postprint

environmental


Introduction
The predominant environmental cue to time the annual cycle in gonadal maturity in birds is photoperiod.However, since gonadal cycles are rarely if ever symmetrical with photoperiod, at least two photoperiodic mechanisms must be implicated.For a considerable time, these have been referred to as photostimulation and photorefractoriness (Burger, 1947), and these were thought to act sequentially to time gonadal maturation and regression respectively.A more recent hypothesis is that there is a stimulatory process and an inhibitory process acting in tandem at all times to control GnRH secretion and the net difference between these two determines the rate of maturation or regression (Dawson, 2015).
Differences between species in terms of the timing and duration of gonadal maturity (e.g.absolute and relative photorefractoriness) can be accounted for entirely by differences in sensitivity of the inhibitory process to photoperiod.The neuroendocrine pathway through which long photoperiods induce stimulation of GnRH secretion is fairly well understood e.g.(Ikegami and Yoshimura, 2016), but this is not true of the inhibitory process.
Although photoperiod is the primary environmental cue, other factors can have modifying effects.These are more important in species with less predictable breeding schedules (Wingfield et al., 1992) and may act directly on GnRH or through GnIH (Bedecarrats et al., 2016;Ernst et al., 2016;Tsutsui and Ubuka, 2016).Many studies have investigated the effects of ambient temperature on photoperiodically induced testicular maturation in birds.In a few cases, testicular maturation was accelerated by higher temperature (Perfito et al., 2005;Silverin et al., 2008;Wingfield et al., 2003); in the majority it was not (Caro et al., 2013;Caro and Visser, 2009;Dawson, 2005;Perfito et al., 2005;Silverin et al., 2008;Soumalainen, 1938;Visser et al., 2011;Wingfield et al., 1996;Wingfield et al., 1997).However, in all of the studies where subsequent regression was also monitored, high temperature advanced regression (Dawson, 2005;Dawson and Sharp, 2010;Visser et al., 2011;Wingfield et al., 2003;Wingfield et al., 1997).The photoperiodic inhibitory process leading to gonadal regression is therefore more amenable to modulation by temperature than is photoperiodic stimulation.
Food restriction also has effects on the reproductive axis (Davies and Deviche 2014).Hahn (1995) photostimulated Red Crossbills Loxia curvirostra but restricted their food consumption to short photoperiod levels.This delayed the increase in LH but did not affect testis size after 30 days.Perfito et al. (2008) found that testis size, but not LH, was affected by similar food restriction in Zebra finches Taeniopygia guttata.Food-deprived house finches Haemorhous mexicanus had under developed testes which was thought to result from decreased GnRH secretion (Valle et al. 2015).The daily timing of food availability does influence the photoperiodic inhibitory process.I did a study a (Dawson, 1986) involving three groups of common starlings (Sturnus vulgaris).One was kept under a short photoperiod 8L:16D with food provided ad lib.Another was transferred to a long photoperiod (16L:8D) with food ad lib.The third group was also transferred to a long photoperiod (16L:8L) but food was removed 8 h after dawn, so a feeding day of the same duration as the short photoperiod group.There was little difference in the rate of testicular maturation between the two long photoperiod groups, demonstrating that food had no influence on photoperiodic gonadal stimulation.However, there was a dramatic delay in regression.For most birds, regression was delayed by about 2 weeks (regression started 6 weeks after transfer to long photoperiods as opposed to 4 weeks in birds with ad lib food).Remarkably, two birds had shown no regression by the end of the study (12 weeks after transfer).Thus, as with temperature, the photoperiodic inhibitory process is more amenable to modulation by food availability than is photoperiodic stimulation.
Lower ambient temperature and decreased food availability had similar effects on the testicular cycle in starlings; neither influenced testicular maturation, but both delayed regression (Dawson, 1986(Dawson, , 2005)).It is perhaps counterintuitive that a decrease in food availability and a decrease in ambient temperature do not have negative effects on the reproductive system; they both extend rather than inhibit it.
Starlings have a robust and marked daily cycle in body temperature (Tb) which is closely related to photoperiod (Dawson, 2017).Tb starts to increase before dawn, remains high throughout daylight hours, and then decreases rapidly at dusk.The amplitude of the cycle increases as photoperiod decreases.Night-time hypothermia is thought to be an energy conserving mechanism and ambient temperature affects depth of nocturnal hypothermia (Reinertsen and Haftorn, 1986).Several studies have shown that food deprivation, or energy deficiency, can also increase the depth of nocturnal hypothermia (Ben-Hamo et al., 2010;Cooper and Gessaman, 2005;McKechnie and Lovegrove, 2002;Noakes et al., 2013;Nord et al., 2009;Pravosudov and Lucas, 2000;Prinzinger et al., 1991;Reinertsen and Haftorn, 1986;Reinertsen et al., 1988;Waite, 1991;Welton et al., 2002).Importantly, food removal in Japanese quail Coturnix coturnix japonica led to diurnal hypothermia (Laurila et al., 2005).
Since food restriction and ambient temperature affect both reproduction by delaying gonadal regression, and both can also affect the daily cycle in Tb, could the effects of food and temperature act through a common mechanism, the daily cycle in Tb, and that this influences the timing of gonadal regression?The aim of this study was to confirm that low temperature and food restriction have similar effects on gonadal cycles i.e. little effect on maturation but delaying regression, and to determine whether either or both affect the daily cycle in Tb.The results show that food restriction and ambient temperature both changed the characteristics of the daily cycle in Tb to that of a shorter photoperiod.A shorter photoperiod would also delay gonadal regression.Thus food and ambient temperature may act through a common pathway to influence the plastic photoperiodic inhibitory process, but not the more resilient photoperiodic stimulatory process.

