Folia Primatol. 24: 203-220 (1975)
Primate Breeding Season: Photoperiodic Regulation in Captive Lemur catta R ichard N. Van H orn Oregon Regional Primate Research Center, Beaverton, Oreg.
Key Words. Lemur catta ■Prosimian reproduction • Photoperiodicity • Primate ■Estrous cycles • Breeding season Abstract. Under natural light in Portland, Oreg., captive ring-tailed lemurs (Lemur catta) experience a breeding season that differs by nearly half a year from the season in Madagascar. A series of experimental day length changes from 1971 to 1974 demonstrated the ability of both temperate and tropical photoperiod cycles to induce estrous cycles in quiescent animals. After photoperiodic activation, most impregnated females failed to resume estrous cycles even after infant separations unless they received additional photoperiod changes. Unimpregnated females, on the other hand, showed no significant decline in the incidence of estrous cycles under prolonged exposure to a constant day length regimen (12.0L: 12.0D) for over a year.
Light cycles have been identified as the primary controllers of breeding seasons in many vertebrate species [H offmann, 1973; Benoit and A ssenmacher, 1970]; however, their role in the regulation of primate seasonality remains equivocal [Lancaster and L ee, 1965]. Field and laboratory studies of rhesus monkeys (Macaca mulatto) suggest that decreasing day length stimulates the onset of reproductive activity [Vandenbergh, 1973] although no experimental studies have verified this regulatory relationship [Birkner, 1970; E rikson, 1964], On the other hand, studies of Japanese macaques ( Macaca fuscata) indicate a more complicated relation between reproductive seasons and environmental or endogenous factors. In their natural habitat, Japanese monkeys show no consistent north-to-south correlation between annual light cycles and the onset of the breeding season in a range over 10° of latitude and 1,200 km [K awai et a i, 1967; Lancaster and L ee, 1965], and
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Introduction
204
V an Horn
E aton [1972] has shown that the presence of an infant and pregnancy influence the termination of the season. Likewise, two populations of African green monkeys ( Cercopithecus aethiops) have breeding seasons 6 months apart even though separated by less than 480 km [G artlan, 1969]. These contradictory results have led some investigators to contend that tropical primates are insensitive to photoperiods since annual light changes are minimal at the equator [Vandenbergh, 1973; Vandenbergh and Vessey, 1968; L ancaster and Lee, 1965; H offmann, 1973], However, the breeding season of the nocturnal mouse lemur ( Microcebus murinus), a small and relatively asocial prosimian, was shifted 6 months when animals were im ported to France, and experiments suggested that increases in artificial day length induced ovulatory cycles and testicular activity [Petter-R ousseaux, 1970, 1972; M artin , 1972a], However, the studies of P etter-R ousseaux [1970,1972] measured only external indices of reproductive activity such as vulvar swelling and perfor ation of the vaginal membrane or changes in testicular size. In addition, neither M artin [1972a,b] nor P etter-R ousseaux [1970,1972] included tests of statistical inference in their studies. Furthermore, observations of photoperiodic regulation in the diminutive and nocturnal Microcebus may not be generally applicable to many other primates because social factors in more gregarious species could mask the effects of light on the breeding season [Vandenbergh, 1973; G ordon and Bernstein, 1973; R ose el al., 1972], Ring-tailed lemurs (Lemur catta), on the other hand, are medium-sized prosimians (mean adult body weight of 2.5 kg) who live year round in multi male groups of 12-24 individuals [Jolly, 1966; P etter, 1965] and display a more complex social organization than most other prosimian species [M artin , 1972b; P etter, 1962; J olly, 1966]. Because of this similarity of social organi zation, findings in L. catta are more likely to be relevant to studies of simian primates. This paper reports evidence that annual photoperiod cycles in both temperate and tropical latitudes provide adequate stimuli for the regulation of seasonal breeding in L. catta.
