Realistic nitrate concentration alters the expression of sexual traits and olfactory male attractiveness in newts

Authors


*Correspondence author. E-mail: jean.secondi@univ-angers.fr

Summary

  • 1The occurrence of endocrine disruption is increasing and may now affect most, if not all, ecosystems. Owing to its impact on survival and reproduction, it is increasingly important to consider this factor in ecological studies. The study of secondary sexual traits is of particular interest because they are under strong hormonal control and play a crucial role in mating. A neglected sub-lethal effect of endocrine disruption is the reduction of male attractiveness, and the distortion of the sexual selection process.
  • 2Nitrate is one of the most widespread endocrine disruptors because of its extensive use in agriculture. Studies, especially in amphibians, have mostly focused on early developmental stages, probably because these were considered critical for population dynamics. However, negative effects on adult survival or reproductive success may also affect population demography.
  • 3We surveyed nitrate concentration in a large sample of ponds. We then exposed male palmate newts, Triturus helveticus, to high (75 mg L−1) but realistic concentration of sodium nitrate to investigate the effect of nitrate on the expression of secondary sexual traits and male attractiveness.
  • 4Exposure to nitrate for 3 weeks was found to limit the expression of sexual traits and to reduce body growth; exposure for 10 days reduced male attractiveness. Female preferred males unexposed to nitrate in olfactory tests but not in visual tests.
  • 5The expression of secondary sexual characteristics and body size is reduced in males exposed to realistic nitrate concentration. However, a drop in male attractiveness may first be caused by disruption of the production of olfactory cues. Depending on the frequency of contaminated breeding sites, the disruption of male ability to express sexual signals might affect the sexual selection process in palmate newts and eventually reduce population reproductive output. We discuss the ecological and evolutionary implications of this result.

Introduction

The ecological impact of endocrine disruptors released by human activities is currently under intense scrutiny (Jones & Reynolds 1997; Crisp et al. 1998; Vos et al. 2000; Giesy et al. 2003; Sumpter 2003; Clotfelter et al. 2004; Rodriguez et al. 2007; Soin & Smagghe 2007). By their action on hormonal systems, these compounds affect key biological functions like development, metabolic activity, reproductive physiology and behaviour (Jones & Reynolds 1997; Park et al. 2001; Dell’Omo 2002; McKenzie et al. 2003). For these reasons they are suspected to negatively affect individual fitness and population dynamics (Sumpter 2003; Palanza & Vom Saal 2002). Major effects like mortality or malformations are usually observed in early developmental stages (Camargo et al. 2005), but non-lethal effects have been reported in later life stages (Park et al. 2001; Dell’Omo 2002; Clotfelter et al. 2004). In particular, the expression of sexual behaviour or secondary sexual traits in vertebrates is usually hormone-dependent and condition-dependent (Pomiankowski 1987; Hill 1990; Andersson 1994; von Schantz et al. 1999) and thus subjected to endocrine disruption (Jones & Reynolds 1997; Halldin et al. 1999; Noriega & Hayes 2000; Park et al. 2001; Lürling & Scheffer 2007). Understanding the effects of sublethal endocrine disruptors, and how they ultimately impact life history, survival or reproduction is a current challenge in the study of natural populations in anthropized environments (Hayes et al. 2006; Relyea & Hoverman 2006; Rohr et al. 2006).

Secondary sexual traits, like vocal signals, colour patterns or pheromones, have a central role in the mating process, and in reproduction in general (Andersson 1994). Endocrine disruption is expected to alter courting behaviour, mate choice and individual breeding success, and eventually to cause a decrease in population reproductive output (Arellano-Aguilar & Macías Garcia 2008). Besides its consequences for population demography, endocrine disruption of sexual communication is of particular interest from an evolutionary point of view. In contaminated sites, secondary sexual traits may reflect resistance to endocrine disruption rather than the quality of the individual (e.g. stamina or pathogen resistance). These two characteristics are not necessarily correlated. Thus, endocrine disruption is a proximal factor that has the potential to alter the sexual selection process by changing physiological trade-offs and altering the information contained in sexual signals. This issue has received little attention so far.

