On the Behavioural Biology of the Mainland Serow: A Comparative Study

Simple Summary Serows Capricornis spp. are solitary, elusive, forest-dwelling goat-antelopes, allegedly the closest ancestral forms to wild sheep and goats (Caprinae). Their behaviour and ecology have been largely overlooked so far, although they could be useful to understand the roots of early ritualisation of weapons, i.e., horns. The activity rhythms, marking behaviour, and social interactions of captive mainland serows have been described and quantified. Activity peaked in mid-afternoon and late night, whereas resting and ruminating were the highest at noon and twilight. The two sexes used different marking sites and marking frequencies. A total of 33 social behaviour patterns were observed: 18 patterns concerned agonistic behaviour, whereas 15 patterns were relevant to courtship behaviour. An evolutionary comparison across Caprinae species with unritualised piercing horns, inclusive of serows, suggests that inter-sexual direct forms of aggressive behaviour are used significantly more often than indirect ones, except for chamois. Thus, Chamois Rupicapra spp. would be confirmed as the most advanced genus in terms of an early ritualisation of weapons, i.e., strongly hooked horns. Conversely, horns of the goral Nemorhaedus spp. and the serow lie on the same plane of the frontal bones, thus making possible the usage of a dominance display through frontal pushing. Abstract Comparative behavioural studies help reconstruct the phylogeny of closely related species. In that respect, the serows Capricornis spp. occupy an important position as they have been assumed to be the closest forms to the ancestors of Caprinae. In spite of that, information on the behavioural repertoire of the mainland serow Capricornis sumatraensis is exceedingly poor. In this paper, we report data on the activity rhythms and social behaviour of rutting mainland serows in captivity (Central Thailand, January 1986; January–February 1987). Activity was bimodal with peaks in mid-afternoon and late night. Resting and ruminating peaked at noon and twilight. Four patterns of marking behaviour were observed out of a total of 1900 events. Males and females were found to use different marking sites and frequencies. A total of 33 social behaviour patterns were observed: 18 patterns concerned agonistic behaviour, whereas 15 patterns were relevant to courtship behaviour. A comparison across Caprinae species with unritualised piercing weapons (i.e., Capricornis, Naemorhedus, Rupicapra, Budorcas, and Hemitragus) has shown that inter-sexual direct forms of aggressive behaviour are used significantly more often than indirect ones, but for chamois, confirming Rupicapra spp. as the most advanced genus among them in terms of an early ritualisation of weapons. Conversely, horns of the goral Nemorhaedus spp. and the serow lie on the same plane of the frontal bones, thus making possible the usage of a dominance display through frontal pushing.

The climate of Thailand is tropical and strongly influenced by the monsoons. Annual rainfall is about 1400 mm, peaking in May-September [32]. The average maximum annual temperature is 36 °C in April; the average minimum is 16 °C in January. In Thailand, free-living mainland serows inhabit steep limestone mountains and cliffs thickly covered with forest that is inaccessible to man and with rocky hollows used by serows as shelter sites [22].

Data Collection
Serows were watched during the day (06:30 a.m.-01:30 p.m.; 02:00 p.m.-06:30 p.m.) and, twice per week round-the-clock (including from 06:30 p.m. to 06:30 a.m.), at a distance of ca. 15−40 m, through binoculars (©Leitz, Wetzlar, Germany, 10 × 40 field glasses, during the day; ©Zeiss, Jena, Germany, 8 × 56 BT*P* during the night) whenever details should be recorded. Observations were conducted by two operators at the same time, with different tasks (see below), for a total of 166 h in January 1986 (Dusit Zoo), and for 98 h in January-February 1987 (Khao Kheow Wildlife Sanctuary), during the rutting season. One observer concentrated on all marking activities, reporting onto checksheets the day, time, mark type, the exact location within the enclosure where marking occurred, and the sex of the marking individual; the other observer recorded all behavioural observations, except for marking. Behaviour patterns and activity rhythms were recorded onto paper check-sheets.

