Abstract
The challenging task of correctly estimating the body size of prehistoric hominins from footprint dimensions has been a matter of long debate, but researchers are still divided about the best methodological approach for obtaining compelling estimates. This research attempts to infer the body size of the trackmakers, who impressed their footprints on the uneven ground slope of a Roccamonfina volcano’s ignimbrite (≈350 ka) at the Foresta/“Devil’s Trails” ichnosite. We used the equations selected among the most adequate for the footprints of this peculiar ichnosite with an exploratory purpose and from a critical estimation point of view. The values of the body size derived from the different equations are highly variable; the minimum and maximum values can differ by up to 50%. The variation range is similarly large when applying the same equations to the footprints from some North African and European Lower Paleolithic ichnosites. The variability of the results mainly depends on the technique of footprint measurements, the environmental context (substrate, surface slope, trackmaker gait, etc.), the equation used, and, to a minor extent, the dimensional slight differences between a fleshy foot and a footprint. This makes comparisons among the sites challenging. All things considered, we could say that the average body mass (about 60 kg) and stature (about 166 cm) of Foresta trackmakers, who may be individuals of different sexes, fall in the range of the Middle Pleistocene Homo heidelbergensis s.l.
1. Introduction
The challenging task of correctly estimating the body size (body mass and stature) of prehistoric hominins and their sex from footprint dimensions has been a matter of a long and contentious debate among anthropologists and paleontologists who are still divided about the best methodological approach for obtaining compelling data when analyzing footprints (e.g., [1,2,3,4,5] and references therein). The growing interest in the issue, triggered by the continuous discovery of new fossil human footprints and trackways, prompted the development and application of several methodological approaches (e.g., [4,5,6] and references therein).
The development of increasingly advanced technologies and ad hoc software has been encouraging and enhancing the experimentation of new methodological approaches, although the application of the most innovative skills may not always correlate to an increase in the result reliability (Ref. [7] and references therein). Several methods and equations have been developed on the basis of data returned from the analysis of directly measured feet or footprints of living Homo sapiens individuals, belonging to populations scattered across the world, for instance, Asia (e.g., [8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28]), Africa (e.g., [29,30,31,32,33]), Europe (e.g., [34,35,36,37,38]), Australia (e.g., [39,40,41,42]), and the Americas (e.g., [43,44,45,46,47,48]). See [5] and references therein for additional information.
Estimating the body mass (BM herein after) and stature (ST herein after) of extinct hominin trackmakers based on their fossil footprints is particularly complex due to several factors (e.g., the nature of the substrate, its consistency, granulometry, content in water, the surface slope, and the trackmaker’s gait) (e.g., [4,49,50,51] and references therein) that variously affect the morphometric characteristics of the footprints and, in turn, the estimation of trackmakers’ body size. Furthermore, some accurate experimental analyses/studies evidenced that the lengths of footprints are greater than those of fleshy feet by about 4% [7,52,53,54]. However, the morpho-classificatory method of some Namibian trackers has been used for the interpretation of some footprint patterns and for an estimation of the body characteristics of human trackmakers [55,56]. Moreover, fossil skeletal evidence indicates that the body proportions of Middle Pleistocene hominins were comparable to those of modern humans, somehow entailing that this relationship could also be applicable to past populations [31,57,58,59]. Accordingly, we attempted to apply a set of equations to the footprints preserved at the Foresta/“Devil’s Trails” ichnosite (F/DT hereinafter) (Tora e Piccilli, Campania, Central-Southern Italy) (Figure 1), although we are aware that the estimates obtained can only be useful for a general evaluation. Indeed, the F/DT ichnosite is unique in the world due to its peculiar geomorphological and ichnological characteristics (see below in Section Why is the Foresta/Devil Trails Ichnosite So Unique?).
Figure 1.
The Foresta/“Devil’s Trails” ichnosite: (a) topographic map where the red hexagon indicates the F/DT ichnosite; (b) southwestern view of the trampled slope; (c) southern photographic view of footprints A22 and A23; (d) stratigraphic sketch (modified after [51] N, p. 26, Figure 3.4).
Therefore, the main purposes of this research are (i) to infer the body size (BM and ST) of each F/DT trackmaker by means of regression equations, selected among those that we regarded as the most adequate for inferring the body size of an hominin walking on a moderately hot, still unconsolidated, slipping, and highly inclined slope; (ii) to critically compare the estimates obtained and evaluate the consistency among the estimates obtained by applying different methods, which should be the best fit regression equations, accounting for a suitable correspondence between body size and footprint dimensions of the Middle Pleistocene hominins; and (iii) to assess the level of error resulting from using the equations experimentally obtained when analyzing footprints left by five modern Homo sapiens groups on sub-planar mainly soft surfaces to infer the body size of pre-sapiens humans, especially if they left imprints on a strongly inclined and unevenly consolidated surface, such as the at least four F/DT individuals who did ([51] and references therein).
Finally, we avoided any attempt to individuate the putative sex of the F/DT trackmakers because the results obtained by applying the algorithms available for modern humans would not be compelling enough. Indeed, inferring the sex of hominin trackmakers from their footprints is a more challenging task because of the high number of variables involved and the inadequate knowledge about the gap between the foot dimensions of males and females and their actual ranges. Hence, the results obtained by applying one or the other algorithm proposed for modern humans to fossil footprints may return untrustworthy results and should be considered with great caution ([4,52,60] and references therein).
Why Is the Foresta/Devil’s Trails Ichnosite So Unique?
Estimating the body mass (BM herein after) and stature (ST herein after) of extinct hominin trackmakers based on their fossil footprints becomes particularly complex in the case of the F/DT ichnosite. At the ichnosite, the human trackways were impressed by hominins (Homo heidelbergensis s.l.) walking on a ≈350 ka old pyroclastic ignimbrite (LS7 unit) of the Brown Leucitic Tuff (385–230 ka) of the Roccamonfina stratovolcano [61], which were deposited during Plinian eruptions of the second period of the volcano activity ([62,63] and references therein) (Figure 1).
The surface on which hominins and some large mammals left their tracks consists of a narrow sub-planar surface at the top, where footprint characteristics and trackways cannot be clearly individuated, and a strongly downward-inclined surface (average slope = 30°, max = 39°) with distinctly impressed footprints [64,65,66,67,68,69,70] (Figure 2 and Figure 3). Rapid erosional processes shaped the slope shortly after the pyroclastic flow had partially filled a pre-existing valley and the ignimbrite was still soft.
Figure 2.
The Foresta/”Devil’s Trails” ichnosite: (a) graphic sketch of the average incline (30°) of the bearing footprint slope; (b) 3D-generated image of the trampled slope with trackway A-B-C-E patterns (diagonal southwestern view). The angle of 30° refers to the average incline of the slope (blue color); 24° refers to the slope incline at the point where the section was drawn.
Figure 3.
The Foresta/”Devil’s Trails” ichnosite: (a) particular A21–A26 footprints of Trackway A. (b) Trackway A is a 3D-generated depth map (zenithal and transversal images) of the human fossil footprint called A25 (right foot) (scale bar: 10 cm).
When at least four hominins climbed down the slope, the temperature of the substrate had significantly decreased, but it was still slippery and soft because the wide circulation of water had soaked the surface in fluids. The slippery, coarse, and uneven ground surface and the strong slope acclivity, which varies from one point to another, strongly influenced the trackmaker’s gait and the trackway direction. Both factors forced the trackmaker to change pace and route and affected the way of placing the foot on the substrate and the stride stability. As a result, the single footprint shape was conditioned not only by the characteristics of the substrate and the stride speed but also by the conscious or unconscious movements of the trackmakers and by the repeated translations, sliding, and oriented sinking caused by the need to maintain balance. As pointed out by Palombo and Panarello [51], the morphometry (shape, length, and width) of each footprint changes along the same trackway. For instance, when the trackmaker climbs down towards the bottom of the slope, the footprint is generally proportionally longer. When the footprint is diagonally oriented to the slope, the footprints of two feet differ with the foot downslope, and the other foot is deformed on its lateral margin and medial margin, respectively. When a footprint orientation is consciously varied in search of greater stability, the preceding or the following footprint (even if left by the same foot) may appear smaller or more evanescent than most of the others.
These few examples show how much the peculiarity of the F/DT floor and the difficulty of descending a slipping slope may have affected the footprints’ shape, varying the footprint proportions along the same path. Furthermore, the study by Palombo and Panarello [51] highlights how much the nature of the F/DT substrate (texture, coarse, and uneven granulometry, presence of large scorias and about 3–4% of lithic fragments, and variable plasticity and slipperiness) along with slope acclivity influenced the trackmaker gait (velocity and stride length), the walking direction and its changes, the pace stability, and the way in which the foot rests against the substrate slope. The combined action of these factors conditioned the shape and determined the dimensional variability of the footprints left by the four trackmakers who zigzagged down the F/DT slope.
The geomorphological characteristics and the hominin footprints’ peculiarities make the F/DT ichnosite unique, and it is difficult to compare its footprint record with those of the other Pleistocene ichnosites bearing human footprints. Indeed, most of them were impressed on the surface of cineritic deposits, alluvial muddy sediments, or have been found in an aeolian context and rarely on slightly inclined surfaces (e.g., Cape South Coast, South Africa; [71,72,73,74]).
2. Materials and Methods
Material
We have analyzed the footprints of the three main F/DT trackways, A, B, and C (Tr A, Tr B, and Tr C hereinafter), the few footprints of the short trackway E (Tr E hereinafter), which were impressed by four different trackmakers [51], and the two very short successions of strides F and D, the latter characterized by its rather small size. We did not analyze the badly preserved footprints of the pathway at the top of the slope, whose preservation and asset did not allow us to recognize any possible trackway [66].
We analyzed two sets of footprints using the dimensional and morphometric data published in 2022 [67,68,69,70]. The first set (Sample A) includes all the footprints for which it was possible to take both the maximum length and width and calculate the area (50 footprints, i.e., 23 of Tr A, 17 of Tr B, 6 of Tr C, and footprints D01, E02, E03, and F02). The second (Sample B) consists of the best-preserved ones, which show the clearest anatomical features (21 footprints, i.e., 9 of Tr A, 9 of Tr B, and the C05, E03, and F02 single footprints) (Table 1).
Table 1.
Measurements of the hominin footprints impressed on the surface of the Foresta/Devil’s Trails ignimbritic slope. * Data are from [68]; ** calculated tracing of the perimeter of the footprint through a polyline of at least 40 points; *** calculated according to [31]. Bold = the best-preserved footprints.
We compare the F/DT BM and ST estimates with those that we have obtained by applying the formulas listed in the following paragraph to the few samples of hominin footprints for which morphometric data are available in the literature, i.e., the latest Early Pleistocene sites of Gombore II-OAM, Melka-Kunture (Ethiopia) [75], and Happisburgh (Great Britain) [76], the earliest Middle Pleistocene Gombore II-2 site (Ethiopia) [77], the about 0.38 Ma old site of Terra Amata (France) [78], the Hungarian site of Vértesszőlős Cave, dated to about 0.31 Ma ([79] and references therein), and the nearly coeval site German site of Schöningen [80,81].
We inferred the body mass (BM), expressed in kg, and the stature (ST), expressed in cm, of the F/DT trackmakers, applying to their measurement dataset [68] the regression equations and formulas that, in our opinion, could be regarded as the most suitable in the case of the F/DT footprints, taking into account the peculiar nature and the slope of the substrate that had influenced the trackmakers’ gait and the footprint shape [51].
Consequently, we estimated the BM of the F/DT hominins by means of the regression formulas proposed by Grivas and colleagues [82] (BM = −71.142 + 5.259 × right rFL; BM = −70.385 + 5.217 × left rFL), Domjanic and colleagues [83] (BM = −80.4 + 15.3 × rFw; BM = −97.4 + 2.44 rFl + 10.7 rFw, where rFw = footprint width), Dingwall and colleagues [31] (BM = 4.71 + 1.82 × average rFL, and BM = 23.64 + 0.11 × average footprint area, Fa, where Fa = footprint area), and Ruff and colleagues [60] (BM = −0.250 × Fa + 0.00099 × Fa2 + 59.1; BM = −21.9 × rFl + 0.546 × rFl2 + 265.4; BM = −25.8 × rFw + 1.84 × rFw2 + 133.3). We analyzed the F/DT datasets using the equations based on the dimensions most objectively measurable in the F/DT footprints (rFl and rFw) among those proposed by Domjanic and colleagues [83]. We applied the [60] quadratic equations to the most reliable dimensions of the F/DT footprints, considering that the body mass index (BMI) of the F/DT hominin trackmakers cannot be confidently determined (Table 2).
Table 2.
Body mass estimates obtained from the dimensions of the footprints impressed on the surface of the Foresta/Devil’s Trails ignimbritic slope. Symbols are shown in Table 1.
We derived the stature (ST) estimates of F/DT trackmakers by means of the regression equations proposed by Grivas and colleagues [82] (ST = 17.369 + 5.879 × right rFl; ST (cm) = 17.592 + 5.861 × left rFl) and Domjanic and colleagues [83] (ST = 47.0 + 5.00 × rFl) and the formulas proposed by Webb ([84] and references therein) (ST = 6.58 rFl) and Fessler and colleagues ([85] and references therein) (assuming that the length would roughly be considered equal to 15,25% of the stature in the F/DT footprints left by trackmakers of unknown sex), and some among the equations proposed by Duveau [52], modifying his dimensions’ abbreviations according to those adopted for the F/DT footprints (ST = 4.1 × rFl + 66.9; ST (cm) = 4.9 × rFw + 121.4; ST = 0.2 × Fa + 137.9) (Table 3). Duveau’s model [52] is based on various anatomical landmarks, which sometimes cannot be identified in the F/DT footprints. Therefore, we used only the equations based on the objectively measurable morphometric dimensions (i.e., maximum rFl, corresponding to the linear distance between the pternion and the akropodion, and maximum rFw, obtained by measuring the linear distance between the 1st and 5th metatarsal joints) and those based on the footprint area. We assumed that the area, calculated by drawing the perimeter of the F/DT footprint with a polyline of at least 40 points [64,86], roughly matches that indicated by Duveau as “Area 5 (A5)” [52] (p. 8, Figure 4).
Table 3.
Stature estimates obtained from the dimensions of the footprints impressed on the surface of the Foresta/Devil’s Trails ignimbritic slope. Symbols are shown in Table 1.
A more detailed discussion of the selected formulas and the reasons behind their choice can be found in Supplementary Information S1.
