Introduction

Compared to land, the visibility in water bodies is generally poorer and variable, predominantly due to attenuation and scattering caused by the substances present in the water and the light loss with depth (Utne, 1997). Light absorption in the aquatic media is attributed to four components: water itself, dissolved yellow substances, photosynthetic biota, and particulate matter (Kirk, 2011). The vertical attenuation coefficient (Kd) is a variable which integrates the absorption of all those components (Kirk, 2011). These factors influence how light penetrates the water and, consequently, how it affects the visual abilities of organisms. Dissolved yellow substances are included in the dissolved organic matter (DOM) or dissolved organic carbon (DOC) complex. DOC in most aquatic ecosystems is composed of up to 50% humic substances (Suhett et al., 2007) and is known to affect light attenuation in water (Curtis & Adams, 1995). DOC can modify the aquatic trophic structure through decreasing primary production by reducing the light that reaches the autotrophs and by acting as a source of energy for bacteria (Jones, 1992). Additionally, humic substances often affect water color, which can be estimated with the water absorbance at 440 nm (g440) in most inland and estuarine waters (Kirk, 2011).

Different life strategies in a given species can favor polymorphism as a result of different evolutionary forces such as genetic drift, migration, or adaptation to different environments. This is especially prevalent in fish, as they are more susceptible than other vertebrates to morphological variation induced by environmental factors (Ferriz & Gómez, 2015). A decrease in water transparency is known to affect; visual feeding (e.g., Vinyard & O’Brien, 1976), predators avoidance (e.g., Ljunggren & Sandström, 2007) and communication (e.g., Maan et al., 2010). Different environments drive adaptations that increase fitness, eventually leading to speciation (McPhail, 1984). The eyes of mammals, birds, amphibians, reptiles, and some elasmobranchs have muscle mechanisms to control the amount of light they detect (Douglas et al., 1998). However, most teleost cannot increase or reduce the pupil’s size because their iris is immobile (Helfman et al., 2009). Thus, the amount of light they detect (i.e., photons) depends on the conditions of the luminosity of the environment which is affected by factors such as depth, water color, or turbidity.

Like any other sensorial system, vision requires high energy expenditure (Niven & Laughlin, 2008). Hence, from an evolutionary point of view, in environments where light penetration is low, there is a trade-off between developing larger eyes or losing them completely (Dugas & Franssen, 2012). Energy acquisition is intrinsically related to feeding (Moran et al., 2015). The development of efficient visual structures contributes to the improvement of efficient feeding systems, such as a larger mouth, or changes in foraging behavior, since larger prey may be more visible (Carter et al., 2010). More so, according to the optimal foraging theory, individuals with a larger mouth can maximize energy allocation by feeding on larger prey items, resulting in increased fitness. Fish can detect their predators through visual, acoustic, or chemical signals (Milano et al., 2010). For an object to be visible, whether prey or a predator, a physiological threshold (ability to detect light) of contrast with the surroundings must be overcome (Aksnes & Giske, 1993). In the case of fish, the threshold depends on factors such as the size of the object and the intensity of the light (Anthony, 1981). For this reason, fish feeding efficiency and capacity to detect predators may decrease with higher turbidity or darker water color (Hecht & van der Lingen, 1992).

Galaxias maculatus (Jenyns, 1842) is one of the world’s most widespread fish and is native to the Southern hemisphere (McDowall, 1968). It has two life strategies: diadromous (amphidromous), where its larval stage develops in the sea; and landlocked, where its entire life cycle develops in lakes, lagoons, or rivers (e.g., Cussac et al., 1992). It is considered a carnivorous species that is a visual predator, mostly on macroinvertebrates (McDowall, 1968). Across its wide geographical range, the species exhibits a great plasticity in its life history including diadromous and landlocked populations, reproduction, maximum size of individuals, morphological variation and behavior (McDowall, 2003; Chapman et al., 2006; Boy et al., 2009; Barbee et al., 2011; Barriga et al., 2012; Rojo et al., 2018). It has previously been described that its foraging efficiency decreases with increasing turbidity (Rowe & Dean, 1998) and that its vision is not only crucial for feeding but also to avoid predators (Milano et al., 2010).

