International Council for the
Exploration of the Sea
ICES CM 2006/H:14
Fisheries-induced evolutionary change
Fisheries-induced evolutionary changes in maturation reaction norms in North Sea
sole (Solea Solea)
Fabian M. Mollet, Sarah B.M. Kraak, Adriaan D. Rijnsdorp
Abstract
Age and size at maturation has been observed to decrease in several commercially exploited fish stocks, which,
according to life history theory, might be due to fisheries-induced evolutionary change. However, an evolutionary
interpretation is not unambiguous as the observed changes may also represent a plastic response to changes in the
environment: e.g. compensatory density-dependence in growth and maturation, or a direct response of maturation to
increased body condition or to temperature. To disentangle phenotypic plasticity from evolutionary change, we employ
the probabilistic reaction norm approach, which allows us to estimate the probabilities of becoming mature independent
of growth and survival. The analysis of 34 cohorts (1963 to 1996) of female North Sea sole from market samples shows
that the reaction norm for age and size at first maturation has significantly shifted towards younger ages and smaller
sizes. Size at 50% probability of maturation at age 3 decreased from 28 cm to 25 cm, whereas this change occurred
mainly in the second half of the analyzed period. The same result is found if changes in condition and temperature are
considered by introducing them as a 3rd dimension in the reaction norm method. Good condition facilitates early
maturation and temperature 3 years prior to maturation was positively related to the probability of maturation. The
results provide support of a fisheries-induced evolutionary change in the onset of sexual maturity.
Key words: maturation, fisheries-induced change, phenotypic plasticity, evolution, reaction norm, growth, condition,
temperature
Contact author: Fabian Mollet, Wageningen IMARES, Institute for Marine Resources and Ecological Studies, P.O.Box
68, 1970 AB IJmuiden, The Netherlands [tel +31 255564646, fax +31 255 564644, e-mail [email protected] ].
1
Introduction
Changes in life history traits such as age and size at first maturation have been reported in several commercially
exploited fish stocks (De Veen 1976, Jørgensen 1990, Rijnsdorp 1993). These changes may be due to: (i) phenotypic
plasticity in response to environmental change; (ii) fisheries-induced evolutionary change (Stokes et al. 1993, Law
2000, Stokes & Law 2000, Heino & Godø 2002). The expectation of a genetic change in life histories stems from the
assumption of individual fitness optimization to a given environment (Roff 1983, Stearns & Koella 1986, Dieckmann
& Law 1996, Geritz et al. 1998). Since fecundity and the viability of eggs and larvae are positively related to maternal
size (Trippel & Neil 2004), the trade off between somatic growth and reproduction (West et al. 1997) translates into a
trade off between current and future reproduction (Heino & Kaitala 1999) over the whole lifespan as well as within a
season. If mortality rates are high, future reproduction may not pay off and earlier maturation and higher reproductive
investment will result in a higher fitness (Charnov et al. 2001, Lester et al. 2004). These predictions have been
confirmed experimentally (Reznick et al. 1990, Conover et al. 2005) and the challenge is now to examine whether
these processes also occur in wild exploited populations. Evolutionary changes, however, may be overshadowed by
phenotypic plasticity (Stokes et al. 1993). For instance, changes in age and size at first maturity can be caused by
changes in growth rate following the reduction of biomass by fishing (Reznick 1993, Law 2000, Heino & Godø 2002).
Fishing typically goes along with a decrease in stock biomass and is therefore likely to weaken density-dependent
effects, such as the competition for food. For sole, there is evidence for density-mediated compensatory growth
(Rijnsdorp & Van Beek 1991, Millner & Whiting 1996), although other factors such as temperature and changes in
food availability might play a role (De Veen 1976, Rijnsdorp & Van Beek 1991).
The effect of phenotypic plasticity can be disentangled from genetic effects by applying the reaction norm approach.
By definition, a reaction norm describes which phenotypes will be expressed by a genotype under a certain range of
environmental conditions (Stearns & Koella 1986). A recently developed method models the probability of the onset of
maturation conditional on age and size (Heino et al. 2002, Barot et al. 2004a) and the resulting predictions do therefore
not depend on variations in growth and survival. This probabilistic reaction norm method (PRNM) has been applied to
several stocks and provided support for a fisheries-induced evolutionary change in the onset of maturity in North Sea
plaice Pleuronectes platessa (Grift et al. 2003), North Atlantic cod Gadus morhua stocks (Barot et al. 2004b, Olsen et
al. 2004, Olsen et al. 2005) and American plaice Hippoglossoides platessoides (Barot et al. 2005). In this approach the
range of environmental conditions is thought to be reflected in differences in somatic growth. The concept can be
extended to a multi-dimensional reaction norm in which also the effect on maturation of other (environmental) factors
can be taken into account (Grift et al. in press).
In this paper we investigate whether the change in maturation in North Sea sole as already described by Veen (1976;
1978), can be attributed to phenotypic plasticity in response to the observed variations in growth rate (De Veen 1976,
Rijnsdorp & Van Beek 1991, Millner et al. 1996) or to a fisheries-induced evolutionary change using the PMRN
described in Heino (2002) and Barot (2004b). As environmental variation may not only have its effect through
variations in somatic growth, we also explore the possible influence of condition and temperature through the method
of multi-dimensional reaction norms. Finally we will discuss the implications of our results for the sustainable
management of this stock.
2
Material and Methods
Distribution. The common sole (Solea Solea) is distributed from the Northwest African coast and the Meditarranean to
the Irish sea through the English Channel and the North Sea up to Skagerrak and Kattegat (Rijnsdorp 1992, Gibson
2005). Spawning of North Sea sole occurs in the southern North Sea with local concentrations in the German Bight
along the Belgian Coast and in the Thames estuary and the Wash (Figure 1). Young soles are distributed in coastal
waters and migrate to deeper offshore waters when they grow older but return every year in spring for spawning (Van
Beek et al. 1989, Van der Land 1991).
Environment. North Sea Sole is mainly caught by the Dutch fleet taking about 75% of the total international landings
(ICES 2006), in a mixed demersal fishery with a minimum mesh size of 80 mm. Whereas e.g. plaice has been
intensively exploited since the late 19th century (Rijnsdorp & Millner 1996), the exploitation of sole has increased since
the 1960s of the 20th century (Figure 2) following the introduction of the beam trawl (De Veen 1976, De Veen 1978).
The exploitation pattern is rather flat for age groups above 3 years when all fish are above the minimum landing size of
24 cm.
Changes in some environmental factors likely to influence maturation are reflected in Figure 4 by the time series of
growth rate, competitive biomass, condition factor and temperature over the analyzed period for the age classes 2, 3 and
4. Competitive biomass was obtained by weighting estimates of stock abundance in terms of mean weight per age class
by the mean crowding coefficients per age class from Rijnsdorp and Van Beek (1991).
