The authors have declared that no competing interests exist.
Conceived and designed the experiments: JW EMG. Performed the experiments: JW. Analyzed the data: JW. Contributed reagents/materials/analysis tools: JW EMG. Wrote the manuscript: JW EMG.
It has been hypothesized that a high reproductive output contributes to the unique gigantism in large dinosaur taxa. In order to infer more information on dinosaur reproduction, we established allometries between body mass and different reproductive traits (egg mass, clutch mass, annual clutch mass) for extant phylogenetic brackets (birds, crocodiles and tortoises) of extinct non-avian dinosaurs. Allometries were applied to nine non-avian dinosaur taxa (theropods, hadrosaurs, and sauropodomorphs) for which fossil estimates on relevant traits are currently available. We found that the reproductive traits of most dinosaurs conformed to similar-sized or scaled-up extant reptiles or birds. The reproductive traits of theropods, which are considered more bird-like, were indeed consistent with birds, while the traits of sauropodomorphs conformed better to reptiles. Reproductive traits of hadrosaurs corresponded to both reptiles and birds. Excluding
The discovery of the gigantic sauropods and other large dinosaurs has stimulated scientists to understand the biology of dinosaurs. Several researchers [
In contrast to any living species, our information on dinosaurs and their reproduction is limited to fossils. Unfortunately, fossils do not allow for the complete reconstruction of an organisms’ traits (e.g. of the life history). Traits are often inaccurately preserved or simply absent from the fossil record (e.g. clutches can be incomplete and breeding frequency is simply not documented in the fossil record). This hampers our understanding of the reproductive strategies employed by dinosaurs.
Equations linking body mass to other traits derived from extant taxa are commonly used to estimate these traits for extinct species, including those dealing with reproduction (e.g. [
To derive information on dinosaur reproduction, the procedure presented by Bryant and Russell [
Our established allometries between body mass and clutch mass might also provide further support for the hypothesis of Seymour [
For our analyses, we selected three extant taxa (birds N=217, crocodiles N=22 and tortoises N=20;
For the crocodile model, we chose all extant crocodilian species (N = 22). Since crocodiles are non-terrestrial, we also included tortoises (N = 20) in our allometric analyses. Molecular data suggest that turtles are more closely related to archosaurs than to lepidosaurs [
We applied established allometries to all dinosaur taxa for which body mass estimates and assignments of fossil eggs or clutches to taxa are currently available (
For all extant species, we gathered data on adult body mass (BM), egg mass (EM), clutch size (CS) and number of clutches per year (CY). When more than one trait value was found in the literature for the same species, the mean value was calculated (
For dinosaurs (for details, see
lambeosaurine | 2390 | 4.737 | 22 | 3.5 | 9.9 | 10.3 | 0.5 | 23.2 | 25.7 | 1.2 |
lambeosaurine | 3344 | 4.737 | 22 | 4.4 | 12.6 | 13.1 | 0.6 | 29.5 | 32.7 | 1.5 |
lambeosaurine | 5057 | 4.737 | 22 | 6.0 | 16.9 | 17.7 | 0.8 | 39.7 | 44.1 | 2.0 |
1500 | 1.023 | 16 | 11.6 | 32.9 | 34.0 | 2.1 | 77.1 | 84.8 | 5.3 | |
2556 | 1.023 | 16 | 16.9 | 48.2 | 50.1 | 3.1 | 112.8 | 124.7 | 7.8 | |
4079 | 1.023 | 16 | 23.7 | 67.3 | 70.2 | 4.4 | 157.6 | 174.9 | 10.9 |
BM: minimum, mean and maximum fossil body mass of a taxon taken from literature in kilograms; CS: mean clutch size observed in fossil record, with minimum and maximum fossil values given in square brackets or CS calculated from an allometric clutch mass (CM) model (CST = tortoise model, CSC = crocodile model, CSB = bird model) using the fossil egg mass (EM, for calculating of fossil egg mass see Material and Method), CS = CM divided by EM. AEN (annual egg number): total number of eggs laid per year, calculated from an allometric ACM (annual clutch mass) model using the fossil egg mass, AEN = ACM divided by EM. CY: number of clutches per year, calculated from an allometric ACM model using the fossil CS and estimated AEN, CY = AEN divided by CS. Minima and maxima are given in brackets. References for fossil data are given in
Given that all studied extant birds and reptiles lay at least one clutch per year, an initial conservative estimate of the unknown annual breeding frequency was assumed to be one clutch per year for all non-avian dinosaurs. Clutch mass is egg mass multiplied by clutch size. Annual clutch mass is clutch mass multiplied by the number of clutches per year.
