Naturwissenschaften DOI 10.1007/s00114-011-0765-4

SHORT COMMUNICATION

Nuptial coloration of red shiners (Cyprinella lutrensis) is more intense in turbid habitats Matthew B. Dugas & Nathan R. Franssen

Received: 8 November 2010 / Revised: 11 January 2011 / Accepted: 12 January 2011 # Springer-Verlag 2011

Abstract Communication is shaped and constrained by the signaling environment. In aquatic habitats, turbidity can reduce both the quantity and quality of ambient light and has been implicated in the breakdown of visual signaling. Here, we examined the relationship between turbidity (quantified with long-term data) and the expression of carotenoid-based nuptial coloration in the red shiner (Cyprinella lutrensis), a small-bodied cyprinid. Males in more turbid habitats displayed redder fins, and an experimental manipulation of adult diet suggested that carotenoid intake alone did not explain among-population color differences. These results run counter to similar studies where signal expression decreased in turbid conditions, and may be explained by the non-territorial red shiner mating system, interactions between the mechanism of coloration and the signaling environment, or reduced cost of color expression in turbid habitats (e.g., reduced predation risk). Our results highlight how the behavioral and ecological contexts in which signals function can shape evolutionary responses to the environment. Keywords Communication . Sexual selection . Turbidity . Visual signaling Introduction The environment through which signals travel constrains communication by shaping the costs and benefits of both Electronic supplementary material The online version of this article (doi:10.1007/s00114-011-0765-4) contains supplementary material, which is available to authorized users. M. B. Dugas (*) : N. R. Franssen Department of Zoology, University of Oklahoma, Norman, OK 73019, USA e-mail: [email protected]

signal production and reception. In aquatic habitats, turbidity can shift the color of ambient light, lower its overall intensity, and, via scattering, compromise the ability of receivers to resolve silhouettes (Levine et al. 1979; UtnePalm 2002). Through such effects, turbidity can reduce the influence of visual signals on intra- (Järvenpää and Lindström 2004; Wong et al. 2007) and inter-sexual (Seehausen et al. 1997, 2008; Heuschele et al. 2009) selection, even limiting species recognition in mate choice (Seehausen et al. 1997). Presumably because receivers respond less to visual signals under turbid conditions, turbidity is also associated with reduced signal expression (Seehausen et al. 1997, 2008; Secondi et al. 2007; Wong et al. 2007). In most fishes in which turbidity influences visual signaling (e.g., Seehausen et al. 1997; Järvenpää and Lindström 2004; Heuschele et al. 2009), males defend spatially fixed resources (e.g., nests), and so at least some signaling (inter- or intra-sexual) takes place over considerable distances. However, turbidity may more sharply restrict long-distance than short-distance visual communication (Utne-Palm 2002), and so examining a variety of social and ecological contexts is needed to broadly assess the effect of turbidity on signal evolution (Fuller 2002). Here, we tested the relationship between local turbidity and the intensity of nuptial coloration in populations of red shiners (Cyprinella lutrensis), a small-bodied (<90 mm) fish in which males are not territorial and actively pursue females prior to mating (references in Vives 1993; personal observations). Males express carotenoid-based (unpublished data) orange–red coloration during the breeding season only, the intensity of which varies across its range (Matthews 1995; online resources, Fig. S1) as does turbidity (Matthews 1988). In addition to field collections from localities for which long-term (∼20 years) turbidity

Naturwissenschaften

data were available, we manipulated the diet of fieldcollected males to examine the potential contribution of carotenoid intake to among-population differences.

testing evolutionary hypotheses in inherently variable environments. Diet manipulation experiment