Data. Fish samples have been collected monthly since 1957 from commercial landings covering the distribution area of
sole in the North Sea and from autumn surveys since 1970. Market samples are stratified into 5 market size-categories
(24-27, 27-30, 30-33, 33-38 and >38cm). Data recorded are: date, position, length, weight, sex, gonad weight, maturity
stage and age. Age (years) is determined from the pattern of growth rings in the otoltih taking 1 January as date of birth.
We restricted the analysis only to females from samples collected in the southeastern North Sea (areas 1 – 5; Figure 1)
in two seasonal sampling windows (see below). The data set was screened for measurement errors using regression
analysis and the most extreme values were removed. Based on the age range of maturation and the number of
observations required for the analysis (Barot et al. 2004a) we restricted our analysis to ages 1 – 6 of the cohorts 1963 –
1996. In total 18037 and 9981 observations were used for the analysis for the two sampling windows respectively
(Table 1).
Maturation. Maturity is determined by macroscopic inspection and 8 maturity stages are distinguished that represent
the seasonal development of the immature and mature gonad (Table 2; Veen, 1970). The immature gonad is classified
as stage I and II. Developing ovaries are classified as stages III – IV and represent the vitellogenic phase. The start of
Vitellogenesis (VTG) in the North Sea sole is variable between years and latitudes and also among individuals, ranging
from July to winter (Deniel 1981, Ramsay & Witthames 1996). Stages V – VI describe the spawning stage when
hyaline eggs are visible, whereas stages VII and VIII represent spent fish. Some fish might undergo abortive maturation,
meaning that VTG is started but ovaries are not developed further than stage III or IV (De Veen 1970, Ramsay &
Witthames 1996). The ovary of a spent female may look quite similar to the immature ovary. Hence for a correct
distinction of immature and mature females, it is important to select the time period when all mature females have
started vitellogenesis or are in spawning stage, while none of them is spent. Therefore, the proportion of matures can
only be reliable determined in the period between January and May (Figure 4, Veen 1976; Ramsay and Witthames,
1996). However, the timing of the spawning season varies annually in response to the ambient water temperature (Van
der Land 1991, Rijnsdorp & Witthames 2005). Therefore, we applied a moving time window to get an unbiased
estimate of the proportion mature fish. The moving time window was determined as the period of 2 months before to 1
3
month after the date of the maximum ovary weight (Table 1a). The date of maximum ovary weight was estimated using
the following model:
log(o) ~ log(l) + d + d2 + d3
(1)
where o is the ovary weight of individual fish, l is its length, d is the day in the year when it was caught. For cohorts for
which this approach was not possible it was modeled as the midpoint between the days at which probability of having
attained maturity stage V and VII respectively, is 50%. We considered a fixed instead of a moving window to analyze
the effect of weight and condition (i.e. relative weight: weight/length3) on maturation because condition is temporally
influenced by many factors such as food availability and temperature. Fish improve their condition as a consequence of
the decision to spawn in the subsequent season (not shown) but the moment of the decision to become mature
determined by the start of VTG, is too early to detect maturity macroscopically. We selected a window comprising the
months November, December and January, as a compromise between the moment of the decision to start maturation
and the reliable detection of maturity (Figure 4, Table 1b).
Probabilistic reaction norm. The probabilistic maturation reaction norm (PMRN) estimates the probabilities with
which an immature individual becomes mature at a particular size and age (Heino et al. 2002). In contrast to the
traditionally used ogive estimation, which estimates the probability of being mature, the PMRN approach overcomes
the confounding effects of growth and mortality. Because for sole, first time and repeat spawners cannot be
distinguished, the probability of becoming mature was estimated by a refinement of the PMRN method (Barot et al.
2004a). The probability of becoming mature p is given by the number of fish that have matured in a given time interval
divided by the number that could have matured in that time interval:
p(a,s,x) = [m(a,s,x) – m(a-∆a,s-∆s,x-∆x)]/[1 – m(a-∆a,s-∆s,x-∆x)]
(2)
where m is the probability of being mature, a age, s size, x any other additional factor possibly affecting maturation
(condition or temperature, see below), ∆a the time interval set to one year for annual spawners, ∆s the age specific
annual growth increment and ∆x the age specific change in x. Equation (2) relies on a simplifying assumption that the
growth rate and mortality rate are the same for mature and immature individuals. This is not expected to apply
accurately; however, Barot et al. (2004a) tested the sensitivity of the estimation method and confirmed that it is robust
to its relaxation.
The probability of being mature m(a,s,x) was estimated by logistic regression with Gaussian errors in dependence of
age and size (Bernardo 1993), condition factor (Kjesbu et al. 1991, Trippel & Neil 2004) and temperature (Dhillon &
Fox 2004, Dembski et al. 2006). Model selection was made by forward selection of variables and was further based on
the significance of their parameters and the Akaike information criterion (AIC). The performance of the selected model
was then checked by determining classification errors by randomly dividing the data set into a training set and a test set
for model building and prediction respectively (Wisnowski et al. 2003) and by comparing it to similar models.
logit(m) ~ c + a + l + c×a + c×l + a×l
(3)
logit(m) ~ c + a + w + c×a + c×w + a×w
(4)
logit(m) ~ c + a + l + k + c×a + c×l + a×l
(5)
logit(m) ~ c + a + l + t*
(6)
4
where m is the probability of being mature, c cohort as factor, a age, l length, w weight, k condition (Fulton’s condition
factor) and t temperature throughout this paper. As it is not clear during which life history stage temperature (measured
daily at Den Helder, the Netherlands) has an effect on the onset of maturation, it was tested as averages over months
July to October for different years (t*) in relation to the moment of catch. Models (3) and (6) are applied to the moving
window data set (Table 1a), models (4) and (5) to the fixed window data set (Table 1b).
Growth increments. The length increments were obtained by fitting Von Bertalanffy growth curves per cohort to
mean length-at-age data from market samples (age ≥ 3) and survey or back-calculation data (age ≤ 3). For age groups
up to age 3, the mean length estimated from market samples will be biased because small individuals may escape
through the meshes or may be discarded if below the minimal marketable size. Pre-recruit survey data and backcalculation data, kindly provided by Millner and Whiting (1996), were used to estimate the mean length of ages 1-3.
The weight increments were obtained by converting mean length-at-age into mean weight-at-length by modeling
log(w) ~ c + log (l) + c×log(l)
(7)
for analysis on the fixed window data set (Table 1b) for ages ≥ 2.75. To account for the length-dependence of the
weight increments in a given age class, we modeled this weight increment (∆w) based on the annual growth matrix as
log(∆w) ~ c + w + l + c×w + c×l+ w×l
(8).
