Introduction

The proper timing of behavioral and physiological events is crucial for survival and reproduction, yet variation in biological timing systems is rarely studied in an evolutionary context. Circadian rhythms (~24 h) are controlled by an internal circadian clock, which allows organisms to anticipate daily changes in the environment and take advantage of favorable conditions. Although a general temporal program is prescribed by the internal clock, rhythmic activity is modulated by external signals such as the natural light–dark cycle, temperature changes, food availability, and social cues. Such stimuli not only influence behavior directly but also adjust the period of the circadian clock to match that of the environmental cycle. This results in a stable phase relationship between the entrained rhythm and the external cycle, which depends on the timing and intensity of the entrainment cue (Pohl 1976; Aschoff and Pohl 1978), the sensitivity of sensory pathways mediating entrainment (e.g., see Mrosovsky and Hattar 2005), and intrinsic parameters of the circadian clock itself (Pittendrigh 1981).

Given that entrainment cues vary extensively across habitats in type, quality, strength, and timing, the circadian system must be tailored to the local environment. This could be achieved through adaptations in circadian sensory mechanisms, in the same way that visual photoreception is adapted to the local sensory environment (Lythgoe 1979; Endler 1992). Alternatively, but not exclusively, selection could act on the internal circadian clock itself, making adjustments to properties of the oscillator that could confer an advantage in a particular environment. There is some evidence, mostly indirect, supporting both of these hypotheses (for review, see Menaker and Tosini 1996; Johnson 2005; Kyriacou et al. 2008; Peirson et al. 2009; Hut and Beersma 2010) but we lack a full picture of the functional, ecological, and evolutionary relationships between the circadian clock and environmental timing cues. The identification of general principles would both broaden and deepen our understanding of an important, intensively studied biological system that pervades nearly all aspects of behavior and physiology.

The daily light–dark cycle is one of the most important cues for circadian entrainment, and the actions of light and dark on the circadian system have been studied in depth (Pittendrigh and Daan 1976; Aschoff 1981b; Pittendrigh 1981; Golombek and Rosenstein 2009). Light modulates the period and phase of the circadian clock in an irradiance-dependent manner. The pathways by which light reaches the circadian system are distinct from those involved in image formation, and the main task of so-called “non-visual” photoreception is thought to be irradiance detection (Peirson et al. 2009). Not surprisingly, circadian responses to photic cues are known to vary, but few serious attempts have been made to identify ecological factors associated with this variation (but see Bertolucci et al. 2003). Across habitats, there is enormous variation in solar irradiance, but the relationship between habitat light and circadian biology has been largely unexplored.

In addition to light, temperature has a major influence on circadian timing, especially in insects and ectothermic vertebrates (Rensing and Ruoff 2002). In fact, in some organisms there is evidence to suggest that temperature cues are just as important, if not more so, than light (Hoffmann 1969; Underwood 1985; Rensing and Ruoff 2002). However, the mechanisms by which temperature influences the clock are not well understood, especially in vertebrates, and virtually nothing is known about the variation in thermal entrainment with respect to variation in thermal habitat. Due to the sun’s radiant energy, the temperature in a given habitat is often highly correlated with the amount of insolation it receives. In the current paper, we investigated the influence of both photic and thermal cues on circadian rhythms and compared these responses among diurnal lizard species whose microhabitats differ in both irradiance and temperature.

Anolis lizards are a good model system for exploring the correlation between photic habitat and the circadian system for a number of reasons. As a group, anoles have undergone extensive adaptive radiation and habitat specialization (Losos 2009). Their phylogenetic relationships are well known (Jackman et al. 1999; Nicholson et al. 2005) and indicate that a suite of morphological and behavioral adaptations have arisen independently due to selection pressure in specific microhabitats, giving rise to distinct “ecomorphs” (Williams 1972). In addition, anoles have diverged into multiple thermal and photic niches, which have been described in detail (Ruibal 1961; Rand 1964; Schoener 1974; Huey and Webster 1976; Roughgarden 1981; Williams 1983; Hertz 1992b; Fleishman et al. 1997), and anoles have been used as a model for both visual and thermal ecology (Gorman and Hillman 1977; Hertz 1992a, b; Fleishman et al. 1997, 2009; Fleishman and Persons 2001; Leal and Fleishman 2002; Leal 2004). Anoles are exclusively diurnal, allowing one to study the causes and consequences of photic niche divergence, independent of divergence in the temporal distribution of activity.

