1. Introduction
Their set of physiological characteristics and social nature has made donkeys very useful to people in many cultures throughout the ages [
1]. These animals nowadays are not an integral part of human life in developed countries; however, currently, a resurgence of interest them can be seen, especially in the dairy industry [
2] and in onotherapy [
3,
4]. As the breeding of donkeys has increased due to different types of use [
5], the donkey’s welfare evaluation has become a challenge.
In the welfare evaluation of lactating donkeys, a functional approach to the body condition assessment was introduced [
5] as the most practical method to describe the body condition of a donkey. Hitherto, in donkeys, a five-point scale of the body condition score (BCS) was used [
6]. Nowadays, the fatty neck score (FNS) enables a complex evaluation of donkeys’ body condition [
5]. It has been concluded that FNS, BCS, and the dental condition score (DCS) are necessary to be evaluated simultaneously as an indicator of the donkey’s welfare [
5]. As body condition can be considered a key criterion of the overall welfare of the animals [
5,
6], it became one of the most frequently used morphometric measurements. On the other hand, it has been shown in the study on horses that other morphometric measurements, such as geometric morphometrics (GM), could reflect the posture and hence welfare state of horses [
7].
When animal-based morphometric indicators are used to describe an equid’s welfare state, many internal and external features should also be considered. In horses, there are substantial differences in morphometrics between breeds [
8], body conditions [
9], and mental states [
7]. Moreover, the association between the basic animal-based morphometric measurements and the body condition has been evaluated in different studies on ponies, horses [
9,
10,
11], and donkeys [
5,
12,
13]. The association between the animal-based morphometric measurements subjected here has not been evaluated yet, as the knowledge concerning donkeys is still scant. Therefore, we hypothesized that the BCS, FNS, and DCS measurements can impact the GM measurement system used in a previous study on this equid’s posture [
14]. Both measurements of the body condition and their potential influence on the donkey’s posture should be considered to understand the complexity of this concept. As the body condition scoring for donkeys [
5,
6] requires a different technique to that used in horses [
9,
10,
11,
15], due to the fat storage in donkeys in more localized areas and a different body shape [
6,
16], similar species-dependent differences may influence donkey posture.
In the case of posture measurement in horses, the repetition of stressful situations over time may lead to chronic states [
17]. Seneque et al. [
7] suggested this repetition may lead to the repetition of the same associated postures, which may also become chronic. Authors have been supporting Broom’s theory, in that the effect of the “mental” state of the animal, also reflecting chronic stress, may compound existing health and physical constraints [
18]. Seneque et al. showed that the dorsal profile of a horse’s back, such as when it is flat or hollow, is related to a compromised welfare state. Horses with an impaired welfare state and a depression-like syndrome can be differentiated through their overall posture [
7]. As the postures and the shape of the dorsal profile of the animals were described as an effective indicator of horses’ [
7,
14] and pigs’ [
19] welfare state, interest in the use of GM in animal breeding has increased. GM is a collection of approaches that provide a mathematical description of the biological forms according to geometric definitions of their size and shape [
20]. The GM method uses Landmarks (LD), which are Cartesian coordinates of points in 2D or 3D that can be localized precisely and without ambiguity on a structure and from one specimen to another. For instance, the anatomical points on a body surface reflecting easy-to-find palpation skeleton structures are suitable landmarks in horses [
7,
14]. Other data points named sliding semilandmarks (SSL) require specific mathematical treatment and are free to slide to capture the geometry of curves and surfaces where landmarks cannot be identified, such as on smooth objects [
21]. LD and SSL configurations have been acquired on sets of digital images, and then a Generalized Procrustes Analysis (GPA) is applied to minimize the sum of squared distances between the corresponding LS to extract the shape data by removing the extraneous information of size, location, and orientation [
20]. In the general patterns of morphological variation in multidimensional data, conventional Euclidean statistical methods are viable, such as Principal Component Analysis (PCA) [
22].
