support, and Gabriella Worwa, Nadine Pfefferkorn, Andrius Juskys and Rahel Berli for their help in sorting particles out of the faeces. M. C., P. B. and J. H. designed the research strategy; T. A. T. designed and supervised the production of the plastic markers; I. L., P. B., W. C. and M. C. performed the trials; I. L. performed the majority of the manual analyses; D. C. and M. C. performed the statistical analyses; M. C. drafted the manuscript; all authors contributed to the interpretation of the results and critical revision of the manuscript, and all of them approved the final version. The authors have no conflicts of interest to declare.
