Many species of the Carnivora consume grass and other fibrous plant tissues

Abstract. Within the Carnivora order, the consumption of fibrous plant tissues (FPT), such as leaves and stems, is only known to serve the nutritional needs of eight species in the Ailuridae and Ursidae. Apart from the Ailuridae and Ursidae, the extent of FPT ingestion in the Carnivora is poorly understood. A literature search was conducted to compile studies containing evidence of FPT consumption in the Carnivora, primarily based on analyses of scats or gastrointestinal tracts. Among 352 studies, there was evidence of FPT consumption in any amount in 124 species, or 41%, of the Carnivora. Grass consumption was documented in 95 species, while ingestion of sedges, marine plants, bryophytes, conifers, and dicots was much less frequent. A few species showed evidence of consuming fungi or soil. Nine studies observed co-occurrences of intestinal parasites with grasses or sedges in the scats of the Carnivora, suggesting these abrasive or hairy plant tissues help to expel intestinal parasites. The relevance of consuming marine plants, bryophytes, conifers, dicots, fungi, or soil has also been underappreciated. Deliberate ingestion of FPT may be more widespread and important than previously realized in the Carnivora.

Introduction fulgens, A. styans, Tremarctos ornatus, Ursus americanus, U. arctos, U. maritimus, and U. thibetanus). Studies were included if they provided evidence of the consumption of algae, bark, flowers, fungi, leaves, soil, stems, or wood. Rarely, studies reported searching for evidence of plant ingestion but found none; these studies were not included here.
For each study, the information compiled included the species of Carnivora, category of plant or fungal tissue ingested, quantitative data on ingestion frequency, and any pertinent notes related to plant, fungi, or soil ingestion. Some studies reported or implied that evidence of FPT herbivory was observed, but explicitly excluded it from the data collection; hence, in these studies quantitative data are not available.
The studies included used a diverse array of terminologies, which were standardized as much as possible to present them as a common category here. The category algae used here includes the terms algae, kelp, and seaweed used in the studies. The category fungi used here includes the terms fungi and mushrooms used in the studies. The category fruit used here includes the terms berries and fruits used in the studies. The category grass here comprises the terms Graminae, graminoids, grass, and Poaceae. Leaves includes the terms leaf or leaves. Moss includes Bryophyta, bryophytes, and moss. Needles includes the terms conifer needles and needles. The category plant includes the terms herbaceous plants, herbs, forbs, plant(s), plant content, plant food, plant fragments, (unidentified) plant material, plant matter, plant remains, plant remnants, and Plantae. The category root includes the terms roots and tubers. Sedge includes the terms Cyperaceae, Cyperales, and sedge. The term Poales in a study was interpreted as including both the grass and sedge categories. Soil includes the terms dirt, sand, and soil. The category stem includes the terms branches, stems, sticks, twigs, and woody material. The category vegetation includes the terms bracts, casuarina needles (presumably actually referring to its stem and whorls of leaves), fibers or fibres, undigested leaves, scales, undigestible plant material, vegetable material, vegetable matter, vegetation, and vegetative. Other categories used here were equivalent to a single term found in the study, such as bark, digested grass, flowers, hair, lichens, miscellaneous, molluscs, plastic, seeds, shells, trap-pan covers, wood, and Zosteraceae. Although categories such as fruit, hair, miscellaneous, plastic, roots, seeds, shells, and trap-pan covers were not the focus of this study, they are included here when they were grouped with other forms of plant eating and not reported individually.
The frequency of occurrence (FO), the most commonly used statistic (Klare et al. 2011a), was the primary quantitative datum compiled from the studies. The FO here is the presence/absence of a plant or fungal category in each sample (usually a scat or stomach) given as a percent of the total number of samples. The FO was sometimes reported as the itemized frequency, being the number of individual food items of one category relative to the total number of food items found.
Sometimes the FO or data to calculate the FO were not provided. In these instances, other data were given such as the mass, relative frequency (RF), relative mass (RM), or relative volume (RV). The mass is the dried mass of a given item. The RF is the FO of one category divided by the sum of all the FOs, which standardizes the FOs so that the sum of all RFs totals 100%. The RM is the mass of one category divided by the sum of all masses, so that the sum of all RMs totals 100%. The RV is the volume of one category divided by the sum of volumes, so that the sum of all RVs totals 100%. Direct observations of animals feeding on FPT were sometimes provided.
Percentages were rounded to the nearest whole number, except anything less than 1% was reported as < 1%. Some studies reported data from different times or locations, but did not summarize the data. In these instances, data were summarized for each species within the particular study.
Personal observations were made of a mixed-breed dog (Canis familiaris) in Florida, USA from the ages of 4-10 during leashed walks or when the dog was roaming freely. Additional observations were made of two adult terriers (C. familiaris) in their yard in Seattle, Washington, USA (Appendix 1). These dogs were all privately owned and were only observed during their normal daily routines. No experimentation was conducted; no manipulation of any sort was enacted. No permissions or licences were necessary.

