**1. Introduction**

Quadrupedal gaits have been variously classified in terms of symmetry (i.e. symmetrical gaits wherein the motion of the limbs on one side of the animal is mirrored by the motion on the other side, versus asymmetrical gaits where it is not); temporal foot sequence (i.e. lateral sequence wherein the sequence of limbs lifting off the ground is left hind-left front-right hind-right front versus diagonal sequence wherein the sequence of limbs is left hind-right front-right hind-left front); temporal couplet pairings (i.e. lateral-couplet wherein the ipsilateral pairs of limbs land closer together in time versus diagonal couplet wherein the diagonal pairs of limbs land closer together in time); coordination of limbs (i.e. laterally coordinated gaits wherein ipsilateral limb pairs move forward together in unison or near unison,

versus diagonally coordinated gaits wherein diagonal limb pairs do so, versus square gaits wherein each limb moves more or less independently); and beats (i.e. two-beat gaits with ipsilateral or diagonal limb pairs contacting the ground close together, versus three-beat gaits involving a hind limb pair contacting the ground followed by independent front limbs contacts, and four-beat gaits wherein all four limbs contact the ground independently). For more on the classification of quadrupedal gaits see [1–7].

The slower gaits are four-beat walking gaits. They are symmetrical, involve a pendulum-like action of swinging legs wherein all four limbs operate relatively independently of each other (i.e. square), lack suspended phases (i.e. have a duty factor above 0.50 as hind limbs are on the ground for more than 50% of the stride cycle), and possess alternating three- and two-limb support structures. Such walks may be lateralsequence diagonal-couplet (as in salamanders and hedgehogs), diagonal-sequence diagonal-couplet (as in crocodilians and primates), or lateral-sequence (ipsi)lateralcouplet (as in carnivores and ungulates). Most medium-speed gaits have periods of suspension wherein all four limbs are off the ground at the same time (and so possess a duty factor of less than 0.50), and are two-beat gaits that involve coordination of ipsilateral or diagonal limbs with bouncing or spring-like mechanics. The diagonally coordinated trot, with a two-limb diagonal support structure and diagonal couplets, is the medium-speed gait of most quadrupeds. A few quadrupeds, such as camels, employ the ipsilaterally coordinated pace, with a two-limb ipsilateral support structure and ipsilateral couplets. The fastest gaits are asymmetrical and involve the coordination of contralateral legs operating in a leaping and strut-like manner, with extensive periods of four-limb suspension. If all four limbs are employed together we have the pronk or stott common to antelopes; if the contralateral hind limbs are employed in unison we have the bound (half or full) of rabbits; and finally if the contralateral hind limbs are employed sequentially then we have the gallop, either transverse as in ungulates with hind and fore legs mirroring each other, or rotary as in carnivores without such mirroring [2–4, 8–13].

Among mammals, horses possess one of the most diverse gait-sets so-far known [5, 6, 14, 15]. Various breeds have been observed employing 14 out of around 17 major quadrupedal gaits [2, 3, 13, 14]. Not only do horses utilize the standard square fourbeat lateral-sequence lateral-couplet walk, along with the diagonally coordinated twobeat diagonal-sequence diagonal-couplet trot, and the asymmetrical three-beat canter and transverse four-beat gallop (with a footfall sequence of left hind-right hind-left front-right front), but select breeds can also employ more unusual gaits. For example, the diagonally coordinated lateral-sequence diagonal-couplet walk (fox walk) and two-beat lateral-sequence diagonal-couplet trot (fox trot) are found in the Missouri Fox Trotter and Walkaloosa [16]. Also occurring are intermediate speed laterally coordinated gaits including the paso corto and paso largo of the Paso Fino Horse [17], the running walk (four-beat) and stepping pace (two-beat) of the Tennessee Walking Horse, the four-beat rack of the American Saddlebred and related tölt of the Icelandic Horse, and the two-beat pace of the Icelandic and Standardbred [18, 19]. Horses also on occasion utilize the asymmetrical rotary gallop (for quick initial bursts) with a footfall sequence of right hind-left hind-left front-right front, the asymmetrical half bound leap (for jumping over obstacles or in the ring) off the back pair of legs, as well as the even more unusual stott (in the ring or when startled) wherein a horse leaps into the air with all 4 ft employed simultaneously as in the Lipizzaner [2, 3, 13, 14].

Traditionally the square lateral-sequence lateral-couplet walk, the diagonally coordinated trot, the asymmetrical gallop, and sometimes the asymmetrical canter,

#### *New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

were considered to be the natural gaits of the horse whereas the laterally coordinated pace, rack, fox trot, and running walk were considered to be "artificial" gaits as they did not occur in all horse breeds and were not always spontaneously expressed in those breeds in which they did occur. It was thus hypothesized that such gaits were introduced by humans, either through breeding or training. For example, Hildebrand, one of the pioneers in the study of animal gaits, wrote that the horse, on account of the training provided by humans, "has learned to be versatile in the selection of gaits and also to use gaits (termed artificial) that are unnatural to the species and unique to itself" ([14], p. 701). Such a view was perhaps influenced by the words of Muybridge [20] who claimed of the rack that "It is an unnecessary and unnatural gait of the horse, and it is scarcely probable that the ancients trained the animal to its use." And more recently the abstract of a study on Tennessee Walking Horse gait genetics implied the running walk's recent development, stating "Following domestication, man selected the horse primarily for the purpose of transportation rather than consumption; this selective strategy created divergent traits for locomotion" ([21], p. 1377).

Be that as it may, there is now solid evidence from fossil horse trackways that ipsilaterally coordinated four-beat running gaits including the rack and perhaps running walk occurred in extinct horse lineages [22–25]. There is also literary and artistic evidence of the presence of lateral gaits in ancient horses. For example, Pliny the Elder, in his *Naturalis historia* 8.57 of 77 CE, talks of theldones [Asturcón] horses from the Asturias of Northern Spain that had an easy gait involving ipsilateral pairs of legs moving in unison [mollis alterno crurum explicatu glomeratio]. In addition, the famous Flying Horse of Gansu statue (ca. 220 CE) and various reliefs from the Chinese Han dynasty show horses in lateral gaits, and a Turkestan painting from ca. 700 CE shows a horse and camel pacing side by side ([18, 26], pp. 291–306; [27]). Hence the nature and occurrence of lateral "artificial" or what we now prefer to call altenative lateral gaits in horses in the present and past is well worth exploring.

We have previously examined contemporary horse gaits and the tracks they leave behind and have employed them to determine the gaits displayed by fossil horses [22–25]. We here refine our methodology and apply it to the study of fossil horse trackways found to be gaitable, i.e. trackways of single individuals with a series of four or more prints.

### **2. Materials and methods**

Previously published photos, diagrams, and data regarding the footprints left by fossil horses were examined in order to isolate or estimate these key linear kinematic values and footprint patterns for fossil horse trackways [22–25, 28–32]. In particular, data from the following 15 trackways were examined: GQ-1 from the Rainbow Basin, Barstow, California of 14.5 Ma ([25], p. 162, Table 5, and p. 163, Figure 5; as well as new material presented below); HSA-9, HSA-10, HSA-11, HSA-12, HSA-13, and HSA-14 from the Hoya de la Sima site in the Jumilla-Ontur Basin near Jumilla, Murcia, Spain formed around 8.7–7.8 Ma ([33], p. 260, Figure 3, p. 263, Table 1, and pp. 264– 265); OS-3, OS-4, and OS-5 from Colle di Osoppo, Osoppo, Italy of ca. 6.0–5.3 Ma ([28], p. 225, Figure 4, and pp. 229–230, Tables 1–3 and 6–8); SC-1 from the Sierra del Colmenar section of the Bajo Segura Basin, outside of Elche, Spain formed 4.9–4.2 Ma ([29], p. 12, Figure 2); LAET-A, LAET-B, and LAET-C from Site G in the Upper Laetoli beds of Tanzania from 3.7 Ma ([22], pp. 472–473, Tables 12.5–12.8,

pp. 476–477, Figures 12.16 and 12.17); and WB-1 from the Wally's Beach deposit at St. Mary's Reservoir, Cardston, Canada formed more recently around 13,000– 11,000 years ago ([30], p. 218, Figure 16).

Key linear parameters and ratios were recorded from the data, sketches, and descriptions of the fossil horse trackways, as were the footprint patterns found therein. Based upon our previous work on modern horse gaits [7, 23–25, 34], and further refined here, the key linear kinematic parameters, ratios, and footprint patterns useful for the determination of horse gaits are as follows (definitions modified from [35]; see **Figure 1**).

Footprint width (FW): Greatest distance in centimeters across the middle portion of the hoof impression taken perpendicular to direction of travel (akin to greatest measurement across quarters, or central portion, of hoof).

Footprint length (FL): Distance in centimeters between anterior and posterior edges of hoof impression taken parallel to direction of travel (akin to measurement between toe and heel of hoof).

Stride length or cycle length (SL): Distance in meters between the anterior portion (i.e. toe of hoof) of successive footprints of the same foot parallel to the orientation of the trackway (ideally between left hind impressions if possible). The stride length tends to increase as a gait gets faster.