The aim of this work is to investigate phenotypic differences in eye and mouth size of G. maculatus in five water bodies of Tierra del Fuego with distinct light climates: two brown humic lakes and three clear water bodies (two estuaries and one river).

Materials and methods

Sampling individuals

Sampling was performed between January 2015 and January 2019 in two estuaries, Arroyo Negro (AN, 54° 50′ 53″ S, 68° 34′ 51″ W) and Bahía Varela (BV, 54° 52′ 22″ S, 67° 16′ 29″ W); two humic lakes, Laguna Cecilia (LC, 54° 49′ 52″ S, 68° 34′ 37″ W) and Laguna Negra (LN, 54° 50′ 45″ S, 68° 35′ 31″ W); and one river, Río Grande (RG, 53° 49′ 16″ S, 67° 52′ 58″ W) (Fig. 1).

Fig. 1
figure 1

Sampling sites of Galaxias maculatus in Tierra del Fuego. AN Arroyo Negro, BV Bahía Varela, LC Laguna Cecilia, LN Laguna Negra, RG Río Grande, TDFNP Tierra del Fuego National Park. Inset: section of the TDFNP showing sampling sites

Full size image

Galaxias maculatus were collected with a trawl net (10 m long, 1 m high, 5 mm mesh size) two hours after high tide in AN (n = 60) and BV (n = 40), and with an electrofishing equipment (Smith-Root, Inc. model LR-24, Vancouver, WA, USA) with 400 V, frequency of 100 Hz and duty cycle of 2 ms in LC (n = 40), LN (n = 60) and RG (n = 45), counting a total of 245 individuals. Fish were taken to the laboratory in aerated 100 l tanks with water at environmental temperature. In the laboratory, fish were euthanized with tricaine methanesulfonate (MS-222, Sigma-Aldrich, St. Louis, MO, USA) in a concentration above 300 mg of MS-222 per liter of water.

Environmental parameters

Dissolved oxygen, conductivity, pH, and water temperature were measured in each water body with a Hanna HI9828 multiparameter probe (Hanna Instruments, Inc., Woonsocket, RI, USA). Photosynthetic active radiation (PAR) was measured every five centimeters in the water column with a light meter Li-250A equipped with a spherical quantum sensor Li-193SA (Li-Cor, Lincoln, NE, USA). The vertical attenuation coefficient, Kd (in m−1; n = 3, 1, 2, 3 and 2 for AN, BV, LC, LN and RG, respectively), was obtained as the slope of the linear regression of the natural logarithm of PAR versus depth (Kirk, 2011).

Water samples (2 in AN, 1 in BV, 2 in LC, 3 in LN and 1 in RG) were taken from each water body for further measurements of absorbance at 440 nm in a GBC Cintra 10e spectrophotometer (GBC Scientific, Braeside, Australia) in water filtered through Whatman GF/F filters. The absorption coefficient at 440 nm is used here as a water color proxy (g440, Kirk, 2011). Suspended solids concentration (SS) was obtained by filtering a known volume of water through pre-dried and weighed MG-C filters of 1.2 μm pore size. The filters with the solids were dried (60°C) until constant weight and weighed again (APHA, 2005).

Eye size and mouth length

Total length (TL) and standard length (SL) of all individuals were measured with an accuracy of 0.01 mm. After the individuals were euthanized, photographs of the left side of each individual were taken with a 16-megapixel digital camera mounted on a photographic stand with a scale reference. The following five homologous points (landmarks) were digitized with the software tpsDig2 v2.22 (Rohlf, 2015): 1. The anterior end of the premaxilla, 2. Posterior end of the mouth corner, 3. The anterior end of the eye, 4. Eye center, 5. Posterior end of the eye (Fig. 2).