We used the following descriptive analysis for summarizing and scrutinizing the BM and ST estimates obtained by applying the above-mentioned methods to Sample A and B data: the descriptive box plot method, which graphically describes the variation range, the presence of outliers, and the skewness of data distribution within each group of estimates, as well as visually compares the difference in the variation ranges among groups of estimates; the statistics summary of the principal statistical indices, in particular, the kurtosis and skewness for testing the symmetry of data distribution; and some tests, such as the Shapiro–Wilk test, which is appropriate, even for small sample sizes, the Kolmogorov–Smirnov test, although it is generally used for large samples, and the less precise Jarque–Bera test, herein given for reference, with the purpose of verifying the null hypothesis that data are normally distributed.
We applied the repeated measures one-way ANOVA method to evaluate whether and how much different from each other the BM and ST average values calculated for the same sample using different methods are, assuming that each sample should be assimilated to a set of measurements repeated in time because it consists of estimates obtained for the same set of data by means of different equations. The repeated measures one-way ANOVA method determines whether there is a statistically significant difference between the means of three or more dependent groups and, therefore, it could be useful in the F/DT case. The ANOVA method requires the following three assumptions to be met: a normal distribution of data (variables); homogeneity of variance (homoscedasticity); and equal variances of the differences between all combinations of groups (sphericity). If one of the three assumptions is violated, the ANOVA probably is not the correct fit for analyzing the dataset because the variance calculations may be distorted and the F test overestimated. A statistic called epsilon (ε) determines the degree to which sphericity has been violated. ε approaches 1 for data, meeting the sphericity assumption. If the sphericity assumption is violated and ε is <1, it is possible to correct the ε value by running the univariate one-way ANOVA analysis, modifying the degrees of freedom so that a valid F-ratio can be obtained and a corrected p-value for the ANOVA analysis provided.
Finally, we used multivariate principal component analysis (PCA) for summarizing with a descriptive and exploratory method, and in a small number of dimensions, all the information that describes the similarities/differences among the variation ranges of the BM and ST estimates that we have obtained for each F/DT trackmaker. This depicts the spatial relationships among BM and ST of the footprints from F/DT and the compared sites.
Most of the analyses we have performed come from equations based on the fleshy feet’s measurements. Their use may raise some approximations of the estimated BM and ST values, even though these dimensions have been thought to be roughly equal to the footprints’ dimensions [54]. Consequently, we used the reduction suggested by Duveau [52] and Ruiz and colleagues [54] to convert the footprints’ length and width into foot dimensions. Since the inferred value would have been determined using data obtained through mathematical processing, we decided not to compute the area since we thought the results obtained would have been unrealistic. Using the BM and ST values derived from the estimated foot width and length, we assessed how much the estimates produced by this method differed from those obtained using the footprints’ dimensions as variables in equations based on those of fleshy feet.
Analyses were executed with PAST (PAleontological STatistics) 4.16 software [87].
3. Results
The dimensions of the footprints impressed by at least four hominins who went down the F/DT ignimbritic slope are moderately variable within each of the three longest trackways (A, B, and C) (Table 1). The variability of the shape and size proportions along the same pathway is mainly related to the substrate characteristics (coarse granulometry, plasticity, and slipperiness) and the slope acclivity that influenced the trackmaker gait (velocity and stride length), the walking direction and its changing, the pace stability, and the way in which the foot rests against the substrate slope [51]. Consequently, BM and ST for the F/DT footprints belonging to the same trackway are rather variable, even if the same methodological approach is applied, but the variation range is considerably wider if we consider the single and average estimated values of BM and ST estimates obtained by applying the different formulas, as specified in the following paragraphs.
3.1. Body Mass and Stature Estimates
3.1.1. Variation Ranges of the Body Mass and Stature Estimates
The variability of the estimate values of BM and ST obtained from the footprint dimensions (i.e., footprints’ length, rFl, width, rFw, and footprint area, Fa) is moderate for the values returned by each of the formulas mentioned above (Table 2 and Table 3). The values of the variation coefficient of both samples (A and B) are lower than 20, which is the maximum value for an acceptable standard deviation to the mean [88], confirming the moderate variability of the BM and ST estimates returned by each method (Table 4, Table 5, Table 6 and Table 7).
Table 4.
Summary of the univariate analysis statistics’ data of the body mass estimates derived from the dimensions of all the measurable footprints (Sample A) impressed on the surface of the Foresta/Devil’s Trails ignimbritic slope.
Table 5.
Summary of the univariate analysis statistics’ data of the body mass estimates derived from the dimensions of all the best-preserved footprints (Sample B) impressed on the surface of the Foresta/Devil’s Trails ignimbritic slope.
Table 6.
Summary of the univariate analysis statistics’ data of the stature estimates derived from the dimensions of all the measurable footprints (Sample A) impressed on the surface of the Foresta/Devil’s Trails ignimbritic slope.
Table 7.
Summary of the univariate analysis statistics’ data of the body mass estimates derived from the dimensions of all the best-preserved footprints (Sample B) impressed on the surface of Foresta/Devil’s Trails ignimbritic slope.
The variation coefficient (CV) is higher for the BM than the ST estimates, reaching its maximum value (18.01) in the case of the BM estimated for the Tr C footprints (Sample A), obtained using rFw as a variable and Domjanic and colleagues’ method [83]. The maximum CV value of the ST estimate (6.51) is significantly lower and has been obtained for all the footprints (Sample A), applying both the Fessler and colleagues [85] and Webb [84] formulas (Table 6).
The values of kurtosis indicate a moderate deviation from the normal distribution of tail values in the Gaussian curve for both BM and ST estimates (Table 4 and Table 5), but the deviation varies in each sample and from one trackway to another. Regarding BM, for instance, in Sample A, the majority of the values are positive in the case of all Tr A footprints (i.e., the extremity of deviations, outliers, is greater than the normal, and the distribution is leptokurtic) and negative for trackways B and C (i.e., the data distribution is platykurtic and produces fewer extreme outliers than the normal distribution) (Table 4), whilst in Sample B, they are only positive if all footprints are considered (Table 5). Regarding ST, in Sample A (Table 6), the excess kurtosis is positive in the case of all Tr A and Tr B footprints, whilst it is only positive for all the footprints in Sample B (Table 7). The skewness values are less than 1 or greater than −1 (except, as expected, for the couple of measurable footprints of the short Tr E), indicating an asymmetry of the distribution with the BM and ST value of the right side of the distribution that, in each sample, taper differently from the values on the left side in a rather variable way (Table 4, Table 5, Table 6 and Table 7). The mean distribution of the best-preserved footprints (Sample B) is skewed to the left of the typical center of the data (negative skewness value, left-tailed distribution of data). Sample A footprints are skewed to the left in the case of the footprints of Tr A and Tr B but are to the right (positive skewness value, right-tailed distribution of data) for all and the Tr C footprints.
Summing up the main results, the BM values returned by the formula proposed by Dingwall and colleagues [31] (44.771 kg and 47.345 kg, using Fa and rFl as variables, respectively) and the BM values returned by Fessler and colleagues’ [85] formulas (153.612 cm) are lower than those obtained by means of Grivas and colleagues’ [82] equations (BM = 53.008 kg and 50.671 kg and ST = 156.153 cm and 153.57 cm, both for the right and left footprints, respectively). The values returned by Domjanic and colleagues’ method [83] are higher (BM = 74.113 kg, ST = 164.020 cm). These differences can be observed by comparing each other the results obtained by analyzing Sample A and Sample B data, as well as those obtained in both samples for each trackway or short footprint sequence (Table 4 and Table 6). Moreover, following Dingwall and colleagues [31], the BM and ST estimates returned by Fa are always lower than those returned by rFl, while the estimates resulting from the right footprint are slightly higher than those from the left ones following Grivas and colleagues [82].
Comparing the BM average values obtained by means of each method, the highest and lowest estimates correspond to those returned by the formulas of Domjanic and colleagues [83] and those derived from those of Dingwall and colleagues [31] and Grivas and colleagues [82], respectively. The first is higher than the second, with a percentage excess spanning from 138.34%, in the case of BM estimates obtained for all the measurable footprints of Tr A, to 190.65% in that of BM estimates for the selected footprint of Tr B. (Table 8). The average ST estimates resulting from Duveau’s [52] equations, in particular, from those based on the Fa2 variable, are higher than those obtained with Grivas and colleagues’ [82] formula, but the percentage excess is less significant than those resulting for BM estimates, ranging from 121.36% (Sample A, all footprints) to 124.96% (Sample A, Tr B) (Table 9). The variations between the average BM estimates obtained comparing each other and the values obtained by means of different methods underpin the highest percentage excess of the average value returned by Domjanic and colleagues’ [83], Dingwall and colleagues’ [31], and Grivas and colleagues’ [82] formulas (δ BM range = 51.14–94.89) (Table 10), while those of the average BM value estimated with Duveau’s [52] and Grivas and colleagues’ [82] formulas are inferior (Table 11). The δ values highlight some similarity among the estimates obtained by means of the equations based on different variables proposed by the same author, possibly due to the uniformity of the footprint measurements taken by a single researcher (Table 10 and Table 11).
Table 8.
Percentage relationships among the average body mass estimates obtained by applying each formula.
Table 9.
Percentage relationships among the average body mass estimates obtained by applying each formula.
Table 10.
Reciprocal percentage excess (δ) of the average stature value returned by each formula applied to the dimension of the footprints impressed on the surface of the Foresta/Devil’s Trails ignimbritic slope.
Table 11.
Reciprocal percentage excess (δ) of the average stature value returned by each formula applied to the dimension of the footprints impressed on the surface of the Foresta/Devil’s Trails ignimbritic slope.
We have computed the average values that arise from applying the various approaches to the F/DT footprints of each trackway or sequence (apart from the single footprints F02 and D02) in order to provide a synthetic comparison view of the BM and ST estimates. The BM estimates based on the dimensions of all the measured (Sample A) and the best-preserved footprints (Sample B) range from 55.757 kg (Tr B) to 69.268 kg (Tr E) and from 56.088 kg (Tr B) to 64.662 (Tr A), respectively. The ST estimates range from 162.278 cm (Tr B) to 180.138 cm (Tr E) and from 164.114 (Tr B) cm to 171.719 cm (Tr A) in the footprints of Samples A and B, respectively.
The means of BM and ST are in line with the size relationships already highlighted between the footprints of the various tracks and directions, as expected since BM and ST estimates were derived from the dimensions of each F/DT footprint. Among the average values of the BM and ST estimates inferred for each trackway, the highest are those of Tr E (BM = 69.268 kg, ST = 180.138 cm). Those of Tr A (BM = 54.67 kg, ST = 159.97 cm) are higher than those resulting from trackways B (BM = 50.82 kg, ST = 151.96 cm) and C (BM = 51.31, ST = 153.30), which are roughly similar to each other.
3.1.2. Quantitative Analysis: Box Plots, Normality Tests, One-Way ANOVA, and Principal Component Analysis
Box Plots
The statistical summary illustrated by box plots (i.e., the diagrams resulting from the explanatory, also known as whisker plots), are particularly suitable for comparing distributions in the analyzed datasets because they provide an easy-to-read visual representation of the variation range in data. The box plots of the BM and ST estimates resulting from the footprint analysis give support to the evidence resulting from the univariate analysis (Table 4, Table 5, Table 6 and Table 7) and the comparison among the average values of the BM and ST estimates returned by applying each method (Table 8, Table 9, Table 10, Table 11 and Table 12). Indeed, the box plots obtained considering the BM and ST estimates obtained from the dimensions of all 50 measurable footprints (Sample A) and the best-preserved footprints (Sample B) highlight the different values returned by the applied formulas and equations, especially in the case of BM estimates. Moreover, the whisker lengths are imbalanced, and the median value frequently does not fall in the middle of the box (which includes the range of the central 50% of the data) but is shifted towards the lowest or, rarely, the highest values of the interquartile range, and, consequently, the distribution of the data (estimates) is skewed (Figure 4, Figure 5, Figure 6 and Figure 7).
Table 12.
Normality test data resulting from the analysis of the body mass estimates obtained from the dimension of all the measured footprints impressed on the Foresta/Devil’s Trails ignimbritic slope (Sample A) and in each main trackway (trackways A, B, and C). Red colors indicate p < 0.05.
Figure 4.
Box plots illustrating the variation ranges of the body mass estimates obtained by applying the different equations based on the dimensions of all the measured footprints (Sample A) impressed on the ignimbritic slope of the Foresta Devil’s Trails ichnosite. Abbreviation: Dingwall = [31]; Grivas = [82]; Domjanic = [83]; Ruff = [60].
Figure 5.
Box plots illustrating the variation ranges of the body mass estimates obtained by applying the different equations based on the dimensions of the best-preserved footprints (Sample B) impressed on the ignimbritic slope of the Foresta/Devil’s Trails ichnosite. Symbols and abbreviations are shown in Figure 4.
Figure 6.
Box plots illustrating the variation ranges of the stature estimates obtained by applying the different equations based on the dimensions of all the measured footprints (Sample A) impressed on the ignimbritic slope of the Foresta/Devil’s Trails ichnosite. Abbreviations: Fessler = [85]; Webb = [84]; Domjanic = [83]; Duveau = [52]. Symbols are shown in Figure 4.
Figure 7.
Box plots illustrating the variation ranges of the stature estimates obtained by applying the different equations based on the dimensions of the best-preserved footprints (Sample B) impressed on the ignimbritic slope of the Foresta/Devil’s Trails ichnosite. Abbreviations as in Figure 6. Symbols are shown in Figure 4.
In the box plots of BM estimates obtained by analyzing the footprints of both Sample A and B, the highest estimates are those returned by Domjanic and colleagues’ [83] methods based on rFw and rFl+rFw. The estimates are higher than those provided by Ruff and colleagues [60] based on rFw. The estimates returned by Dingwall and colleagues’ [31] equations are the lowest, but they are comparable to those returned by Ruff and colleagues’ [60] equation based on Fa1. The BM estimates obtained using the formulas based on rFl that Grivas and colleagues [82] and Ruff and colleagues [60] proposed are rather similar, but the data range of the box and the lines that extend from the box to capture the range of the remaining data in the first box plot are larger and longer, respectively, than in the second. Notably, in the box plots of Sample A, the BM estimates obtained for the D02 footprint using rFl [82] and rFl and Fa1 as variables [31] fall below the lower inner fence (i.e., the values and outliers are lower than 1.5 times the box height from the box). The BM values of E02 and E03 estimated using the rFl variable fall above the lower upper fence in the box plots of both Samples A and B (Figure 4 and Figure 5), while the ST estimates match the lower and upper fences in the box plots of Samples A and B, respectively (Figure 6 and Figure 7). These results are in line with the small and the large dimensions of D02, E02, and E03 footprints, respectively [51]. In the box plots of ST estimates, the lowest values are those obtained by means of the formulas based on rFl [82,84,85], which show a greater variation range but are comparable to each other. Duveau’s [52] equation based on Fa2 returned the highest estimate value, and the box plot obtained using rFw shows the lowest variability range.