Results
The number of published studies included was 357 (Table 1), with some studies including multiple species. Five studies reported only the consumption of fungi but not FPT (Delibes 1978;Grenfell & Fasenfest 1979;Zielinski et al. 1999;Helldin 2000;Mattson et al. 2002). From 352 studies, there were a total of 124 species and one hybrid from 72 genera and 12 families of Carnivora that showed evidence of consuming FPT (Fig. 1). Eight references were provided as dietary reviews of the eight species already well known to consume FPT, the species in the Ailuridae (Ailurus fulgens and A. styani) and Ursidae (Ailuropoda melanoleuca, Tremarctos ornatus, Ursus americanus, U. arctos, U. maritimus, and U. thibetanus). The remaining 344 studies documented FPT consumption in 116 species not generally considered to be folivores or algivores for nutritional needs. For seven species among seven genera  Nyakatura & Bininda-Emonds 2012) with the number of species in each genus (Jackson et al. 2017;Burgin et al. 2018), and the number of species with evidence of fibrous plant tissue (FPT) consumption in green. The number of studies reporting evidence of FPT consumption in each family is given beneath the list of genera. Asterisks denote families well known to consume FPT, for which the number of studies anent FPT is not given.
(excluded from the preceding totals), the only studies identified were too equivocal to conclude if the species had ever eaten any FPT; these species are: Galictis cuja, Herpestes ichneumon, Ictonyx striatus, Leopardus wiedii, Melogale moschata, Neovison vison, and Procyon cancrivorus. For Leopardus and Procyon, other species of these genera were noted to ingest FPT while no other species were included for the other five genera (Galictis, Herpestes, Ictonyx, Melogale, and Neovison). Thirty-eight studies quantifying data on diet mentioned the occurrence of FPT but excluded it from the data collection. No pertinent studies were found for the Caniformia family Odobenidae, and none were found for the Feliformia families Eupleridae, Nandiniidae, and Prionodontidae.
Among the Caniformia, it is well known that the Ailuridae primarily feeds on bamboo grass. Consumption of grass or other FPT was observed in all genera of the Canidae and 75% of its species from 153 studies. Three studies excluded FPT consumption from data collection in Nyctereutes procyonoides.
From six studies on the Mephitidae, unidentified plants were relatively frequent for three species: Conepatus chinga, Mephitis mephitis, and Spilogale putorius. In 50 studies, grass and unidentified plants were found for some species of the Mustelidae. Needles of Pinaceae were common in one study on Pekania pennanti.
Among the Otariidae, eight studies were identified. Algae and Phyllospadix were commonly consumed by three species of Arctocephalus; algae consumption was reported less often in Otaria bryonia and Zalophus californianus. Fourteen studies documented algae consumption in some species of the Phocidae; algae consumption was detected in pups and juveniles of Pagophilus groenlandicus and two species of Phoca.
Bassariscus astutus (Procyonidae) showed evidence of FPT consumption in seven studies and in one study conifer ingestion was frequent. In two studies on Procyon lotor, grass ingestion was apparently quite common. Plant eating is well characterized for most species of the Ursidae, except for the following observations. Consumption of leaves and sticks was documented in Helarctos malayanus. Among three studies, FPT consumption was uncommon for Melursus ursinus. Data from one study showed Ursus arctos ingested several different species of fungi.
Among the Feliformia, 89 studies provided evidence of FPT consumption in the Felidae, covering most of the felid genera and about 57% of its species. Several studies excluded grass from data collection for the genera Herpailurus, Leopardus, Leptailurus, Panthera, and Puma. Consumption of grass and unidentified plants was evidenced in 18 studies on the Herpestidae. From eight studies, FPT consumption was relatively common in all four species of the Hyaenidae. From 22 studies, consumption of grass and other plants was detected in about 24% of the species of the Viverridae.
Grass was the most frequently observed FPT consumed, and usually it was found in relatively small amounts. Kept as pets, several studies on Canis familiaris and one study on Felis catus found grass ingestion was common. In captivity, Chrysocyon brachyurus (Barboza et al. 1994) was observed to eat grass. Several other species in captivity were noted to eat grass, but it was unclear if evidence was based on the direct observation of feeding behaviour and/or samples of scat and vomit (Buck in Lonsdale 2001).