Materials and methods Field collections and turbidity measurements In June to August 2008, we sampled fish from 18 sites on seven rivers in Oklahoma (OK), USA, chosen for availability of long-term turbidity data (Fig. 1); distances between localities (minimum=14.5 km; Fig. 1) greatly exceeded the distances small-bodied fishes typically move (<0.5 km; Skalski and Gilliam 2000). Available habitats within ∼100 m of turbidity-sampling sites were seined for ∼2 h or until ∼20 males were captured. To preserve carotenoids in tissue, fish were immersed in liquid nitrogen in the field and then stored at −80°C until color scoring (Grether et al. 1999). Only males in peak breeding state, defined by the presence of randomly distributed head tubercles (Koehn 1965), were used in the final analysis (peak males were redder than subpeak males; see online resources, Appendix). Turbidity, measured in Nephelometric Turbidity Units (NTU), was sampled following US EPA method 180.1 (EPA 1993) by the OK Department of Environmental Quality and OK Water Resource Board. We calculated median turbidity from April to August (peak breeding in this region, Farringer et al. 1979) for each year sampled and used the mean of these yearly medians to estimate site turbidity (online resources, Table S1). Relative contributions of natural and anthropogenic turbidity are unclear, but presettlement accounts of the region suggest much amongsite variability is natural (Matthews 1988). While this methodology is limited (e.g., it does not allow for visual modeling), such long-term data offer substantial advantages over single or few measurements of transmittance when

Fig. 1 Collection sites coded by median April to August turbidity (NTU). Three populations used for the diet manipulation experiment are indicated by asterisk

In May 2009, fish were captured from three sites on two rivers (Fig. 1) and maintained at natural densities in six indoor tanks (two per population). At the time of capture, males were not colorful (i.e., we could not distinguish males from females). Fish from each population were haphazardly assigned to either a high- (TetraColor® Crisps, Tetra Werke, Melle, Germany) or low-carotenoid (TetraAlgae® Crisps) diet and were fed a fixed quantity of food (per centimeter of fish) daily (aside from differences in carotenoid content, diets were similar). When >10 males reached peak breeding condition in one tank (after ∼45 days), we collected up to 10 per tank (in liquid nitrogen); all peak males were collected in a second sample (after ∼75 days). Redness did not differ between sampling dates (F1,77 <0.01, p=0.968), and body condition (see below) of field-collected, high- and low-carotenoid males did not differ (F2,115 =1.6, p=0.207). Quantifying color We measured caudal fin coloration using a USB4000 spectrometer (Ocean Optics, Dunedin, FL, USA), a deuterium–tungsten halogen lamp (DT-MINI-2-GS), and a 600-μm reflectance probe held at 90° to the fin (∼3 mm from tissue). Reflected light, relative to a white standard (WS-1), was recorded by the SpectraSuite software. Because sparsely pigmented fins were nearly transparent, we measured reflectance with fins placed over Teflon tape, which reflects uniformly across wavelengths used for analysis (online resources, Fig. S2). Using the mean of four measurements per individual (two per caudal lobe), we calculated the wavelength corresponding to the midpoint of reflectance from 450 to 650 nm (lRvis50; hereafter, redness),

Naturwissenschaften

Statistics

580

575 λRvis50 (nm)

a measure positively associated with carotenoid content (Andersson and Prager 2006). In a pilot sample, caudal and anal fin redness were correlated (r=0.81, n=40, p<0.001), suggesting the caudal fin provided a good estimate of overall fin coloration.

570

565

Turbidity, mass, and standard length (SL) were log10transformed to meet the assumption of normality. To assess phenotypic correlates of coloration, we used linear mixed models with redness as the dependent variable, population as a random effect, and either SL or body condition (residual of a mass and SL regression) as a fixed effect. We examined the relationship between color and local signaling conditions with a linear regression comparing mean population redness and site turbidity [we initially included mean SL in a multiple regression, but this effect was nonsignificant (t=0.5, p=0.625) and so was dropped from the final model]. Because we had multiple samples from some rivers, we ran an identical analysis using mean fish redness and mean turbidity of each river. To assess the fixed effects of diet and population in our diet manipulation experiment, we used a general linear model (GLM) with SL as a covariate; we treated population as a fixed effect because populations were selected based on a priori information about average color intensity. We used SPSS v15 for all analyses.

Results Male caudal fin redness was positively associated with SL (F1,181 =6.9, p=0.009, "±SE=25.9±9.8) but not body condition (F1,181 <0.01, p=0.998); the effect of population was significant (p<0.030) in both models. Population mean redness was positively associated with turbidity (R2 =0.34, F1,16 =8.4, p=0.010, "±SE=7.9±2.7; Fig. 2). This relationship was similar, although nonsignificant, when we calculated mean fish redness and turbidity for each river rather than each collection site (R2 =0.33, F1,6 =3.0, p=0.133, "±SE= 8.9±5.1). In the diet manipulation experiment, diet, population, and SL were all significant predictors of redness (Table 1, Fig. 2); no interactions were significant and were dropped from the final model.