Birds
Juvenile starlings were caught at Berwick upon Tweed, UK, 55.8 ⁰N during August 2014.Only males were kept; females were released.Birds were kept in four environmentally controlled chambers (2.5 m x 2.0 m and 2.2 m high) in which they were allowed to fly freely.
They were provided with turkey starter crumbs ad lib.All birds were treated with a single dose of anti-helminthic (Flubenvet) in food to eliminate gut parasites and 1% Ivermectin on the skin to remove ectoparasites.
Temperature was initially kept at 14 ⁰C with 5 changes of air per hour.The time that lights came on and went off each day was controlled by a Lutron GRAFIK Eye QS system.
Light intensity increased from zero to 1000 lux during a 1 min period at dawn and decreased during 1 min at dusk.Photoperiod for all groups was changed daily and simulated the natural changes at 56 ⁰N (minimum at the winter solstice 7L:17D; maximum at summer solstice 17L:7D).
Food was provided in commercial poultry feeders.Each feeder was placed in a solid enclosure (1m x 0.8m x 0.8m) such that access to the food was through two openings at floor level.In two of the chambers the opening could be closed with a metal plate that slid down at pre-determined times (Chicken Guard, used in free-range chicken coups to prevent nocturnal access by predators).

Experimental design
In early December, birds were laparotomized under isofluorane anaesthesia and the dimensions of the left testis measured to the nearest 0.1 mm using a binocular microscope.
Testicular volume was calculated as 4/3 πa 2 b where a is half the width and b is half the length.Birds were laparotomized at approximately three week intervals during the study.
At the winter solstice, temperature was decreased to 6 ⁰C in two chambers and in the other two it was increased to 16 ⁰C.In one of the chambers at 6 ⁰C and one at 16 ⁰C, access to food was prevented by the Chicken Guard door closing 7h after dawn (which was at dusk at the winter solstice).In the other two chambers, food was available ad lib.Sample numbers in the four groups were: 6 ⁰C ad lib food, n = 7; 6 ⁰C restricted food n = 9; 16 ⁰C ad lib food, n = 8; 16 ⁰C restricted food, n = 8.Dawn was kept at a fixed time (08:00 GMT), so that the time when access to food was prevented, also occurred at the same time each day (15:00 GMT).Photoperiod was increased to simulate the natural changes in photoperiod by extending dusk.
The inhibitory photoperiodic process leading to testicular regression (otherwise known as photorefractoriness) starts when photoperiod has increased to 12L: 12D (Dawson, 2015).Changes in the daily cycle in Tb which may be involved should therefore be apparent then.The daily cycle in Tb was assessed using DST nano-T temperature data loggers (Star-Oddi Ltd, Iceland) in which recorded data is stored in the logger's internal memory with a real time clock reference for each measurement.The loggers were synchronised with the same computer as the GRAFIK Eye lighting controls.The loggers were 17 mm x 6 mm, weighing 1 g.Each bird was anesthetised with Isoflurane, a 10 mm incision was made in the skin of the nape of the neck, and a logger inserted sub-dermally.The incision was closed with tissue adhesive (Vetbond).The loggers were implanted 5 days before photoperiod had increased to 12L:12D, and programmed to start recording five days after implantation.
Temperature was recorded every 5 min.Birds were again anesthetised, the loggers removed and data downloaded using the SeaStar program (Star-Oddi Ltd, Iceland).
The start of molt is closely associated with the time of testicular regression (Dawson, 2006).Since the start of molt can be measured exactly, unlike the gradual decrease in testis size, it provides an accurate quantifiable data point which is related to the timing of regression.Birds began to molt during May.The progress of molt was assessed using a molt scoring system which corrects for primary feather mass (Dawson and Newton, 2004).For each bird, a series of linear regressions was used to calculate molt start date and the date that subsequent increments in molt score occurred.
The experimental procedures were licensed by the UK Home Office (Project licence PPL 60/4176).

