Rodents

In many biparental rodent species, including the Mongolian gerbil (Brown, Murdoch, Murphy, & Moger, 1995), Djungarian hamster (Reburn & Wynne-Edwards, 2000), mandarin vole (Wang et al., 2018a), and California mouse (Gubernick & Nelson, 1989), a surge of prolactin is present just prior to or just after the onset of fatherhood. For biparental mandarin voles, paternal experience rather than an underlying basal state plays a role in prolactin levels, as experienced fathers have significantly higher prolactin levels than new fathers (Wang et al., 2018a). In the biparental prairie vole, virgin males had higher levels prolactin than females, but this did not increase with paternal experience (Khatib, 2001; Numan & Insel, 2003). In the polygynous but seasonally paternal striped mouse, prolactin levels were higher during the breeding season than non-breeding season but contact with pups did not change prolactin levels in breeding fathers (Schradin & Pillay, 2004). Overall, these studies suggest that prolactin increases with fatherhood, but the causal mechanism that initiates and maintains this relationship is not yet understood. Though there is a strong association between higher prolactin and paternal care, blocking prolactin with dopamine agonists does not impair either paternal responsiveness or paternal care in Djungarian hamsters or prairie voles (Brooks, Vella, & Wynne-Edwards, 2005; Lonstein & De Vries, 2000). Overall, these rodent studies suggest that there is an association between prolactin and paternal care, but there are likely other neuromodulators and hormones with more direct roles promoting behavioral change.

Primates

In primates, and especially New World monkeys, prolactin also positively correlates with paternal care (Schradin & Azenberger, 2002). In marmosets, hypothalamic prolactin levels are higher in experienced fathers compared to naive inexperience males (Woller et al., 2012). In cotton-top tamarins, prolactin increases mid-gestation in experienced fathers and the month prior to gestation in new fathers (Ziegler, Wegner, & Snowdon, 1996; Ziegler & Snowdon, 2000; Ziegler, Washabaugh, & Snowdon, 2004). Prolactin is also associated with facilitating approach and initiation of paternal care (Storey & Ziegler, 2016) and is increased while fathers interact with their offspring (Dixson & George, 1982). In New World monkeys, the role of prolactin may also have a species-specific functional role in care that varies with paternal responsibilities (Schradin, Reeder, Mendoza, & Anzenberger, 2003). Among titi monkey parents, fathers hold infants almost exclusively and show stable, increased prolactin levels after pair bonding that persists throughout fatherhood. Marmoset fathers are also the primary carriers and show a similar increase in prolactin after pair bonding, but also show an increase in prolactin during their mate’s pregnancy. Goeldi’s monkeys carry infants only when offspring reach three weeks of age and do not show increased prolactin levels after pair bonding but do show a transient peak mid-gestation (Schradin et al., 2003). Overall, these comparative studies suggest prolactin may be involved in organizing male paternal care in titi monkeys and marmosets and activating paternal care in marmosets and Goeldi’s monkeys. However, the role of prolactin in organizing and activating these behaviors may be the first step of many hormonal and neurochemical changes as it occurs so far in advance of the first paternal behaviors expressed. Additionally, similar to rodents, suppression of prolactin is not sufficient to block paternal care in experienced marmosets (Almond, Brown, & Keverne, 2006).

Human fathers also experience changes in prolactin levels. Similar to Goeldi’s monkeys, human fathers experience a prolactin peak prior to the birth of their offspring that drops back down postnatally (Storey et al., 2000). However, another study reported findings more similar to marmosets as fathers had higher prolactin levels than non-fathers and fathers of older children had lower prolactin levels than fathers of young children (Gettler, Mcdade, Feranil, & Kuzawa, 2012). Behaviorally, fathers with higher prolactin are more alert and responsive to infant cries, and this effect is amplified for experienced fathers (Fleming, Corter, Stallings, & Steiner, 2002). This effect may depend on experience with infant contact, as separate study concluded that this variable significantly influenced prolactin reactivity to infant cues (Delahunty, McKay, Noseworthy, & Storey, 2007). Additionally, fathers with higher prolactin engage in more father–infant exploratory play (Gordon, Zagoory-Sharon, Leckman, & Feldman, 2010). Overall, these studies suggest that role prolactin plays in human fathers may be more similar to marmosets and Goeldi’s monkeys than titi monkeys. Additional studies have shown that prolactin in human fathers is not just correlated with experience but may also be activating greater paternal investment and care.

