Temperature and ontogenetic effects on color change in the larval ...

Color profile: Disabled Composite Default screen

710

Temperature and ontogenetic effects on color change in the larval salamander species Ambystoma barbouri and Ambystoma texanum T.S. Garcia, R. Straus, and A. Sih

Abstract: Temperature has been shown to affect body color in several species of amphibians. The interaction between color and temperature may also change over larval ontogeny, perhaps because of age-related or seasonal changes in selection pressures on color. We quantified the effects of temperature on the color of the salamander sister species Ambystoma barbouri and Ambystoma texanum over larval ontogeny. We found that early-stage larvae responded to cold temperatures with a dark color relative to that of the warm temperature response. Both species then exhibited an ontogenetic shift in larval color, with larvae becoming lighter with age. Interestingly, older larvae showed decreased plasticity in color change to temperature when compared with younger stages. Older A. barbouri larvae showed no color response to the two temperature treatments, whereas older A. texanum larvae exhibited a reversal in the direction of color change, with cold temperatures inducing a lighter color relative to warm temperatures. We suggest that the overall pattern of color change (a plastic color response to temperature for young larvae, a progressive lightening of larvae over development, and an apparent loss of color plasticity to temperature over ontogeny) can be plausibly explained by seasonal changes in environmental factors (temperature, ultraviolet radiation) selecting for body color. Résumé : La température affecte la coloration du corps chez plusieurs espèces d’amphibiens. L’interaction entre la coloration et la température peut aussi se modifier pendant l’ontogénèse larvaire, peut-être à cause de changements reliés à l’âge ou à la saison dans les pressions de sélection qui agissent sur la coloration. Nous avons quantifié les effets de la température sur la coloration de deux espèces soeurs de salamandres, Ambystoma barbouri et Ambystoma texanum, au cours de leur ontogénèse larvaire. Les larves des premiers stades réagissent aux températures froides en adoptant une coloration plus foncée que par température chaude. Par la suite, les salamandres des deux espèces subissent un revirement ontogénique par lequel les larves deviennent plus pâles en vieillissant. Étonnamment, les larves plus âgées font preuve d’une diminution de plasticité dans leurs changements de coloration en réaction à la température. Les larves plus âgées d’A. barbouri n’ont pas réagi à deux traitements thermiques, alors que chez les larves plus âgées d’A. texanum, il y a eu revirement dans la direction des changements de couleur et les températures froides ont provoqué l’apparition d’une coloration plus pâle que celle obtenue aux températures chaudes. Nous croyons que, dans l’ensemble, les patterns de changements de couleur (réaction de la coloration à la température chez les jeunes larves, atténuation progressive de la couleur des larves au cours du développement et perte apparente de la plasticité de la coloration en réaction à la température pendant l’ontogénèse) peuvent s’expliquer logiquement par les changements saisonniers des facteurs environnementaux (température, radiation ultraviolette) qui opèrent une pression de sélection sur la coloration. [Traduit par la Rédaction]

Garcia et al.

715

Introduction Color change in larval amphibians occurs in response to several environmental cues, including temperature, background color, ultraviolet radiation, and stress (Bagnara and Hadley 1973; Carey 1978; Hoppe 1979; Kats and Van Dragt 1986; King et al. 1994). Body color in aquatic larval amphibians mediates thermoregulation, increases crypsis, and possibly reduces exposure to ultraviolet radiation (UVR) by screening out harmful wavelengths with dark pigments (Endler 1988; Garcia 2002; Garcia et al. 2003a). However, multiple envi-

ronmental factors can exact conflicting selection pressures on color. For example, dark body color may be appropriate for screening UVR or for thermoregulation but could increase conspicuousness to predators (Garcia et al. 2003a). Color response over two time scales (i.e., an immediate behavioral time scale and an ontogenetic time scale) may help mediate the multiple selection pressures acting on color in aquatic environments. Although temperature influences body color in amphibians, it is also a major factor in determining larval growth rates, development rates, time until metamorphosis, and be-

Received 10 September 2002. Accepted 29 January 2003. Published on the NRC Research Press Web site at http://cjz.nrc.ca on 2 May 2003. T.S. Garcia,1,2 R. Straus, and A. Sih.3 Department of Biological Sciences, University of Kentucky, Lexington, KY 40506, U.S.A. 1

Corresponding author (e-mail: [email protected]). Present address: Zoology Department, Oregon State University, Corvallis, OR 97331, U.S.A. 3 Present address: Environmental Science and Policy Department, University of California, Davis, CA 95616, U.S.A. 2

Can. J. Zool. 81: 710–715 (2003)

J:\cjz\cjz8104\Z03-036.vp Tuesday, April 29, 2003 10:50:12 AM

doi: 10.1139/Z03-036

© 2003 NRC Canada


Garcia et al.

