Growth rates in two natural populations of Gasterosteus aculeatus in northwestern Spain: relationships with other life history parameters

We analysed growth rates of two natural populations of the three–spined stickleback fish, Gasterosteus aculeatus, in Galicia (north–west of Spain) where it has a strictly annual life cycle. We used the von Bertalanffy growth model to estimate nonlinear function for length–at–age data sets. These European peripheral populations reach the highest growth rates (k of the von Bertalanffy model > 0.4 month–1) known for this species. Instantaneous mortality rates and fecundity were computed using von Bertalanffy model parameters for each population. Mortality rates found in Galician populations were 2.0–2.3 higher times than those observed in general for Gasterosteidae. Combining both mortality and fertility, different intermediate fitness optima in each population were obtained for mature females. Overall, these differences in life history compared to other studied populations of sticklebacks can be interpreted as local adaptations to a Mediterranean climate type with high degree–days. Consequently, these populations at the edges of the species’ range may have adapted to the unique environmental conditions and may be of interest in ecology and conservation.


Introduction
The Mediterranean region is a biodiversity hotspot for freshwater ecosystems, harbouring many species (many of them endemic) and genetically distinct lineages that are of conservation concern (Araguas et al., 2012;Sharda et al., 2018). One of these species is the three-spined stickleback Gasterosteus aculeatus, L. It is a small teleost fish and is a model organism in evolutionary biology and ethology (Bell and Foster, 1994;Mäkinen et al., 2006;Cresko et al., 2007;Mäkinen and Merilä, 2008). This fish is widely distributed throughout the northern hemisphere in latitudes ranging from 35º to 70 ºN (Crivelli and Britton, 1987). The Iberian Peninsula is the southern limit in the East Atlantic Ocean (Fish-Base, 2019). The species lives in a variety of habitats (marine, streams, rivers, lakes), leading to a high level of phenotypic variation (Bell and Foster, 1994). Iberian Peninsula populations are mainly located in freshwaters habitats in Portugal and Galicia (a region in northwest Spain). In the rest of Spain, they are limited to fragmented populations (Doadrio, 2002).
The Galician populations of G. aculeatus have been evaluated on the basis of their morphometric and meristic characteristics by multivariate analysis Hermida et al., 2005aHermida et al., , 2005b. Heritabilities for some meristic characters have also been estimated in a natural population (Hermida et al., 2002). More recently, Pérez-Figueroa et al. (2015) obtained the first estimate of Ne / Nc (effective population size/population census) in Galician rivers using molecular markers. The genetic diversity (assessed by nuclear and mitochondrial markers) for local stickleback populations has also been discussed in relation to its phylogeography and conservation in the wider context of Ibero-Balearic populations (Vila et al., 2017).
Galician sticklebacks constitute peripheral populations and are in greater peril than central populations, like various other animal and plant species (Lesica and Allendorf, 1995;Guo et al., 2005). Lying at the southern edge of the species´ European range, they appear to be close to their physiological and reproductive limits, and have undergone local extinction in the recent past (Vila et al., 2017). G. aculeatus is currently classified as endangered under IUCN criteria in both Spain and Portugal (the two Iberian countries). In particular, in Spain the species is classified as vulnerable (Doadrio, 2002) and the local government of Galicia (Xunta de Galicia) includes it in its catalogue of threatened species (CATGEA, 2007).
Any conservation measures aimed at protecting this species require understanding of life history strategies, meaning the evaluation of their life history traits, under natural conditions, if possible. Life history traits are of great interest, especially in this species, G. aculeatus, whose populations have evolved by finding many different ways to combine these traits to affect fitness (Bell and Foster, 1994;Baker et al., 2015). Given the climatic differences between the Iberian Peninsula and northern Europe, the southern limit populations are probably subject to different selective forces and will show adaptations not found in northern populations. For example, G. aculeatus is a strictly annual species in Galicia Pérez-Figueroa et al., 2015): fish breed once in the year following their birth and die shortly after breeding (a semelparous life history), as many dead adults can be found among thousands of young fish. This annual cycle, which is usual in southern populations (Crivelli and Britton, 1987;Clavero et al., 2009), is exceptionally found in some European northern populations (Wootton et al., 2005).
Another topic of long interest in conservation and evolution of life history is the description of ontogenetic growth. Data on growth of G. aculeatus Iberian populations are scarce, but they are necessary to contribute to knowledge of their biology and to design effective conservation strategies in Galicia. One of the purposes of this study was to estimate growth parameters from length-at-age data obtained in natural conditions -in two Galician rivers-applying the von Bertalanffy growth function (VBGF) (Ricker, 1979).
