Breastfeeding women benefit from the period of anovulation induced by lactation (Howie and McNeilly 1982, McNeilly 1993). This period is particularly important in undernourished populations, for whom other forms of birth control are unavailable, expensive or unacceptable. In such populations, this form of natural contraception may be the only constraint on the time between births and, consequently, the number of children a woman may bear (NRC 1989). Because the duration of lactational infecundability increases birth spacing, it has a positive effect on maternal and child health (Prentice et al. 1987). Thus, it is important to understand the effects of any factors that alter the duration of this period of infecundability.

One such determinant is maternal nutritional status, which has been reported to have a negative association with the duration of lactational infecundability (Chávez and Martínez 1973, Delgado et al. 1982, Huffman et al. 1987, Lunn et al. 1980, 1981 and 1984). It may therefore be important to weigh the known positive effects of nutrition intervention against the implications of a potentially shortened period of amenorrhea.

Although shorter periods of amenorrhea have been observed in studies that have provided undernourished women with energy supplements during lactation (Chávez and Martínez 1973, Delgado et al. 1982, Lunn et al. 1980, 1981 and 1984), the mechanisms by which maternal malnutrition delays ovulation during lactation could not be adequately explored. Infants also received food supplements in all of these trials, preventing investigation of the specific role of maternal supplementation. To examine the specific effect of maternal food supplementation on ovulation and hence better understand the mechanism responsible for lactational anovulation, we have found an animal model to be useful.

We have shown that the ovariectomized rat is an effective model for investigating the role of chronic undernutrition in both circulating reproductive hormone levels and the mediating effects of maternal behaviors (McGuire 1994, McGuire et al. 1995, Pachón et al. 1995). Use of the ovariectomized rat allows for a closer simulation of the hormonal milieu of human lactational infecundability because the negative feedback of ovarian steroids to the hypothalamus and pituitary present in intact rats is disrupted. As such, ovariectomized rats were the focus of the hormonal and behavioral aspects of this study.

The present experiment was designed to investigate the effects of food supplementation on the reproductive potential of previously underfed animals. Specifically, we sought to answer three questions: 1) Does food supplementation of undernourished rats during lactation alter their fecundity? 2) Can any observed differences in their reproductive potential be explained by the levels of circulating reproductive hormones? 3) Do maternal behavioral patterns throughout lactation help to explain any of these hormonal differences?

MATERIALS AND METHODS

Rats were acclimated to diet AIN-76A^TM (AIN 1977 and 1980) (Dyets, Bethlehem, PA) for a 1-wk period. At 42 d of age, rats were randomly assigned to one of three dietary treatment groups: control (C; n = 33),⁶ given unrestricted access to diet AIN-76A^TM, food-restricted (FR; n = 36), given 50% of the C intake in the form of a modified AIN-76A^TM with twice the amount of vitamins and minerals (Kliewer and Rasmussen 1987), or food-supplemented (FS; n = 35), FR until d 0 of lactation and offered unrestricted access to AIN-76A^TM thereafter. The amount of food offered to FS dams until lactation and FR dams throughout the study was based on the average daily dietary consumption (by weight) of five randomly chosen C rats that started their dietary treatment earlier than the rest of the animals. Diets were fed to rats daily in the morning, and rats were weighed weekly.

Starting at 65 d of age, animals were checked for signs of estrus by vaginal smears and bred when ready. Once deemed pregnant by the presence of sperm in a vaginal smear and/or the presence of vaginal plugs in the litter pan, rats were systematically assigned to one of two groups: those to be ovariectomized early in lactation (n = 43) and those to remain intact (n = 42). Pregnant rats were weighed every 5 d. In preparation for delivery, rats were moved to polycarbonate cages lined with pine-chip bedding on d 18 of pregnancy.

The day of parturition was designated as d 0 of lactation. On this day, litters were weighed and counted; if necessary, pups were culled or fostered from within dietary groups to have eight pups per dam. Litters were further culled to five pups on d 3 of lactation. Weights for dams and litters were recorded every 5 d. Pups were not offered any food other than their dam's milk and were removed on d 24 of lactation.

Immediately before and after the observation period, the condition of the nest was noted (pups were together at one end, scattered at one end or scattered between both ends). Dam location was characterized as being on the same half of the cage as most of her pups or on the other half.