This study aimed to characterize the dorsal profile of donkeys taking into account the typical donkey body condition conformation. To achieve this goal, the relationship between the body condition score, fatty neck score, dental condition score, sex, breed, and the donkey’s general dorsal profile was evaluated using the GM method. Furthermore, some software modifications were introduced to make the GM method more accessible to a growing community of users in various areas of equid breeding and welfare.
3. Results
The entire dorsal profile of the donkeys was first investigated to try to identify variations in postures associated with the previously reported body condition indicators. The characteristic postures for sex, breed, BCS, FNS, and DCS classes and the direction of changes in the hindquarter, back, and neck and head regions were then examined.
Procrustes coordinates for all individuals are presented in
Figure 3. PCs represent the weight of the partial wraps in the whole warps between all the conformations. The extremum of PC1 supported the dorsal profile with an elevated dorsal line in the hindquarter region, lowered dorsal line of wither in the back region, and elevated occiput in the neck and head region in relation to the consensus dorsal profile. The extremum of PC2 supported the dorsal profile with an elevated dorsal line in the hindquarter region, lowered dorsal line in the back region, and elevated dorsal line in the neck and head region in relation to the consensus dorsal profile. The extremum of PC3 supported the dorsal profile with an elevated and shortened dorsal line in the hindquarter region, lowered dorsal line in the back region, as well as stretched forward and elevated occiput in the neck and head region in relation to the consensus dorsal profile (
Figure 4A). The variance of the first three PCs was as follows: PC1 = 37.41%; PC2 = 23.43%; and PC3 = 13.34% (
Figure 4B).
Five classifiers were used to distinguish the categories of individuals, including sex (21 jennets and 19 males), breed (8 half-breed Andalusian donkeys, 8 half-breed Grigio Siciliano donkeys, 8 half-breed Martina Franca donkeys, 8 half-breed Magyar Parlagi donkeys, and 8 half-breed Romanian donkeys), BCS (12 donkeys with BCS 2, 13 with BCS 3, and 15 with BCS 4, no donkey was rated poor (BCS 1) or obese (BCS 5)), FNS (5 donkeys with FNS 0, 5 donkeys with FNS 1, 6 donkeys with FNS 2, 7 donkeys with FNS 3, 9 donkeys with FNS 4, and 8 donkeys with FNS 5), and DCS (14 donkeys with DCS 0, 15 donkeys with DCS 1, and 11 donkeys with DCS 2). On scatter plots of principal component scores in the PC1 to PC2 orientation, it is easy to see that the scores are not divided into separate categories of sex (
Figure 5A), breed (
Figure 5B), and DCS (
Figure 5E). On the other hand, the scores are partially divided into separate categories of BCS (
Figure 5C) and FNS (
Figure 5D). More donkeys with BCS 2 and FNS 0–1 represented the dorsal profile supported by PC1, whereas more donkeys with BCS 4 and FNS 4 and 5 represented the dorsal profile supported by PC2.
The Procrustes ANOVA was applied to assess the effect of the examined classifiers on both the centroid size and shape of the donkey’s dorsal profile. The donkey’s dorsal profile showed significant differences for size depending only on the BCS (
p = 0.024) and FNS (
p = 0.012), whereas for shape, depending on all classifiers: sex (
p = 0.0264), breed (
p < 0.0001), BCS (
p < 0.0001), FNS (
p < 0.0001), and DCS (
p < 0.0001) (
Table 1).
To better visualize the differences in dorsal profiles between the examined categories of individuals, average observations for consecutive classifiers were executed. These visualizations confirm the results obtained for principal component scores, where the features of the division into separate categories were higher for BSC and FNS (
Figure 6 and
Figure 7) than for sex, breed, and DCS (
Figure 8,
Figure 9 and
Figure 10). It is easy to see that for BCS 2, FNS 0, and FNS 1, a dorsal profile with an elevated dorsal line of the wither in the back region and lowered dorsal line in the neck and head region was observed. On the other hand, for BCS 4, FNS 4, and FNS 5, a dorsal profile with an elevated dorsal line in the neck and head region was noted. No other pronounced deformations for BCS 3, FNS 2, and FNS 3, nor for the sex groups, breed groups, and DCS scores were observed.