Diagnostics. Because 99% of individuals become mature at ages 2, 3 and 4 we only consider these ages for
diagnostics. To visualize the reaction norm and its change over time we use the midpoints, Lp50 and Wp50, the length
and weight respectively at which probability of becoming mature is 0.5. Since the reaction norms appeared to be almost
linear we fitted linear regressions per cohort and checked how the intercepts and slopes of these fitted reaction norms
changed over time. Midpoints of reaction norms derived from maturity models (3) and (4) were obtained by linear
interpolation, whereas those from models (5) and (6) were obtained by selecting combinations of values of variables for
which the probabilities of becoming mature was predicted to be within the range of 49.5-50.5%. Because the
calculation of reaction norm midpoints is based on various steps, it is not possible to obtain confidence intervals
directly and therefore the original data set was bootstrapped by resampling individuals 1000 times with replacement,
separately for each cohort. To test for a trend in the Lp50 at a given age, the reaction norm midpoints were regressed to
cohort as a variable, weighted by the inverse of the variance of each midpoint, estimated by percentiles in the
distribution of replicates in the bootstrap simulation. The time series of environmental factors (Figure 3) not accounted
for in the model were then screened for correlations with the time series of Lp50 and Wp50.
Because the onset of maturation might differ between subpopulations of different areas and latitudes, the analysis was
run separately for the spawning population of the German Bight (sampling areas 1 and 2 in Fig. 1) and the Dutch and
Belgian coast (sampling areas 3, 4 and 5 in Fig. 1) as well as jointly considering the two spawning stocks as one
population. All results shown here in the following are from the joint analysis representing the whole Southern North
Sea. The results are very similar if areas are analyzed separately and do not change qualitatively.
5
Results
Model properties. The maturity models based on the variable time window (Table 1a) gave better fits and predictions,
than those based on the fixed time window (Table 1b). All models had high rates of correct classification 92% for
model (3) and (6) and 89% for models (4) and (5). The sensibility, e.g. the proportion of correctly classified mature
individuals was high: 98% for model (3) and (6) and 96% for models (4) and (5). But specificity, e.g. the proportion of
correctly classified immature individuals, was less: 61% for model (3) and (6) and 55% and 56% for models (4) and (5).
Model (3) explained 53% of the null deviance but model (6) only 38%, whereas the models (4) and (5) explained 40%
and 42%, respectively. These properties are summarized in Table 3.
Two-dimensional reaction norms. The estimated reaction norms appear to be almost linear and tend to have negative
slopes meaning that the length at which sole attains a certain probability to mature decreases with age, that is, at a given
length, old females have a higher probability to become mature than young ones. The age effect is much weaker if
weight is considered as proxy of size, so the probability of becoming mature at a certain weight is almost equal for
younger and older individuals. The width or interquantile range of the reaction norm is narrower for length than for
weight, the average distance between probabilities of becoming mature of 0.1 and 0.9 being 27% of the average Lp50
(model (3)) and 81% of the average Wp50 (model(4)), respectively. The reaction norms have substantially shifted
towards younger ages and smaller sizes and have rotated counter-clockwise. Further, the largest change occurred
between the last decade and the decade before (Figure 5). The intersection of the averaged growth and the reaction
norms shows that the age and length at which 50% of the females become mature shifted from 3.0 years and 27.5 cm in
the 1960s to 2.6 years and 25.0 cm in the 1990s (Figure 5). The intercepts of the reaction norms decreased significantly
(p<0.05) for length but not for weight (p=0.17, but is significant if using data from the moving window instead of the
fixed window) whereas the slope increased but not significantly (p>0.1) over the observed period (Figure 6). The Lp50
and Wp50, weighted by its variances, decreased significantly for age groups 2 and 3 over time (all p<0.05), but not for
age 4 (p>0.1, Fig. 7). For the Lp50, the trends for ages 2 and 3 are even significant on a level of α = 0.01. The estimated
decrease in Lp50 is 0.09 cm/year for age 2 and 0.08 cm/year for age 3, the estimated decrease in Wp50 is 1.5 g/year for
age 2 and 1.4 g/year for age 4. The Wp50 could not be estimated for all cohorts and suggests temporal autocorrelation.
The trends are most significant for age 3, the age at which the highest proportion of individuals matures. Note that the
minimal marketable size of 24 cm seems to operate as an attractor for the Lp50 values.
Time series. Temperature shows a significantly increasing trend in time (0.02ºC/year, p < 0.01), condition a
significantly decreasing trend (- 0.001/year, p < 0.001) as well as growth rates (-0.065cm/year2, p < 1e-08) over the
observed period (Figure 4). Growth rates in the German Bight and the Dutch-Belgian Coast spawning stocks did not
differ substantially (correlation 0.98, R2 = 0.96). Probably due to the relatively high variability in the Lp50 and Wp50,
correlations with these environmental variables were not found to be significant (all p-values > 0.05).
Three-dimensional reaction norms. As expected, condition in the three-dimensional reaction norm derived from
model (5) has a significant positive association with the probability of being mature. Figure 8 shows the reaction norms
over time indicating that the effect of condition on the Lp50 is about 1 cm per unit of 0.1 condition factor and remains
about constant over the observed period. Lp50 is decreasing over time for any given value of condition factor, and
obviously, a change in condition could not explain the observed decrease in the two-dimensional Lp50. The observed
condition factors range from 0.76 to 1.16 (95% confidence interval) and decrease rather than increase over the analyzed
period. The effect of condition seems to be slightly more important for older ages (not shown).
For the temperatures of different periods relative to the time of catch tested in the reaction norm derived from model (6),
higher temperatures of the growing season three years prior to the year of catch were associated with higher
6
probabilities of becoming mature. Higher temperatures one and two years prior were associated with lower
probabilities of becoming mature. The reaction norm displayed in Figure 9 was obtained from mean temperature over
the period June to October, three years prior to sampling for ages 2, 3 and 4. Since about 67% become mature for the
first time at age 3 (Table 1a, Figure 4), the temperature three years prior to sampling can be interpreted as the one
acting during larval and post-larval early life history stage. The effect of temperature three years prior to sampling on
the Lp50 suggests about 1 cm per 1ºC. Lp50 is decreasing over time for any given value of this temperature and
obviously, a change in this temperature could not explain the observed decrease in the two-dimensional Lp50.
Discussion
This study showed that the maturation reaction norm in North Sea sole shifted downwards in the period 1963-1996,
which is in agreement with the theoretical expectations of a fisheries–induced evolutionary change. Evolutionary
models for fitness (Metz et al. 1992, Dieckmann & Law 1996, Geritz et al. 1998) predict the reaction norm intercept to
decrease and the initially negative slope to slightly increase if exposed to size-selective harvesting (Ernande et al. 2004).
The changes in intercept and slope presented here (Figure 6), though some of them on a low level of significance, are
thus consistent with theoretical expectation.
