SEARCH

SEARCH BY CITATION

Keywords:

  • IUCN Red List;
  • habitat loss;
  • von Bertalanffy growth model;
  • modal progression analysis;
  • conservation

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  1. Information on the life-history traits of threatened fishes can inform their conservation as they can indicate resilience to environmental change and vulnerability to extirpation and extinction. Garra ghorensis, a small (< 140 mm total length) riverine cyprinid fish, endemic to the southern Dead Sea area, is critically endangered through habitat loss and invasive species. There are, however, no data currently available on their life-history traits to inform their conservation management.
  2. The age structure and growth rate characteristics of three G. ghorensis populations in Jordan were assessed. Close to the sampling sites, minimum air temperatures approached 0 °C in January but maxima exceeded 40 °C in July and August. Samples collected monthly throughout 2011 contained fish with lengths between 23 and 137 mm, with most less than 100 mm. Monthly length distributions showed three distinct length modes in each population whose mean lengths increased throughout the warmer months.
  3. Growth check formation on scales was annual and their ageing revealed fish in each population present to at least 4 years old, with a maximum of 6 years. Comparison of these data with the length modes indicated that the modes corresponded to ages 0+, 1+ and >2 years. Variability in length at age was apparent within sites, suggesting protracted spawning. Females were significantly larger than males.
  4. Growth rates and lifespans of G. ghorensis were highest at the most disturbed site (habitat loss and the presence of the invasive Oreochromis aureus). This growth plasticity in response to slower flows and deeper water suggests G. ghorensis has some resilience to environmental disturbances and suggests that their conservation management might not have to return their habitats to pristine conditions to avoid impacts on their lifespan and growth parameters. It also shows that further work is needed to identify the issues that are affecting the persistence of their populations.

Copyright © 2014 John Wiley & Sons, Ltd.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

In arid regions the escalating demands for water have resulted in the substantial modification of many river systems (Propst et al., 2008). In conjunction with the widespread invasion of many of these rivers by non-native fishes, this has increased the threat of local native fish populations being extirpated and endemic fishes becoming extinct (Kingsford, 2000; Olden and Poff, 2005; Propst et al., 2008). However, the risks of extirpation and extinction vary between species according to the traits that determine their responses to modified environments (McKinney, 1997; Olden et al., 2006, 2008). These ecological attributes, including their life-history traits, ecological niche, and morphology, have been applied to disturbed rivers in arid regions to identify those species most at risk to extinction so that conservation strategies can focus on their populations (Olden et al., 2006, 2008; Clarkson et al., 2012; Pool and Olden, 2012). Studies suggest that fishes with a ‘slow’ life history, with features such as large body sizes, slow somatic growth rates and delayed maturity (i.e. K-selected traits), tend to have a greater frequency of local extirpation and are more prone to extinction compared with those with the opposite suite of traits (i.e. r-selected traits) (Olden et al., 2008).

In fish populations, lifespan, age structure and somatic growth rates form an important component of their life-history strategy through their relationships with reproductive traits such as length and age of maturity (Winemiller and Rose, 1992). Moreover, fish age and growth data assist the understanding of the basic ecological relationships of fish populations and their interactions with their environments (Beardsley and Britton, 2012; Britton et al., 2012).Given that life-history traits can be a strong predictor of extirpations of fish populations in arid regions (Angermeier, 1995; Reynolds et al., 2005; Olden et al., 2006) the analysis of a species’ age composition and growth rates can help to explain how threatened species might be conserved in these regions by indicating their initial responses to the environmental changes. Where data indicate, for example, significantly reduced lifespans this might suggest the species has relatively narrow tolerances to disturbance that ultimately could result in population decline and eventual loss.

Garra ghorensis (Krupp, 1982) is a small, endemic fish of the Cyprinidae family that occurs in Jordan from south of the Mujib River through to the southern end of the Dead Sea basin (Krupp and Schneider, 1989; Hamidan and Mir, 2003). In recent years this region has been subjected to major river modifications through the construction of dams and impoundments that were designed to increase the supply of potable and irrigation water, and have consequently altered the flow regime and physical habitat structure of the rivers (Hamidan and Mir, 2003). These changes have coincided with the appearance of some non-native species, especially Oreochromis aureus. Overall, G. ghorensis is Red Listed as ‘critically endangered’ because populations are close to extirpation in many areas of their range and there have been general population declines in others (IUCN, 2012). Where the declines have been quantified, these have been up to 90% owing to factors such as habitat destruction and degradation and the introduced Gambusia affinis (Goren and Ortal, 1999; IUCN, 2012). In Jordan, although losses are not known accurately, preliminary, country-wide field surveys suggest that the species has suffered a range contraction to less than 10 km2 (Hamidan and Mir, 2003; IUCN, 2012).