Discussion The carotenoid-based nuptial coloration of red shiners is sexually dimorphic, expressed only during breeding, and positively associated with both tubercle development and body size, supporting the hypothesis that this is a sexually

560

1.0

1.3

1.9

1.6

2.2

Turbidity log10 (NTU)

Fig. 2 Relationship between turbidity and redness (mean ± SE) of field-collected (open circles) male red shiners in peak breeding condition. The regression line shows the relationship for fieldcollected males only. For three sites, redness of laboratory-reared fish on high- (grey triangles) and low- (grey squares) carotenoid diets are also presented

selected trait. Unlike nuptial signals in other systems (Secondi et al. 2007), including red coloration in fish (Seehausen et al. 1997, 2008; Wong et al. 2007), the intensity of this visual signal was positively associated with local turbidity, suggesting at minimum that red shiner coloration is not lost when fish inhabits turbid waters. Three nonexclusive proximate hypotheses might explain this pattern. Among-population differences could result from differences in carotenoid availability. While we cannot exclude this mechanism, it is unlikely to entirely explain among-population differences given that (1) autotroph activity (and thus carotenoid availability, including red carotenoids specifically) is typically negatively associated with turbidity (Kirk 1994) and (2) populations differed in redness even when carotenoid intake was controlled (furthermore, the experimental effect of diet suggests that field-collected fish did not store sufficient carotenoids to develop maximum coloration, Fig. 2). Among-population differences might also arise from genetic differences and/or a plastic response to light environment during development (e.g., Fuller and Travis 2004; Lewandowski and Boughman 2008). Either of these two mechanisms would be consistent with an advantage of intense red nuptial coloration in turbid rivers, and both would presumably be driven by similar selective pressures. To the extent that these pressures act Table 1 Results of a GLM assessing the effects of population of origin, diet (carotenoid rich vs. carotenoid poor), and SL on redness (lRvis50) of lab-reared fish. Df are shown in subscript

lRvis50

Population Diet SL

F 3.742,76 6.581,76 11.091,76

p 0.028 0.012 0.001

Naturwissenschaften

directly on communication, several interacting turbiditymediated mechanisms might shape the costs and benefits of color expression, including features of the signaling environment, balance shifts between natural and sexual selection, and expression of nuptial coloration on otherwise-transparent fins. While our long-term turbidity estimates do not capture the sources of turbidity, red clay is typical of sampled habitats (Matthews 1988; online resources, Fig. S3). Whereas algal turbidity would probably result in red colors appearing black (by removing long wavelengths from ambient light), suspended clay instead shifts the transmittance of water to long wavelengths, conferring upon red colors the dual advantage of producing strong signals and maintaining their qualitative color and any information contained therein (Levine et al. 1979), perhaps explaining why red nuptial colors were still expressed in these habitats. Moreover, the effect of carotenoids on reflectance is not limited to increased redness; relatively carotenoid-rich colors absorb more light and thus appear darker (Andersson and Prager 2006). Carotenoid-rich fins might also, then, be advantageous not because of their color per se but because of this achromatic effect that could increase conspicuousness (Franck et al. 2001) and/or maintain signaling function by accurately revealing pigment density (regardless of the source of turbidity). While other mechanisms of coloration (e.g., melanin; Franck et al. 2001) may be better suited to such achromatic signaling (i.e., higher benefits for lower costs), red shiners might be phylogenetically restricted to carotenoid-based colors. By absorbing and scattering light, turbidity may also reduce the distance over which visual signals can be detected by intended, or unintended, receivers (Utne-Palm 2002). If nuptial color intensity represents a compromise between natural and sexual selection (Endler 1980), a reduction in predation risk driven by turbidity may allow for more intense ornament expression in turbid systems (but see Reimchen 1989). Conversely, when male–male competition and/or female choice occurs primarily at a distance, this effect could be responsible for the reduction of signal expression (i.e., the benefits of expression decrease). Fin coloration might also be subject to different selective pressures and/or constraints than body coloration considered in previous work (Seehausen et al. 1997, 2008; Wong et al. 2007). A turbid (especially reddish) background might obscure differences between transparent and sparsely pigmented fins, reducing receivers' ability to detect subtle differences in pigment density and exaggerating the advantage of preferring and/or displaying especially intense colors. Because morphologically and ecologically similar species with transparent fins (e.g., Notropis stramineus, Notropis atherinoides) are common in these habitats, this