Amphibians are particularly well-suited to address this issue. Many species are sexually dimorphic as males develop either morphological characters (like coloration or crests in Urodeles) or behavioural characters (like calls in Anurans) subjected to sexual selection (Halliday 1975; Klump & Gerhard 1987; Andersson 1994). Many amphibian species spend the breeding season in ponds or water bodies where they are exposed to endocrine disruption by chemicals present in food and water, and by ions exchange through their skin or digestive tract. Surprisingly, adult stages have been almost entirely omitted from endocrine disruption studies so that the actual impact of these compounds on mating behaviour remains largely unknown. A reason for this bias may be that larval survival has often been considered as a more determinant parameter of population dynamics than adult survival. In recent years this view has been criticized and questioned (Biek et al. 2002; Schmidt et al. 2005).

Nitrate is naturally present in many amphibian breeding sites, however it is also one of the molecules most commonly spread by man, and anthropogenic activities like agriculture have recently increased nitrate concentrations at local or regional scales (Vitousek et al. 1997). Effects of nitrates have been observed in several animals groups (Guillette & Edwards 2005; Camargo et al. 2005). Nitrate is transformed into nitrite in the digestive tract. This latter molecule causes methemoglobinaemia in humans, that is, the transformation of haemoglobin into the reduced methemoglobin form that cannot deliver oxygen (Avery 1999; Panesar & Chan 2000). In vertebrates, nitrites are transformed into nitric oxide in the digestive tract, a compound used in cellular signalling. It eventually affects steroidogenesis via the functioning of Leydig cells (Panesar & Chan 2000). It also disrupts thyroid hormone synthesis, which is responsible for metamorphosis and growth in vertebrates (Guillette & Edwards 2005). Most studies on amphibians have focused on larval development (Hecnar 1995; Camargo et al. 2005; Ortiz et al. 2004). They generally described negative effects on survival, time to metamorphosis or behaviour. We are aware of only two studies on adults, in these studies the authors showed that nitrate pellets spread on fields as fertilizers caused severe physiological stress and rapid death (Oldham et al. 1997; Ortiz-Sanliestra et al. 2005). Nitrates are strongly suspected to induce endocrine disruption. However, owing to the diversity of effects and the lack of physiological or biochemical measurements in most studies, whether observed effects reflect mere toxicity or actual endocrine disruption is as yet unknown.

We tested the effect of nitrates on the expression of secondary sexual traits and male attractiveness in adult amphibians using the palmate newt Triturus helveticus as a model (Fig. 1). This species exploits a large range of biotopes, from forests to open landscapes, to complete its life cycle. On land, newts seek habitats with enough moisture and refugia, that is, wooded areas (Nöllert & Nöllert 2003), whereas for breeding they mostly select ponds. These aquatic habitats are subject to nitrate contamination via runoffs or aquifers. Like all other members of this genus, male palmate newts develop large ornaments when they enter water for breeding (Griffiths & Mylotte 1988). The caudal crest, bright ventral coloration, webbing of hind feet and caudal filament are the most dimorphic traits in this species (Griffiths 1996). In this genus, trophic or water conditions influence the expression of male sexual ornaments (Baker 1992; Secondi et al. 2007). Logically, such traits as filament length in T. helveticus (Haerty et al. 2007) or dorsal crest height in other Triturus species are assessed by females during mate selection (Hedlund 1990; Green 1991; Baker 1992). Thus, any factor affecting the expression of secondary sexual traits is expected to have an impact on male reproductive success.

Figure 1.

Picture of a male palmate newt.

Earlier studies showed that nitrates reduced larval survival in T. helveticus (Watt & Jarvis 1997) and size at metamorphosis, a fitness indicator, in T. vulgaris (Watt & Oldham 1995). Ammonium nitrate also caused severe breathing abnormalities in terrestrial adults of T. boscai (Ortiz-Sanliestra et al. 2005). Because secondary traits are grown during the aquatic stage (Griffiths & Mylotte 1988), continuous exposure of adults to nitrate in ponds may limit their growth or their ability to maintain sexual ornaments, and ultimately alter their breeding success. In this study, we surveyed nitrate concentrations in a large sample of ponds to assess the distribution of the risk of exposure to nitrate across our study area. We tested the effect of long-term nitrate exposure on the expression of secondary sexual traits in breeding males. Finally, we measured the sexual attractiveness of males exposed and unexposed to nitrate. Sexual communication often uses different cues in animals and there is no reason to believe that all communication pathways are affected equally by endocrine disruption. We addressed this issue by using an experimental procedure that tested separately the visual and olfactory attractiveness of males.