Statistical analyses
To make the analysis of temperature data more manageable, each block of six 5 min temperature records was averaged to produce a mean for each 30 min period.Data on Tb during 24h and testis volume during the study were then assessed in a repeated measures twoway ANOVA (Graph Pad Prism) followed by Tukey's Multiple Comparison Tests.Molt start dates were analysed using one-factor ANOVA followed by Tukey's Multiple Comparison Test.

Body mass
There were significant differences between the treatment groups: F3, 196 = 3.22, P=0.0386 (Fig. 1).However, there were no consistent differences between any groups.All four groups increased mass significantly between the first two sampling points but there were no subsequent consistent changes.There were two occasions, mid-March and mid-May, when the cold ad lib group had body mass significantly higher than the other three groups, but there were no significant differences between any groups at any other times.Therefore, differences in Tb, testis sizes and molt between the 6 ⁰C ad lib and restricted food groups, and between the 16 ⁰C ad lib and restricted food groups, were not due to under nutrition.

Daily cycle in body temperature under 12L:12D
There were significant differences between the treatment groups: F3, 1175 = 3.52, P=0.0296.In all groups, Tb increased significantly (P<0.001)before dawn and continued to increase after dawn (P<0.001).Also, in all four groups, Tb decreased significantly immediately after dusk (P<0.001).
Comparing the warm and cold ad lib groups (Fig. 2 A), Tb was significantly lower (P<0.01) in the cold ad lib group during the middle of the night but there were no differences at any other time.In the warm restricted group (Fig. 3A) Tb decreased significantly (P<0.05) between access to food being denied (13:00) and 15 min before dusk.Tb in this group was significantly lower than in the warm ad lib group 15 min before dusk (P<0.01) and 15 min after dusk (P<0.001).In the cold restricted group (Fig. 4A) Tb also decreased significantly (P<0.001) between access to food being denied and 15 min before dusk.Tb in this group was significantly lower than in the cold ad lib group 15 min before dusk (P<0.001) and 15 min after dusk (P<0.001).The decrease in Tb following access to food being denied was greater at the lower ambient temperature.Between 15:00 and dusk, Tb was lower (P<0.01) in the cold restricted group than in the warm restricted group (Figs.3A and 4A).
Mean body temperature over 24h did not differ significantly between the groups (Fig. 5) although there was a suggestion that mean body temperature was higher in the two groups at 16 ⁰C.

Changes in testis size
There were significant differences between the treatment groups: F3, 196 = 5.94, P=0.0029.Ambient temperature had a slight effect during the period of testicular maturation.
Between January and April, testicular volume was slightly, but not significantly greater, in the warm ad lib group than in the cold ad lib group (Fig. 2B).During early March, testicular volume was greater (P<0.01) in the warm restricted group than in the cold restricted group, but not at any other time.In contrast, ambient temperature did have a clear effect on the time of regression in the two ad lib groups.Testis volume showed a large decrease (P<0.001) between mid and late April in the warm ad lib group, but only a slight decrease (P<0.05) in the cold ad lib group.There was a greater decrease in the cold ad lib group between late April and mid-May (P<0.001).In both late April and mid-May, testis volume was significantly lower (P<0.001) in the warm ad lib group than in the cold ad lib group.Lower temperature also delayed regression, although less dramatically, in the food-restricted groups.
In mid-May, testis size was lower (P<0.001) in the cold food-restricted group (Fig. 4B) than in the warm food-restricted group (Fig. 3