effect of turbidity could also favor pigment-rich fins in a species-recognition context. Both fin and body coloration are common in fish, and so future comparative work can address this hypothesis. Extensive theoretical and empirical work allows for strong a priori predictions about how visual signals will evolve in different habitats. However, such relationships are subject not only to generalizable and predictable laws of physics, but to species-, community-, and habitat-specific constraints that may shape signal evolution differently in various signaling contexts. Acknowledgments We thank E. Marsh-Matthews, W. Matthews, D. Mock, K. McGraw, G. Rosenthal, K. Hambright, L. Dillow, S. Strickler, and several anonymous reviewers. This study was approved by the University of Oklahoma IACUC (R08-004).

References Andersson S, Prager M (2006) Quantifying colors. In: Hill GE, McGraw KJ (eds) Bird coloration, vol 1: mechanisms and measurements. Harvard University Press, Cambridge, pp 41–89 Endler JA (1980) Natural selection on colour patterns in Poecilia reticulata. Evolution 34:76–91 EPA (1993) Nephelometric method 180.1. 600/R-93-100. Farringer RT, Echelle AA, Lehtinen SF (1979) Reproductive cycle of the red shiner, Notropis lutrensis, in central Texas and southcentral Oklahoma. T Am Fish Soc 108:271–276 Franck D, Dikomey M, Scharti M (2001) Selection and maintenance of a colour pattern polymorphism in the green swordtail (Xiphophorus helleri). Behaviour 138:467–486 Fuller RC (2002) Lighting environment predicts the relative abundance of male colour morphs in bluefin killifish (Lucania goodei) populations. Proc R Soc Lond B 269:1457–1465 Fuller RC, Travis J (2004) Genetics, lighting environment, and heritable responses to lighting environment affect male color morph expression in bluefin killifish, Lucania goodei. Evolution 58:1086–1098 Grether GF, Hudon J, Millie DF (1999) Carotenoid limitation of sexual coloration along an environmental gradient in guppies. Proc R Soc Lond B 266:1317–1322 Heuschele J, Mannerla M, Gienapp P, Candolin U (2009) Environment-dependent use of mate choice cues in sticklebacks. Behav Ecol 20:1223–1227 Järvenpää M, Lindström K (2004) Water turbidity by algal blooms causes mating system breakdown in a shallow-water fish, the sand goby Pomatoschistus minutus. Proc R Soc Lond B 271:2361–2365 Kirk JTO (1994) Light and photosynthesis in aquatic ecosystems. Cambridge University Press, Cambridge Koehn RK (1965) Development and ecological significance of nuptial tubercles of the red shiner, Notropis lutrensis. Copeia 1965:462– 467 Levine JS, Lobel PS, MacNichol EF (1979) Visual communication in fishes. In: Ali MA (ed) Environmental biology of fishes. Plenam, NY, pp 447–476 Lewandowski E, Boughman J (2008) Effects of genetics and light environment on colour expression in threespine sticklebacks. Biol J Linn Soc 94:663–673 Matthews WJ (1988) North American prairie streams as systems for ecological study. J N Am Benthol Soc 7:387–409

Naturwissenschaften Matthews WJ (1995) Geographic variation in nuptial colors of red shiner (Cyprinella lutrensis; Cyprinidae) within the United States. Southwest Nat 40:5–10 Reimchen TE (1989) Loss of nuptial color in threespine sticklebacks (Gasterosteus aculeatus). Evolution 43:450–460 Secondi J, Aumjaud A et al (2007) Water turbidity affects the development of sexual morphology in the palmate newt. Ethology 113:711–720 Seehausen O, van Alphen JJM, Witte F (1997) Cichlid fish diversity threatened by eutrophication that curbs sexual selection. Science 277:1808–1811 Seehausen O, Terai Y et al (2008) Speciation through sensory drive in cichlid fish. Nature 455:620–626

Skalski GT, Gilliam JF (2000) Modeling diffusive spread in a heterogeneous population: a movement study with stream fish. Ecology 81:1685–1700 Utne-Palm AC (2002) Visual feeding of fish in a turbid environment: physical and behavioral aspects. Mar Freshw Behav Physiol 35:111–128 Vives SP (1993) Choice of spawning substrate in red shiner with comments on crevice spawning in Cyprinella. Copeia 1993:229– 232 Wong BBM, Candolin U, Lindström K (2007) Environmental deterioration compromises socially enforced signals of male quality in three-spined sticklebacks. Am Nat 170:184–189

