Almost every process of an animal's body depends for proper functioning on one or more of the mineral elements. Macrominerals such as calcium (Ca++), phosphorus (P+), sodium (Na+), chloride (Cl-), potassium (K+) and magnesium (Mg++) are needed for body structure as well as for maintaining the acid–base balance, fluid balance and transmembrane potential for cellular function, nerve conduction and muscle contraction (Cunha, 1980; Harvey et al., 2005). Minerals gain special importance during specific phases of animal life such as growth, reproduction and lactation (Cunha, 1980). At these stages, maintaining a balanced amount of all minerals is necessary to preserve an animal's health (Lewis, 1995). The increased need for minerals during the peripartum period determines significant changes in mineral metabolism. This may produce a mineral imbalance leading to the onset of pathological conditions, for instance hypocalcaemia (Filipović et al., 2010). A study by Sevinga et al. (2002) found that Friesian mares with retained placenta showed lower serum Ca++ levels compared to mares that shed the placenta spontaneously. A hypomagnesemic tetany responsive to Mg++ and Ca++ administration has been reported in lactating mares as well (Lewis, 1995). A few researchers have dealt with adjustments occurring in a mare's physiology during the peripartum period, focusing on clotting (Bazzano et al., 2014a) and haematological (Bazzano et al., 2014b, 2015) and biochemical profiles (Harvey et al., 2005; Aoki and Ishii, 2012; Satué and Montesinos, 2013; Bazzano et al., 2014c; Mariella et al., 2014). It has been shown that a series of hormonal and metabolic changes occur to modulate the maternal supply of energy and nutrients to the foeto-placental unit (Larsson et al., 2008). As gestation progresses, the direction and magnitude of these changes in maternal organism mainly depends on the increase in foetal metabolic requirements (Satué and Montesinos, 2013). Changes in a mare's metabolism continue during the postpartum period, when a certain amount of minerals is lost through milk production (Filipović et al., 2010): equine milk contains considerable concentrations of macrominerals that vary significantly during lactation (Salimei and Fantuz, 2012).

A general tendency of indiscriminately administrating mineral supplements to pregnant and lactating mares has developed in the last few years. Because little information on serum electrolytes is available for periparturient mares (Rook et al., 1997; Harvey et al., 2005), we aimed to determine whether physiological adjustments in serum macromineral concentrations occur when pregnant mares are fed a balanced diet. In addition, in this study we want to assess whether statistical correlations between studied minerals exist in pregnant and non-pregnant mares.

None of the mares included in Group A showed clinical signs of disease during their pregnancy and postpartum. The mares delivered healthy, viable full-term foals, without human assistance. They passed a normal and intact placenta spontaneously within 2 h and achieved the complete involution of the uterus within 2 weeks after foaling.

The Shapiro–Wilk analysis showed a normal distribution of the data. Figure 1 shows means ± SEM of studied parameters obtained from pregnant mares and the control group. Group A showed significant changes in Cl- concentration (P=0.011) that rapidly decreased during the first 2 weeks following parturition compared to the week before foaling and the time of foaling (-1 and P vs. +1+2). A significant effect of pregnancy was found on serum P+ (P=0.015), pregnant mares showed lower P concentrations than non-pregnant mares throughout the monitored period. Statistical correlations between electrolytes calculated for Groups A and B (Table 1) are displayed graphically in Fig. 2. The linear regression model showed a progressive increase in Na+ concentrations (P=0.002) and a gradual decrease in serum Ca++ (P=0.001) during pregnancy (Fig. 3).

In the present study significant changes in the concentration of Na+, Cl-, Ca++ and P+ occurred during the last 3 months of pregnancy and the first 3 weeks postpartum in the mare.