Stride length/horse height ratio (SL/H): Dimensionless speed ratio found by dividing stride length by horse height at the withers. As gaits get faster this number

#### **Figure 1.**

*Measurements taken of horse trackway (LH = left hind; LF = left front; RH = right hind; RF = right front). FW = footprint width; F = footprint length; ISD = ipsilateral step distance; DSD = diagonal step distance; LO = lateral offset. IS = interior straddle; and SL = stride length.*

*New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

increases. In walking gaits this value is usually less than 1.0, between 1.0 and 2.0 in intermediate speed gaits, while in very fast gaits it can be above 2.0.

Distance between diagonal steps (DSD): Measurement in centimeters between anterior (i.e. toe) and posterior (i.e. heel) edges of contralateral front and hind hoof prints parallel to the orientation of the trackway.

Distance between ipsilateral steps (ISD): Measurement in centimeters between anterior (i.e. toe) and posterior (i.e. heel) edges of ipsilateral front and hind hoof prints parallel to the orientation of the trackway.

Ipsilateral step distance [overstep; overstrike]/Stride length ratio (ISD/SL): Distance between ipsilateral steps divided by the stride length. The ipsilateral overstep [overstrike] can reach as high as 15–30% in fast ipsilaterally coordinated gaits.

Diagonal/Ipsilateral step distance ratio (DSD/ISD): Distance between diagonal steps divided by distance between ipsilateral steps. This value is above 0.5 in square gaits, but increasingly lower than 0.5 in laterally coordinated gaits, and exceeds 1.0 in diagonally coordinated gaits.

Symmetry between diagonal and ipsilateral steps ([DSD1/DSD2 + ISD1/ISD2]/2): ½ (diagonal step distance 1/diagonal step distance 2) + (ipsilateral step distance 1/ipsilateral step distance 2).

Average interior straddle (IS) [gauge]: Average distance in centimeters between quarters of contralateral front and hind hoof prints measured perpendicular to the orientation of the trackway. In the walk and running walk this value is typically positive but in gaits with high-lateral coordination the hind limbs are free to come in or cross the centerline without interference and so this value is often negative.

Interior straddle/hind hoof width (IS/HW): Ratio of average interior straddle divided by average hind hoof width. This value is negative in fast laterallycoordinated gaits such as the rack and pace, around zero in slow walks and trots, and positive in fast walks and trots.

Average foot pair lateral offset (LO): Average distance in centimeters between quarters of closest hoof print pairs whether formed by ipsilateral, diagonal, or contralateral front or contralateral hind measured perpendicular to the orientation of the trackway. This value is low when ipsilateral pairs of feet are close together but high when diagonal pairs are close together.

Foot pair LO/FW: Ratio of average foot pair lateral offset divided by average width of hoof print at quarters. This value is high for gaits with diagonal pairs landing close together in space but low for gaits with lateral pairs landing close together in space.

Angle of hoof impression (HA): Though we did not incorporate this data here as more study is needed, we also recommend measuring the angle each hoof impression makes (line drawn between anterior (toe) center and anterior (heel) posterior margins of hoof impression) in relation to a line drawn parallel to the orientation of the trackway. If 3D image generating software is available and made use of (which would be ideal) it would also be advisable to record any angle of deviation from the horizontal the hoof makes. This may yield important data allowing discrimination of gaits in the future or showing aspects of conformation, slippage, or change of direction.

The first key step in determining the gaits displayed by fossil horses is to determine the footfall sequence found in these trackways, and for this purpose it is necessary to distinguish fore feet (manus) from hind feet (pes). This is possible for modern and fossil horses as from the Miocene onward the front hooves of horses tend to be wider, more isometric (with a length/width ratio closer to 1.00) and circular in shape, and rounder at the tip, whereas the hind hooves tend to be narrower, less isometric and more oval in shape, and possess a more pointed tip (see **Figure 2** below; see [7]).

#### **Figure 2.**

*Bivariate plot of length vs. width for central ungual phalanx (3PhIII) of fossil equids with isometric line. Hooves above line are wider than long (common in monodactyl species) and hooves below line are longer than wide (common in tridactyl species). Front hooves (manus) are solid in color while hind hooves (pes) are cross hatched.*

The second key step for the determination of horse gait from tracks is estimation of the height (at withers) of the horse that made the tracks. This is difficult as one does not always know the species, age, gender, genetics, or development of the trackmaker. This is why trackway data is important, especially the size of the front hoof print. Still correlation of hoof size with horse height is lower (0.41) than that of a correlation of other skeletal elements such as the skull (0.83–91), metacarpal (0.90–0.95), metatarsal (0.87–0.91), and phalanges (0.81–0.91) (see [34, 36–40]). Moreover, the allometric ratios of the different limb bones can vary from one species to another [41–43] and this can affect height estimations. For these reasons, we believe that height of the trackmaker is best estimated by combining data from horse trackways and associated horse species osteology, giving 50% of the weight to the tracks themselves as they are the primary known factor about the print maker, a print maker whose species is a matter of some inference and whose population height can be variable, then giving 25% weight to the skull (if available) which correlates well with the size of fossil horses, and finally giving 25% weight to the other postcranial elements that closely correlate with overall height such as the metacarpal, metatarsal, proximal phalanx, and distal phalanx. This method allows different lines of evidence to estimate printmaker height in cases when the exact size and body ratios in relation to height are unknown for the printmaker.

After determining the measurements of the front hoof track, as well as associated skeletal elements such as the skull, metacarpal, metatarsal, and phalanges, it is necessary to multiply these values height estimation multipliers appropriate to the fossil

*New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

horse tribe or grouping. We have calculated such multipliers elsewhere based upon heights of fossil horse species with complete skeletons, thereby arriving at multipliers for the different fossil horse subfamilies and tries including: Eocene subfamilies Propalaeotheriinae and Hyracotheriinae; the Oligocene subfamily Anchitheriinae; the late Miocene tribe Hipparionini; the Pliocene and Pleistocene tribe Equini; as well as the paraphyletic early Miocene merychippine group consisting of the genera *Acritohippus*, *Merychippus*, and *Scaphohippus* [7]. We also developed scaled hoof height estimation multipliers based upon known values in the modern horse where height at withers is 13.16 times size of front hoof [40] and 21.62 the size of the front distal phalanx or coffin bone. For these multiplier values see Appendix A **Table A1** (as well as [7]). Assuming then the species of the trackmaker can be determined, the following formula is ideal (if not one is forced to rely on trackway data alone):

0*:*25 CL ð � CLMsÞ�

0*:*25 MT ½ � ð � MTMsÞ þ ðMC � MCMsÞ þ ð1P3 � 1P3MsÞ þ ð Þ 3P3 � 3P3Ms *=*4Þ� 0*:*50 FFL ð Þ � FFLMs (1)

where CL = maximal skull length measured from the tip of the incisive bone to the nuchal crest; MT = greatest metatarsal III length, MC = greatest metacarpal III length, 1P3 = greatest proximal phalanx III length, 3P3 = greatest distal phalanx III length; FFL = front footprint length, and CLM, MTM, MCM, 1P3M, 3P3M, and FFLM are the multipliers for the matching fossil horse groupings (s).

With the footfall pattern determined and height of trackmaker estimated one can then reconstruct the gait of the trackmaker using five key linear track ratios (see **Table 1**): Step symmetry calculation (SS), stride length/horse height (SL/HT), interior straddle/hind hoof width (IS/HW), ipsilateral step distance/stride length (ISD/SL), and diagonal/ipsilateral step distance (DSD/ISD). The first three taken together distinguish asymmetrical from symmetrical gaits, and the latter three distinguish symmetrical gates. This can be seen by through Principal Component Analysis (see **Figure 3**). Based upon dimensionless speed, gaits are separated into slow walks (SL/HT = 0.78–1.09) and fast walks (SL/HT = 1.10–1.18); slow trots (SL/HT = 1.32 = 1.51) and fast trots (SL/ HT = 1.55–1.80); slow running walks (SL/HT = 1.13–1.36) and fast running walks (SL/ HT = 1.59–1.60); slow racks (SL/HT = 1.18–1.24) and fast racks (SL/HT = 1.39–1.70).