Discussion
The consumption of leaves or other FPT is widespread in the Carnivora, occurring in at least 124 species (ca. 41% of the Carnivora species). Eight of these species are in the Ailuridae and Ursidae, the only two families generally considered to contain folivores that serve their nutritional needs from FPT consumption. The other 116 species are carnivores and omnivores that are not known to derive nutrition from folivory; their consumption of FPT is here supported by 344 studies. The majority of the studies concerned the Canidae (153 studies), the Felidae (89 studies), and the Mustelidae (50 studies). It is noteworthy that FPT consumption was found not only in omnivores and herbivores of the Carnivora, but also in many predominantly carnivorous species such as those of the Felidae. While the consumption of FPT serves the nutritional needs of the Ailuridae and Ursidae, the purpose of this behaviour in other species of the Carnivora, for the most part, can only be speculated. The consumption of grasses and sedges, marine plants, conifers, bryophytes, dicots, fungi, and soil by species of the Carnivora is most often likely a deliberate behavior. Sometimes these materials were ingested relatively frequently and sometimes in relatively large amounts.

Grasses & Sedges
Grasses (Poaceae) were the most frequently consumed FPT among the Carnivora, being documented in 95 species and one hybrid of the Carnivora (Table 1). Sedges (Cyperaceae) were identified in a few studies but they may have been overlooked in other studies because of their resemblance to grasses. Regardless of the possible confusion between grasses and sedges, grasses are likely ingested more often than sedges in the Carnivora, given that several studies identified the grass genera consumed, but rarely were genera of Cyperaceae indicated (Bothma 1966;Kok & Nel 1992). Presumably the studies included here were reporting observations of the leaves or stems of grasses and sedges, unless their roots or seeds were specifically noted (Table 1).
While grasses are a staple food in some species (of the Ailuridae and Ursidae), in other instances, ingested grass leaves may serve to expel intestinal parasites, which is supported by observations from nine studies on eight species of the Carnivora. The earliest insight into this phenomenom may be that of Murie (1944), who observed that some scats of Canis lupus contained both grass and roundworms (presumably Toxocaridae). Murie thought that the grass may act to scour and remove the parasites. One scat of Panthera tigris contained both grass and tapeworms (Schaller 1967: 280), and Schaller noted the similarity to Murie's earlier observation. Kuyt (1969) found one fresh scat of Canis lupus consisting of a solid mass of grass containing tapeworms (Taenia). One scat of Cuon alpinus had two different kinds of plants, grass and the leaves of Lantana, that were together mixed with three tapeworms (Taenia) and mucus (Johnsingh 1983). Toweill & Maser (1985) observed that some scats of Felis concolor consisted almost entirely of grass entwined with tapeworms. Gilbert (in Huffman 1997) observed in the fall, before hibernation, mature Carex spp. being consumed by Ursus arctos and subsequently the scats being composed of masses of long tapeworms. Makundi (in Huffman & Caton 2001) reportedly observed the expulsion of Ascaris toxicara roundworms after dogs (Canis familiars) consumed grass. Su et al. (2013) found a significant correlation between co-occurrences of grass and Toxocara paradoxura in the scats of Viverricula indica, also providing photographic evidence (Su et al. 2013: fig. 4). Similarly, Laurimaa et al. (2016) found a statistically significant positive correlation between infection with helminths (particularly trematodes) in Nyctereutes procyonoides and consumption of FPT, mostly grasses.
A few other studies hint at a possible relationship between intestinal parasites and grass consumption in the Carnivora. With about 72% of 50 stomachs of Lynx rufus containing intestinal parasites, it was also observed that grass and white cedar leaves occurred in most of their stomachs (Rollings 1945). Urban populations of Canis latrans that had higher intestinal parasite species diversity also consumed vegetation more often (probably grasses but this was not clarified), compared to non-urban populations with lower parasite diversity that consumed vegetation less often (Manning 2007). In the scats of Otocolobus manul, the rates of parasite frequency and grass frequency were very similar, but it was not indicated if these were correlated (Ross 2009).
The ingestion of grasses and sedges may serve to both 1) irritate and dislodge intestinal parasites and 2) stimulate gastric motility and secretion (Huffman & Caton 2001;McLennan & Huffman 2012). The morphological features of grasses and sedges that help to stimulate the gastrointestinal tract and expel parasites are probably the hardened epidermal serrations and trichomes that are mineralized with silica (Mehra & Sharma 1965;Lanning & Eleuterius 1989;Trembath-Reichert et al. 2015). Simpson (1902) stated that cats (Felis catus) "always prefer the coarser kind of grass." Robinette et al. (1959) observed that Puma concolor ingested coarse grasses like Elymus condensatus which even livestock avoid in the winter, reinforcing that it is not the nutritional value of the grass that is important to Puma concolor. Hoppe-Dominik (1988) noted that of the 30 most frequent grass species in the region, Panthera pardus chose to ingest the hairiest two species of grasses. Su et al. (2013) described the ingested grasses as all sharp-edged and covered with trichomes. Additionally, Lantana (Verbenaceae) leaves can be strongly scabrous, and were found together with grass and tapeworms in the scats of Cuon alpinus (Johnsingh 1983). The scats of Canis anthus and Genetta genetta both reportedly contained long leaf blades of Ameplodesmos mauritanicus (Boukheroufa et al. 2020), the blades of which are rather tough and strongly serrated (Anderson & Sigaut 2014). Montalvo et al. (2020) observed three species of wild Felidae consuming Oryza latifolia, which is replete with prickles on the leaf blades (Sánchez et al. 2003). Outside of the Carnivora, other animals appear to favor hairy plant tissues to aid parasite expulsion, e.g., in Ansur caerulescens (snow goose; Holmes in Huffman 1997), Hylobates lar (gibbon; Barelli & Huffman 2016), and Pan troglodytes (chimpanzee; Wranghman & Nishida 1983;Huffman et al. 1996;Fowler et al. 2007;McLennan & Huffman 2012). This behaviour is possibly replicated in some marsupials that show evidence of grass ingestion, in three species of Dasyurus (Green 1967;Blackhall 1980;Glen & Dickman 2006Glen et al., 2009), Didelphis virginiana (opossum;Wood 1954;Hopkins & Forbes 1980), and Sarcophilus harrisii (Green 1967).