In summary, prolactin is positively correlated with fatherhood, paternal responsiveness, and paternal care across most species of paternal mammalian fathers. Because increases in prolactin often occur before paternal experience or increase with paternal experience, prolactin most likely plays a role far upstream in fatherhood. This suggests that there are other neuromodulators involved in organizing and activating paternal care. More manipulations of prolactin levels are required to assess cause and effect.

Rodents

In mandarin voles, fathers have higher OXT serum concentrations than virgin males (Yuan et al., 2019). In prairie voles, fathers have more OXT innervation in the hypothalamus and regions of the brain associated with regulation of the heart than non-fathers but had less OXT in the BNST (Kenkel, Suboc, & Carter, 2014). Similarly, California mice fathers showed less OXTR mRNA in the BNST than non-fathers (Perea-Rodriguez et al., 2015). As OXTRs in the BNST have been associated with anxiety-like phenotypes, it is possible that decreases in OXTR in the BNST function to reduce aversion to pups (Duque-Wilckens et al., 2018). Furthermore, mandarin vole fathers showed greater OXTR expression than virgins in the mPOA, a brain area associated with paternal behavior (Yuan et al., 2019). Taken together, these studies suggest that OXT may be influencing the paternal state transition but that the brain region most sensitive to OXT changes may vary by species in some cases. Studies in mandarin voles have also examined how OXT changes between new and experienced fathers. New fathers had higher OXTR in NAcc compared to experienced fathers and OXTR receptor levels in the NAcc of both groups decreased as pups got older (Wang, Wang, Wang, & Tai, 2018b). Additional changes were also observed in the OXTR expression in the MeA; new fathers showed decreases in OXTR expression while experienced fathers showed increased expression (Wang et al., 2018b). These studies suggest that there may also be activational roles for OXT in fatherhood that differ with offspring needs and parental experience, however, again, hormone manipulations are need to establish cause and effect.

Primates

In non-human primates, OXT may play facilitate paternal care in males with sexual experience. In marmosets, fathers have greater OXT in the hypothalamus than non-fathers (Woller et al., 2012). These differences in OXT also have behavioral effects: in marmosets, OXT decreased food refusal by the fathers to their offspring (Saito & Nakamura, 2011). However, OXT reduced food sharing by male siblings that typically would engage in food-sharing behavior (Taylor, Intorre, & French, 2017). These results suggest that control of paternal and sibling care may differ.

There is mounting evidence that OXT also plays a positive role in paternal care for human fathers. Fathers given OXT had increased blood oxygenation level-dependent (BOLD) activation in the caudate nucleus, anterior cingulate cortex, and visual cortex when seeing images of their child (Li, Chen, Mascaro, Haroon, & Rilling, 2017). This suggests OXT may promote paternal attention toward their own child. Likewise, OXT is associated with father–infant affect synchrony during father–infant social interactions (Gordon et al., 2010). This suggests that OXT may make fathers more sensitive to their child’s emotions. Furthermore, studies indicate that there may be an OXT-mediated positive feedback loop between paternal affection touch and OXT plasma levels (Feldman, Gordon, Schneiderman, Weisman, & Zagoory-Sharon, 2010). However, this effect may depend on T levels. When T was high, there was a negative association between OXT and affectionate touch (Gordon Pratt, Bergunde, Zagoory-Sharon, & Feldman, 2017). Another caveat to consider in these studies is the mechanism of delivery of OXT. Central, peripheral, and intranasal administration routes may activate the oxytocin system in different ways, potentially leading to different behavioral outputs (Bauman, Murai, Hogrefe, & Platt, 2018; Chini, Verhage, & Grinevich, 2017; Grinevich, Knobloch-Bollmann, Eliava, Busnelli, & Chini, 2016; Lefevre et al., 2017).