The progressive increase in serum Na+ which we observed during the last month of pregnancy accords with a study by Harvey et al. (2005), which reported a peak in Na+ concentration around the time of foaling. These changes in Na+ trend may be related to the renin–angiotensin–aldosterone system (RAAS) of the mare that directly controls Na+ excretion, internal homeostasis and blood pressure. The effects of pregnancy on RAAS function have been investigated in humans (Weir et al., 1975; Bentley-Lewis et al., 2005), dogs (Robb et al., 1970) and mares (Satué and Domingo, 2011). In particular, an increase in aldosterone concentration was found in mares during late pregnancy (Satué and Domingo, 2011). This leads to a better renal conservation of Na+ and water to favour the expansion of the foeto-placental unit and to guarantee the correct nutrient intake of the foetus, as well as the appropriate homeostasis and blood pressure between mother and foetus. In our study the highest Na+ concentrations were recorded at the last blood sampling preceding parturition. Since, Na+ occurs in considerable amounts in muscles and plays a vital role in muscle contraction (Cunha, 1980), our finding may represent a physiological adjustment at the time of foaling when forceful abdominal contractions are necessary for the expulsion of the foal (Daels et al., 1991). The significant decrease in Cl- concentration recorded in our study after foaling accords with the findings of Harvey et al. (2005) that showed a progressive reduction in serum Cl- following parturition. This reduction in Cl- levels may be due to the loss of this electrolyte with lactation as a mare's milk contains from 300 to 640 ppm of Cl- (Cunha, 1980). In addition, although not reported in horses, hypochloremia without serum Na+ depletion has been reported in lactating cows (Lewis, 1995). In the present study significant changes in the serum levels of Ca++ and P+ were also found. The progressive decrease recorded in Ca++ concentration until the time of foaling accords with Berlin and Aroch (2009), who found lower serum Ca++ in pregnant mares compared to non-pregnant horses. A possible explanation lies in the equine placenta containing a Ca-binding protein that increases Ca++ transfer from the mare to the foetus in order to meet the needs for the foetal skeleton mineralization (Wooding et al., 2000). Together with Ca++, P+ plays a key role in several body functions such as bone and energy metabolism. These electrolytes comprise about 70 % of the mineral content of the body and up to 50 % of the minerals in milk. Most of the foetal Ca++ and P+ deposition occurs during the last 2 months of gestation, which suggests that most of the skeletal development occurs at that time (Kavazis et al., 2002). This foetal use of minerals and the close relation between the Ca++ and P+ metabolisms (Rosol and Capen, 1997) may explain the lower P+ concentrations in periparturient mares compared to the control group. Although previous studies dealt with the assessment of serum electrolytes in periparturient mares (Rook et al., 1997; Harvey et al., 2005), to the best of the authors' knowledge, this is the first study that statistically assesses the correlation coefficients between serum macrominerals in pregnant and non-pregnant mares. We found that every mineral was significantly correlated to another mineral in non-pregnant mares. In effect, serum electrolytes should be considered as a group rather than individually. As the intake of a mineral increases above that needed, the part not absorbed may bind other minerals, decreasing their absorption and possibly resulting in a deficiency of these minerals (Lewis, 1995). However, Group A did not show the same statistical correlations we found in Group B except Na+/ Cl- and Ca++/ Cl-. Opposite correlations Mg++/ Cl- and Mg++/ P+ were found in pregnant mares compared to the control group. These changes in mineral interrelations may result from physiological adjustments of the mineral metabolism occurring in the mare during both late gestation and lactation. During the peripartum period the mare has to cope with remarkable mineral requirements from the developing foetus as well as the growing foal, particularly in relation to Ca++ and P+ (Summer et al., 2004).

Contrary to the previous studies performed over a limited period of time (Rook et al., 1997) or with a limited frequency of sampling (Harvey et al., 2005), we focused on the last trimester of pregnancy and the first 3 weeks after foaling, using a weekly sampling that allowed us to detect transient changes in mineral serum concentrations. Because the most significant changes in serum electrolytes occurred within 2 weeks before and 2 weeks after foaling, monitoring serum electrolyte concentrations would indicate whether the mare needs further mineral supplementation at this stage. The significant changes in a mare's macromineral profile which we found over the peripartum provide enhanced knowledge about mineral interrelations and their modifications during specific physiological conditions such as late pregnancy and early postpartum.