That the morphological features of grasses or sedges may help to expel parasites in the Carnivora is further supported by the observation that their leaves are often swallowed as large fragments instead of being finely chewed, suggesting they are not being consumed for digesting and assimilating nutrients. Among the studies included on the Carnivora (Table 1), ingested grass was described as undigested (Toweill & Maser 1985;Hoppe-Dominik 1988;Loveridge & Macdonald 2003;Bekele et al. 2008;Bošković et al. 2013), in well-ordered bundles (Gade-Jørgensen & Stagegaard 2000), bundled whole (Su et al. 2013), in wads (Murie 1935;Snead & Hendrickson 1942;Thompson 1952), in short lengths (Lindsay & Macdonald 1986), long blades (Haight 1937), or intact (Barboza et al. 1994;Chuang & Lee 1997;Chua et al. 2016). Other studies noting grass or sedge ingestion (Table 1) generally gave no further description of the plant tissues observed. In the marsupial Dasyurus viverrinus, ingested grass blades to 5 cm long were described as common (Blackhall 1980).
Usually, the amount of grass ingested by the Carnivora was noted to be in relatively small amounts and often the FO was not very high (Table 1). Nonetheless, even rare events of grass ingestion may be purposeful in the Carnivora. For example, the earliest known observed association between parasite expulsion and grass ingestion in the Carnivora found the FO of grasses and sedges to be only about 2% for 1,174 scats of Canis lupus (Murie 1944). On the contrary, in some studies of scats or gastrointestinal contents, grass ingestion was very frequent, with an FO of 50-100%. Further, in other studies the amount of grass found in a single scat or stomach was notably large, sometimes comprising nearly the entire scat or stomach contents (see Results; Table 1). Possibly, small amounts of grass are ingested occasionally for prevention or control of small-scale infestations of intestinal parasites, while a larger amount or more frequent consumption of grass could be indicative of heavier or more persistent parasite loads.
Grass eating may be an innate behavior in some species of the Carnivora (Bjone et al. 2009), as even well-cared for domestic cats (Felis catus) and dogs (Canis familiaris) that might be free of intestinal parasites often regularly consume grass (Hart 2008;Hart & Hart 2018;Hart et al. 2019). A long-held belief is that grasses are consumed by cats or dogs to alleviate nausea or induce vomiting (Huidekoper 1895;Cameron 1927;Powell 1957: 210;Beaver 1981;Bush 1995;Cannon 2013). Culpeper (1666: 89) wrote that "when [dogs] are sick […] they will quickly lead you to [dogs-grass]" which presumably refers to Elymus repens, a grass that can be pilose and scabrous (Szczepaniak 2009). Possibly, symptoms of an illness or nausea might sometimes correlate with intestinal parasite infection (Zanzani et al. 2014). Other historic references associate grass ingestion by dogs with vomiting (Linnaeus 1758: 39, "Vomitu a gramine purgatur"; Morell 1774: "Hound grass" under "Canaria"; Booth 1835: 290; Paulini 1834: 29, "sic canis gramen masticando vomit, luppus a fungo purgatur"). Fenn (1790) wrote that dogs eat grass to vomit, but for cats Fenn stated only that they eat grass as medicine. Recent studies show that when domestic cats or dogs consume grasses or other vegetation, they usually do not vomit nor appear to the owners to be nauseous (Sueda et al. 2007;Hart 2008;Bjone et al. 2009;McKenzie et al. 2010;Hart et al. 2019). Dudley (1892: 87) also noticed that dogs frequently ate grass without vomiting, but rather suggested that grass ingestion prevented vomiting.
Detailed quantitative data collected in controlled conditions found that vomiting is quite rare following grass ingestion in domestic dogs (Bjone et al. 2007(Bjone et al. , 2009, while more subjective reports from surveys to pet owners give a sense that vomiting is more frequent (Sueda et al. 2007;Hart & Hart 2013;Hart et al. 2019). From direct observations of 2,108 total feeding events on grass by 36 dogs (Canis familiaris), only 11 times (0.5%) did a vomiting event follow (Bjone et al. 2007(Bjone et al. , 2009. From surveys, pet owners reported that vomiting after grass consumption was relatively common in about 20-30% of domestic cats (Hart & Hart 2013;Hart et al. 2019) or dogs (Sueda et al. 2007). Possibly a greater amount of variables influences the rates reported in these surveys such as the belief that grass ingestion causes vomiting, a wider variety of breeds, confounding health issues, a wider variety of grass species encountered some of which may be more toxic, and the possibility of toxins like pesticides on grasses causing adverse reactions. In one case, grass ingestion by a poodle always resulted in vomiting, which allegedly was remedied with a high-fiber diet (Kang et al. 2007). Murie (1944) reported one incident of Canis lupus vomiting after consuming grass. Lockhart (1997) mentioned a dog (Canis familiaris) eating a different kind of monocot, chives (Allium sp.), and vomiting afterwards, speculating it was to control parasites.
There has also been the suggestion that grass consumption may help to bind cat hair (Still 1908), possibly to regurgitate hair balls (Barrs et al. 1999) or pass them in scat (Chame 2003), but strong support for this claim is lacking (Donadelli 2019). Grass was reported as a minor component of hairballs in Hyaena hyaena (Alam & Khan 2015). There is some evidence that grass may help to form regurgitated pellets in vultures (Paterson, jr. 1984;Xirouchakis 2005;Houston et al. 2007).