Methods

samples

We captured 36 males and 18 females on 14–16 March 2007 and 38 males and 19 females on 14 March 2008 in a forest pond located in western France (47°33′59″ N, 0°45′44″ W). This pond is not connected to any stream and the surrounding terrain is a habitat of oak trees and bushes. Capture dates correspond with the peak of the breeding period in this part of the breeding range, when male sexual ornaments are displayed (Griffiths & Mylotte 1988).

Nitrate toxicity has been measured in the egg and larval stages of many amphibian species. A recent study reported LC50 values ranging from 17 to 1749·8 mg for the tadpoles of seven amphibian species that were exposed to sodium nitrate or ammonium nitrate for 96 h (Camargo et al. 2005). Such estimates for adults are not available. Taken as a whole, studies reveal our current inability to predict the range of toxicologically active concentrations for a given species. Because large differences in response are observed even within a genus, it is important to determine the actual range of nitrate concentrations in breeding habitats and to compare these values to experimental data to assess the ecological consequence of this molecule. From 2005 to 2007 (November to May) we measured nitrate concentration in 142 ponds distributed within 50 km around our study site. Among these, we checked 55 ponds for the presence of palmate newts in March 2006. Water samples were taken from these ponds between February and May for the purpose of another study. We stored samples at 4 °C and measured nitrate concentration within 2 weeks after their collection. We used a Metrohm (Courtaboeuf, France) ion meter with a nitrate selective electrode (no. 6·0504·120) coupled with a Metrohm reference electrode Ag/AgCl in KCl 3 mol L−1 (no. 6·0733·100) for nitrate measurements. Field sample concentrations were estimated from the regression line obtained from readings of calibration samples of known concentrations.

experiment 1: effect on morphology

We exposed males for 21 days to 0 or 75 mg L−1 nitrate. Watt & Jarvis (1997) showed that concentrations from 50 up to 500 mg L−1 increased larval mortality. We selected a concentration for the experimental treatment that was both realistic for the species to occur and high enough to detect effects. The experimental concentration used was low enough not to cause mortality. We rejected using the highest concentrations for which few larvae survived as such data would carry little information since highly contaminated sites would be unsuitable to establish or maintain a newt population. We housed groups of two males and one female in aquaria (17·5 cm length × 33 cm width × 25 cm height). The presence of females and potential sexual competitors was intended to mimic natural conditions and stimulate sexual activity of males in captivity. Active courting by males was observed daily in several aquaria during caretaking until the end of the experiment. The aquaria contained 0·02 L of a nitrate stock solution (10·27 g NaNO3 L−1) made up to 2·5 L (water depth 5 cm) with re-ionized water for treatment groups, and 2·5 L of re-ionized water for the control group. We used ‘ultra pure’ water (ion exchanger R3 and activated carbon ORC columns; Fisher Bioblock, Illkirch, France). Re-ionized water was prepared by the dissolution of CaCl2, NaHCO3, MgSO4 and NaCl in distilled water so as to match the median ion concentrations observed from a sample of 22 palmate newt breeding ponds in May 2005. Nitrate stock solution was made by dissolving 4·11 g NaNO3 in 0·4 L of re-ionized water. The aquaria contained 0·025 L of nitrate stock solution made up to 2·5 L with re-ionized water for treatment groups, and 2·5 L of water for the control group. Final ions concentrations were 64·1 mg L−1 Ca2+, 16·4 mg L−1 Mg2+, 23·3 mg L−1 Na+, 9·8 mg L−1 K+, 61·7 mg L−1 inline image, 123·0 mg L−1 Cl and 65·5 mg L−1 inline image. The regular change of aquarium water (three times a week) and the large air/water interface, relative to water volume, were expected to keep nitrate concentration constant and limit the formation of nitrite. Nitrate concentrations were measured in aquaria at the beginning of the experiment using a selective electrode. Readings matched the expected concentrations. Additional checks were also carried out regularly using testing sticks (Macherey Nagel, Hoerdt, France). Similarly, testing sticks were used to check nitrite levels of randomly selected aquaria during the course of the two experiments. Nitrite levels were observed to be low.