Fig. 2
figure 2

Morphological measurements and location of the landmarks used in the morphometry analysis of Galaxias maculatus from the five sampling sites in Tierra del Fuego. 1. The anterior end of the premaxilla, 2. Posterior end of the mouth corner, 3. The anterior end of the eye, 4. Eye center, 5. Posterior end of the eye. Total length (TL), standard length (SL), eye diameter (ED), mouth length (ML)

Full size image

The diameter of the eye (ED) and the mouth length (ML) were measured as the Euclidean distance between landmarks 3 and 5 for ED and between 1 and 2 for ML, and their relationships with the standard length (EDR and MLR, respectively) was calculated by \({\text{EDR}} = {\frac{{\text{ED}}}{{\text{SL}}}} \times 100\) and \({\text{MLR}}=\frac{{\text{ML}}}{{\text{SL}}} \times 100.\)

Statistical analyses

All statistical analyses were performed using R (R Core Team, 2018). The comparisons of EDR and MLR among sampling sites were analyzed using a Dunn test after the Kruskal–Wallis test with the dunn.test package (Dinno, 2017), since the assumptions of normality and homoscedasticity were not met, and transformations did not improve the fit.

In addition, Pearson correlations were made between Kd and water temperature and g440. Spearman correlations were made between Kd and EDR and MLR; g440 and EDR and MLR; and between SS and EDR and MLR since the assumption of normality for EDR and MLR was not met.

Results

Dissolved oxygen was similar among sites, ranging between 9.1 and 12.3 ppm. Conductivity was higher in Arroyo Negro, probably due to the influence of the tide during sampling provided its proximity to Beagle Channel (Fig. 1). The pH was also similar among sites, ranging between 6.9 and 8.2 (Table 1). During the study, the Kd ranged between 0.43 and 2.4 m−1 for Arroyo Negro in January and Laguna Cecilia in April, respectively. The absorbance at 440 nm (g440) ranged between 0.07 and 2.52 m−1 for Arroyo Negro in January and Laguna Cecilia in February, respectively. Suspended solids concentration (SS) ranged between 0.44 mg l−1 in Laguna Negra and 7.05 mg l−1 in Bahía Varela (Table 1).

Table 1 Morphometric variables from Galaxias maculatus and environmental parameters from the sampling sites in Tierra del Fuego
Full size table

The Kd was not dependent on water temperature as revealed by the lack of significant correlation between Kd and water temperature (n = 9; r = 0.30; P = 0.43). Hence, we dismiss seasonal effects in the interpretation of our results. Kd and g440 were positively correlated (n = 9, r = 0.77, P < 0.05).

Total length of G. maculatus individuals ranged between 41.7 and 103.4 mm. Relative eye diameter (EDR) ranged between 3.74 for Bahía Varela and 6.16 for Laguna Negra (Table 1). Relative mouth length (MLR) ranged between 3.85 for Arroyo Negro and 7.37 for Laguna Negra. EDR and MLR were significantly different among sampling sites (Kruskal–Wallis, X2 = 179.27, P < 0.05; X2 = 118.48, P < 0.05, respectively) (Fig. 3). Both EDR and MLR were significantly larger in humic lakes (Laguna Cecilia and Laguna Negra) than in the rest of the sites, with clearer waters (Arroyo Negro, Bahía Varela and Río Grande) (X2 = 166.21, P < 0.05 and X2 = 102.97, P < 0.05, respectively), being EDR 20% and MLR 17% higher in humic lakes.

Fig. 3
figure 3

a Relative eye diameter (EDR) and b relative mouth length (MLR) of Galaxias maculatus from sampling sites in Tierra del Fuego. AN Arroyo Negro, BV Bahía Varela, LC Laguna Cecilia, LN Laguna Negra, RG Río Grande. Thin lines represent maximum and minimum values, the bottom and the top of the box represent the 25% and 75% quartile, respectively, the bold line indicates the median and the × indicate outliers. Different lowercase letters denote significant differences between sampling sites (Dunn test, α = 0.05)

Full size image

Both the EDR and MLR were positively correlated with the Kd (n = 11, rs = 0.87, P < 0.05 and rs = 0.84, P < 0.05, respectively) and with g440 (n = 9, rs = 0.76, P < 0.05 and rs = 0.83, P < 0.05, respectively); but not with SS (n = 9, rs = − 0.34, P = 0.38 and rs = − 0.20, P = 0.61, for EDR and MLR, respectively) (Fig. 4). In addition, EDR and MLR were positively correlated (n = 11, rs = 0.72, P < 0.05).