In summary, these studies suggest that OXT may promote human fathers to be more attentive and empathic toward offspring.

Rodents

Many rodent studies have implicated AVP in paternal care (Wang, Liu, Young, & Insel, 2000; Wynne-Edwards, 2001; Yuan et al., 2018). Studies that compare closely related monogamous and non-monogamous rodent species have shown that AVP likely plays an important role in ultimate mechanisms of paternal care. Male monogamous California mice have greater AVP immunoreactivity in the BNST and more V1aR in the LS than their non-monogamous counterparts, the white-footed mice (Bester-Meredith & Marler, 2003; Bester-Meredith, Young, & Marler, 1999). On a proximate level in the California mice, paternal experience is associated with decreased expression of V1aR expression in the BNST (Perea-Rodgriguez et al., 2015). Similar species differences can also be seen in voles. In monogamous prairie voles, fatherhood is associated with a reduction in AVP immunoreactivity in the LS and lateral habenular nucleus, but in the non-monogamous meadow voles, AVP immunoreactivity remains high in fathers (Bamshad, Novak, & De Vries, 1993). Although AVP immunoreactivity in paternal voles decreases, V1aR binding in the LS still has profound effects on paternal care. AVP injected into the LS increased paternal contact and huddling whereas a V1aR antagonist decreased paternal grooming (Wang, Ferris, & De Vries, 1994). This suggests that vasopressin is both necessary and sufficient for paternal care in prairie voles. Overall, the studies in monogamous rodent species show that the vasopressin system may play a key role in the evolution and proximate initiation of paternal care.

Primates

Many studies in primates echo the influence of vasopressin on paternal care. First-time and experienced marmoset fathers show increased V1aRs and spine density in the prefrontal cortex (Kozorovitskiy, Hughes, Lee, & Gould, 2006). This suggests that the AVP system may play a role in remodeling cortical inhibition to facilitate paternal care. Also, in marmosets, AVP increased responsiveness to infant but not juvenile stimuli in fathers (as well as in mothers and alloparents) (Taylor, Carp, & French, 2019). However, this may be specific to certain types of paternal behavior as AVP reduced paternal food sharing in fathers and increased aggressive vocalizations toward the offspring in this context (Taylor et al., 2017). Combined, these studies suggest that the AVP system plays a role in synaptic remodeling that may facilitate certain types of paternal behavior.

The AVP system also likely plays a role in human fatherhood. Genetic variation in the V1aR gene correlates with anterior prefrontal cortex activation while fathers watch a video of their child smiling (Nishitani et al., 2017). This suggests that variation in V1aR genes influences paternal responsiveness in human fathers. There also is evidence that AVP may increase paternal attention; fathers-to-be given AVP spent more time watching baby avatars in a virtual environment (Cohen-Bendahan, Beijers, van Doornen, & de Weerth, 2015). In addition to increasing paternal attention, AVP may be involved in reducing male aggression toward their own offspring. When exposed to their own or unknown infant faces accompanied with infant crying, fathers given placebo had a stronger hand grip (force of squeeze on a handgrip dynamometer) for their own offspring whereas fathers given AVP had stronger hand grip toward unknown infants (Alyousefi-van Dijk, et al., 2019). This suggests that AVP may inhibit paternal aggression toward their own offspring when offspring are displaying distress. Additionally, fathers’ salivary AVP levels negatively correlate with activations in social-cognitive circuits while watching interactions with their child (Atzil, Hendler, Zagoory-Sharon, Winetraub, & Feldman, 2012). One scenario is that AVP may increase attention toward offspring but tone down over-activation of social-cognitive circuits that could lead to aggression.