Marine Plants
The occasional occurrence of marine plant consumption by some Carnivora species suggests that algae have some value, but whether it serves nutritional or medicinal purposes remains uncertain. Algae were the primary FPT ingested in the marine families Otariidae and Phocidae. Algae such as Fucus and Laminaria are somewhat commonly consumed by Ursus arctos (Kistchinski 1972) and U. maritimus (Russell 1975;Stempniewicz 2017). These algal genera are known to contain significant amounts of phenylpropanoids (e.g., phlorotannins) and galactolipids (Tugwell & Branch 1992;Deal et al. 2003). Algae also appear to be regularly consumed by Vulpes lagopus (Fay & Stephenson 1989;Kapel 1999;Pagh & Hersteinsson 2008). The relative importance of algae for Lontra canadensis is difficult to determine since algae were grouped together with other plants (presumably Embryophyta) into one category (Buzzell et al. 2014). Perhaps the consumption of Phyllospadix (Zosteraceae), noted in 40% of the scats of Arctocephalus townsendi (Aurioles-Gamboa & Camacho-Ríos 2007), has nutritive value, as it was also noted in trace amounts in the scats of Ursus maritimus (Russell 1975). The phenylpropanoids of Phyllospadix might also be relevant to their consumption (Choi et al. 2009).

Conifers
Ingestion of the FPT of conifers was noted in eight species of the Carnivora. It is very doubtful that conifer leaves would support the nutritional needs of the Carnivora. The intentional consumption of conifer leaves might be due to their rich terpene content or their phenylpropanoids (Keeling & Bohlmann 2006;Faccoli & Schlyter 2007). The use of turpentine (derived from conifers such as Pinus spp.) has been historically utilized as an anthelmintic (Mclanahan 1918;Hall 1919;Le Roux 1930), which might explain the occurrence of conifers in the gastrointestinal tracts of some species recounted below. Rollings (1945) found that about 72% of bobcats (Lynx rufus) had intestinal parasites and that grass and white cedar (Thuja) leaves were found in most of their stomachs. In the scats of Bassariscus astutus, Alexander et al. (1994) noted the conifer leaves were clearly ingested but were mostly undigested. In Canis latrans, conifer leaves were considered accidentally ingested or incidentally stuck to the scat samples (Souther & Wiggers 2012) but were recorded as rather frequent in other scat samples (Santana 2010;Santana & Armstrong 2017). In Canis lupus, conifer leaves were considered undigestible and unintentionally ingested (Śmietana & Klimek 1993), were relatively frequent in scats (Thompson 1952;Andersone 1998), or were considered non-food items (Müller 2006). Conifer leaves were grouped together with other items into one category in studies on Lutra lutra (Bouroș & Murariu 2017), Martes foina (Apáthy 1998), Martes martes (Pullianinen & Ollinmäki 1996), and Nyctereutes procyonoides (Elmeros et al. 2018), being considered non-food (Elmeros et al. 2018), to be consumed in winter or in mixture with other foods (Apáthy 1998), or to be consumed incidentally with carrion (Golightly et al. 2006).