havior (Wilbur and Collins 1973; Smith-Gill and Berven 1979). Behavioral responses to temperature, such as larval aggregation in warm areas and avoidance of highly variable thermal regions, illustrate the importance of temperature selection to larval fitness (Brattstrom 1962; Navas 1996). Temperature can also have indirect effects on larval amphibians. For example, in predator–prey interactions, warm temperatures increase larval growth rates, creating a size refuge from gape-limited predators (Brodie and Formanowicz 1983; Anderson et al. 2001). Temperature can also indirectly affect mortality, because warm temperatures facilitate drying of ephemeral habitats (Laurila and Kujasalo 1999). Color change as a function of temperature has been observed in several amphibian species. Cold temperatures trigger the dispersal of dark-colored pigment (melanosomes) within color cells (melanophores), causing an overall darkening of the skin (Duellman and Trueb 1986). If color change is a fixed response to temperature, then individuals may have limited ability to modulate color in response to other environmental changes (i.e., predation risk, UVR exposure). For example, Hyla crucifer and Hyla cinerea both darken in cold temperatures and lighten when conditions get warmer. However, neither species effectively changes color to background match against light-colored substrates when temperatures are cold (Kats and Van Dragt 1986; King et al. 1994). Thus, although dark coloring in response to cold temperatures may be a good thermoregulatory defense mechanism against freezing, it can cause larvae to become less cryptic with their background. Although not well studied, body color in salamanders has also been shown to change over ontogeny (Fernandez and Collins 1988). Larval color change over ontogeny could be a response to seasonal variation in selection pressures. For example, if thermoregulatory concerns determine larval color, we predict that as temperatures increase from spring to summer, larvae will get lighter with age. Ultraviolet radiation is another environmental factor showing seasonal variation. If UVR exposure is responsible for a darker color in earlystage larvae (Garcia et al. 2003a), seasonal growth of the UV-filtering overhead canopy may allow for lighter color in late-stage larvae. This study examines larval color response to temperature over ontogeny in two sister species of salamander. We predict a color response to temperature and a species difference in how larvae respond to temperature over ontogeny. Although closely related, these larval salamanders inhabit two distinct aquatic environments with different degrees of predation risk, UVR exposure, and ephemerality. In both habitats, for both species, selection pressures influencing color vary over the course of larval development. Thus, we quantified the effects of temperature on color during early and late larval stages and how larval color changes as a function of ontogeny.

System The smallmouth salamander, Ambystoma texanum, was once classified as having two forms: the pond form, which breeds in ephemeral ponds and is common in much of the eastern United States, and the stream form, confined mostly to cen4

711

tral Kentucky ephemeral streams (Petranka et al. 1982, 1987). Evidence suggests that an A. texanum like ancestor invaded streams and evolved into the streamside salamander, Ambystoma barbouri (Kraus and Petranka 1989). This move into streams resulted in exposure to novel selection pressures, in particular, increased habitat ephemerality, new visual predators, and greater exposure to UVR (Petranka 1983; Petranka and Sih 1987; Sih et al. 1992, 2000, 2003; Garcia 2002; Garcia et al. 2003a). As a result, A. barbouri have adapted with increased activity, feeding, and development rates relative to A. texanum, along with a difference in mean body color (Petranka and Sih 1987; Maurer and Sih 1996; Garcia 2002; Garcia and Sih4). Ambystoma barbouri larvae are significantly darker than larvae of their sister species, A. texanum, and both species respond to varying background colors with cryptic, backgroundmatching color change (Garcia et al. 2003b). The difference in mean color is presumably due to differences between habitats in selection pressures, such as UVR exposure and predation risk. Temperature, however, also affects color, thus limiting larval ability to respond to such environmental factors. Water temperature for both streams and ponds in late winter – early spring is relatively cold, ranging from 5 to 12°C. Limitations on color change because of cold temperature may only affect early-stage larvae; later in development, seasonal warming of water temperature may reduce temperature constraints on color. Ponds and streams in Kentucky receive little shade from canopy cover until middle to late spring (T.S. Garcia, personal observervation). This exposes early-stage A. texanum and A. barbouri larvae to direct sunlight and full, ambient UVR. Previous work has shown that both species get darker when exposed to ambient levels of UVR (Garcia 2002; Garcia et al. 2003a). Ponds, however, have relatively murkier, deeper waters and darker substrates than streams, thus offering more protection from UVR. The fact that early-stage larvae appear to be dark in nature could be a response to either cool temperature or high UVR. In this experiment, we quantify larval color response to temperature over ontogeny to extract the role of temperature in determining larval color. The degree to which temperature affects color change has potentially important impacts on larval ability to avoid predators and to mediate UVR exposure. Previous studies have shown A. texanum to be lighter than A. barbouri (Garcia and Sih4). We predict that under identical temperature treatments, this relationship will continue. Thus, we expect A. texanum to darken in cold temperatures but to maintain an overall lighter mean color relative to their sister species. In addition, as larvae mature, we predict both species will lighten as a result of the evolutionary influence of seasonal increases in temperature and reductions in UVR exposure because of a growing canopy over development. Later-stage larvae may show limited ability to plastically change color in response to temperature, although we expect the direction of color change to stay the same.