Methods
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Historical Background A colony of ring-tailed lemurs (L.catta) was established at the Oregon Regional Primate Research Center in 1963 with the arrival of one breeding male and 6 females. By June, 1974, new acquisitions and births had raised the number to 69: 27 adult females; 27 adult males; 10 juveniles; and 5 infants. Throughout the development of the colony,
Primate Photoperiodicity
205
a seasonal pattern of breeding has persisted although changes have occurred in the timing of the season. The diet of monkey chow, apples, bananas, oranges, canned pineapple, raisins, and grapes has remained uniform over the years and cannot, therefore, account for either the maintenance of the breeding season or alterations in the timing of the season. Rather, differences in the conditions of illumination to which the animals were exposed appear to have influenced the seasonal distribution of conceptions in captivity. Four conditions of illumination associated with seasonality in retrospective studies (1963-1974) and two experimental light regimens (1971-1974) are described below. The relation of these light conditions to reproductive activity is discussed in the Results section.
Retrospective Studies
1 National Geographic Atlas of the World, M. B .G rosvenor (Ed.), National Geographic Society, Washington, D.C., 1963.
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Condition I. In 1966, two pregnant females arrived in our colony from Madagascar and later that year delivered viable infants. Since these females had conceived in the wild under natural light patterns for Madagascar, their conception dates were estimated from birth dates on the basis of a 136-day gestation (n= 14 timed pregnancies) and these dates were then used as an estimate of the Malagasy breeding season. Condition 2. From 1963 to 1966, our captive lemurs inhabited wire-mesh cages in a heated and air-conditioned room. Fluorescent lamps provided light during a 9 h work day, but clear glass windows also exposed animals to natural photoperiod cycles for 45 N. latitude in Portland, Oreg. (fig. 1). Temperatures indoors varied only a few degrees seasonally with, for example, a winter mean (December 21 through March 21) of 75.5 F (SE ±0.1) and a summer mean (June 21 through September 21) of 74.2 F (S E ± 0 .l). Under these conditions, sexually mature females either remained in their home cages with a breeding male or vaginal smears were taken regularly by previous investigators [E vans and G oy, 1968] to insure that isolated females could be exposed to one of a limited number of stud males during estrus. As a consequence of this breeding program, conception dates are known either by the presence of sperm in vaginal smears (first detection of sperm = day 1 of gestation) or by calculations from birth dates on the basis of a 136-day gestation period for this species. Condition 3. In the spring of 1968 and annually thereafter, immature animals were transferred each year from indoor cages to large outdoor enclosures where they grew to maturity in bisexual groups of 3-13. The outdoor enclosures ranged from 1,000 to 1,300 ft3 and contained a heated alcove (100 ft3) 10 ft above the floor. Although the heater maintained a temperature of 68 "F on a windless day in the winter when the rest of the cage was at 32 F, animals frequently experienced much colder temperatures on windy days or on their frequent ventures from the alcove. Animals in these enclosures, then, were exposed to normal seasonal fluctuations of temperature (August: mean high = 73 °F, mean low = 55 F; January: mean high = 45 ’F, mean low = 36 F).1* Condition 4. From 1967 to 1971, sexually mature L.catta inhabited wire-mesh cages in a windowless room with temperatures regulated as in Condition 2. No sunlight entered this room, however, and fluorescent lamps provided a constant day length of only 9 h/day over a 3.5-year period (1967-1971). Under this condition, no year-round studies of reproductive cycles were conducted. Rather, efforts to monitor estrous cycles with vaginal smears and to pair the single stud
206
Van Horn
male with estrous females were either totally abandoned (1968 and 1970) or were confined to the fall and winter months (1967 and 1969) on the premise that the breeding season previously identified by E vans and Gov [1968] was not regulated by light and would, therefore, continue unchanged under the 9L:15D regimen. The consequences of this breeding program are discussed in the Results section.