The two-dimensional reaction norm approach assumes that all environmental variation affecting the maturation process
is reflected in variations in the growth rate and also acts through growth rate. However, other factors such as
temperature or body conditions may also affect the probability of becoming mature, which can be accommodated by
multi-dimensional reaction norms (Grift et al. in press, Kraak in review).
Including condition in our analysis revealed that higher condition is associated with lower Lp50 values (Figure 8), thus
fish in good condition have a higher probability of becoming mature than a similar aged and sized individual in poor
condition. This effect was also suggested in other species (Kjesbu et al. 1998, Kurita et al. 2003, Morgan 2004) and
may be related to the storage of energy during the feeding period that can be metabolized later for reproduction
(Rijnsdorp 1990, Marshall et al. 1999). The decrease in the Lp50 at any given value of condition (Figure 8) indicates
that the earlier maturation is not due to a time trend in the condition factor. Given the positive correlation of condition
and the probability to become mature, the observed decrease in condition would cause the two-dimensional Lp50 not
accounting for condition to increase and the evolutionary shift may therefore be underestimated. Consequently, the
three-dimensional reaction norm (Figure 8) results in a slightly higher estimate of the evolutionary shift as found in the
reaction norm not accounting for condition, from about 28 cm to 25 cm for age 3. In contrast to North Sea plaice (Grift
et al. in press), the effect of condition in sole seems very weakly age-dependent and even more important for older ages.
However, the selection of data to analyze maturity as a function of weight or condition is problematic because the
detection of maturity becomes reliable only after the individual’s decision to become mature is made and because fish
seem to improve their condition as consequence of the decision to become mature in the following season. The
observed weight and condition at the moment when maturity can be detected can therefore not be interpreted as the
cause of maturation. Since condition might change in response to the onset of spawning we selected data from a fixed
window (over months Nov, Dec, Jan), in which the proportions of matures per age class are not very different from
those around peak spawning (see Table 1), taking condition at this moment as a direct proxy for the condition at the
time of the decision. Unfortunately, this ambiguity has no practical solution. The possibly introduced bias is reflected in
the lower fit and performance of the models (4) and (5) (Table 3) and the uncertainty of the Wp50 estimates, making
weight a less robust proxy for size than length. Because of the fixed time window used in data selection the fluctuations
7
in the Wp50 might result from inter-annual differences in the seasonal ovary development. It might also be influenced
by the interplay of condition and growth rate but it is not clear how this could account for the observed patterns.
Temperature has an important effect on the timing of spawning (Rijnsdorp 1992, Rijnsdorp & Witthames 2005); but
may also influence the onset of maturation in an individual lifespan. Temperature affects age at maturity through its
effect on growth rates (Charnov & Gillooly 2004, Baum et al. 2005) but also growth independent effects of temperature
have been suggested (Dhillon & Fox 2004, Dembski et al. 2006). Increases in temperature can accelerate
developmental rates other than growth (Baynes & Howell 1996, Fuiman et al. 1998, Martell et al. 2005). Thus, life
history characteristics, such as the onset of maturation, might be affected at various life history stages, including early
life history stages. We find here evidence for the hypothesis of such early life history influence of temperature as only
the temperature three years before sampling has a positive effect on the maturation reaction norm. Lp50 is decreasing
over time for any given value of this temperature and obviously, a change in this temperature could not explain the
observed decrease in the two-dimensional Lp50. The strength of evolutionary change is somewhat smaller after taking
account for changes in temperature. However, including temperature in the analysis did not improve the model,
suggesting that even more noise is added (Table 3).
By using the reaction norm approach we could disentangle the phenotypic plasticity in maturation due to differences in
environmental conditions as expressed in variations in growth rate. By including condition and temperature in the
analysis, we tested whether environmental variations that did not act through variations in somatic growth contributed
to the variations in the probability of becoming mature. The results of this analysis showed that although both condition
and temperature contributed significantly to the variations in maturation, these factors could not explain the trends in
the maturation reaction norm over time.
Our results do not suggest a linear change, as only in the 1980s the trend in the reaction norm becomes really manifest.
Possible reasons to explain this pattern are (1) that there was still a considerable proportion of old individuals in the
population that contributed to the next generations, which would therefore have delayed or diluted the evolutionary
response, (2) that the selection differential has increased or (3) that there were changes in gene flow between locally
adapted subpopulations (Conover et al. 2005) in the 1980s. Since larger fish have a relatively higher fecundity (Trippel
& Neil 2004) the importance of (1) is likely. Furthermore, the presence of old individuals is likely as the fishing
mortality decreases for older ages (Figure 2). The gradual increase in fishing mortality over time (Figure 2) suggests
that the selection differential did change (2), which at some point may have exceeded a critical threshold. However, (3)
is not straight forward because all stocks have likely experienced the increase in (beam trawl) fishing mortality.
The trend in Lp50 being less significant for older age groups might be explained by the notion that the evolution of the
reaction norm is expected to stop, if it once reaches the size threshold below which selection has no longer any effect
(Ernande et al. 2004), e.g. corresponding to the minimal marketable size of 24 cm in sole. Once reached, the Lp50 does
not decrease any further, so in age 4 the trend might be not significant, because the midpoints have always been similar
to the minimal marketable size.
The intercepts fitted to the estimated reaction norms in this study display substantial variation but a significant
decrease; the increase in the slope is less significant. Because of the difficulty to measure selection differentials
(Sinclair et al. 2002) and heritabilities, especially in the wild, the rate of evolution is not well understood. Experiments
have shown that a change in the pattern and rate of size-selective mortality caused major evolution of life history
characteristics in 18 generations in guppies (Poecilia reticulate) (Reznick et al. 1990) and the growth rate of Atlantic
silverside (Menidia menidia) evolved within just 4 generations (Conover & Munch 2002, Conover et al. 2005, Munch
et al. 2005). In the wild, genetic analyses of rapid life history changes following introduction to novel environments are
well established in several species (Reznick et al. 1997, Hendry et al. 2000, Quinn et al. 2001).
8
Alternative explanations. Maturation could hypothetically also be influenced by factors for which no data is available,
such as social factors or endocrine disruptors. In the southern platyfish (Xiphophorus maculates) the presence of
dominant males seems to suppress maturation of subordinate males (Sohn 1977) but such a mechanism is unlikely to
apply for females. Endocrine disrupting agents present in sewage effluents such as estrogens androgens or progestins
(Vos et al. 2000) would particularly affect nursery grounds and for instance estrogens (Oestradiol or analogues) could
be associated with increased VTG in immature females but also in males (Oberdorster & Cheek 2001, Robinson et al.
2003, Stentiford & Feist 2005). Increased levels of vitellogenin in immature females could possibly accelerate
maturation. Janssen (1997) found elevated concentrations of vitellogenin in premature females of the flounder
(Platichthys flesus) to be associated with pollution in the Dutch Wadden Sea but could not find any effect on
reproduction. However, the reproductive effects of increased vitellogenin concentrations are often associated with
reproductive impairment and reduced fecundity and are therefore more likely to delay maturation (Oberdorster &
Cheek 2001). Furthermore, it is very unclear, how this effect would become manifest on the population level (Robinson
et al. 2003). We are not aware of any study reporting on accelerated maturation due to endocrine disrupting agents.