The comparison among the box plots of the means of the BM and ST estimates obtained for all the measured F/DT footprints and those of Tr A and Tr B (Samples A and B) confirms the previous results. The comparison underlines the asymmetrical distribution of data (the median lines fall below the middle point of the box interquartile data) and the wide variation range of the obtained results. In the BM box plots, the highest and lowest average estimates are those obtained by means of Domjanic and colleagues’ [83] and Dingwall’s [31] formulas, respectively, while the highest average ST estimate is returned by the formula based on Fa2 [52] (Figure 8 and Figure 9).
Figure 8.
Box plots illustrating the variation ranges of the average body mass estimates obtained for all footprints and trackway A and B footprints of Samples A and B impressed on the ignimbritic slope of the Foresta/Devil’s Trails ichnosite.
Figure 9.
Box plots illustrating the variation ranges of the average stature estimates obtained for all footprints and trackway A and B footprints of Sample A and B impressed on the ignimbritic slope of the Foresta/Devil’s Trails ichnosite.
Finally, the distribution of BM and ST estimate scores in the box plots, particularly in those obtained for Sample B, confirm the differences between the footprint dimensions of Tr A (averagely larger) and Tr B, as well as the largest dimensions of the E02 and E03 footprints and the probable presence of a fourth trackmaker at the F/DT ichnosite (see [51]), which is expected given that ST and BM are strictly dependent on the footprint dimensions.
Normality Tests
The normality test is an important step for verifying whether the data are normally distributed and, in turn, whether parametric tests, such as the Analysis of Variance (ANOVA), can be used to compare the groups.
The results of the normality tests (Shapiro–Wilk, Anderson–Darling, and Jarque–Bera tests) give somewhat contradictory information, especially for the BM estimates. The null hypothesis is confirmed (p-values greater than 0.05) for the BM estimates obtained by applying most of the formulas on all the measured footprints (Sample A) [31,82,83], but not for some others [52] (p-values lower than 0.05) (Table 12). Conversely, the null hypothesis is rejected for most samples of the best-preserved footprints (Sample B) by the Shapiro–Wilk and Anderson–Darling tests (Table 13).
Table 13.
Normality test data resulting from the analysis of the body mass estimates obtained from the dimension of the best-preserved footprints impressed on the Foresta/Devil’s Trails ignimbritic slope (Samples B) and the main trackways, A and B. Red colors indicate p < 0.05.
The normal distribution prevails for the ST estimates returned by the analysis of footprint estimates in Sample A, except for the samples obtained using the equation based on Fa2 [52] and those of Tr B (Table 14), whilst in the case of the best-preserved footprints (Sample B), the p-values of the Anderson–Darling test are lower than 0.05, rejecting the null hypothesis (Table 15).
Table 14.
Normality test data resulting from the analysis of the stature estimates obtained from the dimension of all the measured footprints impressed on the Foresta/Devil’s Trails ignimbritic slope (Sample A) and in each main trackway (trackways A, B, and C). Red colors indicate p < 0.05.
Table 15.
Normality test data resulting from the analysis of the stature estimates obtained from the dimension of the best-preserved footprints impressed on the Foresta/Devil’s Trails ignimbritic slope (Samples B) and the main trackways, A and B. Red colors indicate p < 0.05.
One-Way ANOVA Repeated Measurements
We chose to tentatively execute the analysis to assess the difference between the means of the BM and ST estimates returned using the different methods, although the null hypothesis of a normal distribution is rejected for some of the analyzed samples and, therefore, one assumption for running the ANOVA test is at least partially violated. Moreover, we run Levene’s tests for homogeneity of variance (homoskedasticity) for both means and medians to verify the second ANOVA assumption. The first test has more power if the distributions are normal or at least symmetric, while the second is more robust to non-normal distributions. Equal variances of the differences between all pairs of groups (sphericity) are another assumption of the repeated measures ANOVA. If the assumption is violated, Levene’s test and the repeated measures ANOVA must be appropriately corrected, depending on the degree to which sphericity has been violated. The test statistic called epsilon (ε) provides a measure of departure from sphericity. A ε value approaching 1 indicates no departure from sphericity, and the sphericity assumption is confirmed. For smaller ε values, a correction could be applied to the degrees of freedom of the F test that also provides a corrected p-value for the ANOVA test. Among the three proposed corrections, Huynh–Feldt [89], lower bound, and Greenhouse–Geisser [90], we applied the Greenhouse–Geisser and Huynh–Feldt methods, which are more and less conservative, respectively.
The results of the one-way ANOVA (repeated measurements) analysis for testing whether there are statistically significant differences in the mean of the dependent samples of the BM and ST estimates obtained by means of the herein-used formulas clearly indicate that the three assumptions are violated (Table 16 and Table 17). Indeed, the p-value is always inferior to 0.05 in all Levene’s tests from means, while the p-value is only greater than 0.05 in Levene’s test from the medians of the ST estimates of Tr A and Tr B (Sample A). Moreover, both Greenhouse–Geisser and Huynh–Feldt corrections provided ε values largely inferior to 1. Accordingly, ANOVA probably is not the correct fit for analyzing the F/DT dataset because the variance calculations are distorted and the values of the F test (extremely high) and p-values (particularly low) were overestimated and underestimated, respectively (Table 16 and Table 17). Therefore, the high value of F and the low probability returned by the analysis overestimated the probable notable variation among group means (i.e., among the BM and ST estimates returned by each equation).
Table 16.
Results of the one-way ANOVA (repeated measurements) analysis run on the body mass estimates obtained for the footprints impressed on the Foresta/Devil’s Trails ignimbritic slope.
Table 17.
Results of the one-way ANOVA (repeated measurements) analysis run on the stature estimates obtained for the footprints impressed on the Foresta/Devil’s Trails ignimbritic slope.
Principal Component Analysis
We ran the PCA first on the total footprint dataset (Sample A) and then on the dataset of the most compelling footprint measurements (Sample B) to investigate the difference between the BM and ST estimates obtained by applying different methods and formulas to the dimensions of the F/DT footprints belonging to different trackways (Tr A, Tr B, Tr C, and Tr E) and single measurable footprints of the D and F directions.
In the PCA obtained from both the total dataset of BM estimates (Sample A) and that of the best-preserved footprints (Sample B), the eigenvalue and the variance percentage captured by the first component (PC1) are definitely greater than those of the second component (PC2), and, in turn, the variance of axis 1 is greater than that of axis 2 (Figure 9 and Figure 10). PC1 captures nearly all the eigenvalue and the variance percentage in the PCA biplots resulting from the ST estimate dataset of Samples A and B (Figure 11 and Figure 12). The dispersion of scores is greater for Tr A than for Tr B estimates, except for the ST estimates resulting from the analysis of the Sample B dataset. Moreover, the dispersion of the Tr A scores indicates that the BM and ST estimates are, on the whole, greater than those of Tr B and Tr C, with the latter showing the largest dispersion of scores. Moreover, the dispersion of BM and ST scores in the estimates derived from the analysis of the Sample A dataset partially overlaps those of Tr B and Tr C (Figure 10 and Figure 12). The dispersion areas of Tr A and Tr B scores are clearly separate if the PCA is only run for the estimates derived from the best footprint analysis (Figure 11 and Figure 13). The scores of Tr E and Dir F lay in the highest part of the third PCA quadrant and are rather detached from the other scores, while those of D02 lay in the second quadrant in a low position, at the margin of axis 1. Therefore, the score distribution of the BM and ST estimates in the PCA biplots also supports the presence at the F/DT ichnosite of at least four trackmakers going down the slope of the F/DT ignimbrite (i.e., the slightly large trackmaker of Tr A, the trackmakers of Tr B and Tr C, with rather similar body sizes, and the largest trackmaker of short Tr E) [51].
Figure 10.
Biplot diagram produced by the principal component analysis (PCA) using the body mass estimates of each Foresta/Devil’s Trails’ trackway and direction as variables, calculated by averaging the values derived by all the equations based on the dimensions of the measurable footprints (Sample A). The component loadings (below) show the degree to which the different original variables enter into components 1, 2, and 3. A1 = Dingwall Fa1, A2 = Dingwall Fa1 [31]; G2 = Grivas rFl (rFl of both right and left footprints) [82]; D1 = Domjanic rFl + rFw, D2 = Domjanic rFw [83]; F1 = Ruff rFl, F2 = Ruff rFw, F3 = Ruff Fa1, F4 = Ruff Fa2 [60].
Figure 11.
Biplot diagram produced by the principal component analysis (PCA) using the body mass estimates of each Foresta/Devil’s Trails’ trackway and direction as variables, calculated by averaging the values derived by all the equations based on the dimensions of the best-preserved footprints (Sample B). The component loadings (below) show the degree to which the different original variables enter into components 1, 2, and 3. Abbreviations are shown in Figure 8.
Figure 12.
Biplot diagram produced by the principal component analysis (PCA) using the stature estimates of each Foresta/Devil’s Trails’ trackway and direction as variables, calculated by averaging the values derived by all the equations based on the dimensions of the measurable footprints (Sample A). The component loadings (below) show the degree to which the different original variables enter into components 1, 2, and 3. F = rFl [85]; W = Webb rFl [84]; G = Grivas rFl [82]; Do = Domjanic rFl [83]; D1 = Duveau rFl, D2 = Duveau rFl, D3 Duveau Fa1, D4 = Duveau Fa2 [52].
Figure 13.
Biplot diagram produced by the principal component analysis (PCA) using the stature estimates of each Foresta/Devil’s Trails’ trackway and direction as variables, calculated by averaging the values derived by all the equations based on the dimensions of the best-preserved footprints (Sample B). The component loadings (below) show the degree to which the different original variables enter into components 1, 2, and 3. Abbreviations are shown in Figure 10.
3.2. Comparison
The comparison of the average values of the BM estimates obtained by means of the formulas applied to the dataset of the F/DT best-preserved footprints and those of selected African and European ichnosites (the latest Early Pleistocene sites of Gombore II-OAM and Melka-Kunture, Ethiopia [75,77], and Happisburgh, Great Britain) [57,76]; the earliest Middle Pleistocene Gombore II-2 site, Ethiopia [77], Terra Amata, France, 0.38 Ma [78], Vértesszőlős Cave, Hungary, 0.31 Ma [79] and references therein, and Schöningen, Germany, ≈0.30 Ma, [80,81]) show a rather wide range of variability and highlight the significant differences of the estimates obtained for each footprint sample using the various formulas proposed by the authors. Furthermore, some equations [60,83] provide a few unrealistic estimates for the footprints of some Gombore II-OAM levels, whose samples include footprints with a large range of dimensions (Table 18). Similar differences also characterize the ST estimates, though the variation ranges are less wide than those of BM estimates, except for those of most ST estimates obtained for both Gombore II ichnosites by means of Duveau’s [52] equations (Table 19).
Table 18.
Comparison among the average body mass estimated for the best-preserved footprints (Sample B) of the trackways and directions at the Foresta/Devil’s Trails ichnosite and those of selected African and European sites. Table 18 consists of two parts, Part 1 and Part 2. Information about the ichnosites and raw data as provided in the literature are presented in Part 1, and the results of our application of the various equations employing these data as variables are shown in Part 2. * = herein informally adopted to indicate the polymorph Middle Pleistocene populations with diversely advanced characters from those rather typical to those showing a mix of H. heidelbergensis and Homo neanderthalensis features; ** = body mass values available in the literature. Red colors indicate unrealistic estimates.
Table 19.
Comparison among the average stature estimated for the best-preserved footprints (Sample B) of the trackways and directions at the Foresta/Devil’s Trails ichnosite and those of selected African and European sites. Table 19 consists of two parts, Part 1 and Part 2. The raw data as described in the literature is presented in Part 1, and the results of our application of the various equations employing these data as variables are shown in Part 2. Symbols and sources are shown in Table 18. Red colors indicate unrealistic estimates.
The score position in the scatter diagram, visualizing the relationships between the means of the BM and ST estimates obtained for the analyzed ichnosites, suggests that the hominins who left their footprint at the Gombore II-OAM (with the exception of the especially high estimates of the larger footprint of level 7), Happisburgh, and Gombore II-2 ichnosites might have had a body size inferior to that of the trackmakers of Schöningen, Terra Amata, Vértesszőlős Cave, and Tr A and Tr B of F/DT. The F/DT trackmakers who impressed their footprints along Tr A, Dir F, and Tr E could have had the largest body size (Figure 14). The score setting in the PCA biplots gives roughly analogous hints (Figure 15 and Figure 16). In the PCA biplot of BM estimates, in which PC1 captures 78.68% of the variance, the scores are mainly distributed along axis 1, with those of Gombore II-OAM laying in the first quadrant just below the axis and those of Happisburgh and Gombore II-2 in the second quadrant just above axis 1 but close to axis 2. Those of other sites lie in the third quadrant, with the F/DT score position confirming the large BM of the footprints’ trackmakers. The setting is in accordance with the position of the eigenvectors corresponding to the variables based on rFl that are mainly directed toward the left quadrants, while the longest one, which corresponds to the variables based on Fa2, sets in the second quadrant close to axis 2. In the PCA biplot of ST estimates, the score setting is similar. PC1 captures 88.17% of the variance, and all eigenvectors are directed towards the left quadrants, with those based on rFl and Fa set in the third quadrant, while those based on rFw are in the fourth one.
Figure 14.
Scatter diagram of the average stature against the body mass estimates derived from the best-preserved footprints impressed at the Foresta/Devil’s Trails ichnosite and those from the compared African and European sites.
Figure 15.
Biplot diagram produced by the principal component analysis (PCA) using the average body mass estimates derived by all the equations based on the dimensions of the best-preserved footprints (Sample B) impressed at the Foresta/Devil’s Trails ichnosite and those from the compared African and European sites as variables. The component loadings (below) show the degree to which the different original variables enter into components 1, 2, and 3. Abbreviations are shown in Figure 8.
Figure 16.
Biplot diagram produced by the principal component analysis (PCA) using the average stature estimates derived by all the equations based on the dimensions of the best-preserved footprints (Sample B) impressed at the Foresta/Devil’s Trails ichnosite and those from the compared African and European sites as variables. The component loadings (below) show the degree to which the different original variables enter into components 1, 2, and 3. Abbreviations are shown in Figure 8. Colors as in Figure 15.
The average BM and ST values obtained from the dimensions of the best-preserved F/DT footprints of Tr A (BM = 64.0 kg; ST = 166.4 cm), Tr B (BM = 55.5 kg; ST = 152.9 cm), and the single footprints of Tr C (BM = 55.6; 158.1) and Tr E (BM = 64.6 kg; ST = 176.7 cm) are definitely higher than those provided for the footprints of the African ichnosites herein analyzed and of the oldest European Happisburgh (between 1.0 Ma and 800 ka) and slightly more recent Schöningen (~300 ka) sites. Among the other European ichnosites, the single footprint from Terra Amata (~380 ka) returns an ST value comparable with that of Tr A, whilst the BM value is slightly lower but comparable with those obtained for Tr B and Tr C. The footprints from the Vértesszőlős Cave site (~310 ka) returned an average BM estimate lower than those provided by F/DT footprints, while the ST is comparable with that of Tr A (Table 18 and Table 19).