In anoles, the primary circadian clock is the pineal gland, a small organ located underneath the parietal region of the skull, dorsal to the brain. This organ plays a major role in the organization of locomotor activity rhythms (Menaker and Wisner 1983; Underwood 1983a, b; Hyde and Underwood 2000) and contains all the components of a circadian system (sensory pathways for entrainment and a circadian oscillator coupled to an output signal), each of which could potentially evolve in response to selection for optimal circadian entrainment in specific habitats. The pineal synthesizes, in a rhythmic fashion, the hormone melatonin, which serves as a timing signal for the rest of the body. The pineal is directly photo- and thermosensitive, and the melatonin rhythm can be entrained to light–dark cycle as well as thermocycles (Underwood 1985; Underwood and Calaban 1987). Melatonin is not stored in the pineal gland, but is released as it is synthesized, and thus reflects the state of the oscillator itself. Importantly, the melatonin rhythm persists in isolated pineal tissue culture (Menaker and Wisner 1983), making it amenable to experimental manipulation.

The process of entrainment can be characterized quantitatively in the laboratory by administering short “pulses” of the synchronizing agent and measuring the subsequent change in the phase of a free-running circadian rhythm. In this study, we quantified phase resetting of the pineal oscillator in response to light and temperature pulses by measuring melatonin produced by the gland in tissue culture. Anolis species were chosen in order to find out if photic and/or thermal resetting is correlated with photic habitat, while controlling for phylogenetic structure and general habitat differences. Specifically, we tested the hypothesis that species living in more shaded environments would show larger phase shifts in response to light and temperature pulses. Such a correlation would suggest that the perception of light and temperature cues by the circadian system is adapted to the photo-thermal environment.

Materials and methods

Species, phylogeny, and habitats

The phylogenetic relationships of the species used in this study are depicted in Fig. 1. In anoles, habitats are partitioned primarily according to features of the vegetation such as perch height and diameter. Morphological and behavioral adaptations to perch structure have led to the evolution of distinct ecomorphs, named after their preferred perch (see Fig. 1). Phase shifts in pineal melatonin have not been studied previously in these species, so we do not know to what extent they vary or if there is any phylogenetic signal in this trait or association with ecomorph differences. Therefore, these five species cannot be treated as independent populations, but rather, phylogenetic architecture must be considered. To this end, we focused on intra-ecomorph comparisons of photic habitat: trunk-ground species A. cristatellus and A. gundlachi were compared to each other, as were grass-bush species A. krugi and A. pulchellus. Photic habitat and ecomorph classifications are shown in Fig. 1. A. carolinensis, the only previously studied Anolis species, was used as the “outgroup,” that is, we assumed that it represents the ancestral condition.

Fig. 1
figure 1

Phylogeny, ecomorph classifications, and photic habitats of Puerto Rican Anolis. Cladogram was constructed by pruning the phylogeny in Nicholson et al. (2005) and depicts the phylogenetic relationships between a subset of Puerto Rican anole species (branch lengths are not quantitative). The five species included in the current study are highlighted with bold underlined text. Photic habitats are based on descriptions in Loew et al. (2002) and Fleishman et al. (1997)

Full size image

Animal collection and housing

Adult male anoles were captured from Puerto Rico in January–February 2008 and 2010 and May–June 2010 and transported (within 3 days of capture) to the University of Virginia, where all experiments were conducted. Adult male A. carolinensis were obtained from a commercial supplier (Carolina Biologicals, Inc.). Lizards were housed individually in plastic cages equipped with a perch and a carpet floor substrate. Room temperature was maintained at 27.8 ± 0.5°C (A. cristatellus, A. krugi, A. pulchellus, A. carolinensis) or 24.4 ± 0.5°C (A. gundlachi, whose thermal optimum is lower than the other species). For all species, relative humidity was kept at 60 ± 10%. A 12:12 h light:dark (LD) cycle was provided by overhead fluorescent room lighting, as well as UV-enriched fluorescent bulbs (Repti-Sun 10.0 UVB, Zoo Med Laboratories, Inc.) suspended from the top of the cage (~400 μW/cm2). Lizards were misted with water 1–2 times each day and were fed live crickets (gut loaded and dusted with vitamin powder) 2–3 times per week. Water and food were provided during the light period at varying times of day in order to avoid circadian entrainment to these stimuli.