Semi-cylindrical PVC tubes were placed in each aquarium to provide refugia under which subjects can hide and rest. Newts were fed ad libitum with live chironomid larvae once a day. The light regime was adjusted weekly to match the natural photoperiod. We set a curtain in front of the shelves to avoid direct light exposure and ensure limited external visual disturbance. Water temperature was set at 14 °C to limit the potential effect of higher temperatures on the regression of sexual traits. We measured pH changes in control and experimental aquaria that were free of newts. Over 48 h, pH varied from 7·93 to 7·78 in control aquaria and from 7·85 to 7·88 in 75 mg L−1 nitrate aquaria. The experiment started on 16 March 2007 and ended 3 weeks later on 6 April 2007.

experiment 2: effect on male attractiveness

We assigned half of 36 males to 0 mg L−1 nitrate aquaria and the remaining half to 75 mg L−1 aquaria. Housing conditions were the same as in the previous experiment except that we used distilled water instead of ultra pure water and males were housed singly in order to avoid contamination by olfactory cues from another individual. Females were captured on 21 March 2008. After males have been exposed for 10 days to their treatment, we measured female preference to visual and olfactory cues from males in two series of tests. We used an olfactometer (70·5 length × 25 cm width × 15 cm height) with three compartments, a neutral zone (31 cm long) on the distal end of the apparatus and two preference zones (18·5 cm long) on its proximal ends. Females could move freely between the neutral and the two preference zones. For all experiments, we used aged tap water. We kept water flow constant (0·27 L min−1) using a pump in order to accelerate the diffusion of olfactory cues to subjects during chemical tests. Water flow was also generated during visual tests to standardize experimental conditions. The water column was 4 cm deep.

During visual tests, each male was kept in a transparent plastic box (18 cm length × 12 cm width × 7·5 cm height) in one of the preference zones. Black plastic plates ensured that males could not see each other, and that females had visual contact with only one male once they entered a preference zone. Females had no access to chemical information from stimulus males. In each preference zone, water flowed underneath and along the sides of the boxes. Water level inside the box was the same as in the rest of the apparatus. We used a 25 W light tube covering the range of the daylight spectrum (Hagen 98 CRI, 6700 K, Repti Glo 2·0, Montreal, Canada). Because direct light can disturb newt motor behaviour, we shaded the neutral and preference zones using a black plastic plate. We also placed the olfactometer in a chamber with black walls so that no spatial cues could alter subject response. An observation window on the distal end of the apparatus allowed the recording of behaviour.

For chemical tests, we used the same apparatus and the same box model as for visual tests. However, a different set of boxes was used. We allowed water to circulate through the boxes so that olfactory cues emitted by the stimulus males were conveyed to the preference and neutral zones. We covered box fronts with black shutters so that females had no access to visual cues. All other conditions for visual and chemical tests were similar in the two test series.

Each female was tested in each experiment on two different days. The presentation order of experiments (visual or olfactory stimuli) was alternated between individuals. We habituated females to experimental conditions by leaving them in the apparatus for at least 30 min without any male stimulus the day before the first test. During a test, females were placed at the distal end of the neutral zone. They were kept for 5 min in a mesh cage that allowed them to see or smell both males. After that period, the cage was lifted. For the next 15 min we recorded the time it took for a female to enter each preference zone, and the time spent within the zone. At the end of that period, the female was driven back to the cage. The whole apparatus was drained and rinsed and then wiped with ethanol to remove chemical cues. Although this last operation is not strictly necessary for visual tests we carried it out for after all tests. Male position was swapped and once the olfactometer was refilled, we started the second period following the same procedure. The same pair of males was used for a given female for both the visual and the chemical test. Tests were carried out between 25 and 30 March 2008.

morphometric measurements and statistical analyses

We took morphological measurements on males on the first and last day of each experiment. We measured body weight, body length from the tip of the head to the junction of the hind limb, caudal filament length, maximal tail height and hind-foot web area. The two former characters account for body size and body condition whereas the last three are secondary sexual characters. Details of the measurement procedure are described elsewhere (Secondi et al. 2007).