Light regimen. Fluorescent lamps (40 W) with a color temperature of 5,000-6.000 K illuminated the experimental room. For the 4 years preceding this experiment, the subjects in this room had been exposed to short, constant day-length cycles of 9L: I5D from these lamps as described in Condition 4 of the retrospective studies. On June 10, 1971, the photo period cycle was increased in one step to I4L:I0D , a light regimen that continued for 140 days until October 27, 1971 (fig.3). Day length was then progressively decreased overa 44-day period to a pattern of 8.5 L: 15.5 D at the rate of 7.5 min/day for a total decrement
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Experiment 1 In this study, sexually mature female (n =12) and male (n = 7) subjects were housed in individual cages in a windowless room. Female estrous cycles were monitored by the daily examination of vaginal smears [E vans and G o y , 1968; G reenstein , 1964], Smears were initially classified as diestrous (pre dominantly mucus and leucocytes), proestrous (75% or more cornified epi thelial cells), or vaginal estrous (predominantly cornified epithelial cells and absence of leucocytes) [for details see E vans and G oy , 1968, and E ato n et a!., 1973]. Females in vaginal estrus were paired with breeding males at 6 h inter vals in an observation room. If the male intromitted within a 30-min period, the female was reclassified from vaginal estrus to behavioral estrus. Pair tests continued at 6 h intervals until receptivity ceased. Females were paired with more than one male to eliminate the influence of partner preferences. To examine the frequency of estrous cycles within the experimental breed ing season, I permitted only the first 4 females in each of the first two sets of estrous cycles to receive ejaculations. All other copulations were interrupted before ejaculation. A previous study had found that nonpregnant females ex perience three estrous cycles per breeding season [E vans and G o y , 1968]; therefore, all females were allowed to receive ejaculations after the second set of cycles in the experimentally regulated breeding season. Since sexual re ceptivity persists for only 1day(A = 14,X = 26.64h,SE ± 3.57,range = 8-48h) in female ringtails, estrous cycles were expressed with a number that cor responded to the day of the year on which receptivity first occurred. For statistical analyses, nonreceptive estrous cycles were arbitrarily assigned a number that corresponded to the midpoint of full cornification in cycles of vaginal estrus. After parturition, mothers were separated from their infants at widely divergent intervals between May 31, 1972, and March 15, 1973 (fig-4).
Primate Photoperiodicity
207
of 5.5 h. On December 11 and 12, 1971, day length remained at this low point in the regimen (8.5L: 15.5D). On December 13, 1971, progressive day length increments began and continued at a rate of 7.5 min/day over a 46-day period to a high point of 14.5 L :9.5 D for a total increment of 6 h on January 27, 1972. At some time after this date, an equipment failure caused the light cycle to decline at an undetermined rate to a pattern of 12.5 L: 11.5 D on March 11,1972. This photoperiod continued for more than a year until April, 1973 (fig. 3).
Experiment 2 After April 2, 1973, female ringtails (n= 16) were maintained as in Ex periment 1 with minor modifications. In this experiment, females were caged alone during their first two sets of estrous cycles. No ejaculations were per mitted on their first set of estrous cycles in the 1973 season (fig. 3) although females were tested with a male twice daily for receptivity in their home cage. On the second set of estrous cycles, these females were tested twice each day in an observation room as in Experiment 1 and were allowed to receive ejacu lations. On the third and subsequent sets of estrous cycles, the females were caged with a breeding male and were allowed to mate at will. Estrous cycles and ejaculations were detected by the examination of vaginal smears as in the previous experiment. Four females (No. 2370, 4693, 5359, and 5468) who were not included in Experiment 1 were added to the sample for this exper iment (fig. 3). Light regimen. In Experiment I the intermediate day-length cycles of 12.5L continued for 12.5 months until April 2, 1973, when the photoperiod was reduced 30 min from 12.5 L: 11.5D to 12.0L:12.0D. This light-dark ratio was maintained over a 13-month period through April, 1974 (fig. 3).
