1 Introduction
Decline and extinction of amphibian species have been reported in many areas of the world. It has been suggested that infectious diseases [1–3], parasitic infections [4,5], ultraviolet radiation [6], chemical pollutants [7,8], introduced non-native species [9,10], climate change [11] and the destruction of terrestrial and aquatic habitats [12,13] may all be important factors causing the decline of amphibian species. Many urodeles and anurans are extremely dependent on aquatic habitats because of their physiology and natural history. They deposit their eggs in aquatic habitats and the eggs will develop normally only within certain limits of various environmental factors. Moreover, amphibians have highly permeable skin and their larvae may be susceptible to the water chemistry [14]. Thus, some chemicals, such as nitrates and chlorides in water, possibly act as important variables that influence amphibian distribution, abundance and selection of breeding sites [15,16].
Detailed information about the ecological demands of species, including oviposition or breeding sites, is obligatory for successful conservation of endangered amphibians [17–20]. The Caucasian salamander, Mertensiella caucasica is a narrow-ranged species found in the western part of the Lesser Caucasus mountain systems in Georgia and Turkey, and has been categorized as “Vulnerable” by the IUCN [21]. Its red list status stresses the importance of studying the ecological characteristics of this species in order to formulate appropriate management and conservation strategies. In general, salamanders are found only in or near running water such as mountain brooks and streams and, potentially, their distribution can be strongly limited by water quality. Despite the fact that some papers focus on the distribution, taxonomy and ecology of the Caucasian salamander, no detailed investigation about breeding habitat selection has been undertaken [22]. Thus, the aim of this study is two fold; first to determine the characteristics of the water chemistry of breeding habitats and second to determine the relationship between the water chemistry parameters and the species' presence.
2 Material and methods
2.1 Field and laboratory work
The field work was carried out in 2007 and covered throughout the entire distribution range of the Caucasian salamander from Borjomi area in Georgia to Ordu area in Turkey. The field trips covered the second half of June and the first half of July, when it is relatively easy to find both the larvae and adult animals [23].
Water samples were collected from 37 different breeding habitats where both the adult animals and the larvae are found (see Appendix A). Eight sites were sampled as non-breeding habitats where neither the adult animals and nor the larvae were present, although these are occasionally used for breeding by other amphibians found in the region: Rana ridibunda, Rana macrocnemis, Bufo viridis, Bufo bufo, Hyla arborea, and Pelodytes caucasicus, Triturus vittatus, Triturus karelinii [22,23]. The breeding sites sampled during the field work covered the entire range of the salamander territory, including its westernmost and easternmost parts, and a broad range of ecological conditions: the elevation of individual locations varied between 40 and over 2400 m a.s.l., and the locations were distributed in various landscapes, including mixed and broadleaf forests, areas near and above timberline. Both “presence” and “absence” samples were taken from running water sources (small rivers, streams, or brooks) within the extent of occurrence of the species. Water conductivity, pH, dissolved oxygen (DO) and salinity were measured in the field using a Hach Portable pH/Conductivity/Dissolved Oxygen Meter, and water temperature was measured with a digital thermometer. Some water chemistry parameters such as iron, manganese, chloride, ammonium, sulphate, potassium, nitrate and hardness (calcium and magnesium) were measured in the laboratory using a DR 2800 VIS Spectrophotometer, following the manufacturer's procedures. In addition, total alkalinity was measured using test strips (Aquacheck, Hach Company, USA). For the list of analyzed variables, see Table 1.
Descriptive statistics of water chemistry variables in breeding and non-breeding habitats of M. caucasica.