As a summary of the recent results, the distances between the donkeys’ dorsal profiles were compared among the BCS (
Table 2), FNS (
Table 3), sex (
Table 4), breed (
Table 5), and DCS (
Table 6) categories. The highest distances among the categories (Mahalanobis distances ≥ 13.26 and Procrustes distances ≥ 0.044) were noted for FNS between FNS 1 and FNS 5, FNS 0 and FNS 5, as well as FNS 1 and FNS 4. The lowest distances among the categories (Mahalanobis distances ≤ 5.24 and Procrustes distances ≤ 0.018) were noted for sex between jennets and males, for BCS between BCS 1 and BCS 2, as well as for breed between half-breed Martina Franca donkeys and half-breed Romanian donkeys.
4. Discussion
An analysis of the dorsal profiles of the examined donkeys displayed a deformation associated with the body condition indicators, this being higher in the functional one (FNS; effect on size
p = 0.012; effect on shape
p < 0.0001) [
5] rather than the most frequently used one (BCS; effect on size
p = 0.024; effect on shape
p < 0.0001) [
6,
9,
10,
11,
12]. In donkeys, BCS is proposed as an index of the overall adiposity [
6,
12], whereas FNS is a morphometric index of regional fat deposition [
5]. Since the adipose tissue of donkeys tends to cumulate in the neck region and droop on both sides of the crest of the neck [
6], it is not surprising that FNS had a stronger influence on the dorsal profile than BCS. Since in this study no donkey was rated poor (BCS 1) or obese (BCS 5), the effect of BCS may be stronger when the full range of the scale will be used. However, in donkeys, this regional adiposity could play a different role than the indicator of the metabolic status, which is used in horses and ponies [
9,
28], and it can remain even when the overall body weight decreases [
6,
16]. Therefore, the results of the present study supported Valle et al.’s findings that FNS is important for the description of the body conformation of donkeys [
5].
It should be kept in mind that donkeys are not small horses, although they both belong to the Equidae family [
16]. The shape of the neck and back of a donkey is different from that of a horse. The donkey’s general vertebral formula is C7, T18, L5, S5, Cd15–17 [
29], whereas in the horse it is C7, T18, L6, S5, Cd15–21 [
30]. The shorter lumbar vertebrae and relatively shorter neck in the donkey than in the horse support a heavy skull [
16]. Therefore, in donkeys, a remarkably thick cutaneus colli muscle covers the middle one-third of the length of the neck and is evenly developed [
31]. Moreover, donkeys are at greater risk of obesity compared to horses [
6]. Therefore, the enlarged and thickened neck with the longitudinal fat deposits in the crest located on both sides of the neck [
5] may deform the dorsal profile similar to a “depressed” or “abnormal” posture in horses [
7]. In a recent study on horses, a flat or hollow dorsal profile was related to a compromised welfare state [
7], whereas in this study a similar posture was associated with a low body condition. As body condition can be considered a key criterion of the overall welfare of the animals [
5], application of a posture analysis as an additional indicator of equids’ welfare [
7] should in donkeys show the impact of regional fat storage on the neck region, which is reported here.