Empirical studies in teleost fishes have demonstrated a significant connection between growth parameters and life history attributes such as natural mortality and fecundity (Wootton, 1979;Roff, 1984;Gunderson, 1997;Mangel, 2006). Thus, we combined our study on individual growth with the survival exponential distribution and fecundity to obtain lifetime reproductive output. In this way we could describe the lifetime expected reproductive success (ERS) or fitness as the product of survival and fecundity at maturity (Roff, 1984;Mangel, 2006). From this last function, we deduce that fitness will reach a maximum at an intermediate age at maturity (Roff, 1992(Roff, , 2002 and it is relevant for comparative purposes among populations. In general terms, this work aims to increase our knowledge of the biology of semelparous populations of this species in the Iberian Peninsula with a view to establishing appropriate conservation strategies.

Study area and sampling collection
The samples used for this study come from two rivers ( fig. 1): the Rato River (Rato 29T619547, 4761482; 2.8 km from Miño) and the Asma River (Asma; 29T601686, 4717706; 5.5 km from Miño). Both rivers are tributaries of the Miño River (the main Galician river), and are located, and isolated, upstream and downstream by the Belesar dam. Such isolation impedes movement of fishes and configures an important drainage area of Galicia (Pérez-Figueroa et al., 2015). The samples were collected monthly between May and November 1998 using hand nets with a mesh size of 2 mm, selecting specimens greater than 12 mm in length (from tip to tail). All samples were measured using a digital calliper to the nearest 0.01 mm, and then returned to the water. At all times, this work was carried out under the supervision of forest rangers from the autonomous government (Xunta de Galicia).
The annual cycle is usual in southern populations of Gasterosteus aculeatus (Clavero et al., 2009;Fernández et al., 2000;Pérez-Figueroa et al., 2015). Analysis of otoliths (unpublished data 2005, Asma River) revealed that more than 97 % of reproductive adults collected in May or June (with more than 46 mm of standard length) are born in the previous year (age 1+), while fishes between 21 and 25 mm are born in the current year (age 0+). Thus, most fishes captured for our analysis in the two rivers from May to November were cohorts born in 1998, i.e., they belonged to the same generation.
Growth parameters, natural mortality, fecundity and expected reproductive success The length-at-age data were employed to fit the von Bertalanffy growth function nonlinear model, hereafter referred to as VBGF, which has the form: where L t is length (in mm) at age t (in months); L ∞ is the asymptotic length (theoretical final length); k is the growth rate (month -1 ); and t 0 is a constant to improve the fit and it generally takes very small values.
We also note that the rate constant k has units of reciprocal time and is difficult to interpret. It describes the speed at which the maximum size (asymptotic size) is reached; for example (assuming t 0 = 0), when k = 0.5, an individual attains 90% of its asymptotic length in 4.6 months; while if k = 0.3, it takes 7.6 months to reach this percentage.
Natural mortality (m), which is closely related to the growth parameters, and especially to the parameter k of semelparous organisms, can be expressed as (Roff, 1984;Mangel, 2006): where T is the age at maturity (about 12 months for Galician stickleback populations).
On the other hand, supposing that survival to age t is given by the exponential distribution where the natural mortality rate is fixed, and that the organism (female) matures at age t; and that fecundity is given by: where a and b are allometric parameters. Combining both expressions (survival (3) and fecundity (4)), we can now define fitness or lifetime expected reproductive success as Due to legal regulations in Galicia, no more than  Spain two or three females can be collected per river. We thus trapped 28 females throughout the breeding season (May to early August) from ten tributaries of the Miño basin (including Asma and Rato rivers). The ovaries were dissected and all oocytes were counted. Fecundity ranged from 39 to 148 eggs, depending on the size (total length) of the female. All model parameters and their standard errors (SE) were estimated using non-linear regression analysis by means of the GraphPad Prism version 4.0 for Windows (GraphPad Software, La Jolla California USA, www.graphpad.com). Table 1 shows the growth data in length of the cohorts born in 1998 by seven monthly sampling. Table 2 shows the growth parameters and confidence intervals estimated using the von Bertalanffy model.

Results
By averaging r 2 values, we obtained 91 %, which indicates good performance of the model, i.e. VBGF accounted for 91 % of the variance in total length exhibited by these populations. The estimates of t 0 were very small, ranging from -0.03 to 0.37 months.