Flow-through tissue culture

In vitro experiments were conducted between June 2009 and April 2011. Pineal glands were dissected and cultured individually in a flow-through culture system as described previously (Moore and Menaker 2011). The culture apparatus allowed for up to six pineals to be cultured simultaneously; multiple culture trials of n ≤ 6 were performed with an arbitrarily chosen mix of species in each trial (N = 168; sample sizes for each species and treatment group are shown in Figs. 2, 3 and Table 1). Control and treatment cultures were performed separately, in an alternating fashion, using the same apparatus. Dissection took place under standard room lighting 2–4 h before lights-off. Pineal glands were washed in Hank’s balanced salt solution and placed into culture wells containing Gibco Medium 199 (with Hank’s salts, l-glutamine, 25 mM HEPES buffer, and supplemented with penicillin/streptomycin antibiotics) at room temperature. Immediately following dissection (≤1 h for six animals), glands were placed into a light-tight, temperature-controlled (±0.1°C) culture chamber equipped with flow-through tubing primed with oxygenated culture medium. Culture medium was continuously pumped through the culture wells at a rate of 1 mL/h. A fraction collector was used to automate the collection of regularly timed samples of culture medium perfusate (one sample every 2 h). Samples were stored at −20°C until radioimmunoassay.

Fig. 2
figure 2

Effect of light pulse treatment on the pineal melatonin rhythm of five Anolis species. Mean melatonin concentrations (in pg/ml) for each treatment group (solid black circles control groups; unfilled circles light pulse groups) are plotted over the duration of the culture period (~4 days). Time units are in h, with ZT hours used for the first two cycles (light:dark cycle), and cumulative hours in DD used for the remaining two cycles (constant darkness). The lighting protocol is illustrated at the top of each melatonin profile; temperature was held constant at 32°C. Black bars represent periods of darkness, light bars represent periods of light, and gray bars represent dark periods corresponding to the photophase of the previous LD cycle. The 2-h period of light during the second dark period represents the light pulse administered to the light pulse treatment group; note that control groups remained in uninterrupted darkness during this time. Error bars depict standard error of the mean (SEM); sample sizes for each group are summarized in Table 1. Following the light pulse treatment, the pineal melatonin rhythm is clearly phase delayed (shifted to the right) in A. gundlachi and disrupted in A. krugi (see text for details)

Full size image
Fig. 3
figure 3

Effect of warm temperature pulse treatment on the pineal melatonin rhythm of five Anolis species. Mean melatonin concentrations (in pg/ml) for each treatment group (solid black circles control groups; unfilled circles temperature pulse groups) are plotted over the duration of the culture period (~4 days). Time units are in h, with ZT hours used for the first two cycles (32:27°C thermocycle), and cumulative hours in constant conditions used for the remaining two cycles (continuous 27°C). The temperature protocol is illustrated by the dashed line at the top of each melatonin profile, showing the actual variation in culture temperature (from maximum 32 to minimum 27°C) over time. The nocturnal rise in temperature during the second “night” (cool period) represents the warm temperature pulse administered to the temperature pulse treatment group; note that control groups remained in uninterrupted 27°C during this time. Error bars depict standard error of the mean (SEM); sample sizes for each group are summarized in Table 1. From these profiles it can be clearly seen that the warm temperature pulse had very little effect on the pineal melatonin rhythm in any of the species tested

Full size image
Table 1 Summary of circadian parameters obtained by fitting a sine function to individual free-running pineal melatonin rhythms
Full size table
Table 2 Summary of circadian phase analysis and circular statistics
Full size table