Sources of error. In our estimation of growth rates we might have a possible bias due to the fact that fishing will select
only for fish above a certain size and, if the reason for an individual not being on the spawning ground during peak
spawning is related to age or size, also in the maturity ogive. Our results might also be influenced by the assumptions
made to estimate reaction norms, that mature and immature fish have similar growth and mortality rates. However, a
sensitivity analysis has proven that the methodology of the PMRN is robust with respect to differences in growth rate
(Barot et al. 2004a) and these biases would apply for all cohorts equally and thus not change our results qualitatively.
Macroscopical staging of maturity could be erroneous such that spent or not yet ripened females could be classified as
juveniles but the selection of data around peak spawning with a moving window should have removed a big part of this
error. To account for abortive maturation in the sense that fish develop until maturity stage III but not further (De Veen
1970, Ramsay & Witthames 1996) we ran the PMRN by considering maturity stages I, II and III as juveniles instead of
only stages I and II, which did not alter the results qualitatively (not shown).
Implication in management. The evidence for fisheries induced evolution is strong and persistent also if temperature
and condition is accounted for. Fisheries-induced evolution is likely to decrease maximum sustainable yield (Law &
Grey 1989, Heino 1998) and finding practical strategies to reverse the decreasing trend in age and size at maturation
will be increasingly difficult. Recovery from a genetic change will be much slower than could be expected only on
account of the lower abundance (Hutchings 2000, Heino & Godø 2002) because selection pressure toward the original
genotype in the absence of fishing might be much weaker than the selection caused by intensive fishing (Law & Grey
1989, Rowell 1993, Law 2000). If the directional selection is strong, genetic variability of that original phenotypic state
might even have been lost and surviving genotypes might have a lower fitness in a new environment of lower
exploitation (Conover 2000). Advanced maturation will result in a reduced size-at-age of mature fish (Kozlowski 1996)
and thus in a lower biomass per age group (Heino 1998). Whether the reduced growth rates observed in sole (Fig. 4a)
are a direct consequence of the documented shift in the onset of maturation or to a general deterioration of the growing
conditions needs to be analyzed in more detail. A direct selection on growth rates, however, seems unlikely because in
iteroparous species with indeterminant growth the improved survival through slower growth might not offset the loss in
fecundity (Heino & Kaitala 1999, Heino & Godø 2002). Even if earlier maturation will positively affect the spawning
stock biomass, this may not necessarily be translated in an increase in recruitment since fecundity and the viability of
eggs is positively correlated with maternal size (Trippel 1998, Trippel & Neil 2004) and earlier maturing smaller
individuals therefore contribute less to reproduction.
9
Acknowledgements
This research has been supported by the European research training network FishACE funded through Marie Curie
(contract MRTN-CT-2204-005578). We thank R.S. Millner for providing us the data set of 868 back-calculated soles
and the FishACE network for useful discussions.
10
a)
Total number
Dutch-Belgian coast (area 3,4 and 5)
German Bight (area 1 and 2)
Average per cohort
Minimum per cohort
Maximum per cohort
Average percentage mature
Age 1
105
97
8
3
0
24
4
Age 2
1027
843
184
30
0
168
8
Age 3
5893
4177
1716
173
19
508
75
Age 4
5321
3618
1703
156
3
412
97
Age 5
3256
2220
1036
96
3
237
99
Age 6
2435
1707
728
72
0
236
100
All
18037
12662
5375
b)
Total number
Dutch-Belgian coast (area 3,4 and 5)
German Bight (area 1 and 2)
Average per cohort
Minimum per cohort
Maximum per cohort
Average percentage mature
Age 1
240
225
15
0
0
0
0
Age 2
3076
2259
817
7
0
35
8
Age 3
2855
1976
879
90
6
224
63
Age 4
1898
1307
591
84
9
290
91
Age 5
1327
856
471
56
1
185
96
Age 6
585
375
210
39
1
130
94
All
9981
6998
2983
Table 1: Numbers of observations per age class and sampling windows, mean, minimum and maximum of observations
per cohort and the proportion of observed adults for a) the data set selected with the moving window around the peak of
spawning (2 months before and 1 month after the estimated maximum relative GSI) and b) the data set selected with
the fixed window on November, December and January for analysis of effects of weight and condition on maturation.
Maturity stage
interpretation
observation
I
Juvenile (-)
Transparent and homogenous ovary, tight walls and small lumen, eggs
undistinguishable
II
Resting (-)
Reddish translucent ovary, tight walls, eggs are distinguishable, lumen is
filled with liquid
III
Vitellogenic (+)
Reddish gray to dark orange mostly intransparent ovary, bigger than in II
and less tight, rich in blood vessels, Vitellogenesis has started and some eggs
contain yolk, lumen is big.
IV
Vitellogenic (+)
Orange to reddish white, completely intransparent and stiff ovary at half of
its definite size, eggs are polygonal and all contain yolk.
V
Ripe (+)
Orange to reddish white intransparent ovary in its definite size, lumen is
squeezed, eggs are round.
VI
Spawning (+)
Mostly grayish-red translucent hyaline but partly still intransparent ovary,
resilient when pressed together, lumen contains spawn.
VII
Half spent (+)
Gray to dark red completely translucent hyaline ovary, walls are very slack
and bloody, lumen is big and filled with spawn and liquid.
VIII
Spent (+)
Dark red translucent ovary, walls are very slack, lumen is very big mostly
filled with liquid. Very similar to II or in a transition towards II.
Table 2: Classification of maturity stages, biological interpretation and associated macroscopic observations. Maturity
stages I and II are interpreted as immatures (-) and thus juveniles, the stages thereafter as matures (+) and thus adults.
After spawning is completed the adult soles cycle from maturity stage VIII to maturity stage II into the growing phase
of the following season.
11
AIC
Explained deviance
Correct classification
Sensibility
Specificity
Moving window
Model (3) Model (6)
7377
3062
0.53
0.38
0.92
0.92
0.98
0.98
0.61
0.61
Fixed window
Model (4) Model (5)
5818
5745
0.40
0.41
0.89
0.89
0.96
0.96
0.55
0.56
Table 3: Model properties: Akaike’s information criterion (AIC), proportion of the null deviance explained by the
model, proportion of correctly classified observations, sensibility (proportion of correctly classified positives, e.g.
matures) and specificity (proportion of correctly classified negatives, e.g. immatures).
