**4. Discussion**

This is the first study to measure both salivary IgA and salivary cortisol in Asian elephants throughout the day (16-h period), and we found that the trend line from regression analysis followed a diurnal pattern; however, the two were not significantly correlated because cortisol followed a linear pattern, whereas that of sIgA was quartic. The highest concentrations of cortisol were observed at 06:00 h, with the lowest at 22:00 h. By contrast, sIgA concentrations were elevated at 06:00 and 22:00 h, with nadirs at 10:00 and 18:00 h.

Circadian rhythms of IgA are known to vary among species [12]. In humans, sIgA concentrations peak in the morning and decrease throughout the day [38,39]. One study that measured sIgA concentrations over a full 24-h period revealed a gradual increase in sIgA starting at midnight with peak concentrations occurring at 08:00 h the following day [27]. Shirakawa et al. [27] recorded patterns of sIgA in humans coinciding with the sleep-wake time of the subjects, which were between 24:00 and 07:00 h. Our study showed similar peaks in concentrations for sIgA at 06:00 h, which was about an hour after wake time for the elephants. Although the slight increase in sIgA at 14:00 h in our study was not significantly different from 06:00 or 22:00 concentrations, it also was not different from 10:00 and 18:00 h, which agrees with the quartic pattern observed in humans. For elephants in this study, the sleep hours are between 23:00 and 05:00 h. The sIgA increase in samples collected at 22:00 h in elephants corresponded to the time when elephants began their standing sleep period, and agrees with changes associated with the sleep-wake cycle in humans [38]. By contrast, in pigs [21] and dogs [18], peak concentrations are observed during the afternoon. Authors speculate that species-related behavior and differences in daily routines could be the cause of contrasting patterns between species. Like humans, the daily activities of captive elephants are generally fixed, with tourist activities in the morning and afternoon, and a break in the middle of the day. Elephants often sleep in a standing position during these rest periods, which might explain the slight increase in sIgA at 14:00 h. Dogs also show peak concentrations in conjunction with intermittent sleep during afternoon hours [18]. However, the bulls in this study did not interact directly with tourists, ye<sup>t</sup> still showed a quartic pattern, perhaps because although contact was limited, they were still aware of tourist presence. Bull elephants did have a daily routine with mahouts that bathed and fed them at regular intervals, which also could have driven a circadian pattern. Because we were unable to measure a full 24-h cycle, further studies of sIgA concentrations in elephants during sleep would be beneficial to determine the complete cycle and confirm its resemblance to that of humans as compared to other species.

IgA has been measured in multiple sample types (blood, saliva, urine, and feces) across time, including samples from Asian elephants [24]. Samples in the study of Edwards et al. [24] were collected

only once a day, with no apparent attention to time, so circadian patterns were not determined. However, concentrations were highly variable, especially for feces and saliva, with serum having the lowest variability. Concentrations in urine were low, with many being undetectable, suggesting it may not be the best sample type to assess this biomarker. In addition to within animal variability, Edwards et al. [24] also found considerable between animal differences, similar to our study. Overall individual mean concentrations ranged from ~7 to 30 ng/mL saliva for five elephants in Edwards et al. [24], and ~41–70 ng/mL for 10 elephants in the present study. Understanding mechanisms driving this significant intra- and inter-animal variability is key to understanding the utility of IgA as a potential health or welfare biomarker.

sIgA concentrations differed somewhat between the present study and that of Edwards et al. [24] in that our overall mean sIgA was more than double in concentration. Possible causes could be related to minor modifications in the assay protocol, which included using lower antibody (1 mg/<sup>L</sup> versus 10 mg/L) and HRP (1:10,000 versus 1:2500 dilutions) concentrations, although the standard curve range was the same between studies. Climate and daylight hour differences between regions (Washington DC versus Chiang Mai) could have had an effect [40]. For example, Park and Tokura [41] found that brighter light conditions during the day resulted in higher concentrations of sIgA during nocturnal sleep in humans. From Mishra et al. [42], people that travelled from India to Antarctica exhibited increases in sIgA concentrations, which could reflect differences in either climate or day length. No visual differences were evident over six months of longitudinal sIgA data between February and August [24], whereas the Thailand study was conducted in August, so any influence of seasonality may be minimal. Other factors associated with sIgA secretion are age [19,43–47], sex [45], and health status [48,49]. Concentrations of sIgA did not differ between sexes or were related to age in this study, whereas Edwards et al. [24] found the highest sIgA concentrations in the oldest elephant of their study (69 years of age). Another elephant in that study experienced a severe health event indicative of a systemic infection, and showed a four-fold increase in fecal IgA, suggesting it might be a useful health biomarker in elephants [24]. Previous studies in humans have shown decreases in sIgA associated with illnesses, such as upper respiratory tract infections [49] and malignant tumors [48]. Elevated sIgA concentrations also were found after administration of endotoxins to pigs [50] and dairy cows [51], indicating an immune response to pathogens. All elephants in the present study were checked by a veterinarian to ensure there were no underlying health conditions that could interfere with sIgA measures. Although saliva collection generally took less than 3 min, some elephants showed some agitation to the collector's hand swiping the inside the oral cavity. However, that was not reflected in significantly altered sIgA or cortisol concentrations.

Similar to previous studies, our results indicated a diurnal rhythm for salivary cortisol with peak concentrations in the morning (06:00 h) that gradually decreased throughout the day in a linear trend. Salivary cortisol was highest at 08:00 and lowest at 20:00 h in African elephants [31], which was similar to high values at 07:30 compared to 19:30 h for Asian elephants [30]. This is the same pattern observed for urine, where Brown et al. [29] reported a clear diurnal pattern of glucocorticoid excretion in Asian elephants, with the lowest concentrations observed just before midnight and peak concentrations occurring around 06:00–08:00 hours. However, in this study, clear patterns were not always observed during all collection periods. Five elephants exhibited more random cortisol fluctuations on one or more of the collection days, whereas only three showed the clear diurnal pattern on all three days. Various factors can disrupt normal patterns of cortisol, including stressful events during the day (social disputes, physical accidents, physical restraint) [31,52,53], age, sex, parturition, and environmental factors [52–55]. Casares et al. [31] revealed that the diurnal salivary cortisol pattern was disrupted by a fight between two zoo African elephants. The incident took place at 14:30 h and ended without human intervention, but the cortisol concentration of both individuals was increased two-fold at 16:00 h during the time it would normally have been declining. In our study, no obvious social disputes occurred between animals; however, the overall daily cortisol concentrations were highest on the first and lowest on the final day of collection, resulting in a statistically significant day effect. This

suggests that the sample collection might have induced a mild stress response in some elephants, who then acclimated over time. Figure 2 shows that at 18:00 h, cortisol concentrations present the smallest variation. With low baseline concentrations and variations, sample collection is suggested during this time.

Some studies have reported significant correlations between sIgA and cortisol, including in humans [38] and dogs [56], while others found no such relationships. For example, Escribano et al. [22] revealed no significant correlation between the two parameters from pigs experiencing psychological stress in the form of isolation. Edwards et al. [24] also reported no correlation between salivary IgA and cortisol in Asian elephants from longitudinal samples over a six month period. In this study, no significant correlation between the two indicators was found, as evidenced by a linear downward trend in cortisol throughout the day, while sIgA tended to display a quartic pattern, with concentrations higher in the morning and evening.