In summary, rodent, non-human primate, and human primate mammalian species suggest that AVP serves to increase paternal investment and care. AVP may be driving paternal care using mechanisms that differ distinctly from maternal care (Bendesky et al., 2017). Additionally, there are few reported differences in expectant fathers, first-time fathers, and experienced fathers, suggesting that the role of AVP in paternal care begins with pregnancy cues from the mate and does not fluctuate once initiated.

Rodents

In prairie vole fathers, corticosterone levels post stress-test were positively correlated with pup retrievals but were negatively correlated with licking and grooming (Bales et al., 2006). This suggests that immediately after a stressful event, corticosterone levels may influence the type of paternal behavior expressed rather than greater or lesser care overall. However, studies in other rodents have reported no significant findings with corticosterone and paternal care. In California mice, corticosterone injections did not influence paternal care, nor did it influence fitness outcomes for offspring (Harris, Perea-Rodriguez, & Saltzman, 2011). This may be due to species differences or the behavioral assays used. In the prairie vole, there was a behavioral stressor applied and in the California mice, corticosterone was manipulated directly.

Primates

Both non-human and human primates also show a relationship between cortisol and paternal experience and care. In tamarins, experienced fathers showed increases in baseline corticosterone and cortisol during mid-pregnancy of their mate (Almond, Ziegler, & Snowdon, 2008). Similarly, in marmosets, baseline cortisol levels increase just prior to parturition (da Silva Mota, Franci, & de Sousa, 2006). This suggests that it would be valuable to manipulate cortisol levels and examine whether this hormone prepares fathers for paternal care responsibilities.

Humans show similar influences of cortisol on paternal care as non-human primates. Immediately after the birth of a child, human fathers exhibit acute decreased cortisol after holding their baby (Kuo et al., 2018). This suggests infant contact may be a proximate mechanism for influencing cortisol levels in fathers. Higher baseline and reactivity cortisol levels were associated with greater direct and indirect offspring care and greater amounts of offspring play (Kuo et al., 2018). This finding is also supported by another study that showed expectant fathers with higher baseline and reactivity cortisol had higher quality of paternal care with their offspring (Bos et al., 2018). However, there was no effect of postnatal cortisol levels on quality of paternal care (Bos et al., 2018).

In summary, these studies suggest that corticosterone/cortisol positively correlate with paternal care, but that timing of corticosterone/cortisol increase may differ based on experimental setup.

Rodents

In house mice, progesterone agonists increase paternal infanticide but do not have effects on aggression toward adult male conspecifics. Additionally, progesterone antagonists and genetic knockouts both show no infanticidal behavior, decreased aggression toward young and show increased parental behavior, including marked increases in retrievals (Schneider et al., 2003). This suggests that decreased progesterone is both necessary and sufficient to induce paternal care in a non-paternal species.

In naturally paternal species, progesterone is associated with paternal behavior, although species variation exists. In the biparental California mice, peripheral progesterone levels in fathers were lower compared to non-fathers, and fathers showed a decrease in progesterone mRNA in the bed stria of the nucleus terminalis (BNST) and medial preoptic area (mPOA) (Perea-Rodriguez et al., 2015; Trainor, Bird, Alday, Schlinger, & Marler, 2003;). This suggests that both peripheral and central progesterone downregulation is associated with fatherhood in California mice. However, this relationship was the opposite in the biparental hamsters, Phodopus campelli. During late gestation, expectant fathers showed lower levels of progesterone compared to two days later when their first litter of pups were born (Schum & Edwards, 2005). This suggests that progesterone may have opposite functions in different paternal models (Wynne-Edwards & Timonin, 2007). However, the difference between rodent species may be due to the control used, as the California mouse studies found the greatest difference in progesterone levels between virgins and first-time fathers while hamster study examined expectant fathers (pups were due in two days) and new fathers (pups born the day prior). It is possible that males could have a progesterone peak prior to the birth of their offspring, similar to females.

