Sample collection

Foliage from 14 species representing the nearest living relatives of Mesozoic flora from the families of the Equisetaceae (2 species), Marattiaceae (2 species), Osmundaceae (1 species), Cyatheaceae (1 species), and Araucariaceae (8 species), as well as the marine angiosperm family of eelgrasses, Zosteraceae (1 species) (Table 1), was collected. In the case of the Equisetaceae, above-ground green shoots were harvested, while leaf-bearing twigs were collected from the Araucariaceae. Individual fronds were taken from the Marattiaceae, Osmundaceae, and Cyatheaceae. In two species, Marattia attenuata and Cyathea cooperi, a distinction was made between “older” and “younger” leaves. All samples were clipped from the plant using garden shears, except for two clumps of Zostera marina leaves that were picked up during low tide.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Specimens collected and evaluated using the Hohenheim gas test.

https://doi.org/10.1371/journal.pone.0291058.t001

All specimens were collected from the Huntington Library, Art Collections, and Botanical Gardens in San Marino, California, USA, except for Zostera marina, which was collected 50 m north of La Jolla Pier on the southern coast of California. Sampling initially occurred in April 2018 (hereafter called “spring”), then a second collection was gathered from the same individual plants in September 2018 (“fall”). Zostera marina was collected in spring but was not collected again in the fall. Specimens were dried for 48 hours at 60°C as soon as possible after collection, following the protocol established by Menke and Steingass (1988). Four specimens had not dried completely and were thus heated at 60°C for an additional three hours (spec. nos. 18-28-3; 18-28-4; 18-28-7; 18-28-8). The dried samples were then milled successively using 3 mm and 1 mm sieves using a Retsch SM1 cutting mill (Haan, Germany).

Gas production

The digestibility of each species was determined using a modified version of the in vitro Hohenheim gas test [9, 10] at the Institute of Animal Science, University of Bonn, Germany. Rumen fluid was collected from either ruminally cannulated cattle (steer; trials 1 and 2) or ruminally cannulated sheep (trials 3 and 4) fed a standardized diet (hay and concentrate; 70:30) according to the animal maintenance requirements. The animals used for obtaining liquid ruminal content were kept in strict accordance with the German Animal Welfare legislation. All experimental procedures were approved in advance by the Animal Care Committee of the state of North Rhine-Westphalia in Germany (File Number 84–02.04.2017.A247 [cattle] and 81–02.05.40.18.008 [sheep]).

An nXP buffer solution (6 g NH₄HCO₃ + 33 g NaHCO₃ per liter) was prepared according to the mixture outlined by Menke and Steingass [10] but modified to the concentration recommendations of Liu et al. [13] to accommodate the long incubation period. The milled plant material (200 mg ± 10 mg), rumen fluid containing fermentative microbes, and buffer solution were loaded into sealed syringes, then incubated in a rotating syringe rack at 39°C for 96 hours to allow for maximum potential digestion, thereby removing time as a limiting factor. Gas volume, in milliliters, was recorded at the 4, 8, 12, 24, 32, 48, 56, 72, and 96-hour marks and subsequently corrected for blank gas production and expressed as ml/200 mg plant dry matter (DM).

Four total replicates of each sample were incubated in four runs on different weeks, except for the spring Osmunda regalis (spec. no. 18-28-6) and young spring Marattia attenuata (spec. no. 18-28-7), which were each tested in two runs due to low sample volume. One sample, the fall leaves of Araucaria heterophylla “glauca” (spec. no. 18-26-33), leaked during the third run and was retested with two replicates, for a total of five complete data sets. Three blanks containing rumen fluid–buffer solution without samples, three hay standard samples, and three concentrate standard samples were included in each run for the calibration of results. The hay standard and concentrate samples were obtained from the Department of Animal Nutrition at Hohenheim University in Stuttgart, Germany.

Statistical analyses

The parameters for gas production were fitted by a non-linear regression using PROC NLIN in SAS [14], and the following model was used to plot the kinetics of gas production in Excel 2016 [15]: wherein Gp represents total gas production (ml/200 mg DM), a + b is the maximum gas production (ml/200 mg DM), a is the initial gas production of the rapidly digestible fraction (ml/200 mg DM), b is the gas production of the remaining slowly digestible fraction (ml/200 mg DM), c is the rate of gas production (h^-1), and t is time (h) [16, 17]. Differences between net gas production between species, family, season, and maturity were calculated using ANOVAs and Tukey post-hoc tests using R version 4.2.2 [18] with a significance threshold of 0.05 (cf. [19]). Results were plotted using Excel 2016 [15].