In view of the critically endangered status of G. ghorensis, there is an urgent requirement for their remaining populations to be conserved. However, there is little information available on the ecology of the species, with the only study on their Jordanian populations being a distribution survey completed in the early 2000s (Hamidan and Mir, 2003). As there has been no subsequent study that provides data on their life-history traits, the aim here was to start to address this by determining the age composition and growth rates of three Jordanian G. ghorensis populations. By collecting monthly samples throughout 2011 and early 2012, the objectives were to (1) identify the presence of length modes in the samples by month and assess their growth through 2011; (2) quantify the utility of using scales to age individual fish within the populations; (3) determine the age structure and growth rates of the three populations using scale analysis; and (4) identify the conservation implications for G. ghorensis of the outputs of Objectives 1 to 3.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Study area

The three sites were Wadi-al-Burbaita(35°69'E, 30°98'N), Ain al-Haditha (35°54'E, 31°29'N) and Wadi Ibn Hammad (35°38'E, 31°18'N) (Figure 1). All sites were known to contain populations of G. ghorensis that were believed to be sufficiently numerous to allow some fish removals for laboratory analyses and the relatively easy access they provided for sampling expeditions. Ibn Hammad (hereafter referred to as site IB) is a relatively fast flowing and shallow habitat in which G. ghorensis was the only fish present. Wadi-al-Burbaita (hereafter referred to as site BB) has more variable flow rates and some human disturbances through water use for domestic and agricultural uses, with G. ghorensis present together with the native Capoeta damascina. Ain al-Haditha (hereafter referred to site HD) was the most disturbed site, with local impoundments creating sections of slower and deeper water. The invasive cichlid Oreochromis aureus was also present at the site, but no other fishes were recorded. Because of the conditions of each site, water temperature loggers could not be easily deployed and so instead air temperatures from local weather recording stations were used as a surrogate. These data helped in the subsequent interpretation of the temperatures required for initiating the growth of G. ghorensis.

image

Figure 1. Map showing the location of sites in relation to the Middle East (inset) and Jordan (main). The sampling site locations are shown on the main map by the solid black square, where Wadi Ibn Hammad is Site IB, Al Burbaita is Site BB and Al-Haditha is Site HD.

Download figure to PowerPoint

Fish sampling

Fish were sampled (under licence to the Royal Society of Nature Conservation, Jordan) once per month between February 2011 and January 2012 by electric fishing. This was completed in an upstream direction for a standard time of 30 min using handheld Samus 725 MP electro-fishing equipment. Where fewer than 15 fish were captured in this period fishing was continued for 60 min. All of the captured fish were processed in the field before a subsample was taken to the laboratory for further analysis. These subsampled fish were given an overdose of anaesthetic (clove oil; (Soto, 1995)) before being preserved in ethanol. In the laboratory, each fish was assigned a reference number, measured using digital vernier callipers (total length, nearest 0.1 mm) and up to six scales were removed from the area between the dorsal fin and lateral line. These were transferred to paper envelopes for drying and storage.

Age and growth analysis

To identify the presence of length modes in the samples each month and assess their growth through 2011 (Objective 1) modal progression analysis (MPA) was used. Each month, the lengths of all the sampled fish at each site were grouped into length distributions (10 mm increments) and used within decomposition assessment by applying Bhattacharya's method in FISAT (Bhattacharya, 1967; Bolland et al., 2007). This identified the presence of modes in each length distribution by separating them into a series of normal distributions (King, 1998). For each mode, the output was the number of individuals, their mean length and standard deviation (SD) (Bolland et al., 2007). The modes were separated by application of a separation index (SI), calculated as the ratio of the difference between successive means and the difference between their SD modes; values above 2.0 indicate significant difference from the other modes (Bhattacharya, 1967; Bolland et al., 2007). This has advantages over similar methods as it ensures that the identified modes are significantly different; it also means the method is reliable and justifies its use here rather than alternative methods (Bolland et al., 2007). The overall output of MPA for each site per month was the number of modes in the population and their mean length (± SD), which enabled length increases to be identified over the study period.