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Statistical Methods Directional data. One problem in the study of photoperiodocity results from the inability of linear statistics to deal with the directional properties of breeding seasons. For example, a hypothetical breeding season with two conceptions, one on day 364 (December 30) and the other on day one (January 1) of the following year would have a mean direction (season) of day 365 (December 31), although the application of linear statistics would have yielded an erroneous mean of day 182.5 (July 1). Within a single breeding season, this problem is easily circumvented by arbitrarily moving the zero point to the quiescent portion of the year. However, when experimental conditions extend the distribution of conceptions throughout the year in a formerly seasonal species, or when the combined distributions of two or more experimentally modified seasons extend over the entire year, no zero point can be deter mined that will not invalidate the use of linear statistics. Batschelet [1965] and M ardia [1972] discuss this and other problems in analyses of directional data and demonstrate how the use of vectors avoids the problems inherent in the use of linear statistics. In their method, the year is represented as a circle with 365.25 days corresponding to 360° (fig. 1), and each conception within a season is represented by a vector with a magnitude of one and a direc tion equivalent to the time of year on which the conception occurred. Thus, a conception
208
Van Horn
on day 364 (Dec. 30) would have a magnitude of 1 and a direction of 358.77°. Applying the methodsof vector analysis to the entire distribution of conceptions withina season,one sums all vectors and then calculates a mean vector for the sample. The direction of this vector is analagous to the mean of linear statistics and, therefore, reflects the mean of the breeding season. The magnitude of the vector, on the other hand, reflects the concentration of conceptions about the mean and ranges in magnitude from 0 to I. If all conceptions occurred on the same date, the mean vector would have a magnitude of t and a definite season would be demonstrated. In contrast, conceptions that were uniformly distributed throughout the year would yield a mean vector with a magnitude of 0 and therefore would demonstrate no preferred season of conceptions. To examine the retrospective evidence of seasonality and photoperiodicity in captive L. carta, conception dates were either obtained from breeding records or calculated from birthdates (mean gestation = 136 days) and plotted on a circular graph with 1 year (365.25 days) equivalent to 360 degrees. A mean vector and a mean angular deviation (analogous to the standard deviation of linear statistics) were calculated for each distribution (fig. 1-2). The seasonal nature of each distribution was verified with the Rayleigh rest for the signifi cance of deviations from a uniform circular distribution [Batschelet, 1965; M ardia, 1972], The significance of differences in the breeding season under different light conditions and differences in the annual occurrence of the onset of estrous cycles were examined with the Uniform Scores rest for circular distributions [M ardia, 1972]. Linear data. Within Experiment 1, the inhibitory effect of long day length was demon strated with the binomial test. Changes in the incidence of estrous cycles under experimental light regimens were examined with a partitioned Cochrans Q [F leiss, 1973] and the relation between day length and estrous activity was tested with the phi-coefficient, n p [Edwards, 1962], In considerations of the influence of endogenous rhythms on the onset of estrous cycles in Experiment 2, the relation between conception, parturition, mother-infant separation, and the resumption of estrous cycles was examined with the Product Moment Correlation Coefficient.
Retrospective Studies Condition I. The estimated conception dates for the two females who had conceived in Madagascar were on April 16 and April 24 (fig. 2) with a mean date of April 20 (day 110). These dates agree well with the field observations of J olly [ 1966] who estimated that all 9 females in one of her study groups mated between April 16 and April 27. Condition 2. Captive L.catta who were kept indoors with a constant room temperature and with access to natural light cycles through clear glass win dows continued to experience a restricted breeding season (fig. 1). This season extended from September 26 (day 269) to January 25 (day 25) with a mean direction of December 8 (day 342) and a mean angular deviation (MAD) of ±31.45 days (magnitude of the mean vector (/) = 0.851 ). According to the
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Results
Primate Photoperiodicity
209
Rayleigh test [B atschelet , 1965], this distribution differed significantly from a uniform distribution (z = 5.066, />
Primate Breeding Season: Photoperiodic Regulation in Captive Lemur catta R ichard N. Van H orn Oregon Regional Primate Research Center, Beaverton, Oreg.