#### **Table 1.**

*Key linear measurements and ratios of modern horse trackways in different gaits (data from [25, 34]; and original data, 2022).*

#### **Figure 3.**

*Principal component analysis of horse gaits for six key factors: step symmetry (SS); interior straddle/hoof width (IS/HW); stride length/height (SL/HT); ipsilateral step distance/stride length (ISD/SL); diagonal step distance/ stride length (DSD/SL); and diagonal/ipsilateral step distance (DSD/ISD).*

*New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

#### **Figure 4.**

*Footprint patterns of various gaits in modern horses. In the fast walk (A) there is a small stride length with a small overstep of ipsilateral hind feet resulting in distinct lateral pairs of prints in roughly parallel lines and a diagonal step distance much larger than the ipsilateral one. The fox trot, true fast trot, and slow rack (B) forms a trackway similar to that of the walk with lateral pairs of prints lining up more or less in parallel but possesses a greater stride length. In the running walk (C) there is a large overstep yielding no obvious pairs of prints as the ipsilateral step distances and diagonal step distances are roughly equivalent with themselves. This should be contrasted with the gallop (D) which also lacks obvious print pairings but which has a much greater stride length and in which there is greater variance within the ipsilateral and diagonal step distances and a sequence of contralateral feet. In the fast rack or tölt (E) the ipsilateral step length is much greater than the diagonal one resulting in diagonal pairs of prints that form a bowed pattern with a large stride length and hind impressions that often cross over the centerline. In the stepping pace and true pace (F) there is an even greater stride length and the diagonal pairs of prints occur very close together as the ipsilateral step distance is much larger than the diagonal one. The scale in centimeters.*

The asymmetry of the galloping gait is reflected in the step symmetry (SS) calculation (i.e. ½ [DSD1/DSD2 + ISD1/ISD2]). Symmetrical gaits of the horse (walk, rack, trot, running walk) tend to have a step symmetry between 0.80 and 1.00 while asymmetrical gaits of horses (canter and gallop) tend to have a step symmetry between 0.25 and 0.70. In symmetrical horse gaits, the three ratios of ipsilateral step distance/stride length (ISD/SL), diagonal/ipsilateral step distance (DSD/ISD), and interior straddle/hoof width (IS/HW), can go a long way in distinguishing one gait form another. Ipsilateral step distance/stride length (ISD/SL) is around 0.15 to 0.50 in walking and trotting gaits, but around 0.10–0.40 in the rack or pace. The diagonal/ ipsilateral step distance (DSD/ISD) tends to have high negative or positive values in the walk and trot (20 to 3 in slower versions, 3–20 in faster versions) whereas it is usually quite low in ipsilaterally coordinated gaits such as the rack, pace, and running walk (0.20–3.00). Finally, the interior straddle/hind hoof width ratio (IS/HW) tends to be positive in the walk and trot, but negative in the ipsilaterally coordinated gaits of running walk, rack, and pace (see **Figure A1** [7, 25]).

Finally, to determine gait of a fossil horse trackmaker it is useful to compare the trackway with the standard footfall patterns and sequences left by modern horses in various gaits (see **Figure 4**). The galloping trackway is asymmetrical, and so will have a footfall sequence with two contralateral hind limbs landing in succession, whereas the other gaits being symmetrical will contain ipsilateral and diagonal pairs of limbs landing in succession. Walking, trotting, and slow racking gaits leave trackways with ipsilateral pairs of feet landing close together with a slight understep, overstep, or capping. Fast racking and pacing gaits have diagonal pairs of feet landing close together, with hind feet often crossing over the centerline. Galloping gaits have contralateral pairs of feet landing close together, but along with the running walk, often leave trackways of isolated prints without any obvious pairings, or with the canter leave a 1-2-1 pattern with the pair consisting of an overstep of ipsilateral feet.

#### **Figure 5.**

*Photographs of modern horses in various gaits. A. Slow walk of a Shetland Pony (horse 22), stride length = 129 cm; B. Slow Trot of a Sicilian Donkey (horse 25), stride length = 142 cm. C. Pace of a Tennessee Walking Horse (horse 1), stride length = 210 cm; D. Left-lead canter of a Shetland Pony (horse 24), stride length = 188 cm; and E. Running walk of a Tennessee Walking Horse (horse 1), stride length = 180 cm. Black bars are 50 cm long.*

Lastly, the stride length is least with walking gaits, increases with medium speed trotting and racking gaits, and is greatest with galloping gaits (**Figure 5**). Stride length/height, which tends to be 1.10 or lower in a walk, 1.10–2.00 in a medium-speed trotting or pacing gait, and 1.50–2.50 in a gallop (see **Table 1** above).

The last step is to estimate the speed of the trackmaker. For this purpose the authors have come up with a modified Alexander formula based upon study of modern horses in various gaits and speeds [7, 44], which is as follows:

$$w = 0.72 \,\mathrm{g}^{0.50} \mathrm{S}^{0.81} H^{0.21} \tag{2}$$

where *v* is the velocity in meters/second, *g* is the gravitational constant of 9.81 m/s<sup>2</sup> , *S* is the stride length in meters, *H* is the height of the horse at the withers in meters, and 0.23 is the speed multiplier.

### **3. Results**

#### **3.1 Trackways suggestive of diagonally coordinated trotting gaits in horses**

Four fossil equid trackways suggesting a slow trotting gait are found at two different sites (see **Tables 1** and **2** below). Three of these occur near each other at the Hoya de la Sima site in the Jumilla-Ontur Basin near Jumilla, Murcia, Spain from ca 8.7– 7.8 Ma ([31, 33]; see **Figures 6A** and **7A** above and below). Here overlapping tridactyl prints containing a large central hoof along with two smaller lateral hooves were preserved in gypsum layers and set down on a lake shore in the upper Tortonian (European Mammal Neogene Zone 11, hereafter MN 11). The size of the central hoof impression at the Hoya de la Sima site was around 9.3 cm in length and 7.5 cm in width (HSM-2). Three of the trackways (HSA-9, HSA-10, and HSA-11) contain impressions


*New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

> **Table 2.**

*Linear kinematic parameters of fossil horse trackways.*

#### **Figure 6.**

*Trackways of fossil horses in a slow trot (scale in cm). A. Trackway HAS-11 of* Cremohipparion matthewi *from the Hoya de la Sima site near Jumilla, Spain (after [33]) set down on a lake shore around 8.7–7.8 Ma. B. Trackway SC-1 of* Hipparion fissurae *from the Bajo Segura Basin near Elche, Spain (after [29]) made in coastal sands around 4.9–4.2 Ma.*

#### **Figure 7.**

*Photographs of trackways of trotting fossil horses. A. Pliocene trackways HAS-11 and HSA-9 of* Cremohipparion matthewi *from the Hoya de la Sima site near Jumilla, Spain (close-up of Figure 14C from [45], used with permission); and B. Pliocene trackway SC-1 of* Hipparion fissurae *near Elche, Spain (close-up of Figure 4C from [29], used with permission). Arrows show prints and direction of travel.*

in which ipsilateral hind feet understep and partially overlap the front feet as suggested by a hoof impression with a rounded tip and shorter length located anteriorly. Trackway HSA-9 is 9.5 m long and contains 15 impressions of overlapping

### *New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

hooves, each impression around 14.5 cm long and 11 cm wide consisting of a large central hoof around 8.2–10.1 cm long typically followed by lateral toe impressions. The diagonal step distance is around 56.0 cm and the stride length is 137.0 cm. Trackway HSA-10 is 5.5 m long and contains eight impressions of overlapping hooves, each impression around 12.9 cm long and 10.8 cm wide. The diagonal step distance is ca. 62.9 cm and the stride length (steps 1–5) is around 167.0 cm. Trackway HSA-11 is 4.1 m long and contains six impressions consisting of overlapping hooves measuring around 15.1 cm long and 12.3 cm wide. The diagonal step distance is around 63.6 cm and the stride length is 152.0 cm. The interior straddle is around 20 cm.

There is also a trackway consisting of 10 hoof impressions in ipsilateral pairs located in the Sierra del Colmenar section of the Bajo Segura Basin, near the town of Elche, in the Alicante Province of Spain, likely of a trotting horse from around 4.9–4.2 Ma ([29]; see **Figures 6B** and **7B** above). The trackway (SC-1) was laid down in early Pliocene (MN 14) evaporitic coastal muddy sands. The central footprint impressions are around 6.5 cm long and 4.4 cm wide, and the interior straddle between ipsilateral pair prints averages around 13.4 cm (12.7–14.1 cm) indicating a wide gauge trackway. The ipsilateral pairs of the trackway seem to contain an understepping separation of the front foot from the hind foot of 7 cm (13.5 cm total) linked by an odd groove that Lancis and Estevéz [29] postulate is due to hoof sliding, or in our view a toe (side?) scraping across the ground prior to full foot impact. The understep is indicated by the fact that the

*Photograph of modern horse (horse 19) in medium trot with some overstep, stride length = 265 cm. Bar is 50 cm in length.*

anterior print is rounder in front while the posterior one is more oval in front, though the hind print is wider than the front print (perhaps due to partial collapse of sediment around the front foot). Hence we postulate a trotting gait (as seen in **Figure 8** above) here though further study is warranted. As such the diagonal step distance (intercouplet distance) separating the ipsilateral pairs averages around 65.7 cm (65.0 cm, ca. 66.4 cm) and the stride length is about 173.8 cm.

#### **3.2 Trackways suggestive of laterally coordinated gaits in fossil horses**

As we have previously argued there are also trackways suggestive of laterally coordinated intermediate-speed gaits in fossil horses (see **Tables 1** and **2** below).