Finally, we compared the body size of the Tr A trackmaker with that already obtained by Saborit and colleagues [91] by means of a different methodological approach. The authors, who studied the F/DT footprints of Tr A to ascertain the dynamic behavior of the trackmaker, provided a body size estimate of 68.4 kg and 160 cm. These values perfectly match the BM average values that we have obtained for Tr A (Table 18), and they roughly match the BM average values that we have obtained for Tr A using the foot dimensions as variables. Vice versa, the ST average value obtained for Tr A is slightly higher (166.4 cm), but the ST values returned using the formulas based on rFl [82,84,85] are comparable, spanning from 161.131 cm to 162.11 cm, as they are the estimates derived from the inferred foot dimensions, which span from about 155 cm to 165 cm (mean = about158). However, formulas based on the same variable proposed by [52,83] provided higher ST estimates than those obtained by Saborit and colleagues [91] (ST based on fpD = 167.76 cm and 170.00 cm, respectively).
3.3. Remarks on the Differences Resulting from the Comparative Analysis Obtained by Using Footprints and Inferred Fleshy Foot Dimensions
The BM and ST estimates returned by each method using the equation based on the inferred foot dimensions (i.e., footprints’ length, rFl, and width, rFw) (Table 20) are moderately variable (Table 21 and Table 22), as confirmed by the values of the variation coefficient that are significantly lower than 20 in both Samples A and B. The variation coefficient (CV) is higher for the BM than the ST estimates, reaching its maximum values in the case of the BM estimates of the Tr C (Sample A) footprints, which are obtained using Fw as a variable and Domjanic and colleagues’ [83] (CV = 16.789) and Fessler and colleagues’ [85] (CV = 16.635) methods (Table 23). The maximum CV value of the ST estimate (6.51) is significantly lower and has been obtained for all the footprints (Sample A), applying both Fessler and colleagues’ [85] and Webb’s [84] formulas (Table 24). The same occurs regarding the ST estimates obtained from the footprint dimensions (Table 6). The values of skewness and kurtosis indicate a negligible asymmetry in the Gaussian curve, which is slightly greater according to the BM estimates (Table 24 and Table 25).
Table 20.
Dimensions of the footprints impressed by the trackmakers at the F/DT ichnosite converted into foot dimensions. # fleshy foot dimension = footprint dimension −4% [52,54].
Table 21.
Body mass estimates obtained from the inferred dimension of the fleshy foot of the Foresta/Devil’s Trails trackmakers.
Table 22.
Stature estimates obtained from the inferred dimension of the fleshy foot of the Foresta/Devil’s Trails trackmakers.
Table 23.
Summary of the univariate analysis statistics’ data of the body mass estimates derived from the inferred dimension of the fleshy foot of the Foresta/Devil’s Trails trackmakers.
Table 24.
Summary of the univariate analysis statistics’ data of the stature estimates derived from the inferred dimension of the fleshy foot of the Foresta/Devil’s Trails trackmakers.
Table 25.
Reciprocal percentage excess (δ) of the average body mass estimates derived by each equation applied to the inferred dimension of the fleshy foot of the Foresta/Devil’s Trails trackmakers.
The variability of BM estimates derived from equations based on both the Fl and Fw variables is higher than that of the ST ones, which are based only on Fl. The BM variability is higher when the body size estimates obtained by means of all the applied methods to all the footprints are considered. BM estimates range from 38.43 kg to 95.86 kg and from 44.20 kg to 92.92 kg in Samples A and B, respectively (Table 23). The variability slightly reduces for the single trackways. The estimates obtained for Tr B show the lowest range variation, shifting from 41.4 kg to 81.17 kg and 44.9 kg to 75.29 kg in Samples A and B, respectively (Table 23). When all the footprints are considered, ST estimates range from 121.50 cm to 176.60 cm and from 142.27 cm to 176.60 cm in Samples A and B, respectively, while those resulting for each trackway are lower. The ST estimates reach the minimum range in the case of Tr B (Sample A, a range from 132.20 cm to 158.36 cm; Sample B, range = 144.79–158.36 cm) (Table 24).
Comparing the BM average values obtained by means of each method, the highest and lowest estimates correspond to those returned by the formulas of Domjanic and colleagues [83] and Grivas and colleagues [82], respectively. As expected, the same occurs when comparing the BM estimates obtained using the footprint dimensions since the foot dimensions are arithmetically derived from the former. The discrepancies between the average BM estimates obtained by means of different methods are evident when comparing the average values returned by the different equations. The results obtained underpin the highest percentage excess of the average value returned by Domjanic and colleagues’ [83] equations based on Fw and Fl+Fw with respect to those returned by Grivas and colleagues’ [82] equations based on the Fl of the right and left foot (maximum δ = 59.79 and δ = 64.68 for the equation based on Fw and Fw+Fw, respectively, both in the case of Tr A, Sample B). δ is rather inferior when comparing the average BM estimates returned by Grivas and colleagues’ [82] method with Ruff and colleagues’ [60] equation based on Fw (maximum δ = 46.93 and δ = 64.68 in the case of Tr A, Sample B), while δ is inferior when comparing this average value with those resulting using the equations based on Fl (Table 25). Indeed, when comparing the ST average values returned by the four equations based on the Fl variable (Table 26), the δ values between the average estimate obtained by applying Domjanic and colleagues’ [83] equation and those returned by the other formulas are still the highest, but they are negligible if compared with the BM δ between the estimates returned by Fessler and colleagues [85] and those of Webb [84] and Grivas and colleagues [82], which are equal to 0.34 in Samples A and B and range from 0.82 (Sample B, Tr A) to 1.09 (Sample A, all data) (Table 25).
Table 26.
Reciprocal percentage excess (δ) of the average stature estimates derived by each equation applied to the inferred dimension of the fleshy foot of the Foresta/Devil’s Trails trackmakers.
The box plots of the BM and ST estimates resulting from the analysis based on the inferred foot dimension (Figure 17, Figure 18, Figure 19 and Figure 20), which provide an easy-to-read visual representation of the variation range in data, give support to the above-described results. Indeed, in the box plots of BM estimates obtained by analyzing Sample A and B data, the highest estimates are those returned by the equations of Domjanic and colleagues [83] based on Fw and Fl + Fw, even with respect to the estimates provided by the equation of Ruff and colleagues [60], which is also based on Fw, while the BM estimates based on Fl proposed by Grivas and colleagues [82] and Ruff and colleagues [60] are rather similar and definitely lower.
Figure 17.
The box plots illustrate the variation ranges of the body mass estimates resulting from the different equations, which are based on the inferred foot dimensions derived from all the measured footprints (Sample A) impressed on the ignimbritic slope of the Foresta Devil’s Trails ichnosite. Abbreviation: Grivas = [82]; Domjanic = [83]; Ruff = [60].
Figure 18.
The box plots illustrate the variation ranges of the body mass estimates resulting from the different equations, which are based on the inferred foot dimensions derived from the best-preserved footprints (Sample B) impressed on the ignimbritic slope of the Foresta Devil’s Trails ichnosite. Symbols and abbreviations are shown in Figure 17.
Figure 19.
The box plots illustrate the variation ranges of the stature estimates resulting from the different equations, which are based on the inferred foot dimensions derived from all the measured footprints (Sample A) impressed on the ignimbritic slope of the Foresta Devil’s Trails ichnosite. Abbreviations: Fessler = [85]; Webb = [84]; Domjanic = [83]. Symbols are shown in Figure 17.
Figure 20.
The box plots illustrate the variation ranges of the stature estimates resulting from the different equations, which are based on the inferred foot dimensions derived from the best-preserved footprints (Sample B) impressed on the ignimbritic slope of the Foresta Devil’s Trails ichnosite. Abbreviations as in Figure 19. Symbols are shown in Figure 17.
Regarding the ST estimates, which are all based on Fl, those returned by Domjanic and colleagues’ [83] equation are higher than those obtained by means of the other methods [82,84,85], in particular, in the case of those of Sample B (Figure 19 and Figure 20). Moreover, the distribution of the ST estimate scores in the box plots further stresses the rather greater size of Tr A inferred foot dimensions with respect to those of Tr B and further emphasizes, on the one hand, the large inferred foot dimensions of short Tr E and of F02, and, on the other hand, the small size of D02 already suggested by the BM box plots. These results are comparable with those obtained by analyzing the footprint dimensions (Figure 5 and Figure 6). Moreover, the average BM and ST values derived from the inferred fleshy foot dimensions do not differ much from those obtained from the footprint dimensions, especially regarding the ST estimates (Table 27). The latter varies between −2.855% (Sample A, all data) and −4% (Samples A and B, all data). The differences between the BM estimates based on the best-preserved footprints are similar, while those obtained for Sample A are slightly greater than the values returned by Domjanic and colleagues’ [83] and Ruff and colleagues’ [60] equations. They are higher in the case of BM estimates returned by the equations proposed by Grivas and colleagues [82], which range from −6.979% (Sample A, Tr B) to −14.88% (Sample A, Tr A), hypothetically because of the high variation and differences in the length between the left and right footprints caused by the peculiar characteristic of the zigzagging trackways along the F/DT slope.
Table 27.
Percentage difference between the body size estimates derived from the inferred foot dimensions and those of the footprints on the surface of the Foresta/Devil’s Trails ignimbritic slope.
Finally, although the estimated body size values resulting from the inferred fleshy foot dimension are slightly lower than those based on the footprint dimensions, the differences substantially do not alter the results obtained for the footprint samples from the sites of Gombore II-OAM [75], Happisburgh, Gombore II-2 site [77], Terra Amata [78], Vértesszőlős Cave [79] and references therein, and Schöningen [80,81] (Table 28 and Table 29). In particular, the BM estimates obtained for the footprints from the Gombore II-OAM site are sometimes unrealistic with anomalously high and low values and very wide variation ranges, as indicated by the sigma values, though they are lower than those resulting from the estimates based on footprint dimensions. (Table 18). Moreover, the dimensional ratios between the samples from the various sites remain substantially unchanged.
Table 28.
Comparison among the average body mass estimates derived from the foot inferred dimensions (Sample B) of the trackways and directions at the Foresta/Devil’s Trails ichnosite and those of selected African and European sites. Table 28 consists of two parts, Part 1 and Part 2. Information about the ichnosites and raw data as provided in the literature are presented in Part 1, and the results of our application of the various equations employing these data as variables are shown in Part 2. Symbols and sources are found in Table 18.
Table 29.
Comparison among the average stature estimates derived from the foot inferred dimensions (Sample B) of the trackways and directions at the Foresta/Devil’s Trails ichnosite and those of selected African and European sites. Table 29 consists of two parts, Part 1 and Part 2. The raw data as described in the literature is presented in Part 1, and the results of our application of the various equations employing these data as variables are shown in Part 2. Symbols and sources are shown in Table 18.
All things considered, the analyses performed using the inferred foot dimensions of the F/DT trackmakers (e.g., delta percentages and box plots show that although the extent of the difference between the averages of the values obtained by applying each of the formulas used to estimate BM and ST from the foot inferred dimensions of F/DT trackmakers (Table 25 and Table 26) is somewhat inferior to those resulting from the analysis of footprint dimensions (Table 8, Table 9, Table 10 and Table 11), rather significant differences exist regarding the BM estimates). Furthermore, the δ values highlight the increase in the δ value when the equations are based on the Fw variable, as δ values obtained for the ST estimates also suggest. In addition, the box plots obtained considering the BM and ST estimates obtained from the inferred foot and the footprint dimensions of all 50 measurable footprints (Sample A) and the best-preserved footprints (Sample B) do not substantially differ, though the values of the estimates based on the foot dimension (Figure 17, Figure 18, Figure 19 and Figure 20) are slightly lower than those derived from the footprint dimensions (Figure 4, Figure 5, Figure 6 and Figure 7), which is expected due to the greater value of the latter dimensions. Both groups of box plots underline the different values returned by the applied formulas and equations, especially in the case of BM estimates, and the unbalanced length of whiskers and the asymmetric position of the median value frequently shifted towards the lowest or, rarely, the highest values of the interquartile range, implying some skewed distribution of data.
4. Discussion
Body size can be considered the best predictor of most of the fundamental physiological and ecological variables that regulate the role of any organism in its ecosystem (e.g., basal and energetic metabolic rates, reproduction, longevity, species abundance and population density, species home range, consumer resource dynamics, prey–predator relationships, etc.). Since Thomson’s [92] classic work was published, several researchers have scrutinized the multi-faceted relationships between body mass and life history parameters, as they have been documented, described, and extensively discussed in the literature by research dealing with a variety of vertebrate taxa (e.g., [93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108]).
The matter of estimating BM, the best proxy of body size [109], and ST in fossil vertebrate species have been widely investigated and long debated due to their crucial inferences at the biological level and the central role played in the reconstruction of macroevolutionary processes. The authors proposed various methods for obtaining as reliable as possible estimates. The methods are manly based on equations in which the independent variables are cranial, teeth, and bone linear dimensions, but estimates obtained by applying one or the other equations differ from each other, even if derived from either the same measurements of vertebrate bones or from different measurements of a single bone. The inconsistency among the results obtained by means of different formulas basically depends on the differences in the typology of the sample from which each formula was derived. Some authors proposed alternative methods to overcome this basic obstacle and provide more realistic estimates (e.g., [110,111,112,113,114,115,116]).
All things considered, the critical evaluation of the BM estimates obtained for a sample by means of several methods could provide an indicative range of the BM values putatively useful for a comparative evaluation of the average BM in co-specific populations having a different chronology or coming from different paleoecologic contexts. Scrutinizing the BM fluctuations of species in time and across space at a large scale may provide some hints in deconstructing their evolutionary dynamics over time. Consequently, in recent years, interest in estimating human physical parameters, such as weight and stature, and in inferring some characteristics, such as sex and possible pathologies, have been continuously growing and have also extended first to the field of human ichnology, mainly for forensic purposes, and then to hominid paleoichnology (e.g., [117,118]).