Plasma concentrations of prolactin (Prl), luteinizing hormone (LH) and follicle stimulating hormone (FSH) were measured in duplicate when possible with rat ¹²⁵I radio-immuno assay kits (National Hormone and Pituitary Program, NIDDK, MD). For Prl, the interassay CV were 0.13, 0.10 and 0.08 for pools of serum containing mean concentrations of 2.59, 13.93 and 38.84 µg/L, respectively; intra-assay CV were 0.18, 0.02 and 0.23 for pools containing 2.92, 13.59 and 31.67 µg/L, respectively. For FSH, the interassay CV were 0.15, 0.07 and 0.07 for serum pools containing mean concentrations of 4.70, 7.61 and 11.12 µg/L, respectively; intra-assay CV were 0.13, 0.10 and 0.09 for pools containing 4.99, 7.92 and 11.07 µg/L, respectively. LH assay CV were 0.05, 0.11, and 0.19 for pools containing mean concentrations of 0.92, 6.54 and 11.55 µg/L, respectively; intra-assay CV were 0.08, 0.07 and 0.10 for pools of 0.81, 5.97 and 10.80 µg/L, respectively. It should be noted that only endocrine data from ovariectomized dams are presented here.

Data were analyzed with SYSTAT (Version 5.2.1, SYSTAT, Evanston, IL). All analyses for the prepregnant and pregnant period treated FR and FS rats as a single food-restricted group. These analyses also did not differentiate between rats who were or were not going to be ovariectomized because these distinctions were not relevant until after parturition.

Analyses of data acquired during the lactation period first controlled for the ovarian status of rats (intact vs. ovariectomized). As expected, 3-way interactions among dietary treatment, day of lactation and ovarian status were found such that the ovariectomized rats, on average, had higher plasma hormone concentrations with considerably less variability within dietary treatment group; no behavioral variable was found to differ between ovariectomized and intact rats. This further confirmed our use of the ovariectomized rat as the model of choice; therefore, we report hormonal and behavioral analyses only for ovariectomized animals.

The effects of diet on duration of lactational anestrus in intact rats, pup number and litter weight were analyzed by one-way ANOVA. Dam and litter weights during lactation were analyzed by repeated-measures ANOVA with dietary treatment group as a main effect.

The effects of diet and day of lactation on behavioral variables and hormone concentrations were evaluated by 2-way, repeated-measures ANOVA, including the (diet × day of lactation) interaction term for ovariectomized rats. Prl was log transformed before analysis; data reported here are back transformations. When differences were detected (P < 0.05 for main effects and P < 0.1 for interactions), preplanned comparisons were made with Student's t test to examine FS versus C and FS versus FR relationships and/or the significance of changes over the three observation days in lactation (i.e., d 10, 15 and 20).

The effects of behavior on hormone concentrations, while controlling for dietary treatment, were tested by individually adding each behavioral variable into the hormone models described above. This was done separately for each day of lactation during which we observed behaviors in rats, thus resulting in 2-way ANOVA with dietary treatment and behavioral variables as main effects. The same was done for the effects of hormone concentrations on behavior. Data presented here represent means ± SD unless otherwise noted.

RESULTS

DISCUSSION

The delay in onset of proestrus in the FR rats was similar to that reported previously (McGuire et al. 1992, Walker et al. 1995, Woodside 1991, Woodside and Jans 1995). Minor differences between our study and other reports can be attributed to differences in litter sizes (McGuire et al. 1992) as well as to the degree and duration of food restriction (Walker et al. 1995, Woodside 1991, Woodside and Jans 1995).