The lighting and temperature conditions in the culture chamber varied between experiments (see Figs. 2, 3). Light was provided by three white LED bulbs (~800 μW/cm2, positioned 4 cm from culture wells). The spectral distribution of the bulb is fairly broad, characterized by a narrow (~50 nm) peak at 450 nm (peak radiant flux ~1.2 mW/nm) and a broader (~200 nm) peak centered at ~600 nm (~1.6 mW/nm); total radiant flux is ~130 mW. In the first experiment, we examined the effect of a nocturnal light pulse on the melatonin rhythm. For these cultures, the chamber was kept at constant temperature (32°C). Pineals were initially exposed to a light–dark cycle, with the phase of lights on/off matching that of previous housing conditions. Following the light–dark cycle, the pineals were either kept in constant darkness for three cycles (control group) or given a 2-h light pulse during the night (beginning 2 h after the last lights-off) of the second cycle (light pulse group). Light given at this time (early subjective night) is expected to cause a phase delay (Pittendrigh and Daan 1976). For the temperature pulse experiment, glands were given a thermocycle of 32/27°C (with the warm period corresponding to subjective day) for one cycle, after which the culture chamber was either maintained at 27°C (control group) or given a ~2-h warm “pulse” (32°C) during the subjective night of the second cycle (temperature pulse group). Because of the time required for the culture chamber to transition between temperatures, the warm temperature treatment lasted longer than 2 h. The transition to the warmer temperature began 2 hours after the last transition to 27°C, i.e., at ZT15. The warmer temperature (32°C) was reached after about 1 h. At ZT17, the transition to the cooler temperature was initiated and was completed ~3 h later (see Fig. 3). For the temperature pulse experiment, glands remained in constant darkness following a brief period of light on the first day of culture.

Radioimmunoassay

Samples of culture medium perfusate were assayed in duplicate for melatonin content using a modification (Moore and Menaker 2011) of the method described in Rollag and Niswender (1976). Radiolabeled melatonin (125I–melatonin) was prepared by reacting Bolton-Hunter Reagent (Perkin-Elmer Life Sciences) with 5-methoxytryptamine. Aliquots of perfusate samples were diluted into assay buffer (phosphate buffered saline with 0.1% gelatin and 100 mg/L thimerosal, pH 7.4) to a final volume of 100 uL and incubated with antibody (R1055, ~1:25,000 final concentration in 50 uL PBS-gel) and radiolabeled ligand (~25,000–50,000 cpm in 50 uL PBS-gel) for 48–72 h at 0°C. Following ethanol (95%, 4°C) precipitation and centrifugation, bound radioactive precipitate was quantified with a gamma counter (Apex, Micromedic Systems, Inc.). Melatonin content was determined using a standard curve. This procedure has been previously validated (Moore and Menaker 2011) for unextracted culture medium perfusate for three of the species used in the current study (A. carolinensis, A. cristatellus, A. gundlachi). In addition, we validated the procedure for A. pulchellus and A. krugi by evaluating the culture perfusate for parallelism of antibody reactivity and quantitative recovery of added melatonin. For both species, the culture perfusate (twofold geometric series; 40 to 2.5 μl) reacted with the antibody in parallel with a serial dilution of melatonin standards (twofold geometric series; 250 to 0.488 pg). In addition, melatonin standards (0, 25, 50, 75, and 100 pg) added to the media/perfusate were quantitatively recovered.

Data analyses

For visualization purposes, group mean melatonin concentrations (in pg/ml) were plotted as a function of time (“ZT hours” for the entrained condition, “hours in constant conditions” for the remainder, beginning at 0 h) for the duration of the culture, as shown in Figs. 2 and 3. For the analysis, we were mostly interested in melatonin concentrations during the last two cycles, following treatment with light or temperature, compared to respective control conditions. We plotted this portion of the melatonin profile, comprising 26 timepoints over 52 h, for each individual pineal. We employed a custom-written macro in Microsoft Excel to perform sine wave analysis on the melatonin time series data, following a procedure similar to the one described in Lincoln et al. (2006). Briefly, a least-squares approach for non-linear regression was used to fit the following sine function to each individual melatonin profile:

$$ y\left( t \right) \, = l + mt + (a + bt){ \sin }\left[ { 2\pi \, \left( {t - \varphi } \right) \, / \, \tau } \right] $$