Primates

The role of progesterone on paternal care has been sparsely reported in primate literature, but a few studies show that progesterone may have a role in promoting paternal care. In New World monkeys, the amount of progesterone response elements (PREs) found in the OXT receptor promoter region is positively correlated with paternal care (Vargas-Pinilla et al., 2017). This suggests that progesterone binding to its receptor and PREs have the potential to mediate some of the pro-paternal behavior associated with OXT. This is particularly relevant for paternal care in New World monkeys such as marmosets and tamarins but is likely not involved in human paternal care as humans do not have PREs. In human fathers, higher baseline levels of progesterone were associated with a greater feeling of being happy and relaxed while playing with their child, and across all fathers, progesterone decreased over time during father–child play at the 40- and 70-minute time markers (Gettler, Mcdade, Agustin, & Kuzawa, 2013). This suggests that some aspect of playing with their offspring is decreasing circulating progesterone levels. Additionally, higher baseline progesterone levels may facilitate paternal care by keeping mood high during the interaction.

In summary, in most paternal species, the role of progesterone does not appear to be as crucial as it is in mothers. However, in rodents, fatherhood is associated with marked changes in progesterone levels in both the periphery and brain, although the direction of this change varies by species. Studies that include progesterone in primates are sparse, but the strongest evidence is its evolutionarily conserved role in paternal New World monkeys; however, it is not well understood how it changes behaviors at the proximate level.

Rodents

A high level of diversity is found in T control of paternal care in rodents (Bales & Saltzman, 2016), although there seems to be a tendency towards no effect or a positive association between T and paternal behavior. There follows a description of a subset of these studies.

Using castration and hormone replacement, Mongolian gerbils were found to have a negative association between paternal behavior and androgens in one study (Clark & Galef, 1999), no effect in another study (Juana et al., 2010), and a positive association in two other studies (Martínez et al., 2015, 2019). There was variation in methodology between the studies, but it is interesting to note that males that were infanticidal prior to pairing (males become paternal after pairing) were used in the study finding a positive association. Studies with male dwarf hamsters have also displayed varying results that appear to be explained by whether males are experienced or virgin; Wynne Edwards and colleagues found no association between T and paternal behavior in a series of studies in experienced males (Wynne-Edwards & Timonin, 2007). This pattern was mirrored in house mice in which gonadectomy reduces pup retrievals; importantly, however, male response to pup vocalizations did occur in males that were sexually experienced but not virgin males (Okabe et al., 2010). This variable pattern again appears in prairie voles using castration and replacement studies with one study showing a positive association between T and paternal behavior (Lonstein & DeVries, 1999) and another showing the opposite pattern (Wang & De Vries, 1993). Finally, studies with California mice (e.g., Chary, Cruz, Bardi, & Becker, 2015; Trainor & Marler, 2001), bank voles (e.g., Gromov & Osadchuk, 2013), mandarin voles (Wang et al., 2018a) and Volcano mice (e.g., Luis et al., 2009, 2017) also show positive effects of T on paternal care.

As T can be converted into estrogen in the brain, this may explain some species variation in the correlations. In virgin Mongolian gerbils, T, estradiol, and dihydrotestosterone all reduce aggression toward pups, suggesting that T and metabolites may be important for this shift in behavior (Martinez et al., 2015). Recently, estrogen in dwarf hamsters was found to increase paternal behavior when administered to virgin males and induced them to exhibit paternal instead of infanticidal behavior (Romero-Morales et al., 2018). In California mice, estradiol was associated with decreased paternal anxiety and increased hippocampal neurogenesis for fathers but not virgin mice (Hyer et al., 2017). This suggests that a variety of species may use T metabolites to affect different aspects of paternal care.