Repository

The milled plant material from this experiment is archived in the collection of the Institute of Animal Science, at the University of Bonn, Germany.

Digestibility of Equisetum

Equisetum, as previously shown by Hummel et al. [1] and Gill et al. [12], continues to outperform all other groups in energy yield. In the current study, E. hyemale had the highest 96-hour gas production of any species, with a spring volume of 72.0 ml/200 mg DM (Table 2). The maximum gas production values for this species are marginally exceeded only by E. hyemale grown by Gill et al. [12] in 1200 ppm atmospheric CO₂ conditions, which reached 72.4 ml/200 mg DM. However, Gill et al. [12] determined that E. hyemale was not significantly affected by CO₂ concentration, which is further supported by our samples which grew outdoors in ambient CO₂ conditions yet achieved markedly similar digestibility. The consistency of these findings may also suggest that our results approach the upper digestibility potential of E. hyemale. Compared to the other species in this genus, E. hyemale was found to produce significantly more gas than E. giganteum (P < 0.05), which itself produced significantly more gas than all other species included in the current study (P < 0.05).

Although the high silica content of Equisetum [cf. 21] is thought to inhibit the digestibility of foliage [22], the overall high performance of this genus reiterates it would have been a consistently nutritious option for herbivorous dinosaurs. While both E. hyemale and E. giganteum exhibit a reduction in digestibility in the fall, they are still more digestible than any other sample (Fig 3), far exceeding the standards for both modern angiosperm grasses and dicot leaves determined by Hummel et al. [23]. Furthermore, the cosmopolitan distribution of Equisetum and its relatives during the Jurassic and Cretaceous [24, 25] would have made it commonly available to the global fauna of herbivorous dinosaurs in the Mesozoic [26].

Digestibility of ferns

Among the ferns, digestibility is more inconsistent as they tend to be affected by season and leaf maturity. Hence, among the species tested in our study, familial affinity was not a reliable indication of species digestibility. As a whole, the family Marattiaceae performed more poorly than any other family tested (Fig 1), despite Angiopteris evecta being the most digestible fern species overall (Tables 2 and 3) and the third-most digestible species in the fall data set after Equisetum hyemale and E. giganteum. Osmunda regalis, at its peak in the spring, is similarly highly digestible, though it experiences a large reduction in gas production in the fall. The older leaves of Marattia attenuata are consistently the most poorly performing foliage (Fig 2A and 2B and Tables 2 and 3).

Thus, individual fern species may have been extremely nutritious forage for herbivorous dinosaurs, but this would have been greatly influenced by environmental conditions. While a broader survey of ferns could offer deeper insights into digestibility, the results produced in this study show the wide range of energy yield that ferns may have been capable of offering. Massive, bulk-feeding megaherbivores like sauropods likely would have occasionally exploited even poorly fermenting species like M. attenuata, as larger animals are typically more tolerant of poor-quality diets [27–30], although whether this is the result of an actual increase in digestive efficiency due to their large size or a result of necessity has been debated [29–31]. However, it should also be noted that the ecological preference of today’s matoniaceous ferns for closed forests and the extremely slow growing nature of Angiopteris may have made it less likely that the Matoniaceae were frequently sought out as an abundant and readily accessible food source by large, fully grown sauropods [26].

Digestibility of the Araucariaceae

As whole, the Araucariaceae are again confirmed to be among the most energetically nutritious taxa, particularly those of the genus Araucaria. The evergreen foliage of the Araucariaceae may have provided a consistent source of highly caloric food to dinosaurs that could reach the leaves of tall, mature trees, regardless of local environmental conditions. In fact, only one araucarian species, Araucaria bidwillii, was significantly affected by seasonality. Although the araucariaceous leaves are not as immediately digestible as Equisetum, the Araucariaceae, which grew in both the Northern and Southern Hemispheres during the Jurassic [26], would have provided a substantial amount of energy given four or more days for fermentation, if their nutritional qualities were similar to those in the present-day world. Large herbivores such as sauropods likely would have been capable of exploiting the fermentative behavior of araucariaceous species. Hummel et al. [1] found that the Araucariaceae were not only more calorically nutritious than the nearest living relatives of other Mesozoic plants, including cycads, podocarps, and tree ferns, but were also, over time, more digestible than the dicot leaves and grasses on which most modern herbivores feed. The results of our study confirm the high nutritional yield of these conifers, particularly those of the genus Araucaria (Fig 6A and 6B). With the inclusion of Wollemia, all three genera of the Araucariaceae have now been analyzed. While the results indicate that Wollemia is not as digestible as those of Araucaria, it is comparable in its fermentative yield and kinetic behavior to Agathis australis (Fig 6A).