where: y is the melatonin concentration in pg/ml, t is time in h, l is the level of the function in pg/ml, m is the linear trend, a is the amplitude of the waveform (deviation in pg/ml), b is linear damping, φ is the phase (in h) corresponding to the steepest rise of the sine function and τ is the period of the rhythm in h. The “solver” algorithm in Excel was used to find parameter values for l, m, a, b, φ, and τ that minimized the residual sum of squares, given a set of initial parameter estimates, and subject to the following constraints: the amplitude term (a + bt) at any given timepoint t must be greater than or equal to 1, and the period τ must be between 18 and 30 h (to restrict estimates to the circadian range). Goodness-of-fit was evaluated using an F test (null model: linear regression). The final sine model was chosen by performing the regression on a randomly selected subset (n = 10) of melatonin profiles from control groups and, in a stepwise fashion, adding linear and parabolic terms for level trend and damping to the basic sine wave function. The fits were highly significant (p < 0.001) upon inclusion of linear trend and linear damping terms (mt and bt) and not greatly improved with the inclusion of parabolic terms or any terms allowing for change in τ over time. Thus, the final model included the minimum number of parameters necessary to provide a good description of the data. This model was then used to fit sine curves to all the remaining time series. Melatonin profiles were considered rhythmic if the p value was ≤0.05. Using this criterion, some melatonin profiles were found to be arrhythmic and were excluded from further analyses (see Table 1). Examples of curve-fitted melatonin profiles are shown in Online Resource 1.

To determine differences in period (τ) among species and treatment groups, we employed two-way ANOVA (SAS, version 9.1.2) with ‘species’ and ‘treatment’ as main effects. Linear regression was used to calculate Q 10 values for the change in τ in control groups at 32 and 27°C (light and temperature experiments, respectively). To obtain circadian phase from parameter estimates, φ was normalized to τ and converted into circadian-time (CT) hours (circadian phase = φ/τ × 24), determined separately for each individual. Mean vectors of circadian phase angles were plotted in circular phase plots (see Fig. 5 and Online Resource 2) and circular statistics were performed using the software program Oriana (version 3, Kovach Computing Services). Significant phase clustering was determined by applying the Rayleigh uniformity test, and differences in phase between pulse and control groups were determined by the Watson-Williams F test (α = 0.05).

Results

Light pulse experiment

Mean melatonin profiles are shown in Fig. 2. Most individual pineals displayed significant melatonin rhythms under constant conditions (32°C, constant darkness) in both control and treatment groups. Table 1 summarizes the parameter estimates obtained from the sine curve-fitting procedure. Free-running period (τ) was ~23 h for all groups; there were no significant differences in τ among species, nor was there any effect of the light pulse treatment on τ (two-way ANOVA; species p = 0.2487; treatment p = 0.9548).

Trunk-ground species tended to have higher-amplitude rhythms than grass-bush species (Table 1; note also the scale of the y-axis in Fig. 2). Amplitude was lowest in A. krugi and A. carolinensis. In A. krugi, the rhythm appeared to damp in constant conditions, as reflected by higher p values for the curve fits (Table 1), but this was not the case in A. carolinensis. In addition, there was substantial variation in the melatonin profile among A. krugi individuals (see error bars in Fig. 2), especially in the light pulse group. Most of this variation is accounted for by differences in phase, as discussed below.

Circular plots of circadian phase angles are depicted in Fig. 4. All groups showed highly significant clustering of phase angles except for light pulse-treated A. krugi, which were not significantly clustered (Rayleigh uniformity test, p = 0.147). Comparing control and treatment groups within species, it was found that the light pulse treatment caused a phase delay in all species, but this effect was significant only in A. gundlachi (Watson-Williams F test, p = 0.000195) and in A. krugi (p = 0.000195). Regarding the latter, since phases were not significantly clustered in the light pulse-treated group, it was not possible to ascertain the “true” mean phase shift (see “Discussion”). We also found significant differences in phase among species (in control groups). The pairwise comparisons of greatest interest to us were those between photic niche variants of the same ecomorph category; of these comparisons, A. cristatellus and A. gundlachi differed significantly (p = 0.008), whereas A. pulchellus and A. krugi did not (p = 0.086).