The above is not an inclusive list of studies but there are several factors that may contribute to the variation: (1) Whether virgin or experienced males were used, (2) whether the T manipulation were conducted early during development or as adults, and (3) whether males used were prone to infanticidal behavior prior to becoming a father (within population variation). One interesting evolutionary adaptation found in California mice is that T drops as males become parental, but a series of studies shows that T still positively influences paternal behavior and does so via conversion to estrogen through the enzyme aromatase, particularly in the medial preoptic area (MPOA), which is associated with paternal behavior (Trainor & Marler, 2002; Trainor et al., 2003). In summary, there are a surprising number of studies showing a positive association between paternal behavior and T in rodents, despite the early evidence that there are tradeoffs between aggression and paternal behavior that are mediated by T. We hypothesize that infanticide may be a root cause of this potentially emerging difference exhibited by rodents, both the within-species variation in infanticidal behavior and because rodents are subjected to infanticide by conspecifics (Agrell, Wolff, & Ylönen, 1998). The level of infanticide by conspecifics may be a driving force for rodents to maintain high levels of protectiveness of pups via T responsiveness while still suppressing their own infanticidal tendencies toward pups and participating in infant care.

Primates

In primates, there is diversity in the association between T and paternal behavior. For non-human primates, T’s influence on paternal behavior may be directed more towards infant protection than infant care (Muller, 2017). Ziegler and Sosa (2016) administered three doses of T and found no effects on paternal behavior in the cooperatively breeding common marmoset in response to infant distress calls. This manipulation brings into question why T nonetheless can be negatively associated with paternal behavior such as the negative correlation between urinary T and infant carrying in marmosets (Nunes, Fite, & French, 2000) and the decreased T levels associated with a father’s exposure to infant scent in paternal marmosets (Prudom et al., 2008). The root of the difference may be in the type of paternal care such that infant protection may be positively associated with T and more nurturing behaviors negatively associated with T (van Anders et al., 2011). Consistent with this perspective is the finding in wild, red-bellied lemurs, also cooperative breeders, that fecal T is negatively associated with carrying of infants (Tecot & Baden, 2018). More manipulations are required to fully understand the association between T and paternal behavior in primates. Perhaps the answer rests in the levels of aromatase in specific brain regions for explaining responses to T as discussed for rodents or steroid production within the brain (McCarthy, 2012). Consistent with this is the finding that low doses of estrogen in marmosets increased fathers’ responses to infant distress calls. Thus estrogen, possibly neural sources of estrogen, may be more strongly influencing at least non-human primate paternal behavior.

In human males, T has often been associated negatively with paternal behavior (Gordon et al., 2017; Grebe, Sarafin, Strenth, & Zilioli, 2019; Kuo et al., 2018; Mascaro, Hackett, & Rilling, 2013; Saxbe et al., 2017; Weisman, Zagoory-Sharon, & Feldman, 2014; Zilioli & Bird, 2017) as in many avian species (Lynn, 2008; Staley, Vleck, & Vleck, 2011) but with some variation, particularly in non-western cultures (Muller, Marlowe, Bugumba, & Ellison, 2008). We again see some consistency with the hypothesis by Van Anders et al. (2011) that optimal T levels may depend on whether the paternal behavior involved is offspring defense or nurturant intimacy, with high T associated with offspring defense and low T being associated with nurturing paternal behavior. These concepts have been supported recently in the framework of infant cries (Bos et al., 2018; Roellke, Raiss, King, Lytel-Sternberg, & Zeifman, 2019; Zeifman, 2019). However, there is evidence in birds that T pulses may also play an important role in initiating paternal care (Goymann & Davila, 2017), although these manipulations have yet to be done in humans or other mammals.

In summary, while there is some conflicting evidence in rodent and, to a lesser extent, primate literature regarding the influence of T on paternal care, peripheral levels of T tend to negatively correlate with paternal care. However, the exceptions may occur when fathers need to protect young from rival conspecifics or other predators or when T is converted into its metabolite estradiol in the brain. Here, estradiol may exert a positive influence on paternal care in some brain regions.