Araucaria heterophylla was the most digestible member of the Araucariaceae tested in our study, with a maximum gas production of 42.3 ml/200 mg DM in the spring and 40.0 ml/200 mg DM in the fall (Tables 2 and 3). This is comparable to the maximum gas production of 41.9 ml/200 mg DM found in samples collected in 2007 (S1 Fig and S1 Table). Here, Araucaria bidwillii reached a maximum gas production between 27.7 and 37.4 ml/200 mg DM (Tables 2 and 3), whereas the samples collected from Australia had a maximum gas production as low as 24.8 ml/200 mg DM in the mature specimens, but as high as 29.1 ml/200mg DM in saplings (S1 Table) These results highlight the relatively uniform yield that these taxa can have, even between different individuals in variable climates, localities, and collection periods. Overall, the results of the current study confirm the high nutritional yield of these conifers, particularly those of the genus Araucaria (Fig 6A and 6B), though Wollemia and Agathis still provide more energy than other nonangiospermous taxa.

Seasonal effects on digestibility

Overall, four of the 16 plants tested here were found to be significantly affected by seasonality, which encompasses factors such as temperature, the amount of sunlight, day length, and precipitation received during the growing season. Of those that experienced a significant fluctuation in digestibility between the two seasons, all taxa but the younger leaves of Cyathea cooperi decreased production from spring to fall (Fig 3). The species that were affected were not related; one pertained to the conifer family Araucariaceae and one each to the fern families Osmundaceae, Marattiaceae, and Cyatheaceae. In both seasons, Equisetum hyemale and E. giganteum were the most highly digestible, while the older leaves of M. attenuata were the least digestible of all material tested. However, the relative digestibility of all other species varied between the two collection periods. For example, in spring, the most digestible fern was Osmunda regalis, but the more digestible fern in the fall collection was Angiopteris evecta (Fig 2).

Leaf maturity effects on digestibility

Another potential factor for fluctuations in digestibility is leaf maturity. Consistent with studies on the digestibility of angiosperm livestock feed plants, younger leaves are generally more nutritious [32–34]. Thus, leaf maturity may influence digestibility in ferns, as is shown in the species Marattia attenuata (Fig 5). However, maturity effects for Cyathea cooperi were found to be insignificant (Fig 5). Interestingly, maturity may interact with seasonality, as the younger leaves of M. attenuata were not affected by seasonality, although the older leaves were (Fig 3). This combination could result in a large variation in digestibility within this species, as the more mature leaves in the fall produced only half as much gas as the younger leaves did in the spring. Such a nutritional gradient within an individual, poorly digestible plant would favor herbivores that could be more selective about which parts of the plant they consume, allowing them to target the younger, more nutritious leaves and avoiding the largely indigestible mature leaves, particularly late in the growing season. Such herbivores would include younger or smaller dinosaurs that do not need to take in as much fodder each day [31]. A larger survey could be conducted to determine the broader relevance of leaf maturity and the degree to which it interacts with seasonality within these nonangiospermous species, although the inconsistency of the results here may indicate that further experiments will not yield applicable results.

Donor species effects on digestibility

The donor species of rumen fluid was not found to affect gas production. Two donor species—domestic sheep and cattle—were used in this experiment, and all donors followed diets outlined for use in the Hohenheim gas test [9]. These results are consistent with those of Cone et al. [35], who showed that gas production using cow rumen fluid and sheep rumen fluid were comparable after at least 48 hours of production. This supports the concept that the species of the donor does not influence digestive capability; rather, it is the diet of the animal that most strongly influences the composition of the gut biota and therefore efficiency in digestion [21]. As such, Hummel and Clauss [28] hypothesize that an animal that is accustomed to feeding on the plants analyzed here would most likely have gut flora that is even more efficient at fermenting such material than the gut flora taken from animal feeding on standardized angiosperm foliage.

Environmental considerations

Environmental factors, such as precipitation and sunlight related to seasonality, can drastically affect the nutritional yield of plants [36]. In our study, the site of collection at the Huntington Botanical Gardens in San Marino, California, is classified as having a mild temperate climate with warm, dry summers [37]. Climate data from the nearby city of Pasadena document a maximum temperature of 34.4°C and a minimum temperature of 11.1°C in the period between the two collection dates in 2018. Precipitation was low, as 12.8 mm of precipitation was recorded for May and none for the period from June to September [38].