B).
Food restriction had no effect on testicular maturation (Figs.3B and 4B) but it did have an effect on the timing of regression.Testis volume decreased (P<0.001) between mid and late April in the warm ad lib group, and between late April and mid-May in the warm restricted group (Fig. 3B).In both late April and mid-May, testis volume was significantly lower (P<0.001) in the warm ad lib group than in the warm restricted group.Testicular regression began slightly earlier in the cold ad lib group than the cold restricted group (Fig. 4B).There was a slight decrease in the cold ad lib group between mid and late April (P<0.05)but no decrease until later in the cold restricted group.

Molt
There were significant differences between the treatment groups: F3, 25 = 40.28,P<0.0001.The start of molt closely correlated with the time of testicular regression; treatments affected molt in the same way as regression.Ambient temperature affected the start of molt.Molt started 15 days earlier (P<0.001) in the warm ad lib group than in the cold ad lib group (Fig. 2C) and molt started 16 days earlier (P<0.001) in the warm restricted group than in the cold restricted group (Figs.3C and 4C).Food restriction also affected molt.Molt started 7 days earlier (P<0.05) in the warm ad lib group than in the warm restricted group (Fig. 3C) and molt started 8 days earlier (P<0.01) in the cold ad lib group than in the cold restricted group (Fig. 4C).

Discussion
The first aim of this study was to confirm that low temperature and food restriction have similar effects on gonadal cycles i.e. little effect on maturation but delaying regression.
There was indeed little effect of temperature, and no effect of food restriction, on testis size during testicular maturation.In contrast, testicular regression and the start of molt were both significantly delayed by both lower ambient temperature and the restriction of food.The lower ambient temperature (10 degrees lower) delayed molt by 15 days in the food ad lib groups and by 16 days in the food restricted groups.Food restriction (to 7h after dawn) delayed molt by 7 days in the 16 ⁰C groups and by 8 days in the 6 ⁰C groups.The annual cycle in gonadal maturity is thought to be controlled by two photoperiodic mechanisms acting in tandem throughout the year: a positive drive on GnRH secretion which is predominant during increasing photoperiods and leads to gonadal maturation, and an inhibitory process which becomes predominant during long photoperiods and leads to gonadal regression (Dawson, 2015).The effects of ambient temperature and food restriction appear to modulate photoperiodic control of the latter process rather than the former.There is a potential ecological explanation why lower temperature should lead to prolonged breeding.
In cooler years, while food peak food availability may be less, it is likely to be later, and more prolonged.However, it is not possible to be categorical from the data here that the effects were restricted to gonadal regression.Lower temperature caused a slight, although non-significant delay in testicular maturation.Food restriction had no effect on maturation.However, the caveat here is that because of the experimental design, there was little food restriction during the early stages of testicular maturation when photoperiod was still short.Nevertheless, ambient temperature and the time of food availability did both clearly have effects on the photo-induced gonadal cycle.
The second aim was to look for a common mechanism through which the two different environmental factors, ambient temperature and food restriction, could operate to impart their similar effects on the gonadal cycle.Obviously both have energetic implications.
There is evidence that both, independently, can affect the daily cycle in Tb.In birds with ad lib food, the daily cycle in Tb is strictly related to photoperiod; high Tb is normally maintained until dusk and this was true of both ad lib food groups in this study.In the food restricted birds, Tb began to decrease after access to food was blocked.Therefore the duration of maximal Tb was less than the photoperiod, and the timing of testicular regression was also as if photoperiod was shorter.In the ad lib fed birds, lower ambient temperature induced a greater amplitude in the daily cycle of Tb, with lower Tb during darkness.This too is a characteristic of shorter photoperiods (Dawson, 2017).Thus the timing of testicular regression appears, in some way, to be related to the daily cycle in Tb in addition to prevailing photoperiod.
It may be controversial to suggest that the daily cycle in Tb modulates photoperiodic responses in birds since there is a wealth of evidence to show that photoperiod is the major cue used to time gonadal maturation and regression e.g.Dawson (2015), and that light acts directly, in the case of birds, through encephalic photoreceptors (Foster and Follett, 1985;Garcia-Fernandez et al., 2015).However, it is well known in plants that temperature can modulate photoperiodic molecular mechanisms to regulate the timing of flowering (Andres and Coupland, 2012;Song, 2016).Furthermore, the present study led to a subsequent study in which starlings were maintained on ultra-short photoperiods, and gonadal responses related to the daily cycle in Tb much more closely than to photoperiod.Nevertheless, this remains a correlation rather than demonstrating a causal relationship.
Thyroid hormones may be an important link between energetics, Tb and photoperiodic responses.They regulate metabolic rate (e.g.hypothyroidism is associated with low Tb) and also play a critical role in photoperiodic responses (Dawson, 1993;Yoshimura, 2006).
Experimental treatment with exogenous thyroid hormones can mimic long photoperiods (Dawson, 1989;Follett et al., 1988).Wikelski et al (2008) suggested that energy turnover may determine the duration of circannual cycles in house sparrows (Passer domesticus); the lower the rate of energy turnover the longer the cycle length.In the present study, higher mean Tb was associated with shorter gonadal cycle.However, the food-restricted birds were not nutritionally stressedthey maintained their body weight.It was apparently the time of food availability rather than total food intake that was important and in house sparrows periodic food availability can act as a Zeitgeber for the whole circadian system (Hau and Gwinner, 1996).Future studies could directly address the question of whether the timing rather than the general availability of food is important.Starlings held under a simulated natural cycle in photoperiod between January and August.
Birds were held at 16 ⁰C (solid red, n= 8) or 6 ⁰C (solid blue, n=7) and provided with food ad libitum.Each point represents the mean ± S.E. A. Body temperature (Tb) recorded every 30 min for 24 h.Photoperiod was 12L:12D and the duration of darkness is shaded.Tb was lower during the night in birds at 6 ⁰C but there were no differences during the day.B. Changes in testicular volume between January and June.There were no significant differences during testicular maturation, but the birds held at 16 ⁰C showed earlier regression.The vertical line represents when photoperiod reached 12:12D when Tb was recorded (A).C. The progress of moult in the two groups.Each point represent the date ± S.E. that each unit of moult score was reached by each bird.Birds held at 16 ⁰C started to moult sooner.