This parameter can therefore be confidently set to zero with very little effect on the model's performance. The estimates of k showed a marked difference when their confidence intervals were compared, and the Asma population showed the highest k value (0.52 month -1 ). However, a similar comparison for L ∞ values did not show a significant difference between the studied rivers. Figure 2 shows the representative VBGF curves resulting from all these parameters. As age increases the relative differences in size also decrease for populations with different growth functions.
The instantaneous rate of mortality for each population, obtained by the expression (2), were 0.19 and 0.21 for Asma and Rato rivers, respectively. We applied these values to compute the probability of survival (e-mt ) in both populations as described by predicted survival curves in figure 3 (red lines), assuming constant m values throughout the life of mature females. Both curves showed a similar appearance: survivorship dropped precipitously in early sexual maturity until a certain age was reached, at which moment the rate of decline was substantially reduced.
The allometric parameters (a and b) related to fecundity ( fig. 4) took the values 2.7 x 10 -4 (SE = 5.4 x 10 -4 ) and 3.2 (SE = 0.5), respectively (with   Eggs Survival r 2 = 0.62). As many empirical studies in fish species have shown that value of allometric b parameter is about 3 (Gunderson, 1997;Mangel, 2006;Roff, 1984), this value was used in the subsequent analysis.
After applying the expressions (3) and (4), we obtained expected fecundities and reproductive success (expression (5)) as functions of age at maturity for females (see fig. 3). At first, fecundities increased with age and then remained stable. With respect to fitness, both populations showed a maximum at an intermediate age but at different moments. This value was slightly higher for the Asma population (green lines in fig. 3).

Discussion
In a wide sense, the reproductive lifespan in a semelparous species is restricted to a single breeding season, even if age at maturity does not occur until after several years of development, such as in Pacific salmon species (Wootton and Smith, 2015). Life spans (and age at maturity) of G. aculeatus have been reported in many studies using otoliths, spines annuli, annual rings on the operculum and lengthfrequency plots (Bell and Foster, 1994;Yershov and Sukhotin, 2015). These studies comprise natural populations with different lifestyles (marine, stream dwelling, anadromous, lacustrine; revisions by Jones and Hynes, 1950;Wootton, 1984;Bell and Foster, 1994;Yershov and Sukhotin, 2015;FishBase, 2019). In general terms, their findings have concluded that this fish lives for a maximum of 1 to 5 years. Labo-ratory populations can reach 5 years, and Reimchen (1992) reported an exceptional large-bodied natural population that attained eight years. Wootton (1984) suggested a gradual change (a north-south cline) as populations at higher latitudes are more long-lived, but his hypothesis did not reach statistical significance because some high-latitude populations are also short-lived (Giles, 1987). The Galician populations fit this geographical trend, so that a strictly annual cycle can be established as in other populations of sticklebacks at similar latitudes in Spain and France (Crivelli and Britton, 1987;Clavero et al., 2009). The length of time that an organism lives can have a relevant effect on other life history traits such as fecundity and survival.
Values on mortality rates and growth rates on G. aculeatus and nearby species (Gasterosteidae such as Apeltes quadracus, Pungitius pungitius, and Spinachia spinachia) reported to date are expressed in years -1 (table 3). For comparative with our results, we divided these values by 12. Estimates of mortality (the exponential coefficient of mortality) and growth rates (and therefore L ∞ ) in natural conditions are usually hard to obtain. We found only 12 estimates for k and 3 for m (most of them for unsexed populations), considering G. aculeatus and other related closely species (FishBase, 2019;Yershov and Sukhotin, 2015;Roff, 1984;Pauly, 1980: Beverton andHolt, 1959). These estimates (table 3) varied from 0.048 to 0.280 and between 0.075 and 0.120 for k and m, respectively. Therefore, our estimates for Galician populations of sticklebacks are the largest reported so far. However, asymptotic values for Asma and Rato populations   (1), that is, m is k dependent [more theoretical evidence for this can be found in Charnov (1993), Mangel (2006), Roff et al. (2006), and Hamel (2015)]. Furthermore, based on independent estimates of k and m in natural populations of hundreds of fish species, Pauly (1980) found a significant correlation between the two parameters, suggesting that k is a good predictor of m [for example, high k values correspond to high m values (Hamel, 2015)]. Thus, our k values may be sufficient to obtain a rough (approximate) estimate of m, and confirming that high mortality rates will in general be associated with low ages at first reproduction (Roff, 1984;Mangel, 2006). Values m and k are also related to mean environmental temperature (Pauly, 1980). Ziuganov et al. (2010) recorded water temperatures in the Arctic region (1,750-1,850 degree-days) and Galicia (4,750-5,000 degreedays). The growth rate of Galician populations of G. aculeatus was at least 6-10 higher times than those of Artic populations (Yershov and Sukhotin, 2015), and mortality constants found in Galician populations were 2.0-2.3 times higher than those observed in general for Gasterosteidae (Roff, 1984). With higher mean annual temperatures in the south, the increment in metabolic rate would tend to increase growth rate.