Data Analysis
Patterns of activity rhythms and relevant 95% confidence intervals were estimated through the software R 3.6.1 (version 3.5.1., R Foundation for Statistical Computing, Wien, Austria), package "overlap" [33]. The climate of Thailand is tropical and strongly influenced by the monsoons. Annual rainfall is about 1400 mm, peaking in May-September [32]. The average maximum annual temperature is 36 • C in April; the average minimum is 16 • C in January. In Thailand, free-living mainland serows inhabit steep limestone mountains and cliffs thickly covered with forest that is inaccessible to man and with rocky hollows used by serows as shelter sites [22].

Data Collection
Serows were watched during the day (06:30 a.m.-01:30 p.m.; 02:00 p.m.-06:30 p.m.) and, twice per week round-the-clock (including from 06:30 p.m. to 06:30 a.m.), at a distance of ca. 15-40 m, through binoculars (©Leitz, Wetzlar, Germany, 10 × 40 field glasses, during the day; ©Zeiss, Jena, Germany, 8 × 56 BT*P* during the night) whenever details should be recorded. Observations were conducted by two operators at the same time, with different tasks (see below), for a total of 166 h in January 1986 (Dusit Zoo), and for 98 h in January-February 1987 (Khao Kheow Wildlife Sanctuary), during the rutting season. One observer concentrated on all marking activities, reporting onto check-sheets the day, time, mark type, the exact location within the enclosure where marking occurred, and the sex of the marking individual; the other observer recorded all behavioural observations, except for marking. Behaviour patterns and activity rhythms were recorded onto paper check-sheets.

Data Analysis
Patterns of activity rhythms and relevant 95% confidence intervals were estimated through the software R 3.6.1 (version 3.5.1., R Foundation for Statistical Computing, Wien, Austria), package "overlap" [33].
The occurrence of each behavioural type was considered as a percentage of occurrence (frequency). The intersexual difference in the frequency of marking activities, which may support territorial behaviour [6], was tested through a Fisher's exact test. The one-sample Kolmogorov-Smirnov (K-S) test was used to assess if each serow marked sites with the same frequency. Two-sample K-S tests were used to evaluate whether males and females of each pair made use of the same marking sites. Both tests were performed through the software IBM SPSS Statistics 21 [34].
For each study site, the behavioural repertoires of males and females were compared within and between sexes through the Fisher's exact test. If the intrasexual difference between two behavioural repertoires was not statistically significant, we pooled data together to obtain one behavioural repertoire per sex. We used a Z-test to assess differences between the proportions of direct and indirect forms of aggression by serow and other closely related species observed in the wild: the Himalayan goral Naemorhedus goral [19], the Northern chamois Rupicapra rupicapra (Lovari and Albicocco, unpublished data), the Southern chamois Rupicapra pyrenaica [35], the Himalayan tahr [36], and the takin (Lovari and Dahal, unpublished data). All studies included in our comparison were performed by one of us (SL) to reduce observer's bias. For each species, intrasexual differences were compared through Z-tests, whereas differences between sexes belonging to the same species were tested through the Fisher's exact test. The Z-test, the K-S test, and Fisher's exact test are suitable to small sample sizes [37,38].

Activity Rhythms and Marking Behaviour
The activity pattern was bimodal with two peaks, one in the mid-afternoon and the other in the second half of the night, whereas resting and ruminating peaked at noon and twilight ( Figure 2A). The occurrence of each behavioural type was considered as a percentage of occurrence (frequency). The intersexual difference in the frequency of marking activities, which may support territorial behaviour [6], was tested through a Fisher's exact test. The one-sample Kolmogorov-Smirnov (K-S) test was used to assess if each serow marked sites with the same frequency. Twosample K-S tests were used to evaluate whether males and females of each pair made use of the same marking sites. Both tests were performed through the software IBM SPSS Statistics 21 [34].
For each study site, the behavioural repertoires of males and females were compared within and between sexes through the Fisher's exact test. If the intrasexual difference between two behavioural repertoires was not statistically significant, we pooled data together to obtain one behavioural repertoire per sex. We used a Z-test to assess differences between the proportions of direct and indirect forms of aggression by serow and other closely related species observed in the wild: the Himalayan goral Naemorhedus goral [19], the Northern chamois Rupicapra rupicapra (Lovari and Albicocco, unpublished data), the Southern chamois Rupicapra pyrenaica [35], the Himalayan tahr [36], and the takin (Lovari and Dahal, unpublished data). All studies included in our comparison were performed by one of us (SL) to reduce observer's bias. For each species, intrasexual differences were compared through Z-tests, whereas differences between sexes belonging to the same species were tested through the Fisher's exact test. The Z-test, the K-S test, and Fisher's exact test are suitable to small sample sizes [37,38].