Online Resources      Appendix 1. Tubercle development and coloration in male red shiners. We defined peak males as those with head tubercles that were both i) arranged in parallel lines and ii) randomly dispersed. At six sites, we also collected sub-peak males, defined as those with only parallel lines of head tubercles (see Koehn 1965 Fig. 1, tubercles are visible in Fig. S1). Some males collected had not yet developed parallel lines of head tubercles, but these were rare and so we excluded them from analysis. To assess the relationship between coloration and tubercle state, we used a linear mixed model with population entered as a random factor, tubercle state (peak or sub-peak) as a fixed factor, and log-transformed standard length (SL) as a covariate. Peak males were redder than non-peak males (F1,76.8=21.1, p<0.001).

Estimated marginal mean (±SE) redness of peak and sub-peak male red shiners (presented with logSL set to 1.7).

1

Table S1. Site, number of peak males used in analyses, co-ordinates, turbidity estimate (with SE), and turbidity sampling details, including range and number of years sampled, and total April-Aug samples used to calculate the turbidity estimate used in all analyses. Turbidity was significantly different among sites (one-way ANOVA: F17,355=13.3, p<0.001).

Locality Deep Fork River, near Wellston, OK Deep Fork River, near Luther, OK Elk Creek, near Hobart, OK North Canadian River, near Harrah, OK Beaver River (N. Canadian R.), near May, OK North Canadian River, near Prague, OK North Canadian River, near Watonga, OK North Canadian River, near Wetumka, OK North Canadian River, near Woodward, OK North Fork Red River, near Altus Dam, OK North Fork Red River, near Carter, OK North Fork Red River, near Headrick, OK Red River, near Gainesville, TX Red River, near Terral, OK Red River, near Waurika, OK Salt Fork Arkansas River, near Cherokee, OK Skeleton Creek, near Lovell, OK Washita River, near Pauls Valley, OK

Peak males n 12 17 8 11 17 13 18 21 11 12 4 19 8 6 3 9 7 4

Latitude dd 35.69645 35.66673 35.01414 35.50034 36.62433 35.40176 35.84026 35.26565 36.43670 34.88951 35.16811 34.63452 33.72788 33.87871 34.13199 36.82047 36.06004 34.75480

Longitude dd -97.06925 -97.17674 -99.13032 -97.19392 -99.74849 -96.67087 -98.45556 -96.20612 -99.27844 -99.30703 -99.50731 -99.09675 -97.16001 -97.93448 -98.09475 -98.36038 -97.58505 -97.25141

Turbidity NTU (SE) 45.8 (18.4) 92.5 (53.0) 45.4 (5.9) 60.7 (17.8) 29.1 (7.4) 100.6 (42.8) 23.7 (6.0) 172.0 (36.6) 24.7 (4.61) 10.9 (1.97) 27.0 (5.1) 40.5 (7.04) 191.7 (16.0) 193.3 (27.1) 153.4 (17.3) 62.0 (15.5) 88.4 (10.63) 228.4 (19.2)

Start year 1984 1989 1980 1980 1989 1983 1989 1980 1989 1980 1980 1980 1980 1980 1982 1999 1980 1980

End year 2002 2002 2006 2007 2002 2002 2008 2007 2008 1996 2008 2008 2003 2007 2003 2008 2008 2008

Times sampled 53 23 92 108 22 55 39 59 53 57 114 116 78 106 91 40 101 118

Years sampled 17 13 25 28 13 15 18 20 19 17 29 26 21 26 22 10 25 29

2

Figure S1. Red shiner (Cyprinella lutrensis) males with pale (top) and intense (bottom) nuptial coloration. Note that the dorsal fin is uncoloured in both, as it is in all red shiners.

3

Figure S2. Mean±SD reflectance at 20nm intervals (320–700nm) of the 10 most red and 10 least red Cyprinella lutrensis males measured in the field. Also included is reflectance of teflon tape (n=10) used as background for measurement of caudal fin coloration.

4

Figure S3. Example habitats that vary in turbidity.

5