Although MPA showed that the number of length modes in the population and their length increases over time, it could not reveal the age of the fish within the modes. Thus, to complete Objectives 2 and 3, scales were collected from a maximum of 15 fish per month and site. When the number of fish captured exceeded 15 the scaled fish were randomly selected; when the number was below 15 scales were taken from all fish. Their total length (mm) and sex was also recorded (immature, female or male). To determine whether the fish could be aged from their scales (Objective 2), observations were made on whether growth checks were present on scales. As they were, the next step was to determine the frequency and timing of their formation. This required scales to be examined using a projecting microscope (× 48 magnification) following measurements taken from one scale: total scale radius (SR), distance from the focus to the last formed check (LA) and distance to the second-last formed check (LA-1). These data were then subjected to marginal increment ratio analysis (MIRA; Haas and Recksiek, 1995; Vilizzi and Walker, 1999), where the MIRA calculation of the marginal increment ratio (MIR) was determined from MIR = [(SRLA) / (LALA-1)]. When only one check was observed, the denominator was the distance from the scale focus to the check (Vilizzi and Walker, 1999). To test for differences in the MIR data for each month, ANOVA was used where the dependent variable was the MIR for each fish and the independent variable was the month. Tukey's post hoc tests enabled the significant differences to be identified for each month and indicated the timing of when the growth checks were formed.

Once the frequency of check formation had been determined the age of each fish was calculated by counting the number of growth checks, and the scales were measured to enable back-calculation of their lengths (Francis, 1990). Three analyses on their lengths-at-age could then be completed. First, the effect of sex on length at age was determined through building a general linear model (GLM) that tested the effect of sex (male or female; immature fish were excluded from the model) on length at the last annulus while the effects of site and age were controlled in the model. In the model, statistical complications from using repeated measurements on individual fish in the same test (i.e. pseudo-replication) were avoided by using only the back-calculated length at the last growth check for each fish (Beardsley and Britton, 2012). Differences between the sexes were assessed for significance using pairwise comparisons with Bonferroni adjustment for multiple comparisons. Second, the data were analysed for their mean standardized length-at-age residuals for each site (Benstead et al., 2007; Storm and Angilletta, 2007; Beardsley and Britton, 2012; Britton et al., 2012). This required the predicted mean length at each age across all the populations to be determined using the log–log quadratic function of Vilizzi and Walker (1999) as this is the most precise and biologically meaningful growth model. These values enabled the standardized residual of the length at age of each fish to be calculated (Beardsley and Britton 2012), with these compared between sites using a GLM that controlled for the effects of sex. Again, only the back-calculated length at the last annulus was used for each fish to avoid pseudo-replication. Lastly, the length at age data by sex and site were applied to the non-linear von Bertalanffy growth model (von Bertalanffy, 1938) to produce values of the maximum (asymptote) theoretical length at each site (L∞) and K, the annual growth rate towards L∞. All statistics were completed in SPSS v.16.0 and only tests that met all underlying assumptions were used.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Modal progression analysis

Across the study sites, fish were captured to a maximum length of 137 mm (Site HD), although the majority of fish sampled were below 100 mm (Figure 2). Modal progression analysis identified three length modes at each site, although not all modes could be identified every month (Figure 2). The mean lengths of the smallest mode varied between 20 and 30 mm at IB and BB, but up to 48 mm in HD. They only appeared in samples from August at the earliest (Figure 2), suggesting these were young-of-the-year. At each site, there was also a length mode of relatively large fish (generally > 60 mm) whose growth increase was minimal through the year, especially at site HD (Figure 2).

image

Figure 2. Mean lengths (± SD) per month of modes identified by modal progression analysis through the study period, where ▲ = young-of-the-year; ○ age 1+; ● > age 2 years for (a) Site IB, (b) Site BB, and (c) Site HD.