4.1. The Intriguing Matter of the Reliability of Body Size Estimates
The long-debated issue of the reliability of the various equations proposed by the authors and, in turn, body size estimates provided by one or the other method, increases in the case of estimates based on the dimensions of fossil hominin bones and footprints. In the case of fossil human footprints, the equations proposed are based on formulas derived from the statistical elaboration of data experimentally obtained by measuring barefoot shape and dimensions of footprints that numerically congruous groups of modern humans, of known age and sex and belonging to different ethnic groups, had left, mainly on flat substrates (e.g., [31,52,84,119]). Some researchers have critically evaluated the reliability of the obtained results that, in some cases, have been considered not compelling enough (e.g., [120,121,122,123] and references therein). The chance of returning unrealistic estimates is particularly high in the case of equations based on data derived from the analysis of samples either not numerically congruous and/or samples that include trackmakers of either different sexes or different ontogenetic ages (e.g., [83,113] and references therein). Moreover, the values of body size estimates provided by the different equations depend on both the structuring of the sample and the consistence of the single variables on which each formula is found, even in the case of footprint dimensions taken following the first “Standard protocol for the documentation of modern and fossil ichnological data” [124] (see the discussion in Supplementary Information S2). For example, in the case of the F/DT and compared ichnosites (Table 18 and Table 19), the inconsistency among estimates is higher using the equation for the rFw variable, and, in turn, Fa1. The latter has been often chosen to estimate the BM of fossil hominins, such as at Ileret [31], Laetoli [125], Happisburgh [76], and Barcin Hoyük [126].
4.2. Comparison Among Body Size Estimated at Different Sites and Related Issues: A Short Account
Focusing on the comparison among body size estimates obtained from the dimensions of the footprints in various ichnosites, the available evidence stresses the discrepancies not only if the BM and ST values are derived from the application of different formulas but also from applying the same equation to all the compared samples. Moreover, most of the formulas commonly used are based on datasets resulting from the analysis of footprints impressed on flat or gently sloping surfaces (e.g., White Sands National Park, New Mexico; Happisburgh, UK; Willandra Lakes, Australia; Le Rozel, France; etc.) (e.g., [7,53,76,84,127,128,129]) and could be barely used in the case of footprints impressed on uneven ground, coarse, or variously inclined substrates, such as that of F/DT. In fact, the preservation state and the nature and geometry of the substrate may affect the shape and dimensions of some footprints. For instance, several authors experimentally and empirically verified that animal footprints preserved on soft, fine-grained substrates with high water content can have a somewhat different shape from those left on substrates with different grain sizes and low water content, as reported for vertebrate footprints (e.g., [130,131,132,133,134]) and stressed for the human ones ([4,121,135,136], etc.). Furthermore, the pattern of F/DT footprints evidences that footprint shape and size left on a sloped surface by the same trackmaker may vary according to the gait and its direction [51]. The biases limit the possibility of a compelling comparison among data from sites with different characteristics based on both statistical and biomechanical models, substantially downgrading the significance of the obtained results. Hence, to achieve rather suitable results from the comparison of the body size of hominins from their footprints, it would be useful to estimate BM and ST by applying the same formulas to the dataset of each ichnosite and evaluate the obtained results in light of all the factors that could have caused footprint modifications due to any change in the foot–substrate interaction (e.g., presence or absence of relevant anatomical details, substrate nature and acclivity, gait speed, direction strike, step stability, paleoenvironmental context, taphonomic processes, etc.) (e.g., [4,52,135] and references therein), as well as to some footprint characteristics (foot shape, pressure distribution, and walking speed), which, however, may be problematic to evaluate in some cases ([60] and references therein). The same factors account for the discrepancies resulting from applying the same equation to all the compared samples to estimate the body size of the trackmakers at different sites using the inferred foot dimensions as variables.
Comparisons among the estimated body size of hominins from various ichnosites might provide rather compelling results if the estimates have been based on equations experimentally generated from the dimensions of footprints impressed under environmental conditions close to those of the ichnosites yielding the analyzed fossil footprints. This analytic approach was followed with acceptable results by some authors (e.g., [7,60,127,128,137,138,139,140] and references therein), who compared the estimates obtained for the analyzed samples from some ichnosites (Laetoli, Tanzania; Ileret, Kenya; Happisburgh, UK; Le Rozel, France; Barcin Höyük, Turkey; White Sands National Park, New Mexico; and Grotta della Basura di Toirano, Italy).
4.3. Glancing at the Body Size Estimated from Bone Dimensions
The comparative evaluation of the estimates derived from the bone and the footprint dimensions underlines some inconsistencies among the average values of body size estimates obtained by means of different equations, both using the same as well as different variables. Indeed, it is problematic to compare BM and ST estimates derived from equations based on very dissimilar experimental data and measurements. Furthermore, as expected, the body size estimates also vary in the case of values obtained from hominin bone dimensions by means of regression equations based on different variables.
In the case of Homo heidelbergensis s.l. (herein informally adopted to indicate the polymorph Middle Pleistocene populations with diversely advanced characters from those rather typical to those showing a mix of H. heidelbergensis and Homo neanderthalensis features) from Sima de los Huesos (Sierra de Atapuerca, Spain), Carretero and colleagues [58] estimated ST by means of equations based on the maximum length of complete long bones of twenty-seven specimens (twenty-one belonging to males and six to females). The ST values span from 162.3 cm to 181.6 cm. The ST average value obtained for the entire sample of Sima de los Huesos long bones (male mean plus female mean/2) is about 170 cm, while, excluding the fibula estimates, the ST average value is lower (163.3 cm). Later, Carretero and colleagues [141] provided BM values by means of equations based on the head diameter of two male femurs (80.1 kg and 82.1 kg for Femur X plus the pelvis of the same individual, and 69.7 kg and 73.3 kg for Femur XIIIM). The variation range of the resulting estimates is moderately wide, spanning from 67.9 kg to 82.2 kg (mean = 76.3 kg) (80.1 and 82.1 kg for Femur X plus the pelvis of the same individual, and 69.7 and 73.3 kg for Femur XIIIM). The range enlarges if the BM estimates have been derived from the inferred values of stature, femur, and skeleton weight, ranging from 80 kg to 102 kg for Femur X plus the pelvis and from 69.7 kg to 88.2 kg for Femur XIII. Pablos and Arsuaga [142] provided alternative regression equations to estimate the body size of the Sima de los Huesos human sample. The authors [142] derived BM and ST estimates from sex-specific equations based on the mediolateral breadth of the astragalus trochlea (BM) and the maximum length of the astragalus and calcaneus (ST) bones. Both male and female formulas were used for bones of undetermined sex. ST estimates obtained from sexed tarsal bones provide more accurate ST estimates because formulas have a low Estimate Standard Error (SEE). The BM average value calculated for male, female, and unsexed astragalus samples is 78.9 kg ± 14 kg, 64.6 kg ± 7.0, and 69.7 kg ± 10.0, respectively. The average ST estimates obtained using both the maximum length of astragalus and calcaneus are 173.9 cm ± 1.4 cm, 161.9 cm, and 175 cm ± 1.9 cm for males, 161.9 cm ± 2.3 cm and 160.6 cm ± 1.4 cm for females, and 167.4 cm ± 6.0 cm and 179.9 cm ± 6.9 cm for unsexed tarsals and calcaneus, respectively. Considering all the BM and ST values obtained for the Sima de los Huesos samples by means of all the formulas [58,141,142], the variation range of the BM and ST estimates roughly spans from about 65 kg to 102 kg (BM) and from 161 cm to 181 cm, respectively.
The analysis of H. heidelbergensis s.l.-selected bones from some levels of the “Complex moyen III” of Caune de l’Arago (Tautavel, France), dated to about 450 ka and correlated with MIS 12 [143], provided rather narrower variation ranges. Chevalier [144] obtained multiple BM and ST estimates by means of some linear and logarithmic regression equations based on the neck diameter of Femur A57 and the acetabulum height of pelvis A44. The BM values obtained for Femur A57 by my means of formulas derived from samples of Homo neanderthalensis, fossil H. sapiens, and modern H. sapiens span from 80.2 kg to 82.8 kg (mean = 81.57 kg), from 70.6 kg to 72.6 kg (mean = 71.9 kg), and from 77.9 kg to 79.7 kg (mean = 78.6 kg), respectively. The equation based on the A44 pelvis provided an estimate of 80.3 kg. The ST estimates based on femur A141 diaphysis spas are from 157.8 cm to 169.32 cm (mean = 162.1 cm), from 160.0 cm to 168.6 cm (mean =164.6 cm), and from 163.0 cm to 171.7 cm (mean = 167.9), respectively, with an average value of 164.3 cm. All things considered and taking into account the estimates returned by the equations with the best prediction percentage, the author of [144] proposed that H. heidelbergensis s.l. from Caune de l’Arago could have had, on average, a weight of about 80 kg and could be less than 170 cm tall, maybe around 166 cm.
These data confirm the difficulties in obtaining compelling BM and ST estimates using one formula or the other, so the chance of achieving sound results by making comparisons among different sites is further reduced.
4.4. Notes About the Comparison Among Foresta/Devil’s Trails Body Size Estimates of and Those Derived from Bone Dimensions
The problems related to the consistency of the body size estimates derived from both bone and footprint dimensions made a comparison of body size estimates derived from bone and footprint dimensions challenging, giving the results a merely indicative significance. In the case of F/DT estimates, the average BM values obtained from both the dimensions of the best-preserved footprints of Tr A (64.0 kg) and those inferred for the flesh foot, as well as those of Tr E (fpD = 64.7 kg; ftD = 65.05 kg), which are the highest obtained for the F/DT best-preserved footprints, roughly match the average BM value of the females from Sima de los Huesos (about 64.7 kg) but are lower than the BM average value obtained from the entire sample (73.3 kg), which is expected due to the high value provided by the male sample (about 78.9 kg) [142]. The latter is close to the average value estimated for the Caune de l’Arago hominin [144]. The BM average estimates provided by the F/DT footprints of Tr B and T C, using both the footprint and the inferred foot dimensions as variables, are lower than those returned by the bone dimensions of both sites. The ST average value of about 170 cm estimated for H. heidelbergensis s.l. from Caune de l’Arago [145] and Sima de los Huesos based on all the analyzed bones (170. cm) [142] is greater than those estimated for F/DT Tr B (159.2 cm) and Tr C (158.1 cm) and is rather similar to that of Tr A (166.4 cm) and is inferior to that of Tr E (176.7 cm). The latter is close to the ST estimate obtained for the sample of Sima de los Huesos males (175 cm ± 1.9 cm) and slightly inferior to that obtained for the samples of the unsexed calcaneus (179.9 cm ± 6.9 cm) [142]. The average ST values of Tr B and Tr C roughly fall in the range of the estimates obtained for the samples of females using the dimensions of both the astragalus (161.9 tcm ± 2.3 cm) and calcaneus (160.6 cm ± 1.4) as variables. Among the ST estimates derived from the inferred foot dimensions, only the average value obtained for Tr E is comparable to those obtained for the Caune de l’Arago [145] and Sima de los Huesos (all analyzed bones) [141] samples, while those obtained for Tr A (c. 158 cm), Tr B (c. 150 cm), and Dir F (c. 165 cm) are inferior, which is expected due to the small size of flesh feet with respect to that of footprints.
On the whole, the available data enable us to only formulate rather vague conclusions and to observe that the body size of the F/DT trackmakers falls in the wide variation range of H. heidelbergensis s.l.
5. Conclusions
In the present contribution, we investigated the body size of the F/DT hominin trackmakers and critically explored the reliability of the BM and ST estimates derived from the shape and dimensions of fossil hominin footprints.
The results obtained indicate that the body size estimates derived from the shape and dimensions of human footprints basically depend on the characteristics of the footprints impressed at each ichnosite that may limit the comparability of the measured footprint dimensions and the applied method and equations that could return highly dissimilar estimates. Therefore, the body estimates resulting from the F/DT and the compared sites merely represent the midpoint of large variation ranges; thus, their comparison has an indicative value.
The average BM and ST estimates obtained from the dimensions of the best-preserved footprints span from 55.5 kg to 64.6 kg and from 152,9 cm to 176.7 cm, respectively. Among the average estimate values inferred for each trackway, the highest are those of Tr E (BM = 69.268 kg, ST = 180.138 cm). Those of Tr A (BM = 54.67 kg, ST = 159.97 cm) are higher than those resulting from trackways B (BM = 50.82 kg, ST = 151.96 cm) and C (BM = 51.31, ST = 153. 30), which are roughly similar to each other.
Altogether, the results obtained support the presence at the F/DT ichnosite of at least four trackmakers going down the slope of the F/DT ignimbrite: the slightly large-sized trackmaker of Tr A; the trackmakers of Tr B and Tr C, with a rather similar body size; and the largest trackmaker of the short Tr E [51]. Based on the comparison among the average BM and ST estimates of F/DT trackmakers and those of the few sexed H. heidelbergensis s.l. samples [142], we could speculate that the trackmakers of Tr B and Tr C might be females or still young individuals and that of Tr A was a male, but such a hypothesis lacks any firm foundation.
All things considered, given the notable variation among the average body size estimates returned by the herein applied formulas and the objective difficulty in obtaining compelling and fully comparable estimates from bone and footprint dimensions, we could only say that the F/DT BM and ST estimates fall in the wide variation range of H. heidelbergensis s.l.-available samples. Moreover, the relationships between the means of the F/DT body size estimates and those obtained for the African and European ichnosites herein analyzed suggest that most of the hominins who left their footprint at the Gombore II-OAM, Happisburgh, and Gombore II-2 ichnosites might have an average body size inferior to those of the trackmaker of Schöningen, Terra Amata, and Vértesszőlős Cave. In particular, the F/DT trackmakers who impressed their footprints along Tr A, Dir F, and Tr E might have had the largest body size.
This study confirmed the results of previously detailed analyses of footprint preservation, morphology, and spatial arrangement suggested. Among the individuals (A, B, C, and E) who impressed their tracks at the F/DT ichnosite, A and B had a similar stature, as well as probably C, and E was taller; further evidence is needed to confirm the presence of a smaller individual (D) [51].
The results obtained underline the strict dependence of body size derived from the dimensions of fossil human footprints on the methods used for calculating these estimates but also highlight the purely indicative value of the comparison among body size estimates obtained from the hominin footprint dimensions, even applying the same methodological approach and the same measurement protocol. The limited significance of the comparison is due to the number of factors that could influence the mutual consistency of the obtained measurements (e.g., footprint preservation and presence or absence of relevant anatomical details, substrate nature and acclivity, gait speed, direction strike, step stability, paleoenvironmental context, taphonomic processes, etc.). Moreover, they indicate that some differences between the estimates obtained by applying different methods also exist in the case of body size estimates based on inferred foot dimensions, in particular those derived from equations based on different variables (e.g., the length and width of the foot). This evidence underpins the extent to which the peculiarity of the F/DT ichnosite could have influenced the shape and dimensions of each footprint. It also stresses the significant impact that the granularity, plasticity, and slipperiness of the substrate and the slope inclination could have had on the trackmakers’ gait, direction, trackmaker’s balance, and foot placement. These factors influenced footprint shape and size, making the body size estimates derived from any equation approximate and the comparisons with other Pleistocene human footprint sites problematic (see [67,68,69,70] for a detailed description of each footprint and [51] for an accurate discussion their morphometric anomaly). Moreover, the observed inconsistencies in applying the equations based on fleshy foot dimensions to footprints would suggest further experimental investigation of the relationships between the dimensions of footprints of humans walking on different substrates and their body size. As a result, only equations based on measurements of footprints created by a congruent group of persons going down a slope with similar characteristics could provide a more convincing estimate of the body size of the F/DT hominins, but it is difficult to replicate walking conditions that are similar to those that the F/DT trackmakers encountered.