In Experiment 1, we tested the time variation of morphological trait size regardless of treatment using t-tests. We also assessed the overall effect of nitrate on secondary sexual traits by combining the relative variation of the three sexual variables (filament length, tail height and hind-foot web area) in a synthetic variable using Principal Components Analysis. To assess the effect of nitrate on the expression of sexual traits, we transformed each morphological variable to obtain a standardized measure of the relative trait variation during the course of the experiment using the following formula [(trait size at D21 – trait size at D1)/trait size at D1]. Using this formula, we obtained a measure of trait size variation during the experiment. A positive value indicates an increase and a negative value a decrease in trait size. Relative size variation is helpful to compare the responses of individual variables expressed in different units, as here, and to determine which traits are most sensitive to nitrate exposure. We checked variable normality using Liliefors test. All variables were normally distributed. We used mixed linear models with the relative change in trait size as dependent variables and nitrate concentration as a fixed factor. We considered the replication caused by the fact that males were housed in pairs by including the aquarium as a random factor. The effect of nitrate was tested in the same way as for the other variables. All analyses were carried out using jmp 6·0·0 (SAS Institute, Evry-Grégy-sur-Yerres, France).

In Experiment 2, we tested for differences in morphological traits between males of the two groups using t-tests for matched data. Female preferences to male exposed or unexposed to nitrate were measured as the time spent near each male and the mean latency to enter each preference zone. Differences in preference were tested using t-tests for matched data. Square root transformation was applied in the olfactory tests in order to meet normality criteria.

Results

nitrate distribution in ponds

Our pond sample encompassed a large range of terrestrial habitats exploited by, or suitable for, palmate newts (i.e. forest, meadows and pastures). Analyses revealed that 100 mg L−1 was about the upper limit of palmate newt occurrence in this area (Fig. 2). This value is, however, an estimation. Many ponds were not surveyed repeatedly during the breeding period so that newts might occur in sites with higher nitrate concentration, or concentration might have been higher in sites where newts have been observed. Several measurements at our study site, on 24 May 2005 (0·68 mg L−1), 20 February 2006 (7·7 mg L−1) and 16 May 2007 (0·62 mg L−1), indicate that nitrate concentration remains low in that pond.

Figure 2.

Distribution of nitrate concentration in 142 ponds sampled between 2005 and 2007 from November to May. Dark bars represent sites where palmate newts have been observed, shaded bars show sites where no newts were captured, and open bars ponds where presence data were not available. The broken line shows the nitrate concentration to which experimental males have been exposed.

effect of nitrate on male morphology

We assessed the changes in response variables over the 3 weeks of exposure regardless of treatments. Body weight and the two sexual traits, tail height and hind-foot web area, were significantly reduced after 3 weeks. In contrast, body and filament length did not exhibit significant variation (Table 1, Fig. 3).

Table 1.  Size variation of morphological traits between day 1 (D1) and day 21 (D21) in Experiment 1. Tests of trait size variation with a zero mean value have been tested using a t-test. Asterisks indicate significant P-values for α = 0·05
 D1 (mean)D21 (mean)D21 – D1 (mean)SDtd.f.P
Body weight (mg) 1·738 1·486–0·1400·068–12·38435< 0·001*
Body length (mm)32·96933·061 0·0030·009  1·85435  0·072
Filament length (mm) 6·411 6·834 0·0080·146  0·33935  0·736
Tail height (mm) 9·032 8·357–0·0730·061 –7·14035< 0·001*
Hind-foot web area (mm)33·78226·224–0·2090·122–10·22835< 0·001*
Figure 3.

Mean (± SE) relative variation in morphological traits observed after 21 days of captivity in control males (open bars) and males exposed to 75 mg L−1 nitrates (shaded bars) in Experiment 1. Relative trait size variation is expressed as [(trait size at day21 – trait size at day1)/trait size at day1].

Keeping several subjects in the same aquarium could affect the outcome of individual responses. We investigated this hypothesis by considering the effect of size dominance. To do so, we ranked individuals of the same aquarium by increasing body size. We observed no relationship between size rank within the aquarium, estimated by body weight, and trait size variation except for hind-foot web area (mean change ± SD in largest males = –0·168 ± 0·136; change in smallest males = –0·249 ± 0·094; Wilcoxon's test, n = 18, T = 40, P = 0·048). Although a dominance effect might occur, data suggest it does not drastically alter the experiment outcome. We thus considered all males independently in subsequent analyses.