Bryophytes
Reports of moss consumption by the Carnivora are few. Mosses are thought to have low digestibility, even for herbivores (Prins 1982;Ihl & Barboza 2007). Possibly, the secondary metabolites of bryophytes, such as the terpenoids or phenylpropanoids (Peters et al. 2018), have medicinal effects in the Carnivora, or the high concentration of essential fatty acids are nutritionally important (Prins 1982). A few investigations have explored the anthelmintic activity of mosses (Gamenara et al. 2001;Roldos et al. 2008;Kumari 2015), but the pertinence to potential activity in the Carnivora requires further inquiry.
Ursus maritimus occasionally consumes mosses (Russell 1975;Gormezano & Rockwell 2013;Stempniewicz 2017). It was implied that the bryophytes consumed by Bassariscus astutus were well masticated and heavily digested (Alexander et al., 1994). Mosses were only consumed in the spring by Cerdocyon thous (Pedó et al. 2006) and were similarly infrequent in the scats of Canis latrans (Santana & Armstrong 2017). Perhaps the most intriguing report was that in two stomachs of Urocyon cinereoargenteus, mosses made up the entirety of the contents (Hatfield 1939).
Two canids ingested succulent plants in Africa. Canis mesomelas consumed FPT of the plant genera Arthraerua, Psilocaulon, and Zygophyllum (Hiscocks & Perrin 1987) and Cynictis penicillata consumed Chortolirion (Zumpt 1968). Perhaps the water content or the secondary metabolites of these plants are important to these canids.
Unknown woody material was exceptionally frequent in a study of Vulpes vulpes (Stepkovitch 2017). Oxalis bulbs were ingested by Otocyon megalotis . As previously mentioned, scabrous Lantana (Verbenaceae) leaves were found with grass and tapeworms in a scat of Cuon alpinus (Johnsingh 1983). It also well known that the Felidae may ingest Nepeta cataria (catnip), but their behavior appears to be primarily concerned with the odor and not the ingestion of the plant (Tucker & Tucker 1988).

Fungi
Fungi are occasionally consumed by the Carnivora, probably for nutritional properties but their secondary metabolites cannot be discounted. Fleshy fungi were consumed by nine species of the Carnivora and lichens were ingested by five species (see Results).
Lichens typically have tough thalli and presumably they would not be easily digested by the Carnivora. While lichens might provide some nutrients to carnivores (Dubay et al. 2008), the secondary metabolites of lichens may be more relevant (Nybakken et al. 2010), though little is known concerning potential medicinal or anthelmintic activity in the Carnivora. Canis mesomelas was directly observed once ingesting the crustose lichen Caloplaca (Hiscocks & Perrin 1987). Lichens formed the bulk of one scat of Martes americana (Marshall 1946). It was implied that the lichens consumed were well masticated and heavily digested in Bassariscus astutus (Alexander et al. 1994). The amount of lichens consumed by Vulpes lagopus was unclear since lichens were lumped into one category with leaves, mosses, and twigs (Pagh & Hersteinsson 2008).

Soils
Ingestion of soils or rocks were occasionally noted in the Carnivora. Geophagy has been speculated to alleviate gastrointestinal problems such as parasites or toxins, or provide minerals (Beyer et al. 1994;Wilson 2003;Krishnamani & Mahaney 2009). Like grasses or other scabrous plant tissues, rocks and soils could conceivably mechanically irritate and remove parasites. Schaller (1967: 255) described a scat of Panthera tigris consisting of a number of tapeworm segments and a small amount of soil. Soil ingestion appears particularly common in Panthera tigris (Johnsingh 1983), sometimes being noted in most of the scats and some scats having fairly large amounts of soil (Powell 1957: 211;Schaller 1967: 280;Sunquist 1981;Khan 2008). Eadie (1943) noted that occasional scats of Vulpes vulpes were almost wholly soil or gravel. The presence of rocks in the scat of Lynx canadensis was stated to be "most surprising" (Hanson & Moen 2008). Dirt was in 28% of Gulo gulo scats (van Dijk et al. 2007) and in 24% of Panthera pardus scats (Andheria et al. 2007). Earth, gravel, pebbles, and non-grass FPT were encountered with surprising frequency in the scats and stomachs of Canis latrans (Bond 1939), and in one scat was a large amount of dirt (Haight 1937). Kuyt (1969) described that several scats of Canis lupus were entirely made of an unidentified material resembling dried clay, but then speculated it was undigested material from animal prey. The frequent presence of soil in the scats of Procyon lotor was thought to be obtained from the crops and gizzards of bird prey (Thompson 1952). Clearly, some species of the Carnivora consume soil but the reasons for this are unclear.