Download figure to PowerPoint

Marginal increment ratio analysis of scales

In the cooler months, the number of fish sampled tended to be low. In the scale samples per month, analysis revealed a large number of ‘replacement’ scales, i.e. scales that had been lost previously and regenerated. In combination, this resulted in the number of scales suitable for MIR analysis and ageing being variable between months (range 0 to 15). The differences in the mean marginal increment ratios of scales at each site across the study period were significant (IB: F10,76 = 6.61, P < 0.01; BB: F11,92 = 12.91, P < 0.01; HD: F9,88 = 36.21, P < 0.01; Figure 3). At IB and HD, an increase in the marginal increments was apparent from April through to at least October, with post hoc analyses (Tukeys) showing significant differences between March and all subsequent months (P < 0.05; Figure 3). At site BB, growth at the scale margin was apparent from March, with post hoc analyses (Tukeys) revealing significant differences between February and all subsequent months (P < 0.05; Figure 3). This suggests formation of an annual growth check in February/March as the fish started to grow again after the colder winter period (Figures 3, 4).

image

Figure 3. Marginal increment ratio analysis of scales (± SD) across the study period from (a) Site IB, (b) Site BB, and (c) Site HD.

Download figure to PowerPoint

image

Figure 4. Daily maximum and minimum air temperature for (a) the weather station closest to site IB and BB; and (b) closest to site HD.

Download figure to PowerPoint

Age range and structure

Scale ageing, completed by counting the number of annual growth checks (Figure 3), showed that fish present in the samples were between 0+ and 6 years old (Figure 5). This indicated that the modes revealed by the MPA generally did represent discrete age-classes of fish (Figures 2, 5), where the smallest length mode comprised young-of-the-year fish, the next mode comprised fish of age 1 in February/March 2011 and 1+ thereafter, and the largest mode comprised fish > 2 years old (Figures 2, 5). The MPA could not distinguish different age-classes of fish within this largest length mode as the annual growth increments were relatively low compared with their growth earlier in life (Figures 2, 5). There were differences in age structure between the sites, with only one fish above 3 years old at site IB, but with fish aged 4 and 5 present in greater numbers at sites BB and HD (Figure 5). There was only one 6-year-old fish present across all the samples (HD).

image

Figure 5. Length at the last annulus of ○ female and ● male fish at (a) Site IB, (b) Site BB, and (c) Site HD.

Download figure to PowerPoint

Length-at-age

Analysis of the age structure of the populations found a significant difference in the lengths-at-age of female and male fish across all sites (F1,186 = 12.02, P < 0.01; Figure 5), where both site and age had significant effects in the GLM (P < 0.01). Females were the larger sex, with an estimated marginal mean length of 62.7 ± 1.1 mm compared with 57.0 ± 1.2 mm for males; pairwise comparisons with Bonferroni adjustment for multiple comparisons indicated that this difference was significant (P < 0.01). The lengths at each age of fish within each site were variable, with differences at age 1 as much as 40 mm in fish whose lengths did not exceed 70 mm (Figure 5). Length-at-age was also variable between sites, with the mean standardized residual analysis revealing these differences to be significant (F2,185 = 19.19, P < 0.01) when the significant effect of sex (P < 0.01) was controlled for. Pairwise comparisons with Bonferroni adjustment for multiple comparisons showed that the significant differences were between site HD and both IB and BB (P < 0.01), but with no significant difference between IB and BB (P > 0.05; Figure 6). Outputs of the von Bertalanffy growth model also showed that L∞ was highest at site HD (121 ± 2 mm); females also had higher values than males (Table 1).

image

Figure 6. Estimated marginal means (± standard error) of standardized growth residuals at site HD, BB and IB, where the means have been adjusted for the effects of sex.

Download figure to PowerPoint

Table 1. Estimated parameters of the von Bertalanffy growth model for Garra ghorensis at the three study sites. Note at Site IB, values for male fish could not be calculated as fish were present only to age 2 years
 FemaleMale
SiteL∞ (mm)KL∞ (mm)K
IB102 ± 80.36 ± 0.06--
BB112 ± 50.35 ± 0.0480 ± 10.76 ± 0.07
HD121 ± 20.48 ± 0.03109 ± 10.47 ± 0.05

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The ages of G. ghorensis could be estimated from their scales and so provided data on the age composition, lengths-at-age and growth parameters of three Jordanian populations. These data showed that individuals present in the sites were up to 6 years old, with relatively fast growth up to the age of 2 years and the production of relatively small annual growth increments thereafter. The oldest fish and fastest growth rates were recorded at site HD, the most disturbed site because of its habitat modifications (physical and hydrological) and the presence of invasive O. aureus. Thus, the altered hydrological regime that increased depths and reduced flows was insufficient to affect the persistence of G. ghorensis and instead appeared to provide more optimal growth conditions than the less disturbed sites.