We detected a significant effect of nitrate concentration on body length only (Table 2). Yet, all morphological variables showed a concordant variation pattern. Depending on the trait considered, reduced growth (body length or filament length) or accelerated regression (tail height and hind-foot web area) was observed in the nitrate treatment. We used Principal Components Analysis (PCA) to combine the three secondary sexual characters (filament length, tail height and hind-foot web area) into one synthetic variable. We extracted the first axis from the PCA that accounted for 53·81% of the total variance. The three loadings were positive (filament length = 0·550, tail height = 0·543, hind-foot web area = 0·634). Thus, the first PCA axis (PC1) accounted for the development of sexual traits during the experiment. Scores on PC1 of individuals in the nitrate treatment were significantly lower than scores of individuals in the control treatment (Table 2).

Table 2.  Effect of exposure to 75 mg L−1 nitrate on the morphological variation of non-sexual and secondary sexual characters in breeding males in Experiment 1. We tested difference in response between treatments using a mixed linear model where nitrate concentration was considered as a fixed factor and aquarium as a random factor (see Methods for details). Nitrate concentration was set to 0 and 75 mg L−1. Asterisks indicate significant P-values for α = 0·05. d.f.num and d.f.den refer to the degrees of freedom of the numerator and denominator of the F-ratio. PC1, Principal Components Analysis axis 1
 Coefficient (75 mg L−1)SEd.f.numd.f.denFP
Body weight–0·00180·0141161·5040·238
Body length–0·0030·0011165·5020·032*
Filament length–0·0440·0281162·5590·129
Tail height–0·0180·0121162·5610·129
Hind-foot web area–0·0350·0211162·8390·114
PC1–0·5110·2321164·8400·043*

effect of nitrate on male attractiveness

As for the previous experiment, we tested the effect of treatment and captivity (change in trait size between the start and the end of the experiment). We used a repeated measurement anova for each variable (data not shown). We detected no effect of treatment on change in trait size (all traits P > 0·331). In contrast, we observed significant decreases in the size of all sexual characters: filament length (F1,36 = 7·90, P = 0·008), tail height (F1,35 = 73·62, P < 0·001), and hind-foot web area (F1,36 = 75·13, P < 0·001). Body length significantly increased during the experiment (F1,36 = 24·96, P < 0·001). All treatment ×  captivity interactions were not significant (all P > 0·370).

Males were matched for their size to form stimulus pairs. Mean trait size differences (± SE) between experimental and control males at the beginning of the experiment were –0·02 ± 0·19 mm for body weight, –0·17 ± 0·57 mm for body length, 0·39 ± 1·36 mm for filament length, 0·08 ± 0·95 mm for tail height, and –0·68 ± 10·95 mm2 for hind-foot web area. Here a positive value indicates a larger size for unexposed males. We observed no difference in mean trait size for all morphological traits (t-test, all P > 0·20) and not all differences showed the same sign. Similarly, mean trait size differences (± SE) at the end of the experiment were –0·04 ± 0·22 mm for body weight, –0·12 ± 0·93 mm for body length, 0·14 ± 1·59 mm for filament length, –0·21 ± 1·1 mm for tail height, and 0·74 ± 8·46 mm2 for hind-foot web area. We observed no difference at the end of the experiment either (t-test, all P > 0·39) and not all traits exhibited the same sign difference.

During visual tests, females neither spent significantly more time near one male type nor did they enter faster into one preference zone (time spent: n = 19, t = 0·881, P = 0·390; latency: n = 19, t = –0·333, P = 0·743). In contrast, Fig. 4 shows that females spent significantly more time near males unexposed to nitrates (n = 19, t = 3·355, P = 0·004) during olfactory tests (Fig. 4). They stayed on average twice as long in the zone containing unexposed males as in the zone containing exposed males (mean ± SD: 293 ± 148 s for unexposed males and 130 ± 120 s for exposed males). We detected no difference in latency (n = 19, t = 0·853, P = 0·405).

Figure 4.

Female preference to males exposed to 0 mg L−1 nitrate (open bars) or 75 mg L−1 nitrate (shaded bars) during olfactory and visual preference tests in Experiment 2. Box plots represent medians and interquartile ranges. Dots indicate atypical values.