Fig. 4
figure 4

Morphometric variables in relation to environmental variables for Galaxias maculatus of the five sampling sites in Tierra del Fuego. a Relative eye diameter (EDR) and the light attenuation coefficient (Kd), b relative mouth length (MLR) and Kd, c EDR and the absorption coefficient at 440 nm (g440), d MLR and g440, e EDR and suspended solids concentration (SS) and f MLR and SS. AN Arroyo Negro, BV Bahía Varela, LC Laguna Cecilia, LN Laguna Negra, RG Río Grande. For EDR and MLR, the mean ± standard deviation is shown

Full size image

Discussion

In this study, we show that the eye size of G. maculatus, a possible proxy for its visual capacity, is larger in fish inhabiting environments with poor light penetration than in clear water environments. Water transparency, either due to turbidity or color, affects several processes in aquatic ecosystems, such as intraspecific behavior and interactions in the case of fish that depend on their visual capacity (e.g., Heubel & Schlupp, 2006; Ljunggren & Sandström, 2007). Turbidity and color affect water transparency independently (Koenings & Edmundson, 1991) and have different effects on fish survival and feed efficiency. For example, the exogenous feeding of walleye fry [Sander vitreus (Mitchill, 1818)] starts earlier in turbid water than in dye-colored water (Bristow et al., 1996). Due to variations in water transparency, organisms can change their feeding strategy or exploitation of food resources, for instance, from small to larger prey, and this can affect their morphology (e.g., Breck, 1993; Magnhagen & Heibo, 2001). G. maculatus is described as a visual predator (McDowall, 1968) and has larger eyes than other species of Galaxias (McDowall, 1990; Rowe & Dean, 1998). Therefore, its foraging ability is expected to have a close relationship with visibility, since larger eyes often have better resolution (Li et al., 1985). It is also expected that foraging ability is related to reactive distance imposed by the different environments, i.e., the largest distance at which a fish can locate prey or predator (Vinyard & O’Brien, 1976). Hence, Rowe & Dean (1998) found in culture ponds that as turbidity increased, juvenile G. maculatus decreased their feeding ability. Consequently, the larger eye diameter found in Laguna Cecilia and Laguna Negra individuals (20% greater than in Arroyo Negro, Bahía Varela, and Río Grande individuals) could be an adaptation to the reduction of the penetration of light. Similar results were reported by Bartels et al. (2016) which found larger eye size with increasing DOC concentrations (i.e., high concentration of humic substances) in perch (Perca fluviatilis, Linnaeus, 1758). Dugas & Franssen (2012) observed in the cyprinid Cyprinella lutrensis (Baird & Girard, 1853) a positive association between eye size and turbidity (i.e., larger eye size in turbid environments). Lattuca et al. (2007) also found a relationship between lower water transparency of Andean-Patagonian lakes and a larger eye size in juveniles of Aplochiton zebra Jenyns, 1842. In regard to the retina of G. maculatus, Ali et al. (1990) found differences between individuals from light and ‘dark’ running waters, mainly in the location of the cones, although no light measurements from the environment were reported. Thus, individuals of G. maculatus from different light climates would not only present differences in the structure of the retina as suggested by Ali et al. (1990), but also, as our study suggest, in the eye size.

In the studied lakes (Negra and Cecilia), low light penetration is mainly related to the color of the water due to the high amounts of colored dissolved organic carbon (DOC, ca. 8 and 16 mg l−1 in Laguna Negra and Laguna Cecilia, respectively; Rodríguez unpublished) and not due to turbidity or suspended solids concentration. In addition, the positive relationship between Kd and the absorbance at 440 nm shows that humic substances have a direct and positive effect on total Kd (i.e., the higher g440, the higher total Kd) since the total Kd depends on the light absorption of water itself, phytoplankton, tripton and humic substances (i.e., Kd = KdW + KdPhi+ KdTri + KdG). For instance, in Laguna Negra, both turbidity and suspended solids concentrations are in general low (ca. 1.5 NTU and about 0.9 mg l−1, respectively, data not shown). Thus, turbidity (i.e., dissolved or suspended substances which scatter light) might not determine the light environment in these humic lakes. Even though turbidity could not be measured in Laguna Cecilia, it can be suggested that its values are similar to Laguna Negra since suspended solids concentration was also low, and color, and Kd were also high. In these humic lakes, the larger contribution to the water color is probably because the southern lakes of the Isla Grande de Tierra del Fuego usually contain high amounts of humic substances due to the large biomass of Sphagnum sp., which give them their characteristic brown color (Saad et al., 2016). Although in this study, we observed that the smaller the light penetration, the greater the eye size, preliminary results suggest that the DOC threshold of the visual capacity of G. maculatus should be lower than 50 mg l−1. In a highly humic lake from the east of Tierra del Fuego (54° 58′ 1.41″ S, 66° 43′ 54.33″ W) with severe light limitation due to high concentrations of humic substances (Kd = 6 m−1; ca. 50 mg l−1 in DOC, Rodríguez unpublished), no fish were captured. However, as pH in the mentioned lake is around 4, more research is needed to test whether light or light plus another environmental factor besides fish distribution is limiting G. maculatus in highly humic lakes from the east of the island.