All things considered, although studies dealing with the hominin body size estimation have been benefiting from increasingly advanced and complex analytic methods—based on experimental datasets and the critical use of the most appropriate algorithms and formulas—the obtained estimates are merely indicative of the probable body size of trackmakers. On the whole, body size estimates, even if they contribute to increasing our knowledge about Paleolithic hominins, have to be used with caution in any comparisons between multiple ichnosites.
Supplementary Materials
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/quat8010005/s1: Text S1: Methods for Body Mass (BM) and Stature (ST) estimates; Text S2: The intriguing issue of measuring the human footprint dimensions
Author Contributions
M.R.P. conceived and wrote the manuscript with the valuable support of AP and executed the data analyses; A.P. collected the data, calculated the BM and ST estimates, and wrote the Supplementary Information. All authors have read and agreed to the published version of the manuscript.
Funding
This research received no external funding.
Data Availability Statement
The complete F/DT dataset used for this report is available at the following URL: https://editoria.ingv.it/miscellanea/2022/miscellanea64S1/ (accessed on 4 July 2024). The 3D models elaborated for this report are currently used for another ongoing study. Therefore, the files are only available for specific requests.
Acknowledgments
We are grateful to the anonymous reviewers for their constructive suggestions that contributed to improving the scientific quality of the manuscript. We thank the editorial staff for the advice provided during the preparation of the manuscript.
Conflicts of Interest
The authors declare no conflicts of interest.
Acronyms and Abbreviations
| ANOVA | Analysis of variance |
| BM | Body mass |
| CV | Coefficient of variation |
| Dir | Direction |
| F/DT | Foresta/Devil’s Trails |
| Fa1 | Footprint area calculated by tracing the perimeter of the footprint through a polyline of at least 40 points |
| Fa2 | Footprint area calculated according to [31] |
| Fl | Inferred foot length |
| Fw | Inferred foot width |
| fpD | Footprint dimension |
| ftD | Inferred dimension of the flesh foot |
| ka | Kilo years |
| Ma | Million years |
| rFl and Fl (in Figure 8, Figure 9, Figure 10 and Figure 11 and Table 8, Table 9, Table 10 and Table 11) | Footprint length |
| rFw and Fw (in Figure 8, Figure 9, Figure 10 and Figure 11 and Table 8, Table 9, Table 10 and Table 11) | Footprint width |
| PCA | Principal component analysis |
| ST | Stature |
| Tr | Trackway |
References
- Topinard, P. L’Anthropologie, 2nd ed.; Reinwald: Paris, France, 1857; Volume 1. [Google Scholar]
- Martin, R.M. Lehrbuch der Anthropologie in Systematischer Darstellung: Mit Besonderer Berücksichtigung der Anthropologischen Methoden; Fischer: Jena, Germany, 1914. [Google Scholar]
- Pales, L. Les Empreintes de Pieds Humains Dans les Cavernes: Les Empreintes du Reseau Nord de la Caverne de Niaux (Ariege); Masson: Paris, France, 1976. [Google Scholar]
- Bennett, M.R.; Morse, S.A. Human Footprints: Fossilised Locomotion? Springer: Berlin/Heidelberg, Germany, 2014. [Google Scholar]
- Barr, K. Prehistoric Avian, Mammalian and H. Sapiens Footprint-Tracks from Intertidal Sediments as Evidence of Human Palaeoecology. Ph.D. Thesis, University of Reading, Reading, UK, 2010; pp. 1–593. [Google Scholar]
- Bennett, M.R.; Reynolds, S.C. Inferences from Footprints: Archaeological Best Practice. In Reading Prehistoric Human Tracks. Methods & Material; Pastoors, A., Lenssen-Erz, T., Eds.; Springer: Cham, Switzerland, 2021; pp. 15–39. [Google Scholar]
- Bennett, M.R.; Bustos, D.; Odess, D.; Urban, T.M.; Lallensack, J.N.; Budka, M.; Santucci, V.L.; Martinez, P.; Wiseman, A.L.A.; Reynolds, S.C. Walking in mud: Remarkable Pleistocene human trackways from White Sands National Park (New Mexico). Quat. Sci. Rev. 2020, 249, 106610. [Google Scholar] [CrossRef]
- Bavdekar, S.B.; Sathe, S.; Jani, P. Prediction of weight of Indian children aged up to two years based on foot-length: Implications for emergency areas. Indian Pediatr. 2006, 43, 125–130. [Google Scholar] [PubMed]
- Kanchan, T.; Menezes, R.G.; Moudgil, R.; Kaur, R.; Kotian, M.S.; Garg, R.K. Stature estimation from foot dimensions. Forensic Sci. Int. 2008, 179, 241.e1–241.e5. [Google Scholar] [CrossRef]
- Krishan, K. Determination of Stature from Foot and Its Segments in a North Indian Population. Am. J. Forensic Med. Pathol. 2008, 29, 297–303. [Google Scholar] [CrossRef]
- Mohanti, B.B.; Agrawal, D.; Mishra, K.; Samantsinghar, P.; Chinara, P.K. Estimation of height of an individual from foot lenght: A study on the population of Odisha. Int. J. Rev. Life Sci. 2012, 2, 69–74. [Google Scholar]
- Rani, M.; Tyagi, A.K.; Ranga, V.K.; Rani, Y.; Murari, A. Stature estimates from foot dimensions. J. Punjab. Acad. Forensic Med. Toxicol. 2011, 11, 26–30. [Google Scholar]
- Khan, H.B.M.A.; Nataraja Moorthy, T. Estimation of stature from foot outline measurements in Ibans of East Malaysia by regression analysis. Int. J. Biomed. Adv. Res. 2013, 4, 889–895. [Google Scholar]
- Kautilya, D.V.; Bodkha, P.; Poothanathan, P. Determination of stature and sex from anthropometry of the foot among south Indians. Int. J. Rev. Life Sci. 2013, 3, 22–26. [Google Scholar]
- Sherke, A.R.; Tamgire, D.W. Correlations of stature with foot length in Andhra region. Int. J. Biomed. Res. 2013, 4, 494–498. [Google Scholar] [CrossRef]
- Geetha, S.; Swathi, S.; Athavale, A. Estimation of stature from hand and foot measurements in a rare tribe of Kerala State in India. J. Clin. Diagn. Res. 2015, 9, HC01–HC04. [Google Scholar] [CrossRef]
- Malik, A.R.; Akhter, N.; Ali, R.; Farrukh, R.; Aziz, K. A Study on estimation of stature from foot length. Prof. Med. J. 2015, 22, 632–639. [Google Scholar] [CrossRef]
- Nataraja Moorthy, T.; Khan, H.B.M.A. Stature estimation from anthropometric measurements of footprints in Lun Bawang, an indigenous ethnic group of East Malaysia by linear regression analysis. Malays. Appl. Biol. 2016, 45, 69–74. [Google Scholar]
- Nataraja Moorthy, T.; Mostapa, A.M.B.; Boominathan, R.; Raman, N. Stature estimation from footprint measurements in Indian Tamils by regression analysis. Egypt. J. Forensic Sci. 2013, 4, 7–16. [Google Scholar] [CrossRef]
- Nataraja Moorthy, T.; Ling, A.Y.; Sarippudin, S.A.; Hassan, F.N.H. Estimation of stature from footprint and foot outline measurements in Malaysian Chinese. Aust. J. Forensic Sci. 2014, 46, 136–159. [Google Scholar] [CrossRef]
- Krishan, K. Individualizing characteristics of footprints in Gujjars of North India—Forensic aspects. Forensic Sci. Int. 2007, 169, 137–144. [Google Scholar] [CrossRef]
- Krishan, K. Establishing correlation of footprints with body weight–Forensic aspects. Forensic Sci. Int. 2008, 179, 63–69. [Google Scholar] [CrossRef]
- Krishan, K. Estimation of stature from footprint and foot outline dimensions in Gujjars of North India. Forensic Sci. Int. 2008, 175, 93–101. [Google Scholar] [CrossRef]
- Krishan, K.; Kanchan, T.; Passi, N. Estimation of stature from the foot and its segments in a sub-adult female population of North India. J. Foot Ankle Res. 2011, 4, 24. [Google Scholar] [CrossRef]
- Singh, J.P.; Meena, M.C.; Rani, Y.; Sharma, G.K. Stature estimations from the dimensions of foot in females. Antrocom Online J. Anthropol. 2013, 9, 237–241. [Google Scholar]
- Shariff, S.M.; Thamilvaani, M.; Shariff, A.A.; Merican, A.F. Evaluation of Foot Arch in Adult Women: Comparison between Five Different Footprint Parameters. Sains Malays. 2017, 46, 1839–1848. [Google Scholar] [CrossRef]
- Shukla, R.H.; Lodha, A.S.; Das, S. Stature estimation from footprint: A study on Central Indian population. Eur. J. Forensic Sci. 2017, 4, 11–19. [Google Scholar] [CrossRef][Green Version]
- Awais, M.; Naeem, F.; Rasool, R.; Mahmood, S. Identification of sex from footprint dimensions using machine learning: A study on population of Punjab in Pakistan. Egypt. J. Forensic Sci. 2018, 8, 72. [Google Scholar] [CrossRef]
- Abledu, J.K.; Abledu, G.K.; Offei, E.B.; Antwi, E.M. Determination of Sex from Footprint Dimensions in a Ghanaian Population. PLoS ONE 2015, 10, e0139891. [Google Scholar] [CrossRef]
- Akyeampong, J.G. Correlation of Sex, Height and Handedness with Anthropometric Foot and Hand Measurement of Young Adults Ghanaians. Master’s Thesis, University of Ghana, Accra, Ghana, 2017; pp. 1–175. [Google Scholar]
- Dingwall, H.L.; Hatala, K.G.; Wunderlich, R.A.; Richmond, B.G. Hominin stature, body mass, and walking speed estimates based on 1.5 million-year-old fossil footprints at Ileret, Kenya. J. Hum. Evol. 2013, 64, 556–568. [Google Scholar] [CrossRef] [PubMed]
- Fawzy, I.A.; Kamal, N.N. Stature and Body Weight Estimation from Various Footprint Measurements among Egyptian Population. J. Forensic Sci. 2010, 55, 884–888. [Google Scholar] [CrossRef] [PubMed]
- Okubike, E.A.; Ibeabuchi, N.M.; Olabiyi, O.A.; Nandi, M.E. Stature Estimation from Footprint Dimensions in an Adult Nigerian Student Population. J. Forensic Sci. Med. 2018, 4, 7–17. [Google Scholar] [CrossRef]
- Atamtürk, D.; Duyar, I. Age-Related Factors in the Relationship Between Foot Measurements and Living Stature and Body Weight. J. Forensic Sci. 2008, 53, 1296–1300. [Google Scholar] [CrossRef]
- Zeybek, G.; Ergur, I.; Demiroglu, Z. Stature and gender estimation using foot measurements. Forensic Sci. Int. 2008, 181, 54.e1–54.e5. [Google Scholar] [CrossRef]
- Cordeiro, C.; Muñoz-Barús, J.I.; Wasterlain, S.; Cunha, E.; Vieira, D.N. Predicting adult stature from metatarsal length in a Portuguese population. Forensic Sci. Int. 2009, 193, 131.e1–131.e4. [Google Scholar] [CrossRef]
- Reel, S.; Rouse, S.; Vernon, W.; Doherty, P. Estimation of stature from static and dynamic footprints. Forensic Sci. Int. 2012, 219, 283e1–283e5. [Google Scholar] [CrossRef]
- Uhrova, P.; Beňuš, R.; Masnicová, S.; Obertová, Z.; Kramárová, D.; Kyselicová, K.; Dörnhöferová, M.; Bodoriková, S.; Neščáková, E. Estimation of stature using hand and foot dimensions in Slovak adults. Legal Med. 2015, 17, 92–97. [Google Scholar] [CrossRef] [PubMed]
- Ishak, N.-I. Sex and Stature Estimation Using Hand and Handprint Measurements in a Western Australia Population. Master’s Thesis, University of Western Australia, Perth, WA, Australia, 2010; pp. 1–193. [Google Scholar]
- Ishak, N.-I.; Hemy, N.; Franklin, D. Estimation of sex from hand and handprint dimensions in a Western Australian population. Forensic Sci. Int. 2012, 221, 154.e1–154.e6. [Google Scholar] [CrossRef] [PubMed]
- Hemy, N.; Flavel, A.; Ishak, N.-I.; Franklin, D. Sex estimation using anthropometry of feet and footprints in a Western Australian population. Forensic Sci. Int. 2013, 231, 402.e1–402.e6. [Google Scholar] [CrossRef] [PubMed]
- Hemy, N.; Flavel, A.; Ishak, N.-I.; Franklin, D. Estimation of stature using anthropometry of feet and footprints in a Western Australian population. J. Forensic Leg. Med. 2013, 20, 435–441. [Google Scholar] [CrossRef]
- Trotter, M.; Gleser, G. Estimation of stature from long bones of American whites and Negroes. Am. J. Phys. Anthropol. 1952, 10, 463–514. [Google Scholar] [CrossRef]
- Trotter, M.; Gleser, G.C. A re-evaluation of estimation of stature based on measurement of stature taken during life and of long bones after death. Am. J. Phys. Anthropol. 1958, 16, 79–123. [Google Scholar] [CrossRef]
- White, R.W. Comparative Anthropometry of the Foot; United States Army, Natick Research and Development Laboratories: Natick, MA, USA, 1982; 320p. [Google Scholar]
- Gordon, C.C.; Churchill, T.; Clauser, C.E.; Bradtmiller, B.; McConville, J.T.; Tebbetts, I.; Walker, R.A. 1988 Anthropometric Survey of U.S. Army Personnel: Methods and Summary Statistics; United States Army Natick Research, Development and Engineering Center: Natick, MA, USA, 1989; pp. 1–638. [Google Scholar]