In the present study, no significant evidence of an association between deformations in the size of the dorsal profile and sex (
p = 0.100), breed (
p = 0.614), or DCS (
p = 0.554) was noted. On the other hand, evidence of an effect of sex (
p = 0.026), breed (
p < 0.0001), and DCS (
p < 0.0001) on the shape of the dorsal profile were observed. These findings are partially in contradiction and partially in agreement to the Valle et al. study [
5], where dental disorders were considered an indicator of the body condition for lactating donkeys. Other recent studies have demonstrated the association between DCS and BCS or weight loss [
32,
33]. However, based on the results of the present study, no significant association can be established between DCS and deformation in the size of the dorsal profile (
p = 0.554), whereas a significant association can be established between DCS and deformation of the shape of the dorsal profile (
p < 0.0001). Results of the present study are also partially in contradiction and partially in agreement to Mendoza et al.’s study [
34], where jennets had significantly greater body measurements, including BCS, than males, and showed a tendency to have higher triglyceride concentrations. Although the relationship between DCS and BCS as well as between sex and BCS was not determined in this study, the observed lack of deformation in the size of the dorsal profile (
p = 0.100) may be due to the limited size of the groups, which should be considered as one of the limitations of this study. The second limitation of this study is the use of non-purebred donkeys, but a crossbreed of breeds. Therefore, the intraspecific variability in anatomy and physical conformations, which is typical for donkeys [
16,
35], did not appear in this study as a classifier for deforming the dorsal profile. In this study, it is important that the evaluation of the group of donkeys was kept under the same conditions—in the same stable and managed in the same way. Therefore, it was decided to use all available donkeys, which were not representing pure breeds. However, one should not forget that between the donkey breeds, huge interbreed morphotype variability occurred [
35,
36]. Andalusian donkeys are disease and heat resistant and full of energy, with a calm and balanced temperament. These are large-sized donkeys, on average 146–155 cm in height at withers [
35]. They have an impressive head, with roman noses, very large ears, and a muscular neck [
36]. Martina Franca donkeys are relatively large-sized donkeys, on average 127–135 cm in height at withers. They have a large head with a well-developed, strong, and muscular neck, and a large, long, and muscular croup [
35]. Magyar Parlagi donkeys is a Hungarian Steppe Donkey that come in two varieties; a small size, on average 110–115 cm in height at withers, and a large size, on average 136–145 cm in height at withers [
36]. Romanian donkeys are characterized by a large variation in body length with no standardized breed height. Yılmaz et al. [
36] suggested the reason of such variability may be due to the change in confirmation from youth to maturity. In turn, the height of the Grigio Siciliano donkeys is undescribed [
36].
For large population studies and the increasing availability of research methods, some software modifications were introduced here. The labels were digitized following the SSL method described by Seneque et al. [
14]; however, MorphoJ software was used for all analyses [
27], rather than other reported morphometric-based software, such as R Core Team (2018) packages (‘geomorph,’ ‘shapes,’ ‘Morpho,’ and ‘Momocs’) [
20]. MorphoJ software was used as probably the easiest standalone software for the GM method. The graphical user interface is simple and clear, and one can quickly run several analyses and generate fully customized graphs that can be exported as images or vectorized figures [
20]. The only insufficient aspect is a display of one extremum of the PCs rather than two and the inability to cancel the movement of the neck. In recent studies, the movement of the neck, corresponding to a rotation around the withers, was canceled thanks to the R library geomorph [
7,
14]. In this study, the movement of the neck corresponding to a rotation in the articulatio atlantooccipitalis was not canceled; however, no significant deformation of the dorsal profile was found in this region. Since all body condition-related deformations were described for the neck regions, the potential rotation in the atlas area should not affect future research; therefore, the correction of head position, which is recommended in scientific studies, may be denned into this more practical one. This inconvenience is disproportionately small compared to the benefits of facilitating statistical analyses in MorphoJ software. We hope this modification will increase the availability of the GM method to a growing community of users in various areas of management of breeding and welfare evaluation of donkeys.
In summary, it can be stated that the donkey’s dorsal profile allows distinguishing categories of individuals, used so far in the welfare evaluation of lactating donkeys [
5], and thus could be a promising indicator of welfare states, similarly to horses [
7]. Nevertheless, further studies are needed to investigate whether there is a link between the deformation of the dorsal profile and the poor welfare state of donkeys.