In comparison, the climate during the Jurassic was quite different. According to Sellwood and Valdes [39], the Jurassic climate was warm and equable enough that forests grew near both poles, which experienced warm summers but cold, dark, and potentially wet winters. Equatorial regions would have been more even in climate all year-round. In contrast to today’s polar to tropical climate gradient, Cretaceous temperatures were up to 10°C warmer across the globe at its peak in the Cenomanian; warm-climate plants grew in the polar regions, although there is evidence of polar cooling by the Late Cretaceous [40].

Growth rings in Jurassic wood provide direct fossil evidence of Mesozoic plants responding to environmental conditions, along with providing paleoclimatological evidence for moderate seasonality in some localities. Such growth rings have been documented in the Triassic of China [41]; the Jurassic of India [42], England [43, 44], USA [45], and Mongolia [46]; and the Cretaceous of India [42], Brazil [47, 48], and Alaska [49]. In some cases, there is evidence that the trees were subjected to stress or trauma that interrupted growth for unpredictable or irregular periods of time, such as volcanism or localized drought, rather than climatic periodicity [50–52]. Interruptions and changes in the growth in response to these environmental incidents likely would have been reflected in the nutritional capacity of the edible plant parts, at least for short periods of time.

Herbivore physiology and behavior

Although sauropod dinosaurs were not ruminants like the donor species used in this experiment, which have a highly derived digestive system consisting of multiple stomachs to facilitate several oral processing and digestive phases, and it is unlikely that they would have been foregut fermenters [28], the in vitro nature of the Hohenheim gas test minimizes the involvement of the unique digestive anatomy of ruminants while still simulating the fermentative processes that would have occurred within the gut. Because of the rarity of soft tissue evidence in dinosaur fossils, there is no direct evidence as to the nature of sauropod digestive systems; however, Hummel and Clauss [28] suggest that hindgut fermentation is the most likely form of plant digestion in sauropod dinosaurs because a foregut fermenting system is complex to evolve and would not support a high intake diet nor the lack of chewing [28]. Here, the Hohenheim gas test uses foregut-fermenting ruminants as the source of digestive microbes, but it has been found that microbial populations and their functions are highly similar in most animal species, which allows for ruminant results to be applied to non-ruminant species [53]. Furthermore, data from ruminant tests can be reliably transformed to be comparable to results derived from tests using microbes from hindgut fermenters such as horses [54].

While the specific digestive anatomy of the sauropod dinosaurs has not yet been fully determined due to a lack of soft part preservation, the incredible size of these dinosaurs would have allowed for a massive fermentation chamber capable of digesting large volumes of plant material for an extended period of time. Although larger body size does not necessarily lead to increased retention times [30, 55] as once assumed [56], extended retention times would have compensated for the slow rate of digestion of plant species such as Agathis robusta, Wollemia nobilis, or Marattia attenuata (Fig 2A and 2B), as well as for large particle sizes [55] due to the apparent lack of mastication [28, 57], gastric mills [58], or other particle size reduction mechanisms. However, it is noteworthy that even the most slowly fermenting and weakly digestible species, Agathis robusta, achieved at least 56% of gas production within the first 24 hours (Fig 4), although in vivo fermentation and digestion would likely take longer without particle size reduction. Meanwhile, quickly fermenting plants with high digestibility similar to Equisetum spp. and Angiopteris evecta (Fig 2A and 2B) would have been especially important for small-bodied, young, or growing dinosaurs, that would have needed to consume a larger proportion of higher quality material for their body size [31]. Histological evidence also suggests that younger sauropods may have had higher metabolic rates than adult sauropods [59], which would require either more dry matter or more energy-dense food.