| Variables | Breeding habitats ( ) | Non-breeding habitats () | Statistical significance | |||
| Mean | STDEV | Mean | STDEV | |||
| pH ∗ | 7.65 | 0.370 | 8.25 | 0.640 | t = 3.62685 | P-value = 0.0007 |
| DO (mg/L) | 8.33 | 0.816 | 7.98 | 0.820 | t = −1.11917 | P-value = 0.2690 |
| Conductivity (μS/cm) ∗ | 140.42 | 106.228 | 310.24 | 199.521 | t = 3.3418 | P-value = 0.0017 |
| Salinity (%) ∗ | 0.06 | 0.048 | 0.14 | 0.098 | t = 3.51587 | P-value = 0.0010 |
| Iron (mg/L) | 0.03 | 0.052 | 0.01 | 0.018 | t = −0.96489 | P-value = 0.3400 |
| Manganese (mg/L) | 0.46 | 0.175 | 0.55 | 0.151 | t = 1.28599 | P-value = 0.2053 |
| Chloride (mg/L) ∗ | 0.35 | 0.47 | 2.27 | 3.186 | t = 4.05335 | P-value = 0.0002 |
| Ammonia (mg/L) ∗ | 0.01 | 0.032 | 0.41 | 0.761 | t = 3.42121 | P-value = 0.0013 |
| Alkanity (ppm CaCO3) ∗ | 1.21 | 1.522 | 2.87 | 1.642 | t = 3.27325 | P-value = 0.0020 |
| Sulfide (μg/L) ∗ | 2.07 | 2.518 | 7.25 | 9.407 | t = 3.01694 | P-value = 0.0042 |
| Potassium (mg/L) ∗ | 0.84 | 0.452 | 2.02 | 2.240 | t = 3.08045 | P-value = 0.0035 |
| Nitrate (mg/L) | 0.52 | 0.645 | 1.56 | 3.425 | t = 1.79356 | P-value = 0.0797 |
| Hardness Mg (mg/L) | 3.07 | 0.817 | 3.34 | 0.743 | t = 0.81313 | P-value = 0.4206 |
| Hardness Ca (mg/L) | 0.66 | 1.017 | 0.64 | 1.015 | t = −0.04645 | P-value = 0.9631 |
2.2 Statistical analysis
For each variable, as a first step, the statistical significance of differences between the “presence” and “absence” water source were evaluated by means of Students t test. To determine which particular water chemistry variables help to separate presence or absence locations, we used stepwise multiple logistic regression analysis [24]. We used species presence or absence as the dependent variable and water chemistry variables as independent variables. The analyses were performed with SPSS Ver. 13 for Windows (SPSS Inc., Chicago, IL, USA) with a forward stepwise entry of independent variables.
3 Results
Descriptive statistics of water chemistry variables of breeding and non-breeding habitats of M. caucasica, obtained from analyses were given in Table 1.
Although eight of the studied variables showed significant differences between the presence and absence locations, the logistic regression analysis showed the best classification results with only two variables, chloride and O2 retained in the equation at the step 2 (Table 2). Overall, 88.9% of investigated locations were correctly classified (Table 3). The most important parameter was concentration of chloride in water (Table 2).
Results of logistic regression analysis for water chemistry variables; variables in the equation.
| B | S.E. | Wald | df | Sig. | Exp(B) | ||
| Step 1 (a) | Chloride | 1.519 | 0.627 | 5.865 | 1 | 0.015 | 4.569 |
| Constant | −2.692 | 0.661 | 16.577 | 1 | 0.000 | 0.068 | |
| Step 2 (b) | O2 | −0.891 | 0.507 | 3.090 | 1 | 0.079 | 0.410 |
| Chloride | 1.631 | 0.708 | 5.306 | 1 | 0.021 | 5.107 | |
| Constant | 4.467 | 4.001 | 1.246 | 1 | 0.264 | 87.100 |
Classification table of logistic regression analysis for water chemistry variables.
| Predicted | |||||
| Presence | Percentage correct | ||||
| 1 | 2 | ||||
| Step 1 | Presence | 1 | 37 | 1 | 97.4 |
| 2 | 5 | 2 | 28.6 | ||
| Overall percentage | 86.7 | ||||
| Step 2 | Presence | 1 | 37 | 1 | 97.4 |
| 2 | 4 | 3 | 42.9 | ||
| Overall percentage | 88.9 |
4 Discussion
Many factors have responsibilities for declining amphibian populations and it has been shown by a great deal of studies that these factors contribute to their decline alone or synergistically. Of these factors, habitat loss and fragmentation are among the largest threats to amphibian populations. Here we study some aspects of the breeding habitat of Caucasian salamander, which have a crucial part in breeding and survival of their life history.