Methods We collected newly hatched A. barbouri larvae from streams in Raven Run Nature Preserve, Fayette County, Kentucky, and

T.S. Garcia and A. Sih. Species differences in color and color change in the sister species Ambystoma barbouri and A. texanum. Unpublished. © 2003 NRC Canada



712

Wildcat Creek, Anderson County, Kentucky. Ten A. texanum egg clutches were collected from both Beaver Dam, Kentucky, and Livingston County, Kentucky, and reared to early larval stages to ensure correct identification. Both species were held before the experiment in an environmental chamber at the University of Kentucky, Lexington, with a naturally relevant temperature and photoperiod (15°C, 14 h light : 10 h dark). Individuals were held in single-species groups of 20 larvae and fed macroinvertebrates ad libitum. We have complied with all of University of Kentucky Animal Care guidelines and Canadian Council on Animal Care guidelines and obtained all required state and federal permits for the completion of this study. Forty early-stage A. barbouri larvae and 40 early-stage A. texanum larvae (0.1–0.2 g for both species, 1 week after hatching) were randomly placed into two incubation chambers (20 individuals/species per chamber) in individual 1-L Mason glass jars filled with 500 mL of filtered, aerated tap water. Both incubators were set on a 14 h light : 10 h dark photoperiod and randomly assigned a temperature treatment (incubator A at 10°C, incubator B at 20°C). After being held in their temperature treatments for 24 h, digital images were taken of each larva using a Nikon Coolpix 950 digital camera. Immediately after taking pictures, temperature treatments for both incubators were switched (incubator A at 20°C, incubator B at 10°C) and larvae were held in the second temperature regime for another 24 h. Again, digital images were taken of each larva. Following the early-stage color experiment, larvae were held individually in 17.5 cm diameter containers with 1 L of filtered, aerated tap water and fed macroinvertabrates ad libitum. Larvae were kept in constant 15°C temperature and fed every 2 days for a period of 4 weeks before being tested a second time. Holding larvae in a constant 15°C temperature regime before the experiment and for the 4 weeks between trials ensured that during experimental trials, regardless of temperature treatment, all larvae would be exposed to a temperature that was 5°C different from their acclimated temperature. Again, larvae were allocated randomly to temperature treatments (incubator A at 10°C, incubator B at 20°C). Digital images were recorded after 24 h of exposure to these temperatures. As in the earlier experiment, the temperature in each incubator was then switched and larvae were held for another 24 h in the new temperature (incubator A at 10°C, incubator B at 20°C). Digital images were again taken of each larva. Previous color analyses on light and dark A. barbouri and A. texanum larvae showed that larvae vary primarily in brightness values (amount of black versus white), with relatively constant chroma and hue values (Storfer et al. 1999; Grill and Rush 2000). Using digital images of each larva, we quantified brightness from three equal-sized regions of the body. Measurements were taken using black vs. white pixel weights within a size-standardized square on each side of the larval head (approximately 75% of the cheek area). Dorsal coloration was quantified using the same standardized measurement square at the point midway between the snout and vent on the dorsal side of each larva. Because brightness values were correlated for the three regions, we used a prin-

Can. J. Zool. Vol. 81, 2003

cipal component analysis to combine the three measurements into a single measure of larval brightness for each image. Differences between species and effects of temperature and ontogeny on color were tested using repeated-measures ANOVAs. Our design included two levels of repeated measures: measurements at two temperatures for early-stage larvae and for the same two temperatures for the same larvae at a later stage. Results from a single repeated-measures ANOVA using all of these data are difficult to decompose and interpret. To examine short-term color responses to temperature, we ran separate repeated-measures ANOVAs for early- and late-stage larvae, with species as a grouping factor and temperature as the repeated-measure treatment. To address ontogenetic changes in color, we ran separate repeated-measures ANOVAs for 10 and 20°C, with species as the grouping factor and age as the repeated-measures treatment. Mortality was a factor over larval development, thus analyses only included individuals that were tested in both the early and late stages.