Discussion

time variation of morphological traits

On the whole, the size of secondary sexual traits decreased even in individuals not exposed to nitrate. This trend is observed in the wild too. Sexual traits grow when males enter water and regress before they leave it at the end of the breeding season (Griffiths & Mylotte 1988). We kept males in captivity for 3 weeks, which is a relatively long period. We are not aware of individual-based data on the phenology of sexual traits in the field, although we suspect that captivity might accelerate this process. Yet, regression of sexual traits does not imply that males are not sexually active anymore since we observed courtship displays in aquaria during the whole experiment period. Furthermore, not all sexual secondary characters showed the same response. Filament length remained stable whereas tail height and hind-foot web area regressed. This suggests that not all traits have the same phenology. For physiological reasons, it may take more time for a newt to regress its filament than its caudal fin, or the endocrine control of these two traits may simply differ. Alternatively, the caudal filament, a trait under sexual selection (Haerty et al. 2007), is probably less costly to maintain over a long period than the caudal fin. Thus, males might remain attractive over an extended period by maintaining low-cost sexual traits even if other traits are regressed.

effect of nitrate on male attractiveness

Nitrates cause the disruption of metabolic activity or endocrine functions in the larval stages of various amphibian species (Hecnar 1995; Camargo et al. 2005; Ortiz et al. 2004). Effects on adults are expected too but data are generally lacking. Because sexual traits are under steroid hormone control in amphibians (Moore & Zoller 1979; Deviche et al. 1990; Penna et al. 1992; Moore et al. 2005), we expected the expression of secondary sexual traits in breeding adults to be affected by a disruptor of steroid metabolism (Panesar & Chan 2000). Unlike larval stages, no effect on survival was detected during the course of the experiment as all individuals of the two experiments survived. However, we detected a significant effect on the expression of morphological traits of males exposed to 75 mg L−1 nitrate in Experiment 1. On one hand, nitrate impaired body growth (body length); on the other, it accelerated the loss of secondary sexual traits. Different physiological compartments were thus affected as sex hormones control the development of sexual traits and growth hormone regulates body growth in vertebrates (Jørgensen 1992). Nitrate may disrupt independently the control of these hormones. For instance, Ghrelin, a peptide hormone secreted from the stomach and that stimulates secretion of growth hormone, may be affected by the presence of nitrate or nitrites (Kaiya et al. 2001). Yet, body growth and sexual trait expression may also be linked. In humans, both estrogens and androgens contribute to the control of the expression of growth hormone (Leung et al. 2004).

We observed morphological changes in Experiment 1 but not in Experiment 2. Females consistently showed no preference for any type of male during visual tests. In contrast, they showed a clear preference for unexposed males during olfactory tests. We conclude that nitrate affected the production of male olfactory cues or signals that females assess during mating. Thus, exposure to nitrate reduced the sexual attractiveness of male newts.

Our experimental procedure allowed us to test independently the disruptive effect of nitrate on two communication channels: vision and olfaction. Thus, we could determine which component of male attractiveness was more sensitive to disruption. Data suggest that the negative effect of nitrate on the expression of morphological characters seems to be relatively small in comparison to the effect on the expression of olfactory cues. Nevertheless, a decay in male ornaments, as observed in Experiment 1, is likely to affect female choice in newts (Haerty et al. 2007). It is possible that visual attractiveness would be reduced in males exposed over a period longer than the 10 days of Experiment 2. We did not test female preference in Experiment 1 so this question is yet to be answered. It is important to notice that we might have underestimated the effect of nitrate on the expression of visual signals. In both experiments, we caught males in ponds when they had already started growing sexual ornaments. Thus, we did not test the effect on growth itself but rather the effect on trait maintenance. The influence of timing of exposure on phenotype expression is a crucial point still requiring investigation.

The accelerated regression of secondary sexual traits and the alteration of chemical communication are consistent with the hypothesis of endocrine disruption. However, with the data available, we cannot currently preclude the effect of direct toxicity. It is true that the tested concentrations were observed in the field but these are still relatively high. Because nitrates have diverse effects on organisms, we cannot definitively discriminate between endocrine disruption and general toxicity without proper endocrinological investigation. Even if the causes of the observed effects are not known yet and are limited by the current understanding of amphibian endocrinology, this work demonstrates the existence of sub-lethal effects of nitrate in adult amphibians.