Ecological features influencing the occurrences of amphibians in certain habitats have been subjected to numerous studies [25–31]. Some characteristics of water including turbidity, pH, density, chlorine, chloramines, nitrate, nitrite, ammonium, heavy metals and also surrounding vegetation cover, permanency and presence of tadpole predators can influence anuran breeding site choice [32–39]. We used 14 water chemistry variables in order to investigate which ecological parameters better explained Caucasian salamander breeding site selection. Mean values of eight parameters were significantly different between presence and absence locations (Table 1). The values of pH, water conductivity, salinity, alkalinity and concentrations of ammonia, sulfide and potassium of the salamander breeding sites were significantly lower than in streams and brooks where the salamanders do not breed. However, according to the results of the logistic regression analysis, concentration of chloride and, perhaps, oxygen consumption are variables sufficient for separating the sites appropriate and inappropriate for Caucasian salamander reproduction (Tables 2 and 3).
Chloride ions play important role in vertebrate physiology. It is a chemical the body needs for energy metabolism. It also helps to keep the body's acid-base balance. The ebb and flow of chloride ions into and out of cells can alter cells electrical properties, influencing nerve function, muscle contraction and a variety of other processes [40]. According to Kaushal et al. [41] chloride concentrations are increasing at a rate that threatens the availability of fresh water in the northeastern United States. Concentrations of chloride in soils as low as 30 mg/L have been found to damage terrestrial plants. Increased chloride concentrations in surface waters could lead to adverse effects on water quality. Increases in salinity up to 1000 mg/L can have lethal and sublethal effects on aquatic plants and invertebrates [42], and chronic concentrations of chloride as low as 250 mg/L have been recognized as harmful to freshwater life and not suitable for human consumption [43,44]. Most urodeles and anurans have a highly permeable skin, largely unprotected aquatic eggs and aquatic larvae. Their adults use wetlands for foraging, hibernation and breeding. Because of these peculiar characteristics, they are considered as excellent indicators of ecosystem health. In our study, among the investigated water chemistry parameters affecting the presence of Caucasian salamander, chloride concentration was shown to be the most important one. The threshold level of chloride in salamander habitats (range = 0–2.0 mg/L) appears to be lower than in other amphibians of the region that breed, usually or occasionally, in running waters and presence of these animals could be used as a perfect biological indicator of water quality.
According to Cushman [45], in most part of the world, there is very limited knowledge of the species–environment relationships of amphibians, their responses to habitat loss and fragmentation and the factors controlling population connectivity. There are many different studies suggesting the possible ways to protect and conserve amphibian populations, however, the best way to start preserving amphibian populations could be to identify important amphibian habitat characteristics and to prevent human-caused modification of these variables in the first place.
Acknowledgements
Our special thanks are to Irina Serbinova and Giorgi Chaladze for valuable assistance during the field works. This work was supported by CEPF and WWF Caucasus Office (the project “Transboundary Conservation-Oriented Study and Conservation of the Western Lesser Caucasus Endemic species–Caucasian salamander (Mertensiella caucasica)”) and Ege University, Research and Application Center of Science and Technology (EBİLTEM, Project No. 2007/BİL/017).
Appendix A
The sampled breeding and non-breeding sites of M. caucasica in Turkey (TR) and Georgia (GEO) are listed.
Breeding sites: Kovanlık-Giresun (TR), Karapınar (two sampling sites)-Giresun (TR), Bayındır (two sampling sites)-Giresun (TR), Turnalık (three sampling sites)-Ordu (TR), Çambaşı-Ordu (TR), Sümela-Trabzon (TR), Coşandere (two sampling sites)-Trabzon (TR), Kiraz Yaylası-Trabzon (TR), Uzungöl (three sampling sites)-Trabzon (TR), Demirkapı (three sampling sites)-Trabzon (TR), Ayder (three sampling sites)-Rize (TR), Kavrun-Rize (TR), Çamlıhemşin-Rize (TR), Ülküköy-Rize (TR), Fındıklı-Rize (TR), Çarnali-Batumi (GEO), Batumi Botanical Garden-Batumi (GEO), Goderdzi Pass (four sampling sites)-Akhaltsikhe (GEO), Atskuri-Akhaltsikhe (GEO), Kekia-Borjomi (GEO), Savaniskhevi-Borjomi (GEO), Chitakhevi-Borjomi (GEO), Abastumani-Akhaltsikhe (GEO).
Non-breeding sites: Karagöl (two sampling sites)-Artvin (TR), Goderdzi Pass-Akhaltsikhe (GEO), Abastumani-Akhaltsikhe (GEO), Kekia-Borjomi (GEO), Nedzvi (three sampling sites)-Borjomi (GEO).