- Davis, K.T. The Foot Length to Stature Ratio: A Study of Racial Variance. Unpublished. Master of Art Thesis, Texas Tech University, Lubbock, TX, USA, 1990; pp. 1–64. [Google Scholar]
- Bassett, D.R., Jr.; Wyatt, H.R.; Thompson, H.; Peters, J.C.; Hill, J.O. Pedometer-Measured Physical Activity and Health Behaviors in U.S. Adults. Med. Sci. Sports Exerc. 2010, 42, 1819–1825. [Google Scholar] [CrossRef]
- Currie, P.J.; Sarjeant, W.A. Lower Cretaceous dinosaur footprints from the Peace River Canyon, British Columbia, Canada. Palaeogeogr. Palaeoclimatol. Palaeoecol. 1979, 28, 103–115. [Google Scholar] [CrossRef]
- Pasenko, M.R. Quantitative and qualitative data of footprints produced by Asian (Elephas maximus) and African (Loxodonta africana) elephants and with a discussion of significance towards fossilized proboscidean footprints. Quat. Int. 2017, 443, 221–227. [Google Scholar] [CrossRef]
- Palombo, M.R.; Panarello, A. How many hominins walked on the slope of the Foresta ignimbrite deposit (Roccamonfina volcano, central Italy)? J. Mediterr. Earth Sci. 2023, 15, 229–271. [Google Scholar]
- Duveau, J. From footprint morphometrics to the stature of fossil hominins: A common but uncertain estimate. L’Anthropologie 2022, 126, 103067. [Google Scholar] [CrossRef]
- Duveau, J.; Berillon, G.; Verna, C.; Laisné, G.; Cliquet, D. The composition of a Neandertal social group revealed by the hominin footprints at Le Rozel (Normandy, France). Proc. Natl. Acad. Sci. USA 2019, 116, 19409–19414. [Google Scholar] [CrossRef] [PubMed]
- Ruiz, J.; Mansilla, F.; Arsuaga, J.L.; Santos, E.; Jiménez-Díaz, A.; Egea-González, I. The speed and displacement of the Laetoli Site G track-maker hominins. Ichnos 2022, 29, 205–217. [Google Scholar] [CrossRef]
- Pastoors, A.; Lenssen-Erz, T.; Ciqae, T.; Kxunta, U.; Thao, T.; Bégouën, R.; Biesele, M.; Clottes, J. Tracking in Caves: Experience based reading of Pleistocene human footprints in French caves. Camb. Archaeol. J. 2015, 25, 551–564. [Google Scholar] [CrossRef]
- Pastoors, A.; Lenssen-Erz, T.; Breuckmann, B.; Ciqae, T.; Kxunta, U.; Rieke-Zapp, D.; Thao, T. Experience based reading of Pleistocene human footprints in Pech-Merle. Quat. Int. 2017, 430, 155–162. [Google Scholar] [CrossRef]
- Ashton, N. Steps from History. In Reading Prehistoric Human Tracks; Pastoors, A., Lenssen-Erz, T., Eds.; Springer: Cham, Switzerland, 2021; pp. 153–168. [Google Scholar]
- Carretero, J.; Rodríguez, L.; García-González, R.; Arsuaga, J.; Gómez-Olivencia, A.; Lorenzo, C.; Bonmatí, A.; Gracia, A.; Martínez, I.; Quam, R. Stature estimation from complete long bones in the Middle Pleistocene humans from the Sima de los Huesos, Sierra de Atapuerca (Spain). J. Hum. Evol. 2012, 62, 242–255. [Google Scholar] [CrossRef]
- Pablos, A.; Lorenzo, C.; Martínez, I.; Bermúdez de Castro, J.M.; Martinón-Torres, M.; Carbonell, E.; Arsuaga, J.L. New foot remains from the Gran Dolina-TD6 Early Pleistocene site (Sierra de Atapuerca, Burgos, Spain). J. Hum. Evol. 2012, 63, 610–623. [Google Scholar] [CrossRef]
- Ruff, C.B.; Wunderlich, R.E.; Hatala, K.G.; Tuttle, R.H.; Hilton, C.E.; D’Août, K.; Webb, D.M.; Hallgrímsson, B.; Musiba, C.; Baksh, M. Body mass estimation from footprint size in hominins. J. Hum. Evol. 2021, 156, 102997. [Google Scholar] [CrossRef]
- Santello, L. Analysis of a Trampled Formation: The Brown Leucitic Tuff (Roccamonfina Volcano, Southern Italy). Ph.D. Thesis, Padua University, Padova, Italy, 2010; 136p. [Google Scholar]
- Mietto, P.; Avanzini, M.; Rolandi, G. Human footprints in Pleistocene volcanic ash. Nature 2003, 422, 133. [Google Scholar] [CrossRef]
- Di Vito, M.A. Il Geosito Delle “Ciampate del Diavolo”. In 2001–2021: Vent’anni di Ricerche Sulle “Ciampate del Diavolo”. Dalla Leggenda Alla Realtà Scientifica; Mietto, P., Panarello, A., Di Vito, M.A., Eds.; Misc. INGV: Rome, Italy, 2022; Volume 64, pp. 49–56. [Google Scholar]
- Panarello, A.; Palombo, M.R.; Biddittu, I.; Di Vito, M.A.; Farinaro, G.; Mietto, P. On the devil’s tracks: Unexpected news from the Foresta ichnosite (Roccamonfina volcano, central Italy). J. Quat. Sci. 2020, 35, 444–456. [Google Scholar] [CrossRef]
- Panarello, A.; Palombo, M.R.; Biddittu, I.; Mietto, P. Fifteen years along the “Devil’s Trails”: New data and perspectives. Alp. Mediterr. Quat. 2017, 30, 137–154. [Google Scholar]
- Panarello, A.; Santello, L.; Farinaro, G.; Bennett, M.R.; Mietto, P. Walking along the oldest human fossil pathway (Roccamonfina volcano, Central Italy)? J. Archaeol. Sci. Rep. 2017, 13, 476–490. [Google Scholar] [CrossRef]
- Panarello, A.; Farinaro, G.; Mietto, P. L’icnosito Della Località “Foresta” di Tora e Piccilli e le Impronte Umane Fossili. In 2001–2021: Vent’anni di Ricerche Sulle “Ciampate del Diavolo”. Dalla Leggenda Alla Realtà Scientifica; Mietto, P., Panarello, A., Di Vito, M.A., Eds.; Misc. INGV: Rome, Italy, 2022; Volume 64, pp. 123–164. [Google Scholar]
- Panarello, A.; Farinaro, G.; Mietto, P. Il dataset dimensionale completo delle “Ciampate del diavolo”. In 2001–2021: Vent’anni di Ricerche Sulle “Ciampate del Diavolo”. Dalla Leggenda Alla Realtà Scientifica; Mietto, P., Panarello, A., Di Vito, M.A., Eds.; Misc. INGV: Rome, Italy, 2022; Volume 64 S1, pp. 1–38. [Google Scholar]
- Panarello, A.; Farinaro, G.; Mietto, P. Costruzioni geometriche per la creazione del dataset dimensionale completo delle “Ciampate del diavolo”. In 2001–2021: Vent’anni di Ricerche Sulle “Ciampate del Diavolo”. Dalla Leggenda Alla Realtà Scientifica; Mietto, P., Panarello, A., Di Vito, M.A., Eds.; Misc. INGV: Rome, Italy, 2022; Volume 64 S2, pp. 1–46. [Google Scholar]
- Panarello, A.; Farinaro, G.; Mietto, P. Atlante Visuale Delle “Ciampate del Diavolo”. In 2001–2021: Vent’anni di Ricerche Sulle “Ciampate del Diavolo”. Dalla Leggenda Alla Realtà Scientifica; Mietto, P., Panarello, A., Di Vito, M.A., Eds.; Misc. INGV: Rome, Italy, 2022; Volume 64 S3, pp. 1–232. [Google Scholar]
- Helm, C.W.; McCrea, R.T.; Cawthra, H.C.; Lockley, M.G.; Cowling, R.M.; Marean, C.W.; Thesen, G.H.H.; Pigeon, T.S.; Hattingh, S. A New Pleistocene Hominin Tracksite from the Cape South Coast, South Africa. Sci. Rep. 2018, 8, 3772. [Google Scholar] [CrossRef] [PubMed]
- Helm, C.W.; McCrea, R.T.; Lockley, M.G.; Cawthra, H.C.; Thesen, G.H.H.; Mwankunda, J.M. Late Pleistocene vertebrate trace fossils in the Goukamma Nature Reserve, Cape south coast, South Africa. Palaeontol. Afr. 2018, 52, 89–101. [Google Scholar]
- Helm, C.W.; Lockley, M.G.; Cole, K.; Noakes, T.D.; McCrea, R.T. Hominin tracks in southern Africa: A review and an approach to identification. Palaeontol. Afr. 2019, 53, 81–96. [Google Scholar]
- Helm, C.W.; Cawthra, H.C.; Hattingh, R.; Hattingh, S.; McCrea, R.T.; Thesen, G.H.G. Pleistocene vertebrate trace fossils of Robberg Nature Reserve. Palaeontol. Afr. 2019, 54, 36–47. [Google Scholar]
- Altamura, F.; Bennett, M.R.; Marchetti, L.; Melis, R.T.; Reynolds, S.C.; Mussi, M. Ichnological and archaeological evidence from Gombore II OAM, Melka Kunture, Ethiopia: An integrated approach to reconstruct local environments and biological presences between 1.2 and 0.85 Ma. Quat. Sci. Rev. 2020, 244, 106506. [Google Scholar] [CrossRef]
- Ashton, N.; Lewis, S.G.; De Groote, I.; Duffy, S.M.; Bates, M.; Bates, R.; Hoare, P.; Lewis, M.; Parfitt, S.A.; Peglar, S.; et al. Hominin Footprints from Early Pleistocene Deposits at Happisburgh, UK. PLoS ONE 2014, 9, e88329. [Google Scholar] [CrossRef]
- Altamura, F.; Bennett, M.R.; D’Août, K.; Gaudzinski-Windheuser, S.; Melis, R.T.; Reynolds, S.C.; Mussi, M. Archaeology and ichnology at Gombore II-2, Melka Kunture, Ethiopia: Everyday life of a mixedage hominin group 700,000 years ago. Sci. Rep. 2018, 8, 2815. [Google Scholar] [CrossRef]
- de Lumley, M.-A.; Lamy, P.; Mafart, B. Une empreinte de pied humain acheuléen dans la dune littorale du site de Terra Amata. Ensemble stratigraphique C1b. In Terra Amata: Nice, Alpes-Maritimes, France. Tome II: Palynologie, Anthracologie, Faunes, Mollusques, Ecologie et Biogéomorphologie, Paléoanthropologie, Empreinte de Pied Humain, Coprolithes; de Lumley, H., Ed.; CNRS: Paris, France, 2011; pp. 483–507. [Google Scholar]
- Tanaka, I.; Markó, A.; Hyodo, M.; Strickson, C.; Falkingham, P.L. A re-analysis of Chibanian Pleistocene tracks from Vértesszȍlȍs, Hungary, employing photogrammetry and 3D analysis. Ann. Soc. Geol. Pol. 2021, 91, 75–83. [Google Scholar] [CrossRef]
- Altamura, F.; Lehmann, J.; Rodríguez-Álvarez, B.; Urban, B.; van Kolfschoten, T.; Verheijen, I.; Conard, N.J.; Serangeli, J. Fossil footprints at the late Lower Paleolithic site of Schöningen (Germany): A new line of research to reconstruct animal and hominin paleoecology. Quat. Sci. Rev. 2023, 310, 108094. [Google Scholar] [CrossRef]
- Altamura, F.; Serangeli, J. A tale of many tracks: An overview of fossil proboscidean footprints at Paleolithic sites around the world, with a particular focus on Schöningen, in Germany. J. Mediterr. Earth Sci. 2023, 15, 347–368. [Google Scholar]
- Grivas, T.B.; Mihas, C.; Arapaki, A.; Vasiliadis, E. Correlation of foot length with height and weight in school age children. J. Forensic Legal Med. 2008, 15, 89–95. [Google Scholar] [CrossRef] [PubMed]
- Domjanic, J.; Seidler, H.; Mitteroecker, P. A Combined Morphometric Analysis of Foot Form and Its Association with Sex, Stature, and Body Mass. Am. J. Phys. Anthropol. 2015, 157, 582–591. [Google Scholar] [CrossRef]
- Webb, S. Further research of the Willandra Lakes fossil footprints site, southeastern Australia. J. Hum. Evol. 2007, 52, 711–715. [Google Scholar] [CrossRef]
- Fessler, D.M.T.; Haley, K.J.; Lal, R.D. Sexual dimorphism in foot length proportionate to stature. Ann. Hum. Biol. 2005, 32, 44–59. [Google Scholar] [CrossRef]
- Panarello, A. A snapshot on some everyday actions of a Middle Pleistocene hominin: The Trackway B at the Devil’s Trails palaeontological site (Tora e Piccilli, Caserta, Central Italy). J. Anthropol. Sci. 2020, 98, 27–47. [Google Scholar]
- Hammer, Ø.; Harper, D.A.T.; Ryan, P.D. PAST: Paleontological Statistics Software Package for Education and Data Analysis. Palaeontol. Electron. 2001, 4, 1–9. [Google Scholar]
- Freedman, D.; Pisoni, R.; Purves, R. Statistics, 4th ed.; W.W. Norton & Company: New York, NY, USA; London, UK, 2007; 608p. [Google Scholar]
- Huynh, H.; Feldt, L.S. Estimation of the Box correction for degrees of freedom from sample data in randomized block and split-plot designs. J. Educ. Stat. 1976, 1, 69–82. [Google Scholar] [CrossRef]
- Greenhouse, S.W.; Geisser, S. On methods in the analysis of profile data. Psychometrika 1959, 24, 95–112. [Google Scholar] [CrossRef]
- Saborit, G.; Mondanaro, A.; Melchionna, M.; Serio, C.; Carotenuto, F.; Tavani, S.; Modafferi, M.; Panarello, A.; Mietto, P.; Raia, P.; et al. A dynamic analysis of Middle Pleistocene human walking gait adjustment and control. Ital. J. Geosci. 2019, 138, 231–238. [Google Scholar] [CrossRef]
- Thomson, J.A. On growth and form. Nature 1917, 100, 21–22. [Google Scholar] [CrossRef]
- Blueweiss, L.; Fox, H.; Kudzma, V.; Nakashima, D.; Peters, R.; Sams, S. Relationships between body size and some life history parameters. Oecologia 1978, 37, 257–272. [Google Scholar] [CrossRef]
- Lindstedt, S.L.; Calder III, W.A. Body size, physiological time, and longevity of homeothermic animals. Q. Rev. Biol. 1981, 56, 1–16. [Google Scholar] [CrossRef]
- Damuth, J.D.; MacFadden, B.J. (Eds.) Body Size in Mammalian Paleobiology: Estimation and Biological Implications; Cambridge University Press: Cambridge, UK, 1990. [Google Scholar]