Activity Rhythms and Marking Behaviour
The activity pattern was bimodal with two peaks, one in the mid-afternoon and the other in the second half of the night, whereas resting and ruminating peaked at noon and twilight ( Figure 2A). Four types of marking behaviour were observed for a total of 1900 events (preorbital gland, 70% and 90%; forehead/nape, 15% and 5%; horns, 10% and 4%; flank, 5% and 1%, respectively in Dusit Zoo and Khao Kheow) performed by both the male and the female of each pair. In a typical sequence of scent marking, serows sniffed a spot, lay scent there, and alternated marking with licking movements ( Figure 3A,B). Four types of marking behaviour were observed for a total of 1900 events (preorbital gland, 70% and 90%; forehead/nape, 15% and 5%; horns, 10% and 4%; flank, 5% and 1%, respectively in Dusit Zoo and Khao Kheow) performed by both the male and the female of each pair. In a typical As to the pair in Dusit Zoo (hereafter, pair 1, composed by male M1 and female F1), the male was responsible for 80.5% of total marking events (N1 = 1714) (Fisher's Exact Test = 638.34, dof = 1, p < 0.0001). In Khao Kheow (hereafter, pair 2, composed by male M2 and female F2), the male accounted for 58.0% of the events (N2 = 186) (Fisher's Exact Test = 1.742, dof = 1, p = 0.187).
The marking activity was mainly carried out through preorbital glands by both sexes ( Table 1). The intensity of marking activity sharply declined in both sexes in the post-oestrus days of the females ( Figure S1). In pair 1, the male and the female marked 42 and 22 sites respectively, whereas, in pair 2, the male and the female marked respectively 17 and 27 sites. In each pair, males and females were found to use different sites and marking frequencies (K-S tests: 2.26−14.71, p < 0.0001).
The marking activity was mainly carried out through preorbital glands by both sexes ( Table 1). The intensity of marking activity sharply declined in both sexes in the post-oestrus days of the females ( Figure S1). In pair 1, the male and the female marked 42 and 22 sites respectively, whereas, in pair 2, the male and the female marked respectively 17 and 27 sites. In each pair, males and females were found to use different sites and marking frequencies (K-S tests: 2.26-14.71, p < 0.0001).

Behaviour Pattern Description
Approaching (M, F) Threat. The sender walks straight to the receiver. The mane may be erected (D).
Body/Head Shaking (F) Dominance display. The sender vigorously shakes the head several times (I).
Chasing Threat ( Figure 3C). An individual holds its head lower than main axis of body, pointing horns to another one. The mane is erected. It may be performed laterally or frontally (D).
Hooking (M, F) Threat. One attempts to gore another with a horn sweep. It may be frontal (to shoulder/neck of opponent) or side-on (to rump/abdomen of opponent) (D).
Horning Vegetation (M, F) Dominance display. An individual horns a bush vigorously and/or lay scent on it (I).
Licking Body (M, F) Socially explorative behaviour [9]. An individual licks the other's body. Licked areas are generally muzzle, flank, and genitals. Staring (M, F) Threat ( Figure 3G). An individual stares the other one, with mane erected (D).    Table 4. Behavioural repertoire of the mainland serow; male and female behaviour patterns in courtship.

Behaviour Pattern Description
Courtship Foot Stamping (M) Male stamps the ground with one of his fore-hooves, sometimes quickly alternating with both of them, behind the female. Mane can be erected ( Figure 3D).

Croup Touch (M)
Male rests his chin, throat, and lower part of the neck on the female's croup. Often preceded or followed by Kicking.

Twisting (M)
The male lowers his neck into a position that looks like the Low Stretching and then rotates his head up to 90 • so that horns face away from the female ( Figure 3F).
Direct and indirect forms of aggression, as well as socially explorative behaviour built up 24% and 9% of the behavioural repertoire of M1 and M2 respectively, whereas courtship patterns were predominant (M1, 76%; M2, 91%).
The most frequent behaviours were Kicking, Head Up and Mount for males, and Head Down, Licking Genitals/Urine, and Stare for females ( Table 3). As to the intersexual behaviour, the most frequent behaviour was Head Down for both males (53%) and females (46%), followed by Approaching for males (23%) and Staring for females (20%). During courtship, Kicking was performed most frequently by males (M1, 47%; M2, 54%), followed by Head Up for M1 (16%) and Licking Genitals/Urine for M2 (17%).