Fig. 4
figure 4

Rayleigh plots illustrating the effect of light pulse treatment on circadian phase of the pineal melatonin rhythm. Circular plots represent one complete circadian cycle of 24 circadian hours (CT). Phases of individual pineal cultures (in CT h) are represented by small circles on the perimeter of the circular plots (black control pineals; gray light pulse-treated pineals). The arrows represent mean vectors of the circular distributions for each group; the angle of the arrow indicates the mean phase while the length of the arrow is proportional to the phase clustering (a longer arrow reflects more tightly clustered individual phases). The light pulse treatment caused a delay in the mean phase of the pineal melatonin rhythm, and the magnitude of the phase shift varied among species. In A. krugi, the light pulse treatment caused large phase shifts in some individuals as well as increased phase dispersion among individuals. Sample sizes and values for phase, vector length r, and circular statistics are given in Table 2

Full size image

Temperature pulse experiment

Mean melatonin profiles are shown in Fig. 3; circadian parameter estimates are summarized in Table 1. Compared to the light pulse experiment, fewer individuals displayed robust rhythms in constant conditions (27°C, constant darkness), even in control groups (see Table 1). Note that the temperature was 5° lower than that used for the light pulse experiment. Free-running period (τ) differed among species (two-way ANOVA, p = 0.0006), but was not affected by the temperature pulse treatment (p = 0.6247), nor was there a significant species × treatment interaction (p = 0.6575). We specified post hoc comparisons to determine if, in the control groups, τ differed between species belonging to the same ecomorph category, or between A. carolinensis and the Puerto Rican species. These contrasts revealed that there were no differences in τ within ecomorph categories (A. cristatellus vs. A. gundlachi, p = 0.0714; A. krugi vs. A. pulchellus, p = 0.2028), but that A. carolinensis differed from the other species (p = 0.0391). In A. carolinensis and A. gundlachi, peak melatonin levels on the first day of culture were higher compared with the light experiment. However, amplitude in constant conditions was similar between the two experiments, in all species.

Circular plots of circadian phase angles are depicted in Online Resource 2. All groups showed highly significant clustering of phase angles except temperature pulse-treated A. krugi (Rayleigh uniformity test, p = 0.058). The temperature pulse treatment did not cause a significant change in phase in any species (all p values > 0.05, Watson-Williams F tests), although there was a trend for phase delays. In control groups, within ecomorph categories, there were no differences in phase.

Temperature compensation

Melatonin rhythms were temperature compensated in all species, with only small changes in τ across 5°C temperature range (Fig. 5). In all species, τ lengthened under 27°C (temperature experiment) as compared to 32°C (light experiment), but the differences in τ are minor, indicating that the observed oscillations are driven by a temperature-compensated circadian clock. A. gundlachi showed the smallest change in τ across this temperature range, while A. carolinensis showed the greatest change. Q 10 values were determined from the slope of the linear regression of τ versus temperature: A. carolinensis, 1.247; A. cristatellus, 1.076; A. gundlachi, 1.027; A. krugi, 1.173; and A. pulchellus, 1.115.

Fig. 5
figure 5

Temperature compensation of the Anolis pineal melatonin rhythm. Estimates of free-running period (τ) from control pineals are plotted according to the culture temperature during constant conditions (32°C for light pulse experiment and 27°C for temperature pulse experiment). Linear regression (solid line) was used to obtain the Q10 values for the change in rate (period) of the rhythm at different culture temperatures; in all species, this value was close to 1, indicating that the pineal melatonin rhythms are temperature compensated

Full size image

Discussion

This study demonstrates a correlation between photic habitat and the photic control of the pineal clock. We found that, within pairs of ecomorphologically similar Anolis species, sensitivity to light-induced phase shifts of the pineal melatonin rhythm is positively correlated with the amount of shade in the microhabitat. The inclusion of two different pairs of species provides replication of the photic niche comparison, which increases the likelihood that the observed differences are due to divergence in photic habitat rather than some other species difference. Because phase resetting by light reflects the primary mechanism by which circadian clocks are synchronized to external cues, our findings support the hypothesis that the circadian system is adapted to the photic environment. Of course, caution should be used when generalizing results from in vitro experiments, as the conditions are often unavoidably artificial. For example, the intensity of light used in our experiments, although comparable to sunlight levels, would be attenuated by the skull and other overlying tissues before reaching the pineal in an intact animal.