The best preserved of these trackways are those of three hipparionin horses laid down in volcanic ash at Locality 8, Site G, in the Upper Laetoli beds in Tanzania some 3.66 Ma ([22–25]; see **Figures 9B**,**C** and **10E** below). The trackways, 151 to 251 cm in length, contain eight to 13 prints [23, 24]. Trackway LAET-A of an adult horse had ipsilateral step distances (footprints 2–6) that averaged 34.5 cm and diagonal step distances that averaged 18.0 cm, a stride length of 137.0 cm, with the prints forming a bowed or wave-like pattern made by diagonal pairs with lateral offsets around 6.5 cm ([22], 473). Trackway LAET-B of an adult horse (and likely a mare as it parallels and overlaps the trackway of a juvenile), had ipsilateral step distances (footprints 1–5) that averaged 25.1 cm and diagonal step distances that averaged around 27.7 cm, a stride length of 138.9 cm, and consisted of four isolated foot impressions with no obvious pairings. Finally, trackway LAET-C of a juvenile horse, had ipsilateral step distances (footprints 6–10) averaging 23.9 cm and diagonal step distances that averaged 10.8 cm, a lateral offset around 4.0 cm, with a stride length of 94.4 cm, and a trackway that is bowed with distinct diagonal pairs of hoof impressions. The central hoof prints from the Laetoli horse trackways themselves yield the following approximate measurements: LAET-A (adult), front hooves (manus) 8.8 cm long by 8.0 cm

#### **Figure 9.**

*Laterally-coordinated gaits in fossil horses (scale in cm). A. Trackway GQ-1 showing rack or tölt of* Scaphohippus sumani *(after [25]). The impressions (grayed-out where missing) were laid down on a lake shore in the Mud Hills near Barstow, California around 14.5 Ma. B,Trackway LAET-A showing the species* Eurygnathohippus hasumense *in a rack or tölt (after [22–24]). The prints were formed in volcanic ash some 3.7 Ma at Laetoli, Site G, in Tanzania. C. Trackway LAET-B of prior species in a running walk (or rack).*

*New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

#### **Figure 10.**

*Photographs of intermediate speed laterally-coordinated gaits in fossil horses. A. Mold of Miocene trackway GR-1 of* Scaphohippus sumani *made on site at the Mud Hills near Barstow, California; B. Close up of upper pair of GR-1 hoof prints (ruler repositioned from original); C. Cast of upper pair of GR-1 hoof prints (ruler repositioned from original); D. Close up of lower pair of GR-1 hoof prints; E. Pliocene trackways LAET-B and LAET-C of mare in running walk to left and juvenile in rack or tölt beside her to the right made by* Eurygnathohippus hasumense *from Laetoli Site G in Tanzania. The mare prints have posterior distal mud adhesion that adds to impression length.*

wide and hind hooves (pes) 7.5 cm long and 6.5 cm wide; LAET-B (adult), front hooves (manus) 9.1 cm long by 8.0 cm wide and hind hooves (pes) 7.6 cm long by 6.7 cm wide; LAET-C (juvenile), front hooves (manus) 6.4 cm long by 4.0 cm wide and hind hooves (pes) 6.1 cm long by 4.1 cm wide. The adult Laetoli trackways ([22], pp. 476–477, Figures 12.16 and 12.17) possessed negative interior straddle (narrow gauge) measurements of ca. 0.5 cm (LAET-A), 2.0 cm (LAET-B), and 0.0 cm (LAET-C).

Another fossil horse trackway (GQ-1; ichnospecies *Hippipeda araiochelata*), one 61 cm long, also seems to be indicative of a lateral racking gait, this time of a middle Miocene merychippine horse ([25]; see **Figures 9A** and **10A**–**D** below). The trackway consists of four footprints set down at the edge of a lake some 14.5 Ma which were excavated from Greer Quarry in the Barstow Formation of the Mud Hills outside of Barstow, California [25, 32, 46]. The footprints were found on a rock slab exposed in convex epirelief from which a rubber mold was made (SBCM L1816-3436). One of the hoof impressions is incomplete but one can distinguish fore feet (manus) from hind feet (pes) based upon the other three impressions due to the fact that the hind foot impressions are more pointed at the tip and longer while the front foot impressions are more rounded at the tip and have a ratio of length to width

**Figure 11.**

*Laterally coordinated gaits of modern horses. A. Slow rack (show gait) with overstepping lateral pairs in a Rocky Mountain Horse (horse 10), stride length = 180 cm; B. Fast rack (pleasure gait) with understepping diagonal pairs of a Rocky Mountain horse (horse 11), stride length = 201 cm; C. Slow pace with diagonal pairs of a Tennesse Walker (horse 1), stride length = 208 cm. Bars are 50 cm in length. Arrows show direction of travel.*

closer to 1.0. Trackway GQ-1, when reconstructed according to an assumption of constant speed, has a stride length of around 97.2 cm, ipsilateral step distances of 33.3 cm, diagonal step distances averaging 5.3 cm, and lateral offsets of diagonal pairs averaging 1.3 cm in which the hind feet understep the front feet. The interior straddle is 0.1 cm [25]. The tracks themselves possess a front central hoof (manus) impression measuring 5.0 cm long by 3.5 cm wide, and a hind central hoof (pes) impressions measuring around 5.2 cm long by 3.5 cm wide [25, 32, 46]. One can see the resemblance of these trackways to those of modern horses in alternative lateral gaits (see **Figure 11A**–**D**).

#### **3.3 Trackways suggestive of asymmetrical galloping gaits in fossil horses**

Seven fossil horse trackways, from three different sites, are indicative of a galloping gait (see **Tables 1** and **2** below). Three trackways (HSA-12; HSA-13; HSA-14) occur at the same Hoya de la Sima site in the Jumilla-Ontur Basin near Jumilla, Murcia, Spain where the trotting trackways are found ([31]; see **Figures 12A** and **13A** above). They too were originally made on lake shore sediments of the upper Tortonian (MN 11) around 8.7–7.8 Ma. Trackway HSA-12 is 9.6 m long and consists of 21 individual impressions. The manus (front foot) print is around 9.3 cm long and 7.5 cm wide. The step lengths of HSA-12 (steps 1–5) are ca. 58.1, 51.6, 51.6, and 58.1 cm and the stride length is around 277.4 cm. Trackway HSA-13 is 3.3 m long and contains five to seven partly overlapping hoof impressions 14.9 cm long by 13.7 cm wide. The step distances (steps 1–5) are ca. 51.6, 58.1, 51.6, and 58.1 cm and the stride length is around

*New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

#### **Figure 12.**

*Trackways (cm scale) of fossil horses in a gallop. A. Trackway HSA-14 of* Cremohipparion matthewi *from the Hoya de la Sima site near Jumilla, Spain (after [33]) made on a lake shore ca. 8.7–7.8 Ma; B. Trackway OS-3 of* Hipparion fissurae *laid down in a flood plain near Osoppo, Italy some 6.0–5.53 Ma (after [28]). C. Trackway WB-1 of* Equus lambei *made 13–11 kyr in Wally's Beach fluvial sands near Cardston, Canada (agter [30]).*

#### **Figure 13.**

*Photographs of fossil horse galloping trackways. A. Pliocene trackway HSA-14 of* Cremohipparion matthewi *near Jumilla, Spain (close-up of Figure 10 from IGME, PT085, used with permission); B. Pliocene trackways OS-4 and OS-3 of* Hipparion fissurae *near Osoppo, Italy (close-up of Figure 15 from Venturini and Discenza, 2009, used with permission). Arrows show prints and direction of travel.*

277.4 cm. Finally trackway HSA-14 is 2.5 m long and contains four to six impressions 14.4 cm long by 12.6 cm wide. The steps distances (steps 1–5) are ca. 64.5, 64.5, 64.5, and 51.6 cm and the stride length is around 309.7 cm.

Three more likely galloping trackways occur near each other on the Colle di Osoppo, near Osoppo, Friuli Venezia Giulia, Italy [28]. The prints were set down in a flood plain near a river in the late Messinian between 6.0 and 5.3 Ma. The three trackways (OS-3, OS-4, and OS-5) are between 8.4 and 9.2 m long and consist of 11 individual footprints which have large central hooves measuring 7.0–10.0 cm long by 6.0–10.0 cm wide, along with two small lateral hoof impressions located posteriorly. The central hoof foot impressions are around 8.4 (OS-3), 8.3 (OS-4), and 7.0 cm (OS-5) in length (see **Figures 12B** and **13B** below). Trackway OS-3 contained step lengths (steps 4–8) of 75.0, 77.5, 80.0, and 76.5 cm, and a stride length of 344.0 cm. Trackway OS-4 contained step lengths (steps 7–11) of ca. 80.5, 84.0, 84.0, and 81.0 cm, and a stride length of 361.5 cm. Trackway OS-5 contained step lengths (steps 3–7) of 68.0, 81.0, 77.0, and 89.0 cm, and a stride length of 343.0 cm. Trackway OS-5, however, warrants further study as orientation of impressions at odd angles and poor preservation suggests the trackway may not belong to a single individual, nor that prints numbering 6 to 11 cross back and forth over trackway OS-4 several times.

Vincelette [25] also provided footprint parameters, based upon scale drawings [30], for a late Pleistocene set of footprints likely made by *Equus lambei* in fluvial sands and silts at the Wally's Beach deposit (DhPG-8) from the drained bottom of St. Mary's Reservoir near Cardston in Alberta, Canada (see **Figure 12C** below). The trackway (WB-1; ichnospecies *Hippipeda cardstoni*) is 401 cm long and consists of five prints 10.5 cm in both length and width made some 13,000–11,000 years ago. The stride length was around 356.6 cm long, with the distance between isolated footprints being 100.0, 70.9, 56.4, and 87.3 cm. One can see the resemblance of these tracks to a modern horse's gallop (see **Figure 14** below).