We began by separately analysing the relationship between body mass and reproductive traits for extant birds, crocodiles and tortoises. For each taxa and each of the three reproductive traits, we calculated regression slopes and normalization constants using ordinary least square regressions (OLS) on log-log-transformed data (
In these regression analyses, we did not control for phylogenetic effects on reproductive traits. In general, phylogenetic comparative methods perform best when the phylogeny itself and branch lengths are correct [
Next, we tested the homogeneity of the regression lines obtained for the three taxonomic groups within each reproductive trait (
The previous analyses could either reveal no statistical differences in regression slopes between at least two taxonomic groups or a significant difference between groups for a reproductive trait. If slopes were statistically homogeneous but intercepts differed between taxonomic groups, we calculated a new OLS regression function with a common regression slope for each group. We used this common regression slope (average value of taxa) as a fixed parameter in these regression models and only estimated the normalization constant for each group. If both slopes and intercepts of taxonomic groups were statistically homogeneous, we determined a common regression function (with the average slope and average intercept of taxa) for these taxa. If a slope of a single taxonomic group statistically differed from all other groups, the initially found OLS regression function was used as allometry for the respective taxon and reproductive trait (
(
(
(
Finally, as a measure of variability in residuals and the deviation of single species from the expected average, we calculated 95% prediction intervals for each regression line (
Each of the trait pairs (body mass and reproductive trait) of non-avian dinosaurs were compared to the regression lines and to the respective 95% prediction intervals for the bird, the crocodile, and the tortoise model (
Clutch size for non-avian dinosaurs was calculated from fossil egg mass and the clutch mass estimates from the regression lines derived for birds, crocodiles and tortoises. Analogously, the total number of eggs per year (AEN) for dinosaurs was calculated from annual clutch mass (ACM). Since the regression lines for crocodiles and tortoises did not differ statistically (see results), the ACM was estimated from the regression lines of birds and the common regression line for reptiles, and from the fossil egg mass. The number of clutches per year (CY) under the bird and reptile model was calculated from the respective annual egg numbers and from fossil clutch sizes.
The calculations of clutch sizes, number of eggs per year and clutches per year, estimated from the regression models, were done with Excel 2010. All other analyses were carried out in R (Version 2.14.1 [
Reproductive investment in terms of EM, CM, and ACM highly correlated with BM in birds, crocodiles and tortoises (
Except for the prosauropod
All fossil dinosaur CMs fell within the 95% prediction interval of at least one of the extant models (
When assuming one clutch per year, all estimated ACMs of non-avian dinosaurs fell within the 95% prediction interval of at least one of the extant models, with the exception of the sauropod
Independent of the extant model used, the estimated dinosaur AEN did not exceed 850 eggs (75,000 kg sauropod) for any of the taxa studied. According to most estimations, dinosaurs lay less than 200 eggs per year and only some estimates obtained for medium to large sized sauropods were higher (
107 | 0.128 | 34 | 14.0 | 39.7 | 40.2 | 1.2 | 93.1 | 100.0 | 2.9 | |
175 | 0.128 | 34 | 19.9 | 56.5 | 57.3 | 1.7 | 132.4 | 142.9 | 4.2 | |
280 | 0.128 | 34 | 27.8 | 79.1 | 80.6 | 2.4 | 185.3 | 200.8 | 5.9 | |
5000 | 5.211 | 19 [9, 28] | 5.4 | 15.3 | 16.0 | 0.8 [0.6, 1.8] | 35.8 | 39.8 | 2.1 [1.4, 4.4] | |
22399 | 5.211 | 19 [9, 28] | 15.7 | 44.7 | 47.3 | 2.5 [1.7, 5.3] | 104.7 | 118.0 | 6.2 [4.2, 13.1] | |