For adult sauropods, Hummel et al. [1] estimates the amount of consumed dry plant matter sauropods of different body masses and metabolic behaviors would have required based on the energetic density of the plant material, suggesting that a highly digestible diet would allow for 40% less consumed plant mass than a low-quality diet. In their estimates, the number of kilograms of dry matter required varies according to different potential metabolic rates. For example, if they possessed a similar metabolic rate to reptiles (55 kJ ME/kg BW^0.75), a 70 t sauropod would need as little as 26 kg of high-energy dry matter. However, Hummel et al. [1] believe this to be unrealistic. Therefore, metabolic rates of 280 kJ ME/kg BW^0.75 (requiring 64 kg DM/day of high energy food or 106 kg DM/day of low energy food) and a mammalian rate of 550 kJ ME/kg BW^0.75 (requiring 125 kg DM/day of high energy food or 209 kg DM/day of low energy food) were also calculated [1]. Similarly, using an intermediate metabolic rate of 280 kJ ME/kg BW^0.75, Gill et al. [12] calculates that a 10.8 t Diplodocus would only need to need 23.8 kg of Equisetum hyemale per day, as opposed to 33.2 kg of the fern species Polypodium vulgare. However, if digestibility and the energy gained from consumption of those plants were to decrease during part of the year or during times of environmental stress, it would necessitate a greater consumed volume of material or a preference for higher-quality plants to maintain the same energy intake. Conversely, an increase in digestibility would allow for the opposite: a lower consumed volume per day or a wider variety of plants of variable nutritional capacity [31].

Elephants, a modern analog for a bulk-feeding terrestrial megaherbivore, have been found to vary their dietary composition seasonally as different forms of nutrition become more or less available; this has been shown to be true in both the savanna-dwelling African elephant [60] and the tropical Asian elephant [61]. African elephants will consume a wider variety of plants during the dry season [60], whereas Asian elephants will expand or change the habitats they occupy in order to feed on more seasonally nutritional plants [62]. Additionally, while African elephants consume a larger volume of material during the wet season than the dry season, this likely results from the lower availability of resources during the dry season, which is reflected in a corresponding decrease in the African elephants’ physical condition due to nutritional stress [63]. Similar seasonal correlations have been found for other megaherbivores, including black rhinoceroses [64], greater one-horned rhinoceroses [61], giraffes [65], African buffalo [66], and even extinct giant ground sloths [67].

Among birds, which have a strong phylogenetic relationship to dinosaurs, there are few folivores because of the metabolic and behavioral requirements necessary to make such a diet viable in a small-bodied organism [68]. Those that do engage in folivory, however, display seasonal dietary behavioral changes. Ostriches, which have been previously used as an analog for sauropod dinosaurs to investigate digestive anatomy [58], have a strong preference for particular food species, but will greatly broaden their range of food sources when feeding during the dry season [69]. Hoatzin, a South American, foregut-fermenting species, modify their diets in response to seasonal changes in the availability of their preferred food sources; although their diet predominantly consists of leaves, at the end of the dry season when trees lose their foliage, they supplement their diet with flowers [70]. The speckled mousebird also compensates for seasonal food limitations by tolerating a wide variety of resources [71]. Similarly, the white-tipped plantcutter feeds on a much wider variety of species during the winter and spends up to six times longer foraging during this time because of the limited availability of evergreen species [72].

Herbivorous dinosaurs may have exhibited similar dietary behavior changes to compensate for fluctuations in the digestibility of the Mesozoic flora or other biological needs. Isotopes in the tooth enamel of Camarasaurus suggests these sauropods migrated seasonally, shedding light on how they were able to occupy a seasonally dry habitat where they would have otherwise experienced water and nutritional stress [73]. There may be further evidence of seasonally induced dietary plasticity in dinosaurs; for example, Chin et al. [74] interprets coprolite evidence of crustacean consumption by ornithischian dinosaurs as a seasonal feeding habit, possibly in response to nutritional needs induced by reproductive cycles, suggesting that even bulk-feeding herbivorous dinosaurs engaged in targeted feeding when necessary. Particularly if they resulted from reproductive-stage nutritional needs, these behavioral responses to seasonal shifts in available nutrition may have been dependent on age and sex, as they are in modern herbivores. For example, Woolley et al. [75] reports that male African elephants tolerate lower-quality diets better than females do.

Future research

Measuring other aspects of nutritional quality, such as macronutrients and C/N ratios as previously done by Midgley et al. [7], Hummel et al. [1], Wilkinson and Ruxton [8], and Gill et al. [12], may be useful in providing more insight into herbivore nutrition beyond digestibility and relative caloric value. Overall, a broader survey of plants arising from Mesozoic-age lineages would be informative, particularly of the Podocarpaceae and Cycadales. Available data on these plant groups show that the taxa that have been tested are digestively poor [1], and analyzing a broader range of species would confirm or refute these results. Additionally, further experimentation on additional living ferns with close relatives in the Mesozoic may clarify the wide range of digestive behavior seen in the Marattiaceae and Osmundaceae in our study. Testing angiosperm families with deep paleontological roots in the Cretaceous, such as water lilies (Nymphaeaceae) or palms (Arecaceae), may also provide insight into herbivory on Cretaceous angiosperms.