The light pulse treatment caused a phase delay of ~3 circadian h in A. gundlachi, a trunk-ground species occupying heavily shaded habitat, whereas the more heliophilic A. cristatellus showed a smaller phase shift (~1.25 CT h) that was not statistically significant. Because these two species are very similar in ecology, behavior, and morphology, the variation in responsiveness to light probably reflects their different photic habitats. The comparison of grass-bush species A. krugi (partial shade) and A. pulchellus (full sun) provides additional support for this view, although in a slightly different way: while the mean phase delay in A. krugi was quite large (~7.5 CT h), phase clustering in the treatment group was not significant. Thus, it is difficult to distinguish the phase-shifting effect of the light pulse from a more general “phase-disruptive” effect. Since some individuals in the treatment groups (both A. krugi and A. pulchellus) were excluded from the dataset based on their lack of rhythmicity, it was important to determine if the observed phase disruption in A. krugi was due to “poor” rhythmicity. There was no apparent correlation between the p values of the curve-fitting procedure and phase estimates; i.e., individuals with highly significant p values (p < 0.00001) were widely dispersed in phase, as were the two individuals with the lowest p values (p = 0.0055 and 0.001). Therefore, we conclude that the light pulse treatment had a dramatic phase-dispersing effect, in combination with a possible phase resetting effect, in A. krugi. Phase clustering was unaffected in A. pulchellus, which had only a small phase shift (~0.75 CT h, n.s.). As with the comparison between A. cristatellus and A. gundlachi, the ecomorphological similarity of A. krugi and A. pulchellus implies that photic habitat is the most likely explanation for the observed differences in the response to light pulse treatment.

In anoles, photic niche divergence tends to occur at the tips of the phylogeny, following habitat specialization that leads to characteristic ecomorphs. Because A. carolinensis is not known to be a photic habitat specialist (in fact, it is successful in a variety of photic habitats; see Macedonia et al. 2003), we make the assumption that this species represents the ancestral phenotype. In A. carolinensis, the light pulse treatment had no effect on phase (~46 min phase shift, n.s.), and no discernable acute effect on melatonin level, the latter being consistent with several earlier reports (Underwood 1986; Moore and Menaker 2011). Thus, an increased sensitivity to light appears to be derived in some species, perhaps to provide an adaptive advantage in more shaded habitat. Photic niche is not the only factor influencing melatonin rhythms and their response to light, as evidenced by the fact that A. cristatellus and A. krugi occupy similar photic niches, but show different responses. This underscores the need to account for phylogenetic structure in comparative studies, as well as general differences in behavior and morphology (such as those related to ecomorph categories). Indeed, some features of melatonin rhythmicity seem to be primarily associated with phylogenetic and/or ecomorph differences, such as rhythm amplitude and robustness (persistence in constant conditions). Trunk-ground species have high amplitude rhythms, and light pulse treatment tends to slightly increase their robustness. Grass-bush species have lower-amplitude rhythms that appear to be damping in constant conditions. A. carolinensis is somewhat intermediate, with robust low-amplitude rhythms.

It is beyond the scope of the current study to identify the mechanisms responsible for the observed differences in photic phase resetting, but we can identify two main possibilities: (1) variation in pineal photoreception, and (2) variation in the pineal oscillator. (Other possibilities include a combination of (1) and (2), or variation in coupling between photoreception and the clock). Of course, there is no reason to expect that the same mechanism accounts for the differences between both pairs of photic niche variants. Our experiments do not provide a test for the photoreception explanation, but we did uncover some differences in oscillator properties that may be responsible, at least partially, for the differences in phase resetting.