Direct and Indirect Forms
Males and females of Caprinae species with piercing weapons were considered in Figure 4. Only chamois performed direct forms of aggressive behaviour significantly more than indirect forms ( Figure 4; Table 5). There was no significant difference between sexes (Fisher's exact test, mainland serow: p = 0.498; Himalayan goral: p = 0.71) (Figure 4).

Behaviour Pattern
Mainland Serow Japanese Serow

Himalayan Goral
Northern Chamois

Southern Chamois Takin Himalayan Tahr
Approaching

Discussion
Our data show two bimodal peaks of activity, a nocturnal one at ca. 03:00 a.m. and a diurnal peak at ca. 04:00 p.m, in contrast to what has been believed so far of the serow as a mostly nocturnal mammal [21,22,40,41]. In fact, diurnal activity of mainland serows was also recorded by [23] in the wild ( Figure 2B). The activity patterns we recorded in captive conditions are consistent with those recorded by Chen et al. [23], which suggests that confinement in captivity and artificial availability of food resources have not altered their basic activity rhythms. One could object that mainland serows seem to show a few adaptations to nocturnal life, e.g., blackish to black coat colour, acute senses of smell and hearing [22]. However, all these features could also have evolved as adaptations to life in dark habitats, such as dense forests [29].
Large nostrils and particularly big preorbital scent glands, as well as the usage of latrines as territorial landmarks, suggest that this species relies heavily on olfactory signals [22,25,31,42], which would be useful for communication in conditions where visual ones may not be perceived.
Enclosure size may have elicited a different frequency of marking and aggressive activities [43,44], but we detected no other qualitative difference between study sites. In spite of our limited sample size and of captive conditions, a few preliminary conclusions may be drawn. Although intense marking was more evident in the male sex, females showed it too, confirming the prediction of territorial marking in both sexes [8,9]. Most marking occurred using the secretion of preorbital glands, as it has been reported for the Formosan serow C. swinhoei [42]. Serows are sexually monomorphic, also as to horn size and shape: thus, males should be cautious when approaching females, especially in the reproductive period. In fact, females who are not yet fully receptive may be quite aggressive to males, to the point of wounding them seriously (see [45], for the closely related mountain goat Oreamnos americanus). Amongst animals, direct attacks are usually avoided to limit the risk of being physically injured [46][47][48]. The dagger-like, sharply-pointed and potentially lethal horns of serows may have promoted the development of alternative tactics, e.g., marking behaviour and dominance displays, with respect to actual fights. The spatial pattern and the usage of different marking sites for each sex suggest the existence of individual preferred locations, as well as the absence of shared marking sites [17,42].
Intersexual tolerance increased with the oestrus development of the females when the intensity of their marking activity sharply declined ( Figure S1). Kicking was the most used courtship behaviour pattern by males, as it is in the other serow and goral species [17,19,42]. Females may react to Kicking with urination, which provides males with olfactory cues of the female oestrus status [11,49]. The simulation of infant behaviour (i.e., Low-Stretching) is the most used courtship pattern by male serows approaching oestrus females, as in other Caprinae species [9,11,33,50]. The presence of a single male serow in each enclosure determined no competition to female access, thus ruling out the opportunity to record intra-sexual male agonistic behaviour.
The Rupicaprini tribe includes serows, gorals, the mountain goat, and chamois, which are all nearly monomorphic species, but for temporary (chamois) or permanent (mountain goat) body mass dimorphism. The takin (Tribe Budorcatini) and the Himalayan tahr (Tribe Caprini) show some Rupicaprini traits, i.e., they are little sexually dimorphic as to the shape and size of horns.
Among Rupicaprini, the Twist is present-although rare-only in the mainland serow, whereas it is often associated to Low Stretching and Kicking in the bharal Pseudois nayaur [51], as well as in Capra and Ovis spp. [11]. In this Tribe, Frontal Pushing has been reported only for serows, although it has been recorded often in other species of the Caprini Tribe, e.g., the Himalayan tahr [36], the takin [52], and the Mediterranean mouflon Ovis aries [53]. Therefore, both the Twist and Frontal Pushing-rare or absent in Rupicaprini-become much more abundant in the phylogenetically most advanced Caprinae ( Figure S2). Aggressive behaviour builds up over 80% of the repertoire of the mainland serow, and direct forms of aggression are significantly more used than indirect ones. Serows are relatively close to the ancestors of Caprinae species [20]. Aggressive behaviour may be expected to be less differentiated in these phylogenetically ancestral forms of resource defenders, which have not yet developed ritualised weapons, e.g., those of Pseudois, Capra, and Ovis spp. [12]. Beside mainland serows, also gorals show mostly direct aggressive displays [19], whereas chamois, i.e., the most recent genus among Rupicaprini [20,54], may have developed a repertoire privileging indirect forms of aggressiveness. Among Rupicaprini, intrasexual aggressiveness by females is comparable to that shown by males, with some differences in phenology (Southern chamois [55] and mountain goat [56]). Body size and horn size are not different between male and female gorals and serows. The frequent use of intersexual direct forms has also been found in male Himalayan tahrs and takins [30,36], which have not evolved ritualised weapons as Ovis and Capra spp. Lovari (1985) suggested that there could be a direct relationship between well-developed social behaviour and the complexity of behavioural repertoire [35]. This suggestion has not been confirmed through our study. The genus Capricornis was the first to diverge within the Rupicaprini clade, thus qualifying as the most ancestral taxon [20,54]. We recorded 29 behaviour patterns for the mainland serow, as many as those observed in both chamois species, much less solitary ungulates than serows, but more than in the Japanese serow and in the Himalayan goral [17,19].
Among Rupicaprini, Frontal Pushing has been recorded only in serows, whereas it is used rarely by the Himalayan tahr and frequently by the takin and other Caprinae [11,36,47,52]. Conversely, the Head Down is common to all Caprinae species with small horns, including serows, chamois, Himalayan tahr, and takin [11,50]. In all these species, the Head Down is a threat where horns are usually oriented to the target individual: conversely, among sheep Ovis spp. and goats Capra spp., it is a rare behaviour and mostly an indirect display [11]. The mainland serow performs Head Down statically by hitting the fore hooves on the ground in unison. This behaviour increases in dynamism in the Himalayan goral and in the chamois, which perform it while moving and keeping down their head [19,55].