12
1
7
4
2
3
6
5
1957 - 1962
1963 - 1973
1974 - 1985
1986 - 1996
0.4
0.0
0.2
fishing mortality
0.6
0.8
Fig 1: Distribution and main spawning grounds of North Sea sole in terms of egg production density (eggs/(m2*day))
and sampling areas 1 to 7. Kindly provided by Bolle et al. (in prep.).
2
4
6
8
age
Fig. 2: Fishing mortality (yr-1) at age estimated by VPA (ICES 2006) averaged for adjacent periods. A clear increase in
mortality after the 1960s is illustrated. Natural mortality is estimated as 0.1.
13
4 5 6 7 8 9
growth rate
1965
1970
1975
1980
1985
1990
1995
1985
1990
1995
1985
1990
1995
30
20
5 10
0
competitive biomass
cohort
1965
1970
1975
1980
0.95
0.85
0.75
condition
cohort
1965
1970
1975
1980
16.5
15.5
14.5
temperature
cohort
1970
1980
1990
2000
year
Fig. 3: Summary of possibly interacting trends over the observed period: a) growth rate (cm/yr) for ages 2, 3 and 4
(dashed lines) b) individually experienced competitive biomass (tons) for ages 2, 3 and 4 which mostly overlap (dashed
lines) c) condition (100*g/cm3) for ages 2, 3 and 4 (dashed line) and d) temperature per year. The solid line represents
a non-parametric local smoother over the ages in b) and c) or over the years in d) with span set to 0.5. Growth rates
were estimated as described in the text, competitive biomass by weighting the estimated stock number per age group
from VPA (ICES 2006) with the mean crowding estimates per age group from Rijnsdorp and Van Beek (1991) and the
weight-length relationship from eq. (7), condition by the average condition in the fixed window data set (Table 1b) and
temperature as average temperature from June to October measured in Den Helder.
14
1.0
0.8
0.6
0.4
0.0
0.2
proportion mature
age 1-2
age 2-3
age 3-4
age 4-5
age 5-6
age 6-7
jul
aug
sep
okt
nov
dec
jan
feb
mar
apr
mai
jun
0
-2
25
-6
-4
RN slope length
40
35
30
RN intercept length
45
Fig.4: Seasonal development of the average proportion of matures in market samples by age from July in one year to
July in the following year.
1965
1975
1985
1995
1965
1975
1985
1995
1985
1995
cohort
-100
-300
-200
RN slope weight
600 800
400
200
0
RN intercept weight
0
cohort
1965
1975
1985
cohort
1995
1965
1975
cohort
Fig. 6: Change of intercepts (cm) and slopes (cm/yr) of fitted reaction norms based on maturity models (3) (upper
panel) and (4) (lower panel) by linear regression of Lp50 values on ages 2 - 4. The decrease in intercepts is in both
cases significant (p<0.05).
15
35
reaction norms averaged over cohorts
20
25
length
30
1963 - 1973
1974 - 1984
1985 - 1996
1
2
3
4
5
age
Fig. 5: reaction norms for cohorts 1963 – 1973 (solid line), 1974 – 1984 (dashed line) and 1985 – 1996 (dotted line)
obtained from averaged Lp50 (cm) over these cohorts. The thin lines represent the estimates of mean length-at-age
(cm), averaged over the corresponding periods.
16
age = 2
400
300
Wp50
0
20
100
200
30
25
Lp50
35
500
age = 2
1970
1975
1980
1985
1990
1995
1965
1970
1975
1980
cohorts
cohorts
age = 3
age = 3
1985
1990
1995
1985
1990
1995
1985
1990
1995
300
Wp50
0
20
100
200
30
25
Lp50
400
35
500
1965
1970
1975
1980
1985
1990
1995
1965
1970
1975
1980
cohorts
cohorts
age = 4
age = 4
300
Wp50
0
20
100
200
30
25
Lp50
400
35
500
1965
1965
1970
1975
1980
1985
1990
1995
cohorts
1965
1970
1975
1980
cohorts
Fig. 7: Lp50 (cm) and Wp50 (g) estimates (dots), bootstrapped 95%-quantiles (vertical bars), trend regression weighted
by bootstrapped variances (dashed line), fit with a non-parametric smoother (solid line). For the Lp50 trends in ages 2
and 3 are significant on a level of α = 0.01, for the Wp50 on a level of α = 0.05. Trends are not significant for age 4
(p>0.1).
17
age 3
29
29
28
28
27
27
26
26
25
25
Lp50
24
24
0.8
0.9
1.0
1.1
1.2
condition
1965
1970
1975
1980
1985
1990
1995
cohort
Fig. 8: Reaction norm surface of the three-dimensional reaction norm, based on maturity model (5) obtained with a
non-parametric smoother (span = 1) for age 3 over cohorts. Higher conditions are associated with lower Lp50-values
(cm), meaning that condition and the probability to mature are positively related. For all conditions, there is a negative
trend over the time axis (cohort) in the Lp50.
age 3
29
29
28
28
Lp50 27
27
26
26
25
25
14.0
14.5
15.0
15.5
16.0
16.5
17.0
temp
1965
1970
1975
1980
1985
1990
1995
cohort
Fig. 9: Reaction norm surface of the three-dimensional reaction norm, based on maturity model (6) and temperatures
three years prior to maturation, obtained with a non parametric smoother (span = 1) for age 3 over cohorts. Higher
temperatures are associated with lower Lp50-values (cm), meaning that temperatures three years prior to maturation
and the probability to mature are positively related. For all temperatures, there is a negative trend over the time axis
(cohort) in the Lp50.
18
References
1
Barot S, Heino M, Morgan MJ, Dieckmann U (2005) Maturation of Newfoundland American plaice (Hippoglossoides
platessoides): long-term trends in maturation reaction norms despite low fishing mortality? ICES Journal of Marine
Science 62:56-64
2
Barot S, Heino M, O'Brien L, Dieckmann U (2004a) Estimating reaction norms for age and size at maturation when age at first
reproduction is unknown. Evolutionary Ecology Research 6:659-678
3
Barot S, Heino M, O'Brien L, Dieckmann U (2004b) Long-term trend in the maturation reaction norm of two cod stocks.
Ecological Applications 14:1257-1271
4
Baum D, Laughton R, Armstrong JD, Metcalfe NB (2005) The effect of temperature on growth and early maturation in a wild
population of Atlantic salmon parr. Journal of Fish Biology 67:1370-1380
5
Baynes SM, Howell BR (1996) The influence of egg size and incubation temperature on the condition of Solea solea (L) larvae
at hatching and first feeding. Journal of Experimental Marine Biology and Ecology 199:59-77
6
Bernardo J (1993) Determinants of maturation in animals. Trends in Ecology & Evolution 8:166-173
7
Bolle LJ, Eltink ATGW, Rijnsdorp AD (in prep.) The spawning grounds of sole (Solea solea) in the North Sea and English
Channel.