Key Words. Lemur catta ■Prosimian reproduction • Photoperiodicity • Primate ■Estrous cycles • Breeding season Abstract. Under natural light in Portland, Oreg., captive ring-tailed lemurs (Lemur catta) experience a breeding season that differs by nearly half a year from the season in Madagascar. A series of experimental day length changes from 1971 to 1974 demonstrated the ability of both temperate and tropical photoperiod cycles to induce estrous cycles in quiescent animals. After photoperiodic activation, most impregnated females failed to resume estrous cycles even after infant separations unless they received additional photoperiod changes. Unimpregnated females, on the other hand, showed no significant decline in the incidence of estrous cycles under prolonged exposure to a constant day length regimen (12.0L: 12.0D) for over a year.
Light cycles have been identified as the primary controllers of breeding seasons in many vertebrate species [H offmann, 1973; Benoit and A ssenmacher, 1970]; however, their role in the regulation of primate seasonality remains equivocal [Lancaster and L ee, 1965]. Field and laboratory studies of rhesus monkeys (Macaca mulatto) suggest that decreasing day length stimulates the onset of reproductive activity [Vandenbergh, 1973] although no experimental studies have verified this regulatory relationship [Birkner, 1970; E rikson, 1964], On the other hand, studies of Japanese macaques ( Macaca fuscata) indicate a more complicated relation between reproductive seasons and environmental or endogenous factors. In their natural habitat, Japanese monkeys show no consistent north-to-south correlation between annual light cycles and the onset of the breeding season in a range over 10° of latitude and 1,200 km [K awai et a i, 1967; Lancaster and L ee, 1965], and
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Introduction
204
V an Horn
E aton [1972] has shown that the presence of an infant and pregnancy influence the termination of the season. Likewise, two populations of African green monkeys ( Cercopithecus aethiops) have breeding seasons 6 months apart even though separated by less than 480 km [G artlan, 1969]. These contradictory results have led some investigators to contend that tropical primates are insensitive to photoperiods since annual light changes are minimal at the equator [Vandenbergh, 1973; Vandenbergh and Vessey, 1968; L ancaster and Lee, 1965; H offmann, 1973], However, the breeding season of the nocturnal mouse lemur ( Microcebus murinus), a small and relatively asocial prosimian, was shifted 6 months when animals were im ported to France, and experiments suggested that increases in artificial day length induced ovulatory cycles and testicular activity [Petter-R ousseaux, 1970, 1972; M artin , 1972a], However, the studies of P etter-R ousseaux [1970,1972] measured only external indices of reproductive activity such as vulvar swelling and perfor ation of the vaginal membrane or changes in testicular size. In addition, neither M artin [1972a,b] nor P etter-R ousseaux [1970,1972] included tests of statistical inference in their studies. Furthermore, observations of photoperiodic regulation in the diminutive and nocturnal Microcebus may not be generally applicable to many other primates because social factors in more gregarious species could mask the effects of light on the breeding season [Vandenbergh, 1973; G ordon and Bernstein, 1973; R ose el al., 1972], Ring-tailed lemurs (Lemur catta), on the other hand, are medium-sized prosimians (mean adult body weight of 2.5 kg) who live year round in multi male groups of 12-24 individuals [Jolly, 1966; P etter, 1965] and display a more complex social organization than most other prosimian species [M artin , 1972b; P etter, 1962; J olly, 1966]. Because of this similarity of social organi zation, findings in L. catta are more likely to be relevant to studies of simian primates. This paper reports evidence that annual photoperiod cycles in both temperate and tropical latitudes provide adequate stimuli for the regulation of seasonal breeding in L. catta.