In disturbed arid environments, the life-history traits of desert fish that increase their vulnerability to extirpation and extinction relate to those of the periodic life-history strategy of Winemiller and Rose (1992) (Olden et al., 2006, 2008). This combination of large body size, late maturation, and low juvenile survivorship despite high fecundity per spawning event, results in poor adaptation to changing environments. By contrast, the traits of small body size, fast growth to maturation and low fecundity per spawning event are generally well predisposed to ensuring more favourable population responses to highly disturbed and unpredictable environments (Olden et al., 2006). These life-history traits are important to understand given that desert fish conservation management strategies should be based on a fundamental understanding of how the ecological attributes of species interact with fluvial habitats to influence population persistence (Olden et al., 2008). For G. ghorensis, their traits of relatively fast growth and limited lifespan are aligned to the opportunistic strategy of Winemiller and Rose (1992). Thus, their populations do not have ‘slow’ life histories that are congruent with high extinction risk (Olden et al., 2008), although it is acknowledged that data on G. ghorensis reproductive traits are required for this inference to be more robust.

This apparent resilience to disturbance in G. ghorensis was also emphasized by the population comprising the longest-lived and fastest growing individuals being present in the most disturbed site. This is an important outcome given that many studies on threatened desert fishes in arid environments suggest that conservation strategies should focus on the restoration or maintenance of natural flow regimes (Poff et al., 1997; Eby et al., 2003; Richter et al., 2003; Rolls et al., 2013). This is based on the assumption that the restoration of natural flow regimes will provide impaired rivers with the attributes in which the native fauna evolved and so are necessary for the maintenance of robust and healthy populations (Propst et al., 2008). The present study on G. ghorensis found that their populations can at least tolerate some hydrological disturbance, and also the presence of an invasive cichlid. This suggests that their conservation management does not necessarily have to return their habitats to pristine conditions. Instead, the next steps in their conservation should be the identification of those life-history traits (e.g. reproductive traits) and/or ecological associations that reduce population persistence when their environments are disturbed so that these can be mitigated or rehabilitated (Olden et al., 2008). Moreover, some focus on sites where populations are endangered might be necessary to assist identification of their constraints, given that the three populations studied here were all sufficiently numerous to enable destructive sampling. However, this raises ethical concerns of sampling populations that are close to extirpation.

There was high variability in the length-at-age of the fish at each site that was independent of sex, with variation between individuals in lengths at age 1 of up to 40 mm. This was also allied with the regular appearance of new 0+ fish in samples between July and October that were identified in a discrete length mode from August. Although it was not implicitly tested within the study, this suggests that a further favourable trait that provides G. ghorensis with some resilience to environmental disturbances is a protracted spawning period. This is because protracted spawning tends to produce high variation in the lengths of the 0-group cohort (Nunn et al., 2002). Protracted spawning periods are a common feature of many cyprinid fishes. For example, Cyprinus carpio is capable of asynchronous spawning throughout the year in equatorial regions (Britton et al., 2007), and in Lake Naivasha and its tributary rivers in Kenya the cyprinid Barbus paludinous spawns asynchronously with no clear spawning peak (Mutia et al., 2010). This is also the case in temperate cyprinid riverine populations, where protracted spawning periods provide the cohorts of juvenile fish with considerable resilience against stochastic environmental events that generally result in high early-life mortality (Nunn et al., 2002). Here, the apparent protracted spawning by G. ghorensis might be a deliberate strategy that allows greater proportions of their progeny to survive their first year of life in arid regions where high summer temperatures and very low rainfall could lead to periodic drying of some nursery areas and to high mortality of the 0+ cohort.

To conclude, this study revealed the age structure and lifespan of G. ghorensis for the first time. Across the three study sites, fish were relatively long-lived, given their body size and environmental conditions, with the highest growth rate recorded at the most disturbed site. This suggests some trait plasticity that facilitates their adaptation to environmental change and thus provides some resilience to their populations. Conservation programmes can use this knowledge to protect their threatened habitats without necessarily requiring those habitats to be restored to a pristine state.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The authors would like to thank the Royal Society for the Conservation of Nature (RSCN) for making all resources available for this study. The laboratory assistants Mr. Anas Abu Yehya, and Eiz al-Deen al-Aqeel are thanked for their help in scale preparation and in measuring specimens.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References