Figure 3.
Starlings held under a simulated natural cycle in photoperiod between January and August.
All birds were held at 16 ⁰C.One group was provided with food ad libitum (solid red, n= 8).
For the other group, access to food was prevented 7h after dawn (open red, n=8).Each point represents the mean ± S.E. A. Body temperature (Tb) recorded every 30 min for 24 h.Photoperiod was 12L:12D and the duration of darkness is shaded.The vertical line represents when access to food was prevented, at 13:00 under 12L:12D.Tb began to decrease after access to food was prevented.B. Changes in testicular volume between January and June.
There were no significant differences during testicular maturation, but the birds with ad libitum food showed earlier regression.The vertical line represents when photoperiod reached 12:12D when Tb was recorded (A).C. The progress of moult in the two groups.Each point represent the date ± S.E. that each unit of moult score was reached by each bird.Birds with ad libitum food started to moult sooner.Starlings held under a simulated natural cycle in photoperiod between January and August.
All birds were held at 6 ⁰C.One group was provided with food ad libitum (solid blue, n= 7).
For the other group, access to food was prevented 7h after dawn (open blue, n=9).Each point represents the mean ± S.E. A. Body temperature (Tb) recorded every 30 min for 24 h.Photoperiod was 12L:12D and the duration of darkness is shaded.The vertical line represents when access to food was prevented, at 13:00 under 12L:12D.Tb began to decrease after access to food was prevented.B. Changes in testicular volume between January and June.There were no significant differences during testicular maturation, but the birds with ad libitum food showed earlier regression.The vertical line represents when photoperiod reached 12:12D when Tb was recorded (A).C. The progress of moult in the two groups.Each point represent the date ± S.E. that each unit of moult score was reached by each bird.Birds with ad libitum food started to moult sooner.

Figure 5.
Mean body temperature in the four groups of starlings when photoperiod was 12L:12D: A Birds held at 6 ⁰C with food provided ad lib (n = 7); B birds held at 6 ⁰C with access to food restricted to 7h after dawn (n = 9); birds held at 16 ⁰C with food provided ad lib (n = 8) and birds held at 16 ⁰C with access to food restricted to 7h after dawn (n = 8).Each point represents the mean ± S.E.The differences were not significant.