8
Charnov EL, Gillooly JF (2004) Size and temperature in the evolution of fish life histories. Integrative and Comparative
Biology 44:494-497
9
Charnov EL, Turner TF, Winemiller KO (2001) Reproductive constraints and the evolution of life histories with indeterminate
growth. Proceedings of the National Academy of Sciences of the United States of America 98:9460-9464
10
Conover DO (2000) Darwinian fishery science. Marine Ecology-Progress Series 208:303-307
11
Conover DO, Arnott SA, Walsh MR, Munch SB (2005) Darwinian fishery science: lessons from the Atlantic silverside
(Menidia menidia). Canadian Journal of Fisheries and Aquatic Sciences 62:730-737
12
Conover DO, Munch SB (2002) Sustaining fisheries yields over evolutionary time scales. Science 297:94-96
13
De Veen JF (1970) On some aspects of maturation in the common sole Solea Solea (L.). Ber dt Kommn Meeresforsch 21:78-91
14
De Veen JF (1976) Changes in some biological parameters in North-Sea sole (Solea-Solea L). Journal du Conseil 37:60-90
15
De Veen JF (1978) Changes in North Sea sole stocks (Solea Solea L.). Rapp P-v Réun Cons perm int Explor Mer 172:124-136
16
Dembski S, Masson G, Monnier D, Wagner P, Pihan JC (2006) Consequences of elevated temperatures on life-history traits of
an introduced fish, pumpkinseed Lepomis gibbosus. Journal of Fish Biology 69:331–346
17
Deniel C (1981) Les poissons plats (Téleostéens-Pleuronectiformes) en Baie de Douarnez., Université de Bretagne Occidentale
18
Dhillon RS, Fox MG (2004) Growth-independent effects of temperature on age and size at maturity in Japanese medaka
(Oryzias latipes). Copeia:37-45
19
Dieckmann U, Law R (1996) The dynamical theory of coevolution: A derivation from stochastic ecological processes. Journal
of Mathematical Biology 34:579-612
20
Ernande B, Dieckmann U, Heino M (2004) Adaptive changes in harvested populations: plasticity and evolution of age and size
at maturation. Proceedings of the Royal Society of London Series B-Biological Sciences 271:415-423
21
Fuiman LA, Poling KR, Higgs DM (1998) Quantifying developmental progress for comparative studies of larval fishes.
Copeia:602-611
19
22
Geritz SAH, Kisdi E, Meszena G, Metz JAJ (1998) Evolutionarily singular strategies and the adaptive growth and branching of
the evolutionary tree. Evolutionary Ecology 12:35-57
23
Gibson RN (2005) Flatfishes biology and exploitation, Vol. Blackwell Science Ltd
24
Grift RE, Heino M, Rijnsdorp AD, Kraak SBM, Dieckmann U (in press) The role of length, weight and condition in the
maturation of North Sea Plaice
25
Grift RE, Rijnsdorp AD, Barot S, Heino M, Dieckmann U (2003) Fisheries-induced trends in reaction norms for maturation in
North Sea plaice. Marine Ecology-Progress Series 257:247-257
26
Heino M (1998) Management of evolving fish stocks. Canadian Journal of Fisheries and Aquatic Sciences 55:1971-1982
27
Heino M, Dieckmann U, Godo OR (2002) Measuring probabilistic reaction norms for age and size at maturation. Evolution
56:669-678
28
Heino M, Godø OR (2002) Fisheries-induced selection pressures in the context of sustainable fisheries. Bulletin of Marine
Science 70:639-656
29
Heino M, Kaitala V (1999) Evolution of resource allocation between growth and reproduction in animals with indeterminate
growth. Journal of Evolutionary Biology 12:423-429
30
Hendry AP, Wenburg JK, Bentzen P, Volk EC, Quinn TP (2000) Rapid evolution of reproductive isolation in the wild:
Evidence from introduced salmon. Science 290:516-518
31
Hutchings JA (2000) Collapse and recovery of marine fishes. Nature 406:882-885
32
ICES (2006) Report of the Working Group on the Assessment of Demersal Stocks in the North Sea and Skagerrak (WGNSSK).
ICES CM 2006/ACFM:09
33
Janssen PAH, Lambert JGD, Vethaak A, Goos HJT (1997) Environmental pollution caused elevated concentrations of
oestradiol and vitellogenin in the female flounder, Platichthys flesus (L). Aquatic Toxicology 39:195-214
34
Jørgensen T (1990) Long-term changes in age at sexual maturity of Northeast Arctic cod (Gadus-Morhua L). Journal du
Conseil 46:235-248
35
Kjesbu OS, Klungsoyr J, Kryvi H, Witthames PR, Walker MG (1991) Fecundity, atresia, and egg size of captive Atlantic cod
(Gadus-Morhua) in relation to proximate body-composition. Canadian Journal of Fisheries and Aquatic Sciences 48:23332343
36
Kjesbu OS, Witthames PR, Solemdal P, Walker MG (1998) Temporal variations in the fecundity of Arcto-Norwegian cod
(Gadus morhua) in response to natural changes in food and temperature. Journal of Sea Research 40:303-321
37
Kozlowski J (1996) Optimal allocation of resources explains interspecific life-history patterns in animals with indeterminate
growth. Proceedings of The Royal Society of London Series B-Biological Sciences 263:559-566
38
Kraak SBM (in review) Does the probabilistic maturation reaction norm approach disentangle phenotypic plasticity from
genetic change?
39
Kurita Y, Meier S, Kjesbu OS (2003) Oocyte growth and fecundity regulation by atresia of Atlantic herring (Clupea harengus)
in relation to body condition throughout the maturation cycle. Journal of Sea Research 49:203-219
40
Law R (2000) Fishing, selection, and phenotypic evolution. ICES Journal of Marine Science 57:659-668
41
Law R, Grey DR (1989) Evolution of yields from populations with age-specific cropping. Evolutionary Ecology 3:343-359
42
Lester NP, Shuter BJ, Abrams PA (2004) Interpreting the von Bertalanffy model of somatic growth in fishes: the cost of
reproduction. Proceedings of The Royal Society of London Series B-Biological Sciences 271:1625-1631
43
Marshall CT, Yaragina NA, Lambert Y, Kjesbu OS (1999) Total lipid energy as a proxy for total egg production by fish stocks.