The differences between the responses of A. gundlachi and A. cristatellus could be due to differences in phase of the pineal clock at the time of light pulse administration. It is well known that phase shift magnitude depends on oscillator phase (reviewed in Pittendrigh 1981); this relationship can be plotted as a “phase-response curve,” the properties of which are species specific and vary according to the type, strength, and duration of the resetting stimulus. Phase estimates in control groups reveal that the clock in A. gundlachi was at a significantly earlier phase compared to A. cristatellus by ~1.75 CT h. Thus, even though the pulse was given at the same external time, in A. gundlachi the pulse occurred at a later internal oscillator phase. The difference in underlying oscillator phase might be sufficient to explain the observed differences in photic resetting, although it would be necessary to obtain a complete phase-response curve for each species in order to test this hypothesis. Furthermore, it would be interesting to find out if species differ in pineal oscillator phase under natural conditions, as this would be an elegant mechanism by which the sensitivity of the circadian system could be adapted to habitat light.

For grass-bush species, we did not find any differences in oscillator phase, but differences in oscillator “robustness” (the extent to which an oscillator is self-sustaining) could explain the results. As can be clearly seen in the melatonin profile for the control groups (Fig. 2), the rhythm in A. krugi is much less robust than that in A. pulchellus; the amplitude is lower in A. krugi and the rhythm damps in constant conditions. Compared to self-sustained oscillators, damped oscillators can be entrained to relatively weak cues (e.g., low-amplitude light cycles; Hau and Gwinner 1994) across a wider range of periods (Aschoff 1981a); this could result, in part, from larger phase shifts in response to entrainment stimuli. However, inadequate nutrients, oxygen, or temperature can cause damping in tissue culture, so it would be necessary to test each species under a variety of culture conditions in order to exclude this possibility. It is also possible that phase shifting in response to light is temperature dependent, and the nature and extent of temperature dependency could vary among species.

In contrast to our findings from the light pulse experiment, we did not observe significant thermal resetting in any of the species studied, precluding any attempt to correlate response to temperature pulses with thermal habitat. This was somewhat surprising and differs from findings in chick pineal cell culture, in which warm temperature pulses cause acute changes in melatonin levels as well as significant phase shifts of the rhythm (Barrett and Takahashi 1995). We did, however, observe small phase delays, even though they were not significant, and the differences among species parallel the results from the light pulse experiment. That is, the phase shift in A. gundlachi (~1.25 CT h) was larger than A. cristatellus (~0.75 CT h), and A. krugi had a larger shift (~1 CT h) than A. pulchellus (~0.25 CT h). Thus, it appears that thermal resetting (to the extent that it occurs) might be correlated with photo-thermal niche, but our protocol may not have been sufficient to induce large changes in phase. The temperatures used for this experiment were chosen for two reasons: (1) both temperatures (27 and 32°C) are known to sustain robust rhythms in A. carolinensis pineal culture (Menaker and Wisner 1983), and (2) these temperatures reflect typical day/night temperature cycles in Puerto Rico, at least during part of the year, although both higher- and lower-amplitude temperature cycles also occur. Recently, it was found that thermal entrainment of locomotor activity rhythms in Drosophila occurs slowly, requiring several days (Currie et al. 2009), in contrast to the rapid resetting (in response to 6-h temperature pulses) observed in chick pineal (Barrett and Takahashi 1995). It has been proposed that, in ectotherms, gradual entrainment to temperature cues would be advantageous in nature, as it would ensure that phase changes are not caused by “anecdotal” fluctuations in temperature (Currie et al. 2009). If this is also the case in lizards, a pulse-resetting paradigm may not be adequate for characterizing effects of temperature on the circadian clock. Even so, it is surprising that we did not find any notable effects (not even transient ones) of temperature on the level, amplitude, or phase of the melatonin profile. It may be that temperature effects are light dependent; in flies, constant darkness interferes with thermal resetting (Glaser and Stanewsky 2007).

In summary, photic resetting of the pineal oscillator is correlated with photic habitat in Puerto Rican Anolis lizards. Variation in phase resetting is not due to differences in free-running period (τ) of the pineal oscillator, but may be due to differences in oscillator phase and/or robustness. Regardless of the mechanism, our findings clearly demonstrate species variation in photic resetting of the pineal oscillator and suggest an adaptive relationship between the circadian system and habitat light.