Is ingestion of FPT accidental?
An overwhelming majority of the 344 studies provided almost no interpretation or discussion concerning the evidence observed of FPT consumption in the Carnivora. About one-sixth of the studies suggested that the consumption of FPT by the Carnivora was accidental or incidental. In contrast, about one-sixth of the studies suggested or concluded that FPT consumption was intentional.
In the studies that favored interpreting the ingestion of FPT as unintentional, the most common explanation given was that the predator incidentally consumed the FPT present in the gastrointestinal tracts of prey. Other reasons given were that the FPT was consumed from herbivore dung, during grooming, or from material near prey. It was occasionally speculated that FPT detritus on the ground had become externally stuck to a scat. One study described "digested grass" to imply it originated from the digesta of the prey, contrasting it with the undigested grass consumed (Loveridge & Macdonald 2003). Nonetheless, in the two Carnivora species studied, the digested grass had a FO of 2%, while undigested grass had a FO of 45-47% (Loveridge & Macdonald 2003).
Behavioral observations indicate carnivores typically avoid the gastrointestinal tracts of prey (Thompson 1952;Schaller 1967;Jobin et al. 2000;Buck in Lonsdale 2001: appendix B) or eat the tissues of the gastrointestinal tract but avoid the digesta of large herbivorous prey (Johnsingh 1983;Fabregas et al. 2016). Peterson & Ciucci (2003: 123) stated that the digesta "is of no interest to" Canis latrans, but that they may consume the stomach lining and intestinal wall. However, Wade & Bowns (1985) stated "the milk-filled stomach is a preferred item" for Canis latrans. Among several captive species of Carnivora, it was observed that the gastrointestinal tract and its contents are typically avoided, except it was alleged that Lycaon pictus may eat a small amount of the digesta (Buck in Lonsdale 2001: appendix B). Nonetheless, in captivity, Lycaon pictus was also presumably observed to eat grass (Buck in Lonsdale 2001: appendix B). Black bears (Ursus americanus) were described as "clean […] delicate feeders [whereby] most debris is either spat out or avoided" (Bacon & Burghardt 1976). For the Felidae, it had been stated that their "feeding pattern is relatively neat" (Wade & Bowns 1985). As some smaller prey may be consumed whole (Buck in Lonsdale 2001: appendix B), possibly digesta and FPT consumed from small prey could have been detected in carnivore scats or stomachs. The corms of a grass (Melica) in the scats of Canis latrans were thought to derive from the cheek-pouch contents of rodent prey (Murie 1935). The consumption of seeds from prey intestines, a form of diploendozoochory, has been considered plausible, although proof that seeds have actually been consumed from prey intestines by wild animals is wanting (Hämäläinen et al. 2017).
It also been said that the ingestion of FPT or other items was due to the animals being trapped. Gipson (1974) stated that trapped Canis latrans "tend to chew and swallow almost anything within reach" as a reason to exclude collecting data on ingestion of FPT. Similarly, in the marsupial Sarcophilus harrisii, it was explained that the animals probably chew and ingest grass while trying to escape; on the contrary, the same study implied that the evidence of grass ingestion by Dasyurus viverrinus was derived from prey, i.e., the stomach contents of wallabies (Green 1967).
The feeding habits of the Carnivora suggest that accidental ingestion of FPT is a poor explanation for its frequent occurrence in scats or gastrointestinal tracts of the Carnivora, especially when there is a lack of direct evidence that FPT are indeed accidentally consumed. Moreover, there are direct observations of species of Carnivora deliberately eating FPT (e.g., Montalvo et al. 2020;see Results). Further, it would be disadvantageous for carnivores to be imprecise in their eating habits (e.g., incidentally consuming prey digesta), which could potentially increase their exposure to infectious diseases or toxins.