Methods
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Historical Background A colony of ring-tailed lemurs (L.catta) was established at the Oregon Regional Primate Research Center in 1963 with the arrival of one breeding male and 6 females. By June, 1974, new acquisitions and births had raised the number to 69: 27 adult females; 27 adult males; 10 juveniles; and 5 infants. Throughout the development of the colony,
Primate Photoperiodicity
205
a seasonal pattern of breeding has persisted although changes have occurred in the timing of the season. The diet of monkey chow, apples, bananas, oranges, canned pineapple, raisins, and grapes has remained uniform over the years and cannot, therefore, account for either the maintenance of the breeding season or alterations in the timing of the season. Rather, differences in the conditions of illumination to which the animals were exposed appear to have influenced the seasonal distribution of conceptions in captivity. Four conditions of illumination associated with seasonality in retrospective studies (1963-1974) and two experimental light regimens (1971-1974) are described below. The relation of these light conditions to reproductive activity is discussed in the Results section.
Retrospective Studies
1 National Geographic Atlas of the World, M. B .G rosvenor (Ed.), National Geographic Society, Washington, D.C., 1963.
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Condition I. In 1966, two pregnant females arrived in our colony from Madagascar and later that year delivered viable infants. Since these females had conceived in the wild under natural light patterns for Madagascar, their conception dates were estimated from birth dates on the basis of a 136-day gestation (n= 14 timed pregnancies) and these dates were then used as an estimate of the Malagasy breeding season. Condition 2. From 1963 to 1966, our captive lemurs inhabited wire-mesh cages in a heated and air-conditioned room. Fluorescent lamps provided light during a 9 h work day, but clear glass windows also exposed animals to natural photoperiod cycles for 45 N. latitude in Portland, Oreg. (fig. 1). Temperatures indoors varied only a few degrees seasonally with, for example, a winter mean (December 21 through March 21) of 75.5 F (SE ±0.1) and a summer mean (June 21 through September 21) of 74.2 F (S E ± 0 .l). Under these conditions, sexually mature females either remained in their home cages with a breeding male or vaginal smears were taken regularly by previous investigators [E vans and G oy, 1968] to insure that isolated females could be exposed to one of a limited number of stud males during estrus. As a consequence of this breeding program, conception dates are known either by the presence of sperm in vaginal smears (first detection of sperm = day 1 of gestation) or by calculations from birth dates on the basis of a 136-day gestation period for this species. Condition 3. In the spring of 1968 and annually thereafter, immature animals were transferred each year from indoor cages to large outdoor enclosures where they grew to maturity in bisexual groups of 3-13. The outdoor enclosures ranged from 1,000 to 1,300 ft3 and contained a heated alcove (100 ft3) 10 ft above the floor. Although the heater maintained a temperature of 68 "F on a windless day in the winter when the rest of the cage was at 32 F, animals frequently experienced much colder temperatures on windy days or on their frequent ventures from the alcove. Animals in these enclosures, then, were exposed to normal seasonal fluctuations of temperature (August: mean high = 73 °F, mean low = 55 F; January: mean high = 45 ’F, mean low = 36 F).1* Condition 4. From 1967 to 1971, sexually mature L.catta inhabited wire-mesh cages in a windowless room with temperatures regulated as in Condition 2. No sunlight entered this room, however, and fluorescent lamps provided a constant day length of only 9 h/day over a 3.5-year period (1967-1971). Under this condition, no year-round studies of reproductive cycles were conducted. Rather, efforts to monitor estrous cycles with vaginal smears and to pair the single stud
206
Van Horn
male with estrous females were either totally abandoned (1968 and 1970) or were confined to the fall and winter months (1967 and 1969) on the premise that the breeding season previously identified by E vans and Gov [1968] was not regulated by light and would, therefore, continue unchanged under the 9L:15D regimen. The consequences of this breeding program are discussed in the Results section.