Natural mortality, as defined in the literature, is made up by all possible causes of death except for fishing. These can include, for example, mortality caused solely by disease, by old age, or by both, or mortality proportional only to the number of potential predators. Causal (direct) and non-causal relationships between mortality and temperature in fish have been proposed. For example: temperature determines m via k increasing physiological mortality by aging; fishes living at higher temperatures have more chances to have encounters with predators (Charnov, 1993). We cannot specifically assign mortality to any of the aforementioned causes, and several of them may be acting at the same time on Asma and Rato stickleback populations. Therefore, our Iberian (Galician) populations may have adapted to local conditions within the European population context of G. aculeatus. These local populations are also peripheral populations of G. aculeatus. In ecology, central (core) and peripheral populations are key components to be considered. Peripheral populations are essential in terms of species biogeography, evolution and conservation (Lesica and Allendorf, 1995). Peripheral populations in natural conditions exist under different environmental conditions and are distinct from core populations. Thus, marginalized populations at the edge of the species distribution range must be integral parts of the conservation efforts for global biodiversity (Johannesson and André, 2006). Asma and Rato populations may represent particular genetic (Hermida et al., 2002;Vila et al., 2017) and phenotypic  Hermida et al., 2005b) adjustments to their environments, and variations in growth and mortality rates here reported with respect to northern populations can enhance their conservation value. Expression (4) predicts fitness in adult females of a semelparous organism (most species of fish are iteroparous) using two pieces of information: probability of survival to reproduce decrease with a constant mortality rate (e -mt ) and the assumption that fecundity is a power function of length. Mathematical models making these assumptions accounted for more than 80 % of the variation in natural populations of lizards, salamanders, and fish (Stearns and Koella, 1986;Stearns and Hoekstra, 2005). Thus, age at maturity appears to be adjusted to different intermediate fitness optima in each population ( fig. 3), that is, the peaks of the curves are linked to specific age at maturity. The age at maturity in this mathematical model is the age at which the survival fecundity curve has a turning point or peak, that is, up until this age the product survival fecundity increases while after this age it decreases. Age at maturity and fitness are plastic traits that can respond to natural selection because they can vary among close related species, among populations within species, and among individuals within populations.
It is widely accepted that global climate change (warming or cooling) is impacting on marine and freshwater fish and will continue to do so (Moss, 2010). Most data on climate-induced effects have been obtained for commercially relevant species, while those that are not targeted by fisheries have received less attention. Non-commercial short-lived species are of special interest because climate-induced effects on their populations could be more clearly recognized than in exploited species (Yershov and Sukhotin, 2015). Data trends show climate change effects ranging from fish growth, digestion physiology and performance in marine and freshwater ecosystems: the literature is replete with studies on increased growth rates at elevated temperatures between and within fish species (Mazumder et al., 2015). As ectotherms fish cannot thermoregulate physiologically, but only behaviourally by moving to areas with appropriate temperatures, even a small increase in temperature may thus put them at high risk of extinction (Tewksbury et al., 2008). Cross-comparisons of fish populations in similar systems in South America and Europe and within Europe have shown that lower-latitude fish species are often not only individually smaller but also grow faster, mature earlier, have shorter life spans and allocate less energy to reproduction than species at higher latitudes (Jeppesen et al., 2012). As we have indicated there is a relationship between m and environmental temperature, but m is also connected to the half-life of our female cohorts. Half-life is a statistic used by demographers to measure the time it takes for half of a cohort of organisms to die: the equation used to calculate the time taken by the population to decrease to half its initial size is t = ln 0.5/m (Gotelli, 2008), and applying our estimates of instantaneous mortality rates the half-lives are about 3.6 and 3.3 months (Asma and Rato populations respectively). Climate-induced variation (elevated or low temperatures) in life history traits (such as mortality) will have consequences in expected reproductive success, and therefore there will be a relevant probability of local extinctions. In Gasterosteids it is difficult to find experiments on a link between climate and populations dynamics, but it has been shown that modestly rising temperatures result in fewer males (G. aculeatus) making nests, and less time spent tending the nests (by fanning oxygenated water through them) by those that make them. As a result, fewer young are produced with more than a 2 ºC rise in temperature over current values (Moss, 2010). Studies are needed to predict future effects, and highly resilient species, such as the three-spined stickleback, provide indications of what might happen to less robust species.