*New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

### **4. Discussion**

Modern horses (unless possessing a rare mutation) possess only one toe per leg, which consists of the previous central metapodials (metacarpals and metatarsals III) and phalanges (first, second, and third phalanges III). They evolved from early Eocene horses which were tetradactyl in the front limbs and tridactyl in the hind limbs, and from late Miocene tridactyl horses. The earliest horses studied here, from the Miocene and Pliocene, were all tridactyl, though the second and fourth digits were reduced in size. Quite often the lateral toes do not make impressions in the substrate, suggesting the horses are "functionally monodactyl" [46], and that the lateral toes have specialized functions, though this has been much debated [47]. The Pleistocene horses studied here were monodactyl.

Comparison of the data obtained from the fossil horse trackways with key kinematic parameters (**Table 3**) and footprint patterns (**Figures 4** and **5** above) of modern horses leads to the recognition of four different gaits in fossil horse trackways, the trot, the running walk, the rack, and the gallop.

#### **4.1 Diagonal trotting gaits in fossil horses**

As noted earlier, four fossil equid trackways best match a slow trotting gait (see **Tables 2**–**4**, and **Figures 6**–**8** above). This is based upon the fact that they are set down in a pattern of capping or slightly overstepping ipsilateral prints (ipsilateral step distance/stride length ranging from 0.05 to 0.08), at a moderate speed (stride length 130–180 cm; stride length/height ratios of 1.36–1.66), with a large negative


#### **Table 3.**

*Key ratios for fossil horse lateral gait trackways.*


#### **Table 4.**

*Major gait characteristics and velocity for fossil horse trackways.*

diagonal/ipsilateral step length ratio of 5.0 to 10.0. Though a slow rack can also set down prints in ipsilateral pairs this usually occurs at shorter stride lengths (stride length/height around 1.20) and is accompanied by an interior straddle that is close to zero or negative as hind feet can cross the center line as there is less foot interference [25, 34]. Here though the interior straddle of the tracks varies from around 0.6–1.1 cm.

Three of these trackways (HSA-9, HSA-10, and HSA-11) are from the Hoya de la Sima site near Jumilla, Spain [31, 33] consisting of overlapping tridactyl prints of ipsilateral feet formed around 8.7–7.8 Ma (see **Figures 6A** and **7A** below).

Three species of tridactyl hipparionins are found in MN 11 strata in nearby Spain, a smaller form *Cremohipparion matthewi*, a larger form *Hipparion laromae*, and even larger forms *Cremohipparion mediterraneum* and *Hipparion longpipes* [48–50]. In Pleistocene and extant horses the length of the hoof is around 1.8–2.0 times the length of the distal phalanx bone, in Miocene horses the hoof length is around 1.0–1.3 times the length of the distal phalanx III bone [7, 32]. In Pliocene horses the hoof size would presumably land between these two values (see [46], 27, **Table 1**, who notes changes

#### *New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

in hoof wall thickness over time). Hence we can estimate the hoof size to be around 1.4–1.7 times the length of the distal phalanx III. As *H. laromae* from Rome had a distal phalanx III (i.e. 3PhIII) 6.5 cm long by 6.2 cm wide, albeit from slightly later MN 10 strata [50], this suggests it would make a central hoof impression around 9.1–11.1 cm long whereas the impressions made by the central hoof in the substrate at the Hoya de la Sima site were around 8.0–9.3 cm long. This would seem to rule out *H. laromae* as the trackmaker, as well as the slightly larger forms of *C. mediterraneum* and *H. longpipes*. The smaller hipparionin species then, *Cremohipparion matthewi* [periafricanum], which had a size around 79% that of *H. laromae* and an estimated hoof size of 7.2–8.8 cm long would seem to be the best candidate for the trackmaker and is tentatively assigned here.

*Cremohipparion matthewi* from MN 11 strata at the Puente Minero site, Teruel Province, Spain, has a row of upper cheek teeth (PM-1350; P2 –M<sup>3</sup> ) 11.3 cm long, a metatarsal III (PM-960) 19.2 cm long, and a hind first phalanx III (PM-771) 4.7 cm long [50]. The type specimen from Samos had a skull (LGPUT OK-557) around 31.9 cm long with a row of upper cheek teeth around 11.2 cm long, associated metacarpals that averaged around 20.5 cm long, along with a metatarsal 21.4 cm long [51], and footprints around 9.3 cm long by 7.5 cm wide.

Using the hipparionin multipliers (**Table A1**) we arrive at the following estimated heights: 82.1 cm (from the skull), 97.5 cm (from the metatarsal), 94.5 cm (from the hind first phalanx), and 112.3 cm (from the footprints). Assigning 25% weight to both the skull and postcrania and 50% to the footprints gives a height estimation of 100.7 cm for the trackmaking *Cremohipparion matthewi*. This in turn would yield the following stride length/height ratios: 1.36 (HSA-9), 1.66 (HSA-10), and 1.51 (HSA-11), indicative, along with the overlapping or understepping ipsilateral foot pairs, of a slow trot at around 2.9–3.4 m/s using our modified Alexander formula.

The other trackway (SC-1) suggesting a trotting gait was located in the Sierra del Colmenar section of the Bajo Segura Basin, near Elche, Spain and formed around 4.9– 4.2 Ma ([29]; see **Figures 6B** and **7B** above). The trackway was likely made by the tridactyl horse species *Hipparion fissurae* found in similar strata [52–55]. Postcranial remains of *H. fissurae* from the MN 14 zone in Spain [52] include third metacarpals (GL 1+68, GL 370, GL 92+127, VAR 1-43) averaging 24.3 cm in length, third metatarsals (GL 142, GL 304, GL 391+390, FSL-SN, LA 90, LCA 81-42; VAR 132, VAR 144) averaging 26.8 cm in length, and first phalanges III averaging 6.4 cm in length in the front legs (GL 133, GL 374, GL 409, LA 127, LCA 8168, NM 18078) and 6.1 cm in the hind legs (GL 22, LA 91, LA 93, LA 137, LA 138, NM 18079, NM 18087, ORR 32, ORR 48, VAR 1-112, VA 31). The front central hoof impressions were around 6.5 cm long. A skull of the species found near Pavlodar, Kazakhstan, from ca. 5.5–5.3 Ma, measured about 38.4 cm long [55].

If we use the hipparionin multipliers (**Table A1**), we arrive at the following height estimations for the trackmaker: 97.9 (from the skull), 139.7 cm (from the third metacarpal), 136.1 cm (from the third metatarsal), 131.4 cm (from the anterior first phalanx III), 127.3 cm (from the posterior first phalanx III), and 78.5 cm (from the footprints). The footprint height estimation is here much lower than the skeletal estimations suggesting the presence of a young individual here, different bone length proportions than the norm, or perhaps incorrect species assignment. In any case, by assigning 50% weight to the footprints, 25% to the postcrania, and 25% to the skull, we arrive at a height estimate of 99.8 cm for the *Hipparion fissurae* trackmaker. This in turn yields an ipsilateral step/stride length ratio of 0.08, a diagonal/ipsilateral step distance ratio of 4.87, and a stride length/height ratio of 1.74, which along with the

understepping ipsilateral pairs strongly suggest a slower trotting gait, one which occurred at around 3.5 m/s using our modified Alexander formula. This trackway did have an unusually large interior-straddle for a trot, however.

#### **4.2 Lateral gaits in fossil horses**

As we have previously noted, there are also trackways suggestive of laterally coordinated intermediate-speed gaits in fossil horses (see **Tables 2**–**4**, and **Figures 9**–**11** above). This gait identification is suggested by footprint patterns wherein diagonal feet land near each other (low diagonal/ipsilateral step distance ratio of 0.15–0.95) at an intermediate speed (stride length of 90–150 cm; 1.20–1.50 stride length/height), with much overstepping (a large ipsilateral step distance/stride length ratio of 0.25–0.59). The other key identifying feature is the possession of a negative interior straddle (narrow gauge trackway), as laterally coordinated gaits have less potential limb interference and allow the hind limbs to cross the center line of the gait (possibly for balance and/or increased speed).

The tridactyl trackways (LAET-A; LAET-B; LAET-C) of the Upper Laetoli beds in Tanzania that formed some 3.66 Ma ([22–24]; see **Figures 9B, C** and **10E** above) were likely made by the species *Eurygnathohippus hasumense* found in nearby strata of a similar geological age [25, 56–61]. Two metacarpals (WM 1635/92 and WM 1669/92) from the Kiloleli Member (ca. 4-3 Ma) of the Manonga Valley, Tanzania, assigned to *E*. *hasumense* averaged 24.1 cm in length [57]. There is also a skull (AL 340-8) assigned to *E*. *hasumense* from the younger Denan Dora II beds (ca. 3.2 Ma) of Hadar, Kenya that measures ca. 51.5 cm, along with metacarpals (AL 155-156) averaging 26.2 cm [41, 62]. An incomplete skull (WM 1528/92) from the Manonga Valley, Tanzania seems slightly larger in size (1:1.04 ratio) than the one from Hadar, Kenya based upon muzzle-length comparisons [57], which would make it 53.6 cm long if fully proportional. There is also a front proximal phalanx III from Laetoli (LAET 2357, Loc. 10) 6.8 cm long that is likely from this same species ([22], 475, Table 12.9). Central hoof impressions were 8.8 and 9.1 cm long for the adult trackways A and B, and 6.4 cm long for the juvenile trackway C. Finally radii of the species from Manonga (WM 368/94) and Hadar measure around 31.2 and 32.5–34.5 cm, respectively. Intraspecific height variation is also observed here and likely elsewhere as can be seen in the variation in size of horse astragali from the Upper Laetoli Bed localities 2, 6–8, 14, 16, and 22 (plotted in [22], p. 475, Figure 12.15). The postcranial material from the Upper Ndolanya Beds noted in our previous publications [22, 25] has now been more definitively assigned to the more gracile hipparionin species *Eurygnathohippus cornelianus* and not included herein.