Light regimen. Fluorescent lamps (40 W) with a color temperature of 5,000-6.000 K illuminated the experimental room. For the 4 years preceding this experiment, the subjects in this room had been exposed to short, constant day-length cycles of 9L: I5D from these lamps as described in Condition 4 of the retrospective studies. On June 10, 1971, the photo period cycle was increased in one step to I4L:I0D , a light regimen that continued for 140 days until October 27, 1971 (fig.3). Day length was then progressively decreased overa 44-day period to a pattern of 8.5 L: 15.5 D at the rate of 7.5 min/day for a total decrement
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Experiment 1 In this study, sexually mature female (n =12) and male (n = 7) subjects were housed in individual cages in a windowless room. Female estrous cycles were monitored by the daily examination of vaginal smears [E vans and G o y , 1968; G reenstein , 1964], Smears were initially classified as diestrous (pre dominantly mucus and leucocytes), proestrous (75% or more cornified epi thelial cells), or vaginal estrous (predominantly cornified epithelial cells and absence of leucocytes) [for details see E vans and G oy , 1968, and E ato n et a!., 1973]. Females in vaginal estrus were paired with breeding males at 6 h inter vals in an observation room. If the male intromitted within a 30-min period, the female was reclassified from vaginal estrus to behavioral estrus. Pair tests continued at 6 h intervals until receptivity ceased. Females were paired with more than one male to eliminate the influence of partner preferences. To examine the frequency of estrous cycles within the experimental breed ing season, I permitted only the first 4 females in each of the first two sets of estrous cycles to receive ejaculations. All other copulations were interrupted before ejaculation. A previous study had found that nonpregnant females ex perience three estrous cycles per breeding season [E vans and G o y , 1968]; therefore, all females were allowed to receive ejaculations after the second set of cycles in the experimentally regulated breeding season. Since sexual re ceptivity persists for only 1day(A = 14,X = 26.64h,SE ± 3.57,range = 8-48h) in female ringtails, estrous cycles were expressed with a number that cor responded to the day of the year on which receptivity first occurred. For statistical analyses, nonreceptive estrous cycles were arbitrarily assigned a number that corresponded to the midpoint of full cornification in cycles of vaginal estrus. After parturition, mothers were separated from their infants at widely divergent intervals between May 31, 1972, and March 15, 1973 (fig-4).
Primate Photoperiodicity
207
of 5.5 h. On December 11 and 12, 1971, day length remained at this low point in the regimen (8.5L: 15.5D). On December 13, 1971, progressive day length increments began and continued at a rate of 7.5 min/day over a 46-day period to a high point of 14.5 L :9.5 D for a total increment of 6 h on January 27, 1972. At some time after this date, an equipment failure caused the light cycle to decline at an undetermined rate to a pattern of 12.5 L: 11.5 D on March 11,1972. This photoperiod continued for more than a year until April, 1973 (fig. 3).
Experiment 2 After April 2, 1973, female ringtails (n= 16) were maintained as in Ex periment 1 with minor modifications. In this experiment, females were caged alone during their first two sets of estrous cycles. No ejaculations were per mitted on their first set of estrous cycles in the 1973 season (fig. 3) although females were tested with a male twice daily for receptivity in their home cage. On the second set of estrous cycles, these females were tested twice each day in an observation room as in Experiment 1 and were allowed to receive ejacu lations. On the third and subsequent sets of estrous cycles, the females were caged with a breeding male and were allowed to mate at will. Estrous cycles and ejaculations were detected by the examination of vaginal smears as in the previous experiment. Four females (No. 2370, 4693, 5359, and 5468) who were not included in Experiment 1 were added to the sample for this exper iment (fig. 3). Light regimen. In Experiment I the intermediate day-length cycles of 12.5L continued for 12.5 months until April 2, 1973, when the photoperiod was reduced 30 min from 12.5 L: 11.5D to 12.0L:12.0D. This light-dark ratio was maintained over a 13-month period through April, 1974 (fig. 3).