75000 | 5.211 | 19 [9, 28] | 37.3 | 106.1 | 113.6 | 6.0 [4.1, 12.6] | 248.6 | 283.1 | 14.9 [10.1, 31.5] | |
5000 | 1.741 | 28 [15, 40] | 16.1 | 45.7 | 47.8 | 1.7 [1.2, 3.2] | 107.2 | 119.1 | 4.3 [3.0, 7.9] | |
22399 | 1.741 | 28 [15, 40] | 47.1 | 133.8 | 141.7 | 5.1 [3.5, 9.4] | 313.4 | 353.0 | 12.6 [8.8, 23.5] | |
75000 | 1.741 | 28 [15, 40] | 111.8 | 317.6 | 340.1 | 12.1 [8.5, 22.7] | 744.0 | 847.3 | 30.3 [21.2, 56.5] |
BM: minimum, mean and maximum fossil body mass of a taxon taken from literature in kilograms; CS: mean clutch size observed in fossil record, with minimum and maximum fossil values given in square brackets or CS calculated from an allometric clutch mass (CM) model (CST = tortoise model, CSC = crocodile model, CSB = bird model) using the fossil egg mass (EM, for calculating of fossil egg mass see Material and Method), CS = CM divided by EM. AEN (annual egg number): total number of eggs laid per year, calculated from an allometric ACM (annual clutch mass) model using the fossil egg mass, AEN = ACM divided by EM. CY: number of clutches per year, calculated from an allometric ACM model using the fossil CS and estimated AEN, CY = AEN divided by CS. Minima and maxima are given in brackets. References for fossil data are given in
34 | 0.329 | 23 [22, 24] | 2.4 | 6.8 | 6.8 | 0.3 [0.3, 0.3] | 15.9 | 17.0 | 0.7 [0.7, 0.8] | |
44 | 0.329 | 23 [22, 24] | 2.9 | 8.2 | 8.2 | 0.4 [0.3, 0.4] | 19.2 | 20.4 | 0.9 [0.9, 0.9] | |
51 | 0.329 | 23 [22, 24] | 3.2 | 9.1 | 9.1 | 0.4 [0.4, 0.4] | 21.3 | 22.8 | 1.0 [0.9, 1.0] | |
33 | 0.262 | 24 [20, 30] | 2.9 | 8.4 | 8.4 | 0.3 [0.3, 0.4] | 19.6 | 20.8 | 0.9 [0.7, 1.0] | |
37 | 0.262 | 24 [20, 30] | 3.2 | 9.1 | 9.1 | 0.4 [0.3, 0.5] | 21.3 | 22.6 | 0.9 [0.8, 1.1] | |
40 | 0.262 | 24 [20, 30] | 3.4 | 9.6 | 9.6 | 0.4 [0.3, 0.5] | 22.5 | 24.0 | 1.0 [0.8, 1.2] | |
79 | 0.473 | 22 [15, 30] | 3.0 | 8.7 | 8.7 | 0.4 [0.3, 0.6] | 20.3 | 21.7 | 1.0 [0.7, 1.4] | |
176 | 0.602 | 63 [25, 100] | 4.2 | 12.1 | 12.2 | 0.2 [0.1, 0.5] | 28.3 | 30.5 | 0.5 [0.3, 1.2] |
BM: minimum, mean and maximum fossil body mass of a taxon taken from literature in kilograms; CS: mean clutch size observed in fossil record, with minimum and maximum fossil values given in square brackets or CS calculated from an allometric clutch mass (CM) model (CST = tortoise model, CSC = crocodile model, CSB = bird model) using the fossil egg mass (EM, for calculating of fossil egg mass see Material and Method), CS = CM divided by EM. AEN (annual egg number): total number of eggs laid per year, calculated from an allometric ACM (annual clutch mass) model using the fossil egg mass, AEN = ACM divided by EM. CY: number of clutches per year, calculated from an allometric ACM model using the fossil CS and estimated AEN, CY = AEN divided by CS. Minima and maxima are given in brackets. References for fossil data are given in
Our results corroborate that body mass and reproductive investment (in terms of egg mass, clutch mass or annual clutch mass) are highly correlated in extant reptiles and birds [
In summary, our results revealed four important insights into dinosaur reproductive biology. First, corroborating our hypothesis (i), the reproductive traits of dinosaurs that are considered to be more bird-like (theropods) did indeed coincide with reproductive traits of birds. Similarly, those traits of dinosaurs that were probably more reptile-like (prosauropods, sauropods) coincided with those of reptiles. Second, although the size difference between a dinosaur egg and the egg-laying female is very impressive, for all dinosaurs studied the egg to body mass relationship was similar to similar-sized or scaled-up extant reptiles (in
Our results suggest that the egg masses of most dinosaurs match neither the egg masses of similar-sized or scaled-up birds nor those of reptiles, but were in fact in-between (
In contrast to the egg masses, all dinosaur (mean) clutch masses matched the masses of similar-sized or scaled-up birds or reptiles.