If we take the above values and multiply them by the hipparionin height estimation multipliers (**Table A1**), we end up with the following height estimates in centimeters: 136.7 (from the skull), 138.6 (from the third metacarpal), 139.6 (from the front proximal phalanx III), and 106.2 (A), 109.8 (B), and 77.2 (C) from the hoof impressions. If, following the method noted above, we assign the 50% weight to the footprint, 25% to the skull, and 25% to the postcranial measurements, we arrive at the following size estimations for the adult *Eurygnathohippus hasumense* track makers: adult horse A, 122.1 cm; adult horse B, 123.9 cm. The juvenile horse C had a size of around 87.9 cm if its skeleton was proportionate in size to the adult horses.

This yields the following key ratios for the three trackways. For trackway LAET-A: diagonal/ipsilateral step distance, 0.52; ipsilateral step distance/stride length, 0.25; stride length/horse height, 1.12; lateral offset/hind hoof width, 1.00; and for trackway

*New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

LAET-B: diagonal/ipsilateral step distance, 0.90; ipsilateral step distance/stride length, 0.18; stride length/horse height, 1.12; and lateral offset/hind hoof width, 0.81. And for trackway LAET-C: diagonal/ipsilateral step distance, 0.45; ipsilateral step distance/ stride length, 0.25; stride length/horse height, 1.07; lateral offset/hind hoof width 0.98.

The gait displayed by the three individuals (two adults and one juvenile) in these trackways have been variously described as a running walk [23–25], a singlefoot [22], or a tölt (Islandpferde Reitbuch, 286). Part of the inspiration for this study was to be able to reexamine these trackways and see if we could distinguish exactly which lateral gait(s) were displayed in the Laetoli hipparionin horses. We now believe we can. The values and characteristics noted above suggest that *Eurygnathohipus hasumense* adult horse A and juvenile horse C were engaged in a rack or tölt. This is indicated by the gaits having the diagonal/ipsilateral step distance ratios around 0.50 or less, as well as the bowed or wave-like pattern of prints formed by diagonal pairs, and a very narrow gauge trackway with negative interior straddles for adult horse A (2.0 cm) and foal C (0.0 cm). Adult horse B, likely a mare traveling near her offspring (horse C), instead seems to have employed a running walk gait (though a rack at an intermediate speed is also a possibility). This is suggested by the footprints lacking readily identifiable pairs, a diagonal/ipsilateral step ratio of 0.90, and a slightly negative interior straddle (0.5 cm). It may be that the mare utilized a running walk instead of a rack in order to more easily travel at a velocity matching her foal. There is a possibility that the mare employed an intermediate-speed rack which can lay down isolated prints resembling a running walk, but one would tend to expect an even more negative interior straddle (see [7, 34]). Based upon our modified Alexander formula these gaits would have taken place at around 3.0–3.1 m/s for the adults (LAET-A and LAET-B) and 2.1 m/s for the foal (LAET-C).

Another fossil horse trackway (GQ-1; ichnospecies *Hippipeda araiochelata*), one 61 cm long, of a middle Miocene merychippine horse from Greer Quarry outside of Barstow, California, also arguably displays a laterally coordinated gait ([25]; see **Figures 9A** and **10A–D**). Previously the four hoof impressions (GQ-1), laid down on the edge of a lake some 14.5 Ma which, and found at Greer Quarry outside of Barstow, California, had been assigned to *Scaphohippus intermontanus*. This was based upon its distal phalanx III matching the hoof impression closely, albeit a bit snuggly suggesting a very narrow hoof wall [25, 32], as opposed to the slightly smaller species *Scaphohippus sumani*, or the much larger *Hypohippus affinis* and *Megahippus mckennai*, found nearby in strata of a similar geological age. However, a recent study with which we agree argues that the *S. intermontanus* material should be reassigned to *S. sumani* due to intraspecific variability in tooth complexity between the two populations [63]. In fact, Merriam [64] only distinguished the postcrania of the two merychippine species by size, wherein larger material was assigned to *S. intermontanus* and smaller material to *S. sumani* [64]

The size of the *Scaphohippus sumani* printmaker can be estimated from cranial and postcranial material, as well as the prints, according to the method described above. A skull of the species from the Barstow Formation (Ba2), California (UMCP 21386) measures 33.6 cm, another skull (UCMP 21385) from the same locality is of proportionate length, and a third skull AMNH 87301) from the slightly older Olcott Formation (Ba1), Nebraska, measures around 31.5 cm [64, 65]. Barstow formation postcranial material includes a metacarpal (UCMP 22372) measuring 16.5 cm, metatarsals (UCMP 19817 and 23,130) measuring 18.1 and 18.3 cm in length, a front proximal phalanx III (UCMP 22372) around 3.9 cm long, a hind proximal phalanx III around 3.8 cm long (UCMP 19817), a front distal phalanx III (UCMP 22372) that

measured around 3.5 cm long by 3.2 cm wide, a hind distal phalanx III (UCMP 19817) measuring around 3.0 cm long by 2.7 cm wide [32, 64], and a front hoof impression 5.0 cm long by 3.5 cm wide.

If we take these measurements and multiply them by the protohippin multipliers (**Table A1**), we arrive at the following height estimations in centimeters: 91.4 (from the skull), 90.3 (from the metacarpal), 89.5 (from the metatarsals), 93.9 (from the front proximal phalanx III), 89.1 (from the rear proximal phalanx III), 76.1 (from the front distal phalanx III), 78.6 (from the hind distal phalanx III), and 66.2 (from the front hoof). The height estimation from the front hoof is here somewhat lower than that derived from the other dimensions suggesting perhaps a young individual, bone proportionality differences from those found in *Protohippus*, or perhaps misidentification of species. By again giving 50% weight to the footprints, 25% to the skull, and 25% to postcranial measurements we arrive at a height estimation for the *Scaphohippus sumani* printmaker of 77.5 cm.

This yields the following ratios for the GQ-1 trackway: stride length/height, 1.25; diagonal/ipsilateral step distance, 0.16; ipsilateral step distance/stride length, 0.34; lateral offset/hind hoof width, 0.37. The trackway also seems to display a somewhat bowed pattern of prints with diagonal pairs of feet landing close together. Due to the large lateral offsets that occur in different directions within the print pairs and form a somewhat bowed pattern, we interpret the gait in trackway GQ-1 to again be a laterally coordinated rack or tölt. However, given that there are only four impressions (one incomplete) and that the lateral offset is not large, it is possible, though less likely, for the trackway to be formed by a slow trot (in a horse with poor conformation or changing directions). A rack is more likely though, due to the bowed pattern, the fact the hind hooves both cross the centerline, and there is a lateral offset/hind hoof width ratio above 0.25 (see **Table 3**). The large understep, 1.06 times the hoof length, also better matches that of a rack or stepping pace, wherein the understep is 0.60–1.50 the hoof length, versus a slow trot, wherein the understep is usually 0.35–0.55 the length of the hoof, even if it can reach 0.80–0.85 the length of the hoof at times [7, 25, 34]. On the whole then, the trackway is similar to ones made by living horses in a fast rack (or tölt) or stepping pace (see **Figure 11** above). The value of stride length/height of 1.25, however, somewhat puzzlingly more resembles that of a slow rack than a fast one as is also the case with the *Eurygnathohippus* values, even as the footprint patterns match those of a fast rack. It could be that the horse heights are slightly overestimated here (due to the tracks being by a young individual or species with different limb proportions than *Protohippus*), or that horses are capable of a greater overstep at slower speeds than seen in our study. Further study of the *Scaphohippus* prints has also revealed that the first hind impression points slightly to the right rather than paralleling the center line and that the expected stride length/height is a bit lower than expected for a fast tolt. This is a bit puzzling but can occur in the rack or trot due to change of direction or poor limb conformation [22–25, 32]. The evidence here, therefore, best fits the gait of a rack or tölt, one at a speed of around 2.1 m/s using the modified Alexander formula, made by a member of the *Scaphohippus sumani* species. However, the evidence is not as conclusive as with the Laetoli prints noted above and it is possible that this trackway is from the gait of a tro.