In addition, the mouth length was larger in the environments with less light penetration. This could be an adaptation related to the size of the prey and feeding efficiency (e.g., Karpouzi & Stergiou, 2003). Larger mouth length in darker environments would increase feeding efficiency, increasing the capacity of individuals to ingest larger prey, that we presume more detectable than smaller ones in such environments. Vinyard & O’Brien (1976) working with a turbidity gradient of clay sediments in the bluegill Lepomis macrochirus Rafinesque, 1819, found at low turbidity conditions (6.25 or 10 Jackson Turbidity Units, JTU) a positive linear relationship between reactive distance and prey size, which diminished in high turbidity conditions (30 JTU), where reactive distance was independent of the prey size. However, we cannot expect the same response as turbidity in our dark sites was low. Mouth length, among other morphological characteristics (e.g., fins shape and location, body depth, coloration), in many cases, influence the niche occupied by organisms (Karpouzi & Stergiou, 2003). Our results regarding mouth length —larger in Laguna Cecilia individuals, intermediate in individuals from Laguna Negra and smaller in Arroyo Negro, Bahía Varela, and Río Grande—could be related to feeding ecology and behavior, as has been described in other species (e.g., Keast & Webb, 1966; Norton, 1991). For instance, in A. zebra, a relationship has been found between the shape of the mouth and the gape size with the geographic distribution in different lakes and diet (Lattuca et al., 2007). According to the optimal foraging theory, those individuals who feed more efficiently (i.e., ingesting larger prey) and maximize their energy allocation, have greater fitness and thus contribute more to offspring (e.g., Pyke, 1984; Tytler & Calow, 1985).

We observed a similar eye and mouth length in Arroyo Negro, Bahía Varela, and Río Grande. These sites present different limnological conditions but similar light climates, which widen the extent of our results. For example, as Arroyo Negro and Bahía Varela are estuaries, differences in tide influence or current velocity can be found between these two clear water systems and Río Grande, which is a clear water river. These differences did not hinder us from detecting a clear pattern of smaller eyes with clear waters. This strengthens the hypothesis that the water color could be one of the strongest modeling factors in the environment regarding eye size and mouth length of G. maculatus.

We conclude that individuals of G. maculatus that inhabit environments with less light penetration have larger eyes, probably related to improved feeding efficiency and predator avoidance. We also observed individuals with larger mouths in darker environments, which we discuss as an advantage to catch larger prey. Our results also indicate that color, due to higher concentrations of humic substances, was the driving force influencing light attenuation in these environments. Furthermore, the largest relative eye diameter was observed in more isolated populations, potentially related to a lesser gene flow and genetic drift in the landlocked populations (Laguna Cecilia and Laguna Negra) than in the diadromous ones (Arroyo Negro, Bahía Varela, and Río Grande), due to less migration or smaller population size. The results found could be strengthened through experimental laboratory corroboration to evaluate the larger eye-better vision assumption. For example, by measuring the reaction time of adult individuals with different eye size, or by measuring the eyes diameter in individuals reared at different light conditions. In addition, further studies of the retina, as well as diet and food availability, are needed in these populations of G. maculatus to fill the gaps in the knowledge of the species. Also, the relationship between morphology and population genetic structure is worth to study in search of a better comprehension of the wide polymorphism of G. maculatus’ populations.