Theropods. As expected under our initial hypothesis, the bird model was the best model for theropods. Fossils indicate that at least some avian reproductive characteristics, such as adult brooding [
Varricchio et al. [
Sauropodomorpha. Contrary to our initial hypothesis (ii) clutch masses of sauropods were consistent with an extant species model, the tortoise model (
Hadrosaurs. The applicability of allometric models for clutch mass differed between the two hadrosaurs. For the lambeosaurine hadrosaur, the bird model was best, but the crocodile model was also applicable. For
A reliable estimate of the number of clutches per year and the annual egg number for dinosaurs is uncertain because of the high variability observed in traits of extant species. Furthermore, there is a high inaccuracy in body mass and clutch mass estimates of dinosaurs, making it difficult to completely rule out any of the extant models for dinosaur taxa. Irrespective of all these limitations, we are able to provide qualitative estimates for the number of clutches per year and the annual egg number laid by dinosaurs. By using further information from the fossil record, we were able to identify the most likely extant model for a taxon.
Theropods. Fossil egg masses and clutch masses of theropods are consistent with the bird model, which suggests a general applicability of allometries on reproductive traits of birds to theropods. Assuming one clutch per year for theropods, the theropod ACMs match the bird model (
Sauropodomorpha. ACMs of sauropodomorphs were lower than all “average” extant species studied when assuming the studied sauropodomorphs (
Under the reptile model, an ”average”
Altogether, our results imply that sauropods probably had several clutches per year, resulting in more than one hundred eggs per year. However, sauropod clutch and egg numbers probably did not exceed the numbers found in some recent reptile species (e.g. sea turtles [
Hadrosaurs. We could not clearly identify the most likely extant model for the ACM of hadrosaurs. The ACM to BM relation of the lambeosaurine hadrosaur was intermediary to both the bird and reptile model; for
Why should some dinosaurs have several clutches per year and others not? As seen in extant species, they could have lived in different environments, each favoring different reproductive strategies. Producing several small clutches within a breeding season could reflect a bet-hedging strategy [
Since birds are dinosaurs, there must have been an evolutionary shift in the reproductive mode from the basal reptilian/non-avian dinosaur mode to that currently observed in birds. This shift might be observable in the studied dinosaurs. As expected, all studied reproductive traits of Sauropodomorpha were more reptile-like, whereas traits of studied theropods conform well to those of recent birds. Furthermore, it is likely that within the dinosaur lineage (including birds), an increase in egg size was linked to a decrease in egg numbers per clutch/year and vice versa. The prosauropod had many small eggs for its body mass. The two sauropods had in fact larger eggs than the prosauropod, but they still had many eggs in comparison to other dinosaurs because of their large size (
From our study we conclude i) that allometric regression functions are a suitable approach to describe the relation between body mass and the studied reproductive traits in birds or reptiles. It is ii) appropriate to transfer these established allometries to specific taxa of extinct non-avian dinosaurs. Although we found a high variability in reproductive traits around the (average) allometric regression lines in extant species, we think that the results provide new testable hypotheses about dinosaur reproduction, its evolution and their ecological implications, especially for reproductive traits that are insufficiently documented or lacking in the fossil record.
(DOCX)
(DOCX)
(DOCX)
EM = egg mass, CM = clutch mass, ACM = annual clutch mass, group = categorical variable, coding whether the species is a bird, crocodile or tortoises, log = logarithm to the base 10. n.a. = not applicable. For sample sizes refer to
(DOCX)
EM = egg mass, CM = clutch mass, ACM = annual clutch mass, group = categorical variable, coding whether the species is a crocodile or tortoises, log = logarithm to the base 10. n.a. = not applicable. For sample sizes refer to
(DOCX)
(DOCX)
We thank two reviewers for their valuable comments and all members of the DFG Research Unit 533 and the participants of the 2nd workshop on sauropod gigantism for valuable discussions, especially Martin Sander who introduced us to dinosaur reproductive biology. We are grateful to Rebecca Nagel for some linguistic improvements. This paper is part of the PhD thesis of JW. this is contribution number 145 of the DFG Research Unit 533 “Biology of the Sauropod Dinosaurs: The Evolution of Gigantism”.