#### **4.3 Galloping gaits in fossil horses**

Trackways evidencing a galloping gait are somewhat common in fossil horses, perhaps as such forceful hoof impressions are more likely to be preserved (see **Tables 2**–**4**, *New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

and **Figures 12**–**14** above). As noted above, evidence for a galloping gait occurs in seven fossil horse trackways from three different sites. All of these trackways had long stride lengths (270.0–370.0 cm) and stride length/height ratios above 2.75, suggesting high speed and presented four independent hoof impressions (i.e. no obvious foot pairings). The trackways did sometimes display more symmetry than is typical for the gallop in modern horses where it averaged 0.63 (extrapolated from [25]). Here the symmetry [(DSD1/DSD2 + ISD1/ISD2)/2] varied between 0.88 to 0.90 at the Hoya de la Sima site (HSA-12 and HSA-14) to 1.00 (HSA-13), between 0.90 (OS-5) to 0.96 (OS-3 and OS-4) at the Colle di Osoppo site, and was 0.69 (WB-1) at the Wally's Beach site. Still, assuming proper individuation of footprints in the trackway, the long stride length and lack of obvious foot pairs strongly suggests a galloping gait.

Three galloping trackways (HSA-12; HSA-13; HSA-14) occur at the same Hoya de la Sima site near Jumilla, Spain where the trotting trackways are found ([31]; see **Figures 12A** and **13A**) and were likely made by the same tridactyl species of the horse *Cremohipparion matthewi* described above, which had an estimated height of 100.7 cm. This would yield stride length/height ratios of 2.75 (HSA-12), 2.75 (HSA-13), and 3.08 (HSA-14), indicative of left-, right-, and left-lead gallops, respectively, at speeds of around 5.2–5.6 m/s using the Original Alexander formula for such a fast gait.

The three galloping trackways located near each other on the Colle di Osoppo, near Osoppo, Friuli Venezia Giulia, Italy ([28]; see **Figures 12B** and **13B** above) were probably formed by *Hippotherium malpassii*, known from late Messinian deposits in Italy [66, 67]. *Hippotherium malpassii* had an incomplete skull (IGF 5286 V) of a juvenile specimen measuring ca. 22.5 cm, metacarpals (IGF 8192 V and 9397 V) that were 21.4 and 21.6 cm long, a metatarsal (IGF 8193 V) measuring 24.0 cm long, proximal phalanges III (IGF 9390 V and 9391 V; NHMB JH129, JH 158, Nonnumb1) that averaged 6.3 cm long (58.4–67.0), and left central hoof impressions 8.4 (OS-3), 8.3 (OS-4), and 7.0 (OS-5) cm long. Utilizing the *Hippotherium* multipliers (**Table A1**) yields heights of the trackmaker of 131.2 cm (from the metacarpals), 128.4 cm (from the metatarsals), and 128.0 cm (from the proximal phalanges III). The *Hippotherium* hoof multiplier (**Table A1**) results in estimated heights of 91.0 cm (OS-3), 89.9 (OS-4), and 75.8 (OS-5). Lacking a good skull-length measurement, applying the formula of 50% weight given to the hoof values and 50% weight to the postcrania, we arrive at height estimations of 110.1 cm (OS-3), 109.6 cm (OS-4), and 102.5 cm (OS-5) for the trackmaking *Hippotherium malpasii* individuals.

This gives very high values of stride length/height of 3.12, 3.30, and 3.35, respectively. The high ratio of stride length/height as well as the trackway consisting of four independent prints, plus, the fact that the distance between the prints varies, consisting of short, long, and intermediate steps, strongly suggest a galloping gait here, not the trot as suggested by Dalla Vecchia and Rustioni [28] who perhaps misinterpreted the capped trotting trackway shown in Renders [23, 24] as being isolated impressions. Two of the gallops possessed a left-lead (OS-3 and OS-4) and the other a right-lead (OS-5) as reconstructed by Dalla Vecchia and Rustioni [28] and occurred at speeds of 6.2–6.5 m/s using the Modified Alexander Formula.

The trackway WB-1 formed by the late Pleistocene monodactyl horse in fluvial sands and silts at the Wally's Beach deposit (DhPG-8) of St. Mary's Reservoir near Cardston in Alberta, Canada, has previously been referred to horse skeletal remains found at Wally's Beach assigned to the species *Equus conversidens* ([25, 30, 68]; see **Figure 12C** above). Further study, however, has caused us to question that specific identification. The skeletal morphology of the Wally's Beach *Equus* contains features now excluded from the definition of *E. conversidens* [69], namely, infundibulae on at least one of the lower incisors (typically I<sup>3</sup> ), and lower molars with a moderately-long ectoflexid that approaches the isthmus but does not penetrate it. The Wally's Beach horses, however, do have more of a deep U-shaped linguaflexid rather than a Vshaped or shallow U-shaped one, cheek teeth with a moderate anterior heel and moderately long protocones, and thick, short legs rather than long stilt-like ones. Such characteristics match closely with the specimens of *Equus lambei* found in Gold Run Creek in the Yukon Territory and in the Bluefish Caves near Edmonton, Ontario, Canada [70–75], as well as with populations formerly classified as *E. conversidens* from a site in the city of Canyon, Randall County, Texas, and a few other Texas locations [70, 76]. McNeil [68], after a careful comparison, placed the Wally's Beach individuals in *E. conversidans* rather than *E. lambei*, but this was based mainly on size ratio differences rather than discrete characteristics, apart from the smaller and less common canines in the Wally's Beach populations, more complex plications on upper cheek teeth of the Wally's Beach specimens, and more prominent and angular masseter ridges on the skulls of the Wally's Beach horses as compared to *E. lambei* specimens from the Bluefish Caves in the Yukon Territory. We, however, do not consider these differences profound enough to differentiate the two populations of horses on a specific level and indeed *E. lambei* found at Canyon, Randall County, Texas have somewhat complex plications in the upper cheek teeth [70, 76]. The Yukon *E. lambei* skull (USNM 8426) has a fairly prominent masseter ridge as well [75], and McNeil [68] admits that other Yukon specimens of *E. lambei* have well-defined masseter ridges approaching those from Wally's Beach. Hence we consider the Wally's Beach fossil horses to be members of the species *E. lambei* rather than *E. conversidens* (see also [77]). A third option would be to group the Wally's Beach horses with modern horses in *Equus ferus* [70], but we believe the shorter size of the Wally's Beach horses overall, along with their possession of short but moderately-stout metapodials, a broad skull with a quite convex ventral border of the mandible, and the greater presence of canines in certain populations (such as in the Bluefish Caves though not at Wally's Beach) preclude that assessment. Hence we assign the prints to *Equus lambei*, preferring to keep this taxon distinct from that of *Equus alaskae* which some authors combine it with [78].

In any case, skeletal remains are found in the same Wally's Beach strata as the footprints, and the phalanges match the prints closely. Such horses from Wally's Beach possessed a skull that averaged 41.3 cm in length, a metacarpal averaging 21.6 cm, a metatarsal averaging 26.1 cm, a hind proximal phalanx III (1PhIII) that averaged 8.1 cm, and a hind distal phalanx III (3PhIII) that averaged 5.7 cm in length, along with a humerus that averaged 28.4 cm in length and the radius 31.5 cm [68], and the central hoof impressions were 10.5 cm long. Using the horse height estimation multipliers for extinct *Equus* species (**Tables A1** and **A2**), and the skeletal material from Wally's Beach itself, would yield the following height estimations: 90.9 (from the skull), 117.1 (from the third metacarpal), 127.9 (from the third metatarsal), 129.7 (from the proximal phalanges), 133.5 from the distal phalanges, and 138.2 cm from the front hoof impression. Using our height estimation formula of assigning 50% weight to the prints, 25% to the skull, and 25% to the postcrania would yield an estimated height of 120.5 cm for the trackmaker of the species *Equus lambei*. This would give a large ratio of stride length/height of 2.96 and be indicative of a gallop, along with the occurrence of short, intermediate, and long step distances rather than more uniform ones. Such a gallop would have been a left-lead one that occurred at a speed of 6.6 m/s using the Modified Alexander Formula.

*New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

#### **4.4 Walking gaits in fossil horses**

So far very few walking equid gaits have been identified from fossil trackways, perhaps because the horse is less likely to leave a trail in the softer impact of the walk. Recently what does seem to be such a walking trackway [DR-1] was uncovered in regard the recently extinct Giant Cape Zebra (*Equus capensis*) from Driefontein, South Africa, 14 km east of Still Bay. The trackway was made in coastal sands around 109– 161 ka. The trackway consists of 12 tracks and is 320 cm long in toto. One cycle of understepping prints has a stride length of ca. 144 cm, an ipsilateral step distance [ISD] of 27.5 cm, a diagonal step distance [DSD] of ca. 33.3 cm, an average lateral offset of ca. 5.1 cm, and an interior straddle of ca. 2.0 cm. The average SDL though was 147 cm, the average ISD was 25.7 cm, and the average DSD was 23 cm. The front hoofprint measured ca. 13.74 cm in length and 12.44 cm in width and the hind footprint (noticeably more oval in shape) measured ca. 15.44 cm in length and 11.15 cm in width ([79], pp. 6–7 and Figure 5).