Nature 402:288-290
20
44
Martell DJ, Kieffer JD, Trippel EA (2005) Effects of temperature during early life history on embryonic and larval development
and growth in haddock. Journal of Fish Biology 66:1558-1575
45
Metz JAJ, Nisbet RM, Geritz SAH (1992) How should we define fitness for general ecological scenarios. Trends in Ecology &
Evolution 7:198-202
46
Millner R, Flatman S, Rijnsdorp AD, vanBeek FA, DeClerck R, Damm U, Tetard A, Forest A (1996) Comparison of long-term
trends in growth of sole and plaice populations. ICES Journal of Marine Science 53:1196-1198
47
Millner RS, Whiting CL (1996) Long-term changes in growth and population abundance of sole in the North Sea from 1940 to
the present. ICES Journal of Marine Science 53:1185-1195
48
Morgan MJ (2004) The relationship between fish condition and the probability of being mature in American plaice
(Hippoglossoides platessoides). ICES Journal of Marine Science 61:64-70
49
Munch SB, Walsh MR, Conover DO (2005) Harvest selection, genetic correlations, and evolutionary changes in recruitment:
one less thing to worry about? Canadian Journal of Fisheries and Aquatic Sciences 62:802-810
50
Oberdorster E, Cheek AO (2001) Gender benders at the beach: Endocrine disruption in marine and estuarine organisms.
Environmental Toxicology and Chemistry 20:23-36
51
Olsen E, Lilly GR, Heino M, Morgan MJ, Brattey J, Dieckmann U (2005) Assessing changes in age and size at maturation in
collapsing populations of Atlantic cod (Gadus morhua). Canadian Journal of Fisheries and Aquatic Sciences 62:811-823
52
Olsen EM, Heino M, Lilly GR, Morgan MJ, Brattey J, Ernande B, Dieckmann U (2004) Maturation trends indicative of rapid
evolution preceded the collapse of northern cod. Nature 428:932-935
53
Quinn TP, Kinnison MT, Unwin MJ (2001) Evolution of chinook salmon (Oncorhynchus tshawytscha) populations in New
Zealand: pattern, rate, and process. Genetica 112:493-513
54
Ramsay K, Witthames P (1996) Using oocyte size to assess seasonal ovarian development in Solea solea (L). Journal of Sea
Research:275-283
55
Reznick DA, Bryga H, Endler JA (1990) Experimentally induced life-history evolution in a natural-population. Nature 346:357359
56
Reznick DN (1993) Norms of reaction in fishes. In: Stokes TK, McGlade, J.M., Law, R. (ed) The exploitation of evolving
resources. Springer-Verlag, p 72-90
57
Reznick DN, Shaw FH, Rodd FH, Shaw RG (1997) Evaluation of the rate of evolution in natural populations of guppies
(Poecilia reticulata). Science 275:1934-1937
58
Rijnsdorp AD (1990) The mechanism of energy allocation over reproduction and somatic growth in female North-Sea plaice,
Pleuronectes-Platessa L. Netherlands Journal of Sea Research 25:279-289
59
Rijnsdorp AD (1993) Fisheries as a large-scale experiment on life-history evolution - disentangling phenotypic and geneticeffects in changes in maturation and reproduction of North-Sea plaice, Pleuronectes-Platessa L. Oecologia 96:391-401
60
Rijnsdorp AD, Millner RS (1996) Trends in population dynamics and exploitation of North Sea plaice (Pleuronectes platessa L)
since the late 1800s. ICES Journal of Marine Science 53:1170-1184
61
Rijnsdorp AD, Van Beek FA (1991) Changes in growth of plaice Pleuronectes-Platessa L and sole Solea-Solea (L) in the
North-Sea. Netherlands Journal of Sea Research 27:441-457
62
Rijnsdorp AD, Witthames P (2005) Ecology of reproduction. In: Gibson RN (ed) Flatfishes, Biology and Exploitation.
Blackwell Science Ltd., p 68-93
63
Rijnsdorp ADea (1992) Recruitment of sole stocks, Solea Solea (L.), in the Northeast Atlantic. Netherlands Journal of Sea
Research 29 (1-3):173-192
64
Robinson CD, Brown E, Craft JA, Davies IM, Moffat CF, Pirie D, Robertson F, Stagg RM, Struthers S (2003) Effects of
sewage effluent and ethynyl oestradiol upon molecular markers of oestrogenic exposure, maturation and reproductive
success in the sand goby (Pomatoschistus minutus, Pallas). Aquatic Toxicology 62:119-134
21
65
Roff DA (1983) An allocation model of growth and reproduction in fish. Canadian Journal of Fisheries and Aquatic Sciences
40:1395-1404
66
Rowell CA (1993) The effects of fishing on the timing of maturity in the North Sea cod (Gadus morhua L.). In: Stokes TK,
McGlade, J.M., Law, R. (ed) The Exploitation of Evolving Resources. Springer-Verlag, p 44-69
67
Sinclair AF, Swain DP, Hanson JM (2002) Measuring changes in the direction and magnitude of size-selective mortality in a
commercial fish population. Canadian Journal of Fisheries And Aquatic Sciences 59:361-371
68
Sohn JJ (1977) Socially Induced Inhibition Of Genetically Determined Maturation In Platyfish, Xiphophorus-Maculatus.
Science 195:199-201
69
Stearns SC, Koella JC (1986) The evolution of phenotypic plasticity in life-history traits - predictions of reaction norms for age
and size at maturity. Evolution 40:893-913
70
Stentiford GD, Feist SW (2005) First reported cases of intersex (ovotestis) in the flatfish species dab Limanda limanda: Dogger
Bank, North Sea. Marine Ecology-Progress Series 301:307-310
71
Stokes K, Law R (2000) Fishing as an evolutionary force. Marine Ecology-Progress Series 208:307-309
72
Stokes TK, McGlade JM, Law R (1993) The exploitation of evolving resources, Vol. Springer-Verlag
73
Trippel EA (1998) Egg size and viability and seasonal offspring production of young Atlantic cod. Transactions of the
American Fisheries Society 127:339-359
74
Trippel EA, Neil SRE (2004) Maternal and seasonal differences in egg sizes and spawning activity of northwest Atlantic
haddock (Melanogrammus aeglefinus) in relation to body size and condition. Canadian Journal of Fisheries and Aquatic
Sciences 61:2097-2110
75
Van Beek FA, Rijnsdorp AD, Declerck R (1989) Monitoring Juvenile Stocks Of Flatfish In The Wadden Sea And The Coastal
Areas Of The Southeastern North-Sea. Helgolander Meeresuntersuchungen 43:461-477
76
Van der Land MA (1991) Distribution of flatfish eggs in the 1989 egg surveys in the southeastern North-Sea, and mortality of
plaice and sole eggs. Netherlands Journal of Sea Research 27:277-286
77
Vos JG, Dybing E, Greim HA, Ladefoged O, Lambre C, Tarazona JV, Brandt I, Vethaak AD (2000) Health effects of
endocrine-disrupting chemicals on wildlife, with special reference to the European situation. Critical Reviews in
Toxicology 30:71-133
78
West GB, Brown JH, Enquist BJ (1997) A general model for the origin of allometric scaling laws in biology. Science 276:122126
79
Wisnowski JW, Simpson JR, Montgomery DC, Runger GC (2003) Resampling methods for variable selection in robust
regression. Computational Statistics & Data Analysis 43:341-355
80
81
22