Results from this study and from an earlier work in another urodele give some insight into the differential response of visual and chemical communication systems to chemical disruption by natural or exogenous compounds. Park et al. (2001) observed changes in attractiveness and mating success of female red-spotted newts, Nopthopalmus viridescens, exposed to a pesticide. Olfactory attractiveness was altered whereas visual attractiveness was not. The authors connected this effect to a decrease in pheromone production. Both studies consistently suggest that chemical communication is more rapidly affected by chemical disruptors than visual communication in amphibians. Differences in the expression of both types of sexual traits may account for this pattern. Pheromones are not remnant and have to be produced continuously, whereas visual signals (structural or pigment based) are often continuously expressed during the breeding period once developed or grown. For this reason, the higher sensitivity to disruption of chemical communication systems might be a common phenomenon in animals.

ecological and evolutionary implications

The present work shows that realistic nitrate concentrations alter adult phenotypic characteristics by reducing the ability of males to both maintain secondary sexual traits and to produce attractive olfactory cues or signals involved in the mating process. Newts tend to mate repeatedly during the course of the breeding season (Gabor & Halliday 1997; Osikowski & Rafinski 2001). In the closely related smooth newt, T. Vulgaris, female choosiness increases with time as females tend to remate with high quality males displaying the highest dorsal crests (Gabor & Halliday 1997). Similarly, female palmate newts prefer males with longer caudal filaments (Haerty et al. 2007). Disrupting the expression of sexual signals is expected to affect mate choice and thus pairing patterns. In support of this hypothesis, a recent study showed that endocrine disruption lowered the availability of attractive males in a fish (Arellano-Aguilar & Macías Garcia 2008). Disruption could have broader consequences for sexual behaviour and mating strategies of both sexes (Rohr et al. 2005). For instance, lower attractiveness could displace trade-offs between predation risk and mate search or courting effort (Rohr & Madison 2001). In the long run, sub-lethal effects of chemicals like nitrate might have demographic or evolutionary implications for populations that are chronically exposed to nitrate or any other compound producing similar effects.

From an evolutionary point of view, endocrine disruption might have implications for sexually selected traits. Considering sexual traits under Fisherian runaway sexual selection, disruption would increase habitat-induced selection pressure and would thus displace the equilibrium with sexual selection pressure towards males with less well-developed traits. This situation is likely to reduce trait variance and the intensity of sexual selection in contaminated areas. Disruption can also change the information content conveyed by trait indicators of male quality. There is no reason to believe that male performance, like resistance to pathogens or foraging ability, for instance, is correlated with the physiological capacity to cope with a novel factor disrupting trait expression. As a consequence, trait size may eventually reflect male resistance to the disrupting factor rather than the original quality it was advertising for. Such a situation can have several evolutionary outcomes depending on whether resistance to disruption confer a selective advantage. If not, the trait is likely to degenerate over generations because it does not relate to male quality anymore. If so, the trait might be maintained but its meaning would change as large trait values in a nitrate-contaminated environment would reflect resistance to disruption more than the capacity to exploit resources. The shift of signal function or meaning is a process known as co-option (Borgia & Coleman 2000). Co-option is not an unlikely outcome of endocrine disruption if resistant males also produce more resistant progeny (higher larval survival).

Over shorter time intervals, the fact that males can no longer reliably express their quality through secondary sexual traits can lead females to make suboptimal mating choices and cause the overall population breeding success to decrease. Such effects driven by the disruption of the sexual selection process could be particularly deleterious for small populations. As for larval stages, negative effects on adults could reduce overall population reproductive output. The relative contribution of negative effects on larval and adult stages are not known, but the importance of this question is heightened by recent studies which showed that, contrary to the classical view, adult stages play an important role in the dynamics of amphibian populations (Biek et al. 2002; Schmidt et al. 2005). These studies referred to survival and not to breeding success, which is difficult to estimate in the wild. Nevertheless, any process reducing breeding success, i.e. the number of matings for males or the number of eggs for females, ought to be considered.

Anthropogenic changes to ecosystems impact many aspects of organism biology. Because they target crucial functions, the study of endocrine disruptors would benefit from large-scale experiments where one can observe evolutionary processes at work. Investigating the responses of communication systems to new selection pressures is one of the lines of research that offers us an opportunity to understand and predict how species adapt, if at all, to new conditions.

Acknowledgements

We thank Viviane Hardouineau, Florian Cazimajou and Anne-Sophie Gobin who assisted in animal care taking. We are grateful to Steeve Thany and Christophe Lemaire for comments on an early draft of this manuscript. A permit provided by Préfecture du Maine-et-Loire allowed for the capture and testing of individuals.

Ancillary