- McNab, B.K. The Physiological Significance of Body Size. In Body Size in Mammalian Paleobiology: Estimation and Biological Implications; Damuth, J.D., Damuth, J., MacFadden, B.J., John, D., Eds.; Cambridge University Press: Cambridge, UK, 1990; pp. 1–23. [Google Scholar]
- Brown, J.H.; Marquet, P.A.; Taper, M.L. Evolution of body size: Consequences of an energetic definition of fitness. Am. Nat. 1993, 142, 573–584. [Google Scholar] [CrossRef]
- Brown, J.H.; Kodric-Brown, A.; Sibly, R.M. Chapter Nine. On Body Size and Life History of Mammals. In Body Size: Linking Pattern and Process Across Space, Time and Taxonomic Group; Smith, F.A., Lyons, S.K., Eds.; University of Chicago Press: Chicago, IL, USA, 2013; pp. 206–234. [Google Scholar]
- Brown, J.H.; Burger, J.R.; Hou, C.; Hall, C.A. The pace of life: Metabolic energy, biological time, and life history. Integr. Comp. Biol. 2022, 62, 1479–1491. [Google Scholar] [CrossRef]
- Marquet, P.A.; Navarrete, S.A.; Castilla, J.C. Body size, population density, and the energetic equivalence rule. J. Anim. Ecol. 1995, 64, 325–332. [Google Scholar] [CrossRef]
- Sibly, R.M.; Brown, J.H. Effects of body size and lifestyle on evolution of mammal life histories. Proc. Natl. Acad. Sci. USA 2007, 104, 17707–17712. [Google Scholar] [CrossRef]
- Smith, F.A.; Lyons, S.K. (Eds.) Body Size: Linking Pattern and Process across Space, Time and Taxonomic Group; University of Chicago Press: Chicago, IL, USA, 2013. [Google Scholar]
- De Magalhaes, J.P.; Costa, A.J. A database of vertebrate longevity records and their relation to other life-history traits. J. Evol. Biol. 2009, 22, 1770–1774. [Google Scholar] [CrossRef]
- Szekely, P.; Korem, Y.; Moran, U.; Mayo, A.; Alon, U. The mass-longevity triangle: Pareto optimality and the geometry of life-history trait space. PLoS Comput. Biol. 2015, 11, e1004524. [Google Scholar] [CrossRef]
- Ofstad, E.G.; Herfindal, I.; Solberg, E.J.; Sæther, B.E. Home ranges, habitat and body mass: Simple correlates of home range size in ungulates. Proc. R. Soc. Ser. B Biol. Sci. 2016, 283, 20161234. [Google Scholar] [CrossRef] [PubMed]
- Hopkins, S.S.B. Estimation of Body Size in Fossil Mammals. In Methods in Paleoecology. Vertebrate Paleobiology and Paleoanthropology; Croft, D., Su, D., Simpson, S., Eds.; Springer: Cham, Switzerland, 2018; pp. 7–22. [Google Scholar]
- Lyons, S.K.; Smith, F.A. Macroecological Patterns of Body Size in Mammals Across Time and Space. In Animal Body Size: Linking Pattern and Process Across Space, Time, and Taxonomic Group; Smith, F.A., Lyons, S.K., Eds.; University of Chicago Press: Chicago, IL, USA, 2019; pp. 116–146. [Google Scholar]
- Kozłowski, J.; Konarzewski, M.; Czarnoleski, M. Coevolution of body size and metabolic rate in vertebrates: A life-history perspective. Biol. Rev. 2020, 95, 1393–1417. [Google Scholar] [CrossRef] [PubMed]
- Gingerich, P.D.; Smith, B.H.; Rosenberg, K. Allometric scaling in the dentition of primates and prediction of body weight from tooth size in fossils. Am. J. Phys. Anthr. 1982, 58, 81–100. [Google Scholar] [CrossRef] [PubMed]
- Reynolds, P.S. How big is a giant? The importance of method in estimating body size of extinct mammals. J. Mammal. 2002, 83, 321–332. [Google Scholar] [CrossRef]
- Sellers, W.I.; Hepworth-Bell, J.; Falkingham, P.L.; Bates, K.T.; Brassey, C.A.; Egerton, V.M.; Manning, P.L. Minimum convex hull mass estimations of complete mounted skeletons. Biol. Lett. 2012, 8, 842–845. [Google Scholar] [CrossRef]
- Brassey, C.A.; Gardiner, J.D. An advanced shape-fitting algorithm applied to quadrupedal mammals: Improving volumetric mass estimates. R. Soc. Open Sci. 2015, 2, 150302. [Google Scholar] [CrossRef]
- Larramendi, A. Shoulder height, body mass, and shape of proboscideans. Acta Palaeontol. Pol. 2016, 61, 537–574. [Google Scholar] [CrossRef]
- Romano, M.; Manucci, F. Resizing Lisowicia bojani: Volumetric body mass estimate and 3D reconstruction of the giant Late Triassic dicynodont. Hist. Biol. 2019, 33, 474–479. [Google Scholar] [CrossRef]
- Campione, N.E.; Evans, D.C. The accuracy and precision of body mass estimation in non-avian dinosaurs. Biol. Rev. 2020, 95, 1759–1797. [Google Scholar] [CrossRef]
- Romano, M.; Antonelli, M.; Palombo, M.R.; Rossi, A.; Agostini, S. Drone testing for 3D reconstruction of massive mounted skeletons in museums: The case of Mammuthus meridionalis from Madonna della Strada (Scoppito, L’Aquila). Hist. Biol. 2022, 34, 1305–1314. [Google Scholar] [CrossRef]
- Robbins, L.M. Footprints: Collection, Analysis, and Interpretation; Charles, C., Ed.; Thomas: Springfield, IL, USA, 1985. [Google Scholar]
- Bennett, M.R.; Budka, M. Digital Technology for Forensic Footwear Analysis and Vertebrate Ichnology; Springer: Cham, Switzerland, 2019. [Google Scholar]
- Tuttle, R.; Webb, D.; Weidl, E.; Baksh, M. Further Progress on the Laetoli Trails. J. Archaeol. Sci. 1990, 17, 347–362. [Google Scholar] [CrossRef]
- Morse, S.A.; Bennett, M.R.; Gonzalez, S.; Huddart, D. Techniques for verifying human footprints: Reappraisal of pre-Clovis footprints in Central Mexico. Quat. Sci. Rev. 2010, 29, 2571–2578. [Google Scholar] [CrossRef]
- Morse, S.A.; Bennett, M.R.; Liutkus-Pierce, C.; Thackeray, F.; McClymont, J.; Savage, R.; Crompton, R.H. Holocene Footprints in Namibia: The Influence of Substrate on Footprint Variability. Am. J. Phys. Anthropol. 2013, 151, 265–279. [Google Scholar] [CrossRef] [PubMed]
- DeSilva, J.; McNutt, E.; Benoit, J.; Zipfel, B. One small step: A review of Plio-Pleistocene hominin foot evolution. Am. J. Phys. Anthropol. 2019, 168, 63–140. [Google Scholar] [CrossRef]
- Belvedere, M.; Budka, M.; Wiseman, A.; Bennett, M.R. When is enough, enough? Questions of Sampling in Vertebrate Ichnology. Paleontology 2021, 64, 661–672. [Google Scholar] [CrossRef]
- Falkingham, P.L.; Bates, K.T.; Avanzini, M.; Bennett, M.; Bordys, E.M.; Breithaup, B.H.; Castanera, D.; Citton, P.; Díaz-Martínez, I.; Farlow, J.O.; et al. A Standard Protocol for Documenting Modern and Fossil Ichnological Data. Palaeontology 2018, 61, 469–480. [Google Scholar] [CrossRef]
- McNutt, E.J.; Hatala, K.G.; Miller, C.; Adams, J.; Casana, J.; Deane, A.S.; Dominy, N.J.; Fabian, K.; Fannin, L.D.; Gaughan, S.; et al. Footprint evidence of early hominin locomotor diversity at Laetoli, Tanzania. Nature 2021, 600, 468–471. [Google Scholar] [CrossRef]
- Atamtürk, D.; Özbal, R.; Gerritsen, F.; Duyar, İ. Analysis and Interpretation of Neolithic Period Footprints from Barcın Höyük, Turkey. Mediterr. Archaeol. Archaeom. 2018, 18, 163–174. [Google Scholar]
- Bustos, D.; Jakeway, J.; Urban, T.M.; Holliday, V.T.; Fenerty, B.; Raichlen, D.A.; Budka, M.; Reynolds, S.C.; Allen, B.D.; Love, D.W.; et al. Footprints preserve terminal Pleistocene hunt? Human-sloth interactions in North America. Sci. Adv. 2018, 4, eaar7621. [Google Scholar] [CrossRef]
- Bennett, M.R.; Bustos, D.; Pigati, J.S.; Springer, K.B.; Urban, T.M.; Holliday, V.T.; Reynolds, S.C.; Budka, M.; Honke, J.S.; Hudson, A.M.; et al. Evidence of humans in North America during the Last Glacial Maximum. Science 2021, 373, 1528–1531. [Google Scholar] [CrossRef]
- Webb, S.; Cupper, M.L.; Robins, R. Pleistocene human footprints from the Willandra Lakes, southeastern Australia. J. Hum. Evol. 2006, 50, 405–413. [Google Scholar] [CrossRef] [PubMed][Green Version]
- Milàn, J. Variations in the morphology of emu (Dromaius novaehollandiae) tracks reflecting differences in walking pattern and substrate consistency: Ichnotaxonomic implications. Palaeontology 2006, 49, 405–420. [Google Scholar] [CrossRef]
- Fanelli, F.; Palombo, M.R.; Pillola, G.L.; Ibba, A. Tracks and trackways of Praemegaceros cazioti (Dépèret, 1897) (Artiodactyla, Cervidae) in the Pleistocene coastal deposits from Sardinia (Western Mediterranean, Italy). Boll. Soc. Paleontol. Ital. 2007, 46, 47–54. [Google Scholar]
- Eren, M.I.; Durant, A.; Neudorf, C.; Haslam, M.; Shipton, C.; Bora, J.; Korisettar, R.; Petraglia, M. Experimental examination of animal trampling effects on artifact movement in dry and water saturated substrates: A test case from South India. J. Archaeol. Sci. 2010, 37, 3010–3021. [Google Scholar] [CrossRef]
- Platt, B.F.; Hasiotis, S.T.; Hirmas, D.R. Empirical determination of physical controls on megafaunal footprint formation through neoichnological experiments with elephants. Palaios 2012, 27, 725–737. [Google Scholar] [CrossRef]
- Falk, A.R.; Hasiotis, S.T.; Gong, E.; Lim, J.D.; Brewer, E.D. A new experimental setup for studying avian neoichnology and the effects of grain size and moisture content on tracks: Trials using the domestic chicken (Gallus gallus). Palaios 2017, 32, 689–707. [Google Scholar] [CrossRef]
- Marty, D.; Strasser, A.; Meyer, C. Formation and Taphonomy of Human Footprints in Microbial Mats of Present-Day Tidal-flat Environments: Implications for the Study of Fossil Footprints. Ichnos 2009, 16, 127–142. [Google Scholar] [CrossRef]
- Hatala, K.G.; Roach, N.T.; Ostrofsky, K.R.; Wunderlich, R.E.; Dingwall, H.L.; Villmoare, B.A.; Green, D.J.; Harris, J.W.K.; Braun, D.R.; Richmond, B.G. Footprints reveal direct evidence of group behaviour and locomotion in Homo erectus. Sci. Rep. 2016, 6, 28766. [Google Scholar] [CrossRef]
- Scales, R. Footprint-Tracks of People and Animals. In Prehistoric Coastal Communities: The Mesolithic in Western Britain; Bell, M., Ed.; CBA Research Report: York, UK, 2007; Volume 149, pp. 139–159. [Google Scholar]
- Citton, P.; Romano, M.; Salvador, I.; Avanzini, M. Reviewing the upper Pleistocene human footprints from the ‘Sala dei Misteri’ in the Grotta della Bàsura (Toirano, northern Italy) cave: An integrated morphometric and morpho-classificatory approach. Quat. Sci. Rev. 2017, 169, 50–64. [Google Scholar] [CrossRef]
- Romano, M.; Citton, P.; Salvador, I.; Arobba, D.; Rellini, I.; Firpo, M.; Negrino, F.; Zunino, M.; Starnini, E.; Avanzini, M. A multidisciplinary approach to a unique palaeolithic human ichnological record from Italy (Bàsura Cave). eLife 2019, 8, e45204. [Google Scholar] [CrossRef]
- Avanzini, M.; Salvador, I.; Starnini, E.; Arobba, D.; Caramiello, R.; Romano, M.; Citton, P.; Rellini, I.; Firpo, M.; Zunino, M.; et al. Following the Father Steps in the Bowels of the Earth: The Ichnological Record from the Bàsura Cave (Upper Palaeolithic, Italy). In Reading Prehistoric Human Tracks. Methods & Material; Pastoors, A., Lenssen-Erz, T., Eds.; Springer: Cham, Switzerland, 2021; pp. 251–276. [Google Scholar]
- Zunino, M.; Starnini, E.; Arobba, D.; Avanzini, M.; Caramiello, R.; Citton, P.; Clementi, L.C.; Firpo, M.; Giannotti, S.; Negrino, F.; et al. Toirano revisited: Nuove ricerche geoarcheologiche e paleontologiche nella Grotta della Bàsura (Toirano, SV). Primi risultati dello studio multidisciplinare delle tracce antropiche e dei depositi a fauna del Pleistocene superiore. Riv. Di Sci. Preist. 2023, 73, 371–382. [Google Scholar]
- Carretero, J.-M.; Rodríguez, L.; García-González, R.; Quam, R.-M.; Arsuaga, J.L. Exploring bone volume and skeletal weight in the Middle Pleistocene humans from the Sima de los Huesos site (Sierra de Atapuerca, Spain). J. Anat. 2018, 233, 740–754. [Google Scholar] [CrossRef] [PubMed]
- Pablos, A.; Arsuaga, J.L. Tarsals from the Sima de los Huesos Middle Pleistocene site (Atapuerca, Burgos, Spain). Anat. Rec. 2024, 307, 2635–2664. [Google Scholar] [CrossRef] [PubMed]
- de Lumley, M.-A. Chapitre 3. In ventaire. Répertition stratigraphique. Rèpresentation anatomique. In Les Restes Humaines Pléistocèns de la Caune de l’Arago; Tome, I.X., de Lumley, M.-A., Eds.; CNRS Editions: Paris, France, 2022; pp. 21–53. [Google Scholar]
- Chevalier, T. Chapitre 15. La corpulence de l’Homme de la Caune de l’Arago: Stature et masse corporelle. In Les Restes Humaines Pléistocèns de la Caune de l’Arago; Tome, I.X., de Lumley, M.-A., Eds.; CNRS Editions: Paris, France, 2022; pp. 615–693. [Google Scholar]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).