Conclusions
Plant eating is widespread in the Carnivora, and includes frugivory, granivory, rhizovory, nectarivory, and folivory. Well over 100 species of the Carnivora deliberately ingest leaves or other FPT, for a variety of purposes. Grasses and sedges are especially useful to the Carnivora, in many cases ostensibly to manage intestinal parasites, as plant leaves with abrasive or hairy structures mineralized with calcium or silicon (Lanning et al. 1958;Lanning 1961;Kaufman et al. 1981;Dayanandan 1983;Lanning et al. 1980;Lanning & Eleuterius 1989;Weigend et al. 2018) appear to be most sought after to mitigate intestinal infections. While control of intestinal parasites is a plausible explanation for the ingestion of abrasive or hairy plants (or perhaps soil), additional focused research is desirable to corroborate this. Fresh and old scats might both be useful to observe this potential association (Napoli et al. 2016).
The FPT of marine plants, bryophytes, conifers, and dicots are deliberately consumed by some species of the Carnivora, but it is unclear how it may affect their fitness. Many of these plants lack scabrous structures and probably are relative undigestibile, giving cause to consider that their secondary metabolites may have some value, such as anthelmintic properties (e.g., Quinlan et al. 2002;Katiki et al. 2011;Ndjonka et al. 2014;Romero-Benavides et al. 2017;Spiegler et al. 2017;Liu et al. 2020). Several other reasons were provided to explain the ingestion of FPT in the included studies (Table 1), such as the FPT acting as a food, source of minerals or vitamins, toxin elimination, water source, antiinflammatory, hair elimination, maintenance of the gastrointestinal tract during starvation, or a digestive aid (e.g., for bones, food, hair, or skin). These possibilities also bear consideration. The consumption of fungi or soil also requires further investigation to understand their role and value. The consumption of FPT, fungi, or soil could also be an exploratory behavior that does not always increase fitness. Since diet influences the gut microbiome (Nishida & Ochman 2018), it would be of interest to explore how the consumption of FPT, fungi, or soil could influence the gut microbiome of the Carnivora.
That about 41% (123 species) of the Carnivora may consume FPT is probably an underestimate for several reasons. First, its occurrence has definitely been underappreciated and in many cases probably ignored altogether. Most studies included here were focused on the dietary analysis of carnivory and frugivory, and usually showed negligible interest concerning the consumption of FPT. It is likely that many other studies on the Carnivora found FPT but never reported it. Indeed, 38 studies reported here explicitly excluded FPT evidence from data collection. Some studies stated that the evidence of FPT consumption was only recorded if there was a relatively large amount of FPT detected (e.g., Eadie 1943;Schaller 1967;Andelt et al. 1987). Second, potential FPT consumption by other species is unknown because the ecology of many species (e.g., scat analyses) is poorly known; examples include numerous viverrids (Papeş & Gaubert 2007), Bornean felids (Mohamed et al. 2009), the canid Vulpes pallida (Brito et al. 2009), and the mustelid Bdeogale jacksoni (De Luca & Rovero 2006) which are all understudied. Lastly, this study undoubtedly failed to include all pertinent studies ever published.
To better understand FPT consumption in the Carnivora, it is requisite that more attention is paid to the species, amounts, and parts of plants ingested (for exceptionally detailed analyses of consumed plant tissues in the Carnivora see Scott 1942;Thompson 1952;Alexander et al. 1994;Santana 2010: 24), as well as the health of the animal, such as intestinal parasites. The same is true for investigating the role of fungi (Claridge & May 1994) and soil in the Carnivora. Even the absence of plants, soil, or fungi in scats or stomachs is useful information if it is explicitly stated these items were searched for but not found. If the methodologies are standardized and results are more detailed, then it will be possible to compare across studies and make inferences about conditions that lead to FPT, fungi, or soil consumption. Direct comparisons are not practical nor statistically logical among the 344 studies here (Table 1) because the methodologies and results are excessively heterogeneous. About half of the studies (Table 1) either nebulously defined what kinds of FPT were consumed (e.g., using the vague category "plant material" or "vegetation") or combined multiple discrete items into one category (e.g., grasses and fruits combined). Very rarely (~5%), was plant material treated in detail to identify the genera or species consumed (e.g., Scott 1942;Thompson 1952;Lever 1959 Sánchez et al. 2008;Nakwaya 2009;Ramesh et al. 2009;Su et al. 2013;Habtamu et al. 2017;Akrim et al. 2019;Boukheroufa et al. 2019;Montalvo et al. 2020). It is realized that often times the material may be very scant or extremely difficult to identify morphologically without intensive efforts. Availability of DNA sequencing resources will certainly be useful to identify plants or fungi consumed by the Carnivora. Plastid primers were used by Xiong et al. (2016) to identify plants in the scats of Prionailurus bengalensis, although they were unable to confidently identify plants to the species level.  Sc G eaten at all seasons; G in Sc sometimes with round worms, seemed to act as a scour; observed a male eat G, leaving a watery Sc and later vomiting some of the G.

C Ca
Canis lupus G exc --Sm 300 and 350 cubic cm of G found each in two Sm; gravel also mentioned; pine Nd and shredded W not counted. as other vegetative matter.
Stepkovitch 2017 Sc G, pebbles, and debris found substantially in all Sc and were the major category in ca. 25% of Sc (Fig. 1).

Wilson & Dookia 2019
Vulpes zerda   presence of rocks in Sc was most surprising.

Hanson & Moen 2008
Lynx lynx Sc G (Imperata) and rarely Lv found in a number of Sc, but only one Sc had G being more than 50% of the Sc volume; 51% of Sc had large quantities of soil, being more than 50% of the volume, and 80% of these found in winter.

Buck in Lonsdale 2001
Puma concolor