Conclusions
Among Rupicaprini, the most efficient weapons are those of the mountain goat, which are stiletto-like with an excellent inclination to stab an opponent [50]. Conversely, the horns of the goral and the serow lie on the same plane of the frontal bones, thus making possible the usage of an alternative dominance display (i.e., Frontal Pushing). The high proportion of indirect forms of aggressive behaviour in both species of chamois suggests Rupicapra spp. as the most advanced genus among those considered in Figure 4 [20,54]. In fact, horns with hooked tips are unlikely to be efficient weapons, as the hook prevents a direct stab and forces to strike at an odd angle. Furthermore, thinly shaped hooks break easily inside the body muscles of an opponent [56].
Chamois are the most social Rupicaprini, and increasing gregariousness is likely to generate a change in weapon system from "daggers" to-ultimately-wrestling-type weapons [56]. Therefore, chamois may have evolved hooked horns, without yet having developed wrestling-or butting-type horns and behaviour, cf. Capra spp. Lovari and Apollonio (1994) argued that sociality may have triggered the evolution of ritualised horns within the Rupicaprini Tribe [19]. Yet, if so, no evolutionary push to develop a ritualised usage of horns should have occurred in the solitary serow. While detailed information on the social organisation of the mountain goat and of both chamois species is available [57-59], we still lack detailed information on the social behaviour of the other Caprinae with near-monomorphic horns between the two sexes. This information is crucial to further develop hypotheses on the evolution of aggressive behaviour, in turn horn shape and size, of little dimorphic caprids.
Author Contributions: S.L. conceived the study, collected data and participated in data analyses and in writing up all drafts; E.L.P. participated in data analysis and wrote the first draft; E.M. participated in data analysis and in writing up all drafts. All authors have read and agreed to the published version of the manuscript.
Funding: S.L. was financially supported by a grant from the National Geographic Society, USA.