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Statistical Methods Directional data. One problem in the study of photoperiodocity results from the inability of linear statistics to deal with the directional properties of breeding seasons. For example, a hypothetical breeding season with two conceptions, one on day 364 (December 30) and the other on day one (January 1) of the following year would have a mean direction (season) of day 365 (December 31), although the application of linear statistics would have yielded an erroneous mean of day 182.5 (July 1). Within a single breeding season, this problem is easily circumvented by arbitrarily moving the zero point to the quiescent portion of the year. However, when experimental conditions extend the distribution of conceptions throughout the year in a formerly seasonal species, or when the combined distributions of two or more experimentally modified seasons extend over the entire year, no zero point can be deter mined that will not invalidate the use of linear statistics. Batschelet [1965] and M ardia [1972] discuss this and other problems in analyses of directional data and demonstrate how the use of vectors avoids the problems inherent in the use of linear statistics. In their method, the year is represented as a circle with 365.25 days corresponding to 360° (fig. 1), and each conception within a season is represented by a vector with a magnitude of one and a direc tion equivalent to the time of year on which the conception occurred. Thus, a conception
208
Van Horn
on day 364 (Dec. 30) would have a magnitude of 1 and a direction of 358.77°. Applying the methodsof vector analysis to the entire distribution of conceptions withina season,one sums all vectors and then calculates a mean vector for the sample. The direction of this vector is analagous to the mean of linear statistics and, therefore, reflects the mean of the breeding season. The magnitude of the vector, on the other hand, reflects the concentration of conceptions about the mean and ranges in magnitude from 0 to I. If all conceptions occurred on the same date, the mean vector would have a magnitude of t and a definite season would be demonstrated. In contrast, conceptions that were uniformly distributed throughout the year would yield a mean vector with a magnitude of 0 and therefore would demonstrate no preferred season of conceptions. To examine the retrospective evidence of seasonality and photoperiodicity in captive L. carta, conception dates were either obtained from breeding records or calculated from birthdates (mean gestation = 136 days) and plotted on a circular graph with 1 year (365.25 days) equivalent to 360 degrees. A mean vector and a mean angular deviation (analogous to the standard deviation of linear statistics) were calculated for each distribution (fig. 1-2). The seasonal nature of each distribution was verified with the Rayleigh rest for the signifi cance of deviations from a uniform circular distribution [Batschelet, 1965; M ardia, 1972], The significance of differences in the breeding season under different light conditions and differences in the annual occurrence of the onset of estrous cycles were examined with the Uniform Scores rest for circular distributions [M ardia, 1972]. Linear data. Within Experiment 1, the inhibitory effect of long day length was demon strated with the binomial test. Changes in the incidence of estrous cycles under experimental light regimens were examined with a partitioned Cochrans Q [F leiss, 1973] and the relation between day length and estrous activity was tested with the phi-coefficient, n p [Edwards, 1962], In considerations of the influence of endogenous rhythms on the onset of estrous cycles in Experiment 2, the relation between conception, parturition, mother-infant separation, and the resumption of estrous cycles was examined with the Product Moment Correlation Coefficient.
Retrospective Studies Condition I. The estimated conception dates for the two females who had conceived in Madagascar were on April 16 and April 24 (fig. 2) with a mean date of April 20 (day 110). These dates agree well with the field observations of J olly [ 1966] who estimated that all 9 females in one of her study groups mated between April 16 and April 27. Condition 2. Captive L.catta who were kept indoors with a constant room temperature and with access to natural light cycles through clear glass win dows continued to experience a restricted breeding season (fig. 1). This season extended from September 26 (day 269) to January 25 (day 25) with a mean direction of December 8 (day 342) and a mean angular deviation (MAD) of ±31.45 days (magnitude of the mean vector (/) = 0.851 ). According to the
Downloaded by: King's College London 137.73.144.138 - 11/21/2017 2:54:40 PM
Results
Primate Photoperiodicity
209
Rayleigh test [B atschelet , 1965], this distribution differed significantly from a uniform distribution (z = 5.066, />