The height of Equus capensis can be estimated on the basis of comparisons with modern zebra skeletal dimensions (see **Tables A3** and **A4**). The hippotigrine multipliers are 2.64 for the cranium, 6.15 for the third metacarpal, 5.44 for the third metatarsal, 16.36/17.29 for anterior and posterior first phalanx, 22.95 for anterior third phalanx, and 17.32/17.11 for manus and pes hoof measurement, and 13.74 and 12.72 for manus and pes print measurement. Equus capensis has the following skeletal maximal lengths: cranium = 56.0 cm, metacarpal = 21.6 cm, metatarsal = 25.4 cm, phalanx I (manus/pes) = 8.4/8.1 cm, phalanx III manus = 7.7 cm, and the print of the fossil specimen was ca. 13.74 cm long for the manus and 15.44 cm long for the pes. These values yield the following height estimations in cm of *Equus capensis*: 147.8 (from cranium), 132.8 (from third metacarpal), 138.2 (from third metatarsal), 137.4/140.0 (from first phalanx), 176.7 (from third phalanx), 188.8/196.4 (from footprints), or yielding a final height estimate of 169.5 cm (from our formula). This is a bit larger than the estimation of 135.0–159.0 cm in Eisenmann [42], due to the very large nature of the footprints.

With this height estimation we have the following key trackway parameters of SL/ HT = 0.85; ISD/SL = 0.18; DSD/SL = 0.23; DSD/ISD = 1.30; LO/HHW = 0.46; IS/ HHW = 0.18. These parameters most closely agree with the gait of a walk (taking place at around 0.77 m/s via the original Alexander formula for slow gaits), though the negative interior straddle (and wavy print pattern) have some resemblance to a slow ambling gait such as a running walk (see **Figures 4** and **5E**). Might it be that when traveling over slippery terrain horses tend to shift their limbs interiorly and closer to or straddling the centerline for balance? If so some gaits that appear to be very slow racking ones would actually be trotting ones. More study of this issue is warranted. Alternatively, ancient equines might be capable of very slow alternative lateral gaits, perhaps on slippery substrates. Still trackway DR-1 seems to most closely match that of a walking gait.

### **5. Conclusion**

We find that horses displayed a great variety of gaits in the past, perhaps matching their current repertoire. The footprints studied above show they were capable of both intermediate-speed laterally coordinated gaits such as the running walk and rack or tölt, the diagonally coordinated trot which is the common current medium-speed gait in horses and zebras, and the transverse gallop. All these gaits are found in tridactyl Miocene horses (15–3.5 Ma) and provided a variety of options for locomotion. Laterally coordinated gaits allow surer footing in intermediate speed gaits when traveling over uneven or muddy terrain; for they provide a diagonal base of support (pace) or continual ground contact throughout the stride (rack or running walk). Following Sondaar [84], such lateral gaits would have been easier than diagonal gaits on the more flexible fetlock joints of tridactyl horses and so helped to prevent injury [22]. Intermediate speed diagonal gaits, such as the trot, are more efficient than lateral ones for longer periods of travel, and so would have been favored for migratory needs.

If these laterally coordinated gaits were common in fossil horses, exactly when horses lost the ability to perform laterally coordinated gaits and retained only the trot, as in zebras, is hard to determine. Janis and Bernor [47] argue that the evolution of monodactyl limbs in *Dinohippus*, *Pliohippus*, and *Equus* in the Miocene built upon a pre-existing suspensory apparatus system in the horse limbs (first seen in tridactyl members of the subfamily Equinae) that allowed elastic return of bouncing energy during the stride cycle, and that this in turn would favor development of the trot for more efficient locomotion over long distances. They speculate that as horses began to adapt to the more open plains they required longer dietary migrations, which in turn favored the development of monodactyly and a trotting gait. Tridactyl horses, even functionally monodactyl ones with reduced lateral legs, would have had mechanical reasons, according to Janis and Bernor [47], to utilize lateral gaits as these allowed for fast locomotion over shorter distances with less hyperextension and compression of the limb joints and for great stability on uneven terrain with a diagonal base of support. This may well be one reason horses and zebras began to lose the ability to engage in lateral gaits. We do find, however, trotting gaits present in tridactyl horses as well as lateral ones in this study. In any case the function of the reduced lateral digits in the Miocene and Pliocene is still in dispute as lateral hoof impressions occurred in some of the trackways studied above (HSA-9 to 14; OS-3 to 5; LAET-B) but not in others (LAET A; GQ-1), and in the LAET-B impression (n. 7) the side toe impressions occur in a more anterior location and more deeply, suggesting lots of flexing of the fetlock joint and use of the side toes for extra support, whereas the side toe impressions in trackways HSA-9 to 14 and OS-3 to 5 tend to occur at the rear of the central hoof impression and less deeply, suggesting light contact during extreme flexion or deep impressions. The lateral digits likely had more than one function such as aiding in limb stabilization, balance, and proprioception, preventing overcompression of the joints, and giving increased traction on uneven or slippery surfaces and during rapid turns [23, 24, 47, 85].

Moreover, the evolution of the passive stay apparatus is also complex as it occurs in the monodactyl *Dinohippus* (ca. 10.3 Ma) horse to the greatest degree but is also found in the earlier tridactyl horses *Acritohippus stylodontus* (ca. 15.3–14.9 Ma) and *Hypohippus equinus* (ca. 20.4–16.0 Ma), perhaps having arisen independently in the latter [86–89]. Hence we are still in the early stages of untangling the biomechanics and functional anatomy of fossil horse feet.

Genetic studies have concluded that laterally coordinated gaits in modern horses trace back to a gene mutation that occurred around the time of horse domestication prior to 10,000 years ago [21], or, for some breeds, to a genetic mutation arising in ninth century England [90]. Genetical studies suggest then that the modern ability to perform lateral gaits in select horse breeds is a parallel evolutionary innovation rather than a reversion to an ancestral condition. This would make sense, as after

#### *New Studies on the Gaits Displayed by Miocene, Pliocene, and Pleistocene Fossil Horse… DOI: http://dx.doi.org/10.5772/intechopen.112157*

domestication riders would have preferred smoother more comfortable gaits over less comfortable ones; hence laterally coordinated gaits would have been prized. Indeed fossil horse trackways show lateral gait parameters indicative of higher-speed racks even though they seem to have been performed at slower speeds. So the particular mechanisms and limb coordinations used in fossil horses may have been different than in modern horses, but this is not clear.

Also of interest is the social behavior evident in the Pliocene trackways. Four zones display multiple horses engaging in the same or similar gaits nearby each other, probably at around the same time. We see three series of prints (HSA-9, HSA-10, HSA-11) of the species *Cremohipparion matthewi* located next to each other and orientated in the same general direction displaying a trot at the Hoya de la Sima site near Jumilla, Murcia, Spain around 8.7–7.8 Ma [31, 33]. Three other print series (HSA-12, HSA-13, HSA-14) located near each other and orientated in the same general direction at the same Hoya de la Sima all show horses in a gallop, as do two or three similar trackways (OS-3, OS-4, and OS-5) of *Hippotherium malpassii* found at the Colle di Osoppo site, near Osoppo, Italy from ca. 6.0–5.3 Ma [28]. Finally, three horse trackways of the species *Eurygnathohippus hasumense* orientated in the same direction and in similar medium-speed gaits [two in a rack or tölt (LAET-A and LAET-C) and another in a running walk (LAET-B)] located near each other occur at Laetoli site 8-G in Tanzania were laid down around 3.7 Ma. This shows a common occurrence found in wild horses and zebras today of herding behavior displayed during locomotion. Hence it appears that horses have long been a social species and move together in a herd. We also see evidence of differing gaits within the herd (assuming the trackways were made close in time to each other). At the Hoya de la Sima site in Jumilla, Spain, there are not just three sets of galloping prints side by side proceeding in the same direction but there are also three trackways of horses engaged in a trot (HSA-9, HSA-10, HSA-11) located side by side and going in the same direction but crossing over the galloping tracks at a perpendicular angle (HSA-12, HSA-13, HSA-14). This could be due to some of the younger horses playing while the older horses were traveling to a certain location, as happens in modern herds (Renders, pers. observations). Two of the trotting trackways (HSA-9-10) had slightly smaller hind footprints averaging 13.7 cm in length and 10.9 cm in width, consistent with juvenile trackways, as compared to the galloping trackways (HSA 12-14) which averaged 14.6 cm in length and 13.4 cm in width [31, 33]. The hind footprint of trotting trackway HSA-11, however, was fairly large at 15.1 cm long by 12.3 cm wide so might have been that of an adult. In the Laetoli *Eurygnathohippus* site we see two horses, an adult (LAET-A) and a juvenile (LAET-C), with trackways displaying a rack or tölt, while the trackway of what is likely a mare (LAET-B), in that it slows down and crosses over the juvenile trackway, displays a running walk or intermediate-speed rack. Here as noted earlier the mare may have used a slightly different laterally coordinated gait to match the speed of her foal or avoid a collision.

There are a few other fossil horse sites that constitute trample grounds with multiple individuals leaving footprints. It is possible that further fieldwork may be able to isolate individual trackways for which gait can be determined [91–95]. Additional study would also be beneficial in regard to unshod horse hoofs of modern horses and the impressions they make in various substrates such as sand, mud, ash, and snow. Regarding fossil horses, more work correlating front and hind central coffin bone measurements with horse height would be helpful, especially for Eocene and Oligocene horses.