Prolactin has been implicated widely as the chief inhibitor of ovarian activity during lactation through a suckling-induced hypothalamic response. As such, it is expected that the initiation of ovarian activity would be associated with relatively low concentrations of plasma prolactin. When the effects of varying nutritional status have been investigated in humans, a clear increase in plasma Prl concentration has been detected in those of lower nutritional status (Delvoye et al. 1978, Lunn et al. 1980 and 1984). Rat studies, however, have not produced the same trends (Kliewer and Rasmussen 1987, McGuire 1994, Schulze and Rasmussen 1993, Walker et al. 1995). This investigation also found no consistent effect of diet on Prl. However, we did observe the expected trend of higher Prl concentrations in FR rats compared with both the FS and C treatment groups; no differences were observed between FS and C rats. The trend directly corresponds to the duration of lactational anestrus. However, the characteristic decline of Prl throughout lactation reported for humans (Delvoye et al. 1978, Lunn et al. 1980 and 1984) and rats (Mattheij et al. 1979, McGuire 1994, Schulze and Rasmussen 1993, Walker et al. 1995) was present in our study. This abatement is interesting, because pups were not weaned until 4 d after the last blood sample was taken, indicating continued suckling in all groups. It has been suggested that although the suckling stimulus may still be present, it is probably insufficient to maintain high levels of Prl in late lactation (Mattheij et al. 1979, Schulze and Rasmussen 1993). Additionally, all of the control animals displayed proestrus before their pups were weaned, suggesting that ovulatory function was not dependent solely on the presence of suckling and/or accompanying Prl secretion in these groups.

As expected, we found that supplementing previously food-restricted rats increased LH concentrations to at least those of C rats by d 15 of lactation. Further, FR rats showed no signs of elevated LH by d 20, supporting the observed longer period of anestrus in the intact animals. No direct relationship was found between LH and rat behaviors, as we had observed previously (McGuire et al. 1995). However, this may not be surprising given the finding by others (Taya and Sasamoto 1991) that suckling is mainly responsible for LH changes only in the first half of lactation.

As was the case for LH concentrations, we also found that food supplementation ameliorated the effects of previous food restriction on FSH concentrations by d 15 of lactation. With the exception of the FR group, mean plasma FSH increased across lactation, as observed in previous reports (Smith and Neill 1977, Taya and Sasamoto 1991). As we documented previously (McGuire et al. 1995), we again found no mediating role for behavior on the effect of dietary treatment on FSH concentrations.

When we looked to maternal behavior to help explain the differences in duration of infecundability, a picture similar to that revealed by hormone data emerged. As expected, food supplementation during lactation led to behaviors in rat dams that were overall less "maternal" than their chronically food-restricted counterparts. FS dams were less apt to stay close to their pups than both FR and C rats during mid-lactation, the time at which hormone differences were becoming pronounced. As a result, FS rats had fewer nursing opportunities and hence a shorter duration of anestrus in the absence of the suckling stimulus. Additionally, the mere absence of the young, irrespective of suckling, may explain the observed shorter periods of anestrus in FS rats; it has been shown in cows that being in close proximity of the calf, without suckling, can result in longer periods of postpartum anovulation (Hoffman et al. 1996). When compared with FR dams, both FS and C dams failed to keep pups together in their nest. FS and C rats had relatively heavier litters by d 10 of lactation, indicating the potential for more precocious pups. FS dams and pups would not be expected to have the same thermal requirements as undernourished animals and consequently may not need to remain as close together as FR rats (Jans and Woodside 1990).

Further support for the diminishing effect of food supplementation on maternal behavior comes from consideration of what we have termed the "suckling intensity score," a variable designed to capture both the duration of time suckled and number of pups engaged in suckling. FS dams experienced less suckling than FR dams during mid-lactation. This is in keeping with previous reports investigating the behavioral effects of food restriction (Crnic 1980, Pachón et al. 1995, Smart 1976). However, we did not detect an association between suckling intensity and Prl.

We show here that food supplementation hastens the postpartum return of anestrus in previously food-restricted rats; prolactin and gonadotropin concentrations as well as maternal behaviors were useful in interpreting this effect. Although the differences in fecundability could be ascribed to relative hormone concentrations alone, we were unable to successfully unravel the interrelationships of these reproductive hormones with maternal behaviors. These data represent novel findings in this animal model and are consistant with those collected from other species including the high producing dairy cow (Canfield and Butler 1990), sow (Cosgrove et al. 1995) and ewe (Rhind et al. 1989). We consider these data important both biologically and logistically, because of the frequent use of the laboratory rat as a model for human reproductive physiology and the ease with which both diet and suckling intensity can be manipulated in this animal.

Although the direct applicability of our results to human lactational infecundability requires further study, our findings emphasize the importance of considering the potential consequences of providing food supplements to those previously undernourished without additional family planning advice. This information, as well as consideration of socioeconomic status, specific breastfeeding patterns and time of introduction of supplementary foods should be given serious thought when initiating food intervention programs in malnourished populations.

ACKNOWLEDGMENTS