Results Temperature and color change Initial-analyses blocking by incubator showed no significant incubator effects on larval color or color change; thus to retain degrees of freedom, we did not include incubator as a blocking factor in the analyses shown below (Table 1). For both early- and later-stage larvae, we quantified a significant difference in color between the two species (Table 1, A and C); A. texanum larvae were consistently considerably lighter in color than larvae of its sister species A. barbouri (Fig. 1). In early stages of larval development, temperature had a significant effect on larval color for both species. Early-stage larvae of both A. texanum and A. barbouri were lighter in the 20°C treatment than in the 10°C treatment. Although both species reacted in the same direction to warm temperatures, young A. texanum lightened to a greater degree than young A. barbouri (Table 1, B). In contrast, in late-stage larvae, warm temperatures had different effects on the two species. Although temperature had no significant effect on larval color in A. barbouri, warmer temperatures induced darker color in A. texanum. Color change over ontogeny Species differences in mean color were maintained over ontogeny and temperature; A. texanum larvae were always lighter in color than A. barbouri larvae, regardless of age or temperature (Table 2, A and C; Fig. 1). For both species, larval color became lighter over ontogeny (Table 2, B and D; Fig. 2). For A. barbouri, this was true regardless of temperature. For A. texanum, however, color change was significantly greater when temperatures were cold, whereas color change at 20°C was relatively minimal. Apparently, at warmer temperatures, because early-stage A. texanum larvae were already light in color, they showed little tendency to grow even lighter over ontogeny.

Discussion Our study showed that early-stage A. barbouri and A. texanum larvae were lighter in the warmer temperature © 2003 NRC Canada



Garcia et al.

713

Table 1. Repeated measures ANOVA for temperature effects (10 and 20°C) on early- and late-stage larval color response (Ambystoma barbouri and Ambystoma texanum). df Early-stage larvae (A) Between subjects Species Error (B) Within subjects Temperature Temperature × species Error Late-stage larvae (C) Between subjects Species Error (D) Within subjects Temperature Temperature × species Error

MS

F

Table 2. Repeated measures ANOVA for ontogenetic effects on larval color (A. barbouri and A. texanum) at two temperatures (10 and 20°C).

P

1 44

120.49 0.23

524.76

0.0001*

1 1 44

11.95 5.91 0.19

61.07 30.19

0.0001* 0.0001*

1 44

66.72 0.71

94.1

0.0001*

1 1 44

0.94 3.55 0.21

4.46 16.76

0.04* 0.0001*

df 10°C (A) Between subjects Species Error (B) Within subjects Time Time × species Error 20°C (C) Between subjects Species Error (D) Within subjects Time Time × species Error

MS

F

P

1 44

86.47 0.29

289.21

0.0001*

1 1 44

148.21 0.57 0.38

385.91 1.48

0.0001* 0.231

1 44

96.94 0.42

233.74

0.0001*

1 1 44

60.01 12.68 0.25

241.86 51.12

0.0001* 0.0001*

Note: Between subjects shows differences in color between species; within subjects indicates color differences over ontogeny; df = degrees of freedom; asterisk (*) indicates significance.

Note: Between subjects shows differences in color between species; within subjects indicates color differences over ontogeny; df = degrees of freedom; asterisk (*) indicates significance.

Fig. 1. Principal component analysis on mean body color for young and old Ambystoma texanum and Ambystoma barbouri larvae over two temperature regimes (cold and warm).

Fig. 2. Principal component analysis on mean body color for young and old A. texanum and A. barbouri larvae over two temperature regimes (10 = 10°C, 20 = 20°C).

treatment relative to the colder temperature treatment. In addition, we found an ontogenetic change in mean body color for both A. barbouri and its sister species A. texanum. Over larval development, both species became lighter in color. However, temperature did not have an effect on the color of late-stage A. barbouri larvae. Furthermore, warm temperatures induced a darkening response in older A. texanum. Although this darkening by late-stage A. texanum goes against our predicted response, the body color of these late-stage A. texanum larvae in warm temperatures was still significantly lighter than most other treatments. Temperature is one proximate mechanism controlling color in early-stage larvae, with cold temperatures inducing a dark

color. Cold temperatures trigger the release of melanocyte stimulating hormone (MSH), which disperses melanosomes, causing an overall darkening of the skin (Duellman and Trueb 1986). Both A. barbouri and A. texanum darkened in the cold temperature treatment in the early stages, which is consistent with the hypothesis that MSH is triggered by cold temperatures and plays a role in Ambystoma larval color change. One possible selective force explaining dark early-stage larval coloration is the need for fast development rates to cope with habitat ephemerality. Both A. barbouri and A. texanum live in ephemeral habitats in which habitat drying can cause substantial larval mortality. Field surveys sug© 2003 NRC Canada



714

gest that streams inhabited by A. barbouri are even more likely than ponds used by A. texanum to dry before larvae have metamorphosed into the terrestrial stage (Petranka and Sih 1987). Developmental rates can be enhanced by increasing body temperatures. In terrestrial frogs, dark coloration has been shown to increase body temperature, possibly helping with digestion and metabolic activity (Carey 1978; Hoppe 1979). Water, however, has such a high specific heat capacity that for animals as small as these salamander larvae (