Fisheries Research 80 (2006) 158–168 Effects of temperature on development and mortality of Atlantic mackerel fish eggs D. Mendiola a,∗ , P. Alvarez a , U. Cotano a , E. Etxebeste a , A. Martı́nez de Murguia b b a AZTI-Tecnalia, Marine Research Division, Herrera Kaia, Portualdea s/n, 20110 Pasaia, Spain Aquarium Donostia-San Sebastián, Marine Biology Laboratory, Karlos Blasco de Imaz Plaza 1, 20003 Donostia, Spain Received 1 December 2005; received in revised form 26 April 2006; accepted 4 May 2006 Abstract The development of Northeast Atlantic mackerel (Scomber scombrus) eggs, obtained from artificial fertilisation during the spawning season in the Biscay Bay area, was monitored at five temperatures (ranging from 8.6 to 17.8 ◦ C). The times to intermediate stages (III–V) and total hatching, obtained in this study, agree with the results of previous studies undertaken some years ago. However, the times over stages IA, IB and 50% hatching indicate that development rates differed significantly between the studies; this could be related to an effect of the previous thermal history of the eggs, or to experimental biases. Daily mortality (Z) during the embryonic period was found to vary between 0.17 and 0.38, using traditional exponential decay models. The estimates of mortality rate were found to range consistently with those derived from previous studies on Northwest Atlantic mackerel eggs without predation. However, the shape of the survivorship-curve during the development pursued in this study has indicated that mortality is not constant in time. Similarly, the suitability of an exponential model to describe daily mortality should be considered. © 2006 Elsevier B.V. All rights reserved. Keywords: Atlantic mackerel; Scomber scombrus; Eggs; Development; Mortality; Temperature models; Stock assessment 1. Introduction High mortality in the embryonic and larval stages is an important factor in determining the success of a year-class, and thus, the recruitment to the fishery (Hjort, 1914; Blaxter, 1974). Small differences in instantaneous mortality rates (Z) during the early life stages can generate large differences in final abundance or survivorship (Houde, 1987; Chambers et al., 2001); but these changes in the magnitude of Z are often difficult to quantify from field estimates of abundance (Dickey-Collas et al., 2003); this is mainly because of magnitude and variability of the factors involved (Pepin and Myers, 1991). ∗ Corresponding author. Tel.: +34 943 00 48 30; fax: +34 943 00 48 01. E-mail addresses: dmendiola@pas.azti.es (D. Mendiola), palvarez@pas.azti.es (P. Alvarez), ucotano@pas.azti.es (U. Cotano), eetxebeste@pas.azti.es (E. Etxebeste), depinvestigacion@aquariumss.com (A.M. de Murguia). 0165-7836/$ – see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.fishres.2006.05.004 Proper estimates of egg mortality in the field require accurate ageing of the eggs, which implies a good knowledge of the egg developmental rates at different temperatures. Many studies have examined the influence of temperature on the development and mortality rates of fish embryos (Lasker, 1964; Lo, 1985; Pauly and Pullin, 1988; Pepin, 1991; Van der Land, 1991; Le Clus and Malan, 1995), but these temperaturerelated mechanisms remain unclear for many species (Legget and Deblois, 1994). In this way, only limited information is available for the southern spawning component of the Northeast Atlantic mackerel, Scomber scombrus L. (herein NEA mackerel). Only Lockwood et al. (1977) have described the development of NEA mackerel eggs from the Bay of Biscay, Celtic Sea and West of Ireland at temperatures ranging from 7.4 to 17.8 ◦ C. Similarly Thompson (1989) fitted mortality curves to NEA mackerel egg counts, from the northern spawning components of the Bay of Biscay and Celtic Sea. However, both studies have limitations, since Lockwood’s experimental design might not be optimal for the study of development; as such, Thompson (1989) concluded that erro- D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 Nomenclature a A b h(t) i I Nt N0 S S(t) t T Z parameter of temperature total mortality parameter of scale hazard function of the Weibull distribution (i.e. age-specific instantaneous mortality rate) egg stage (i = 1, . . ., Nstages ) egg incubation time to reach the end of each stage of development number of surveyed eggs initial number of eggs fertilised successfully and incubated per temperature treatment survival survival function of the Weibull distribution (i.e. probability that an individual survives longer than an age) incubation time (i.e. h, days) egg incubation temperature (◦ C) instantaneous rate of mortality, derived through exponential models fitted in log scale Greek letters α scale parameter, in units of event observations ␤ vector parameter, defining the age range within stage and temperature for the event analysis γ shape parameter of the event analysis equation 159 Egg production in many surveys is estimated by fitting a mortality exponential decay model to the data for each stage, extrapolating back to the number of eggs at the time of spawning, then calculating the area under the seasonal production curve (Bunn et al., 2000). However, there are indications that eggs without predation do not die at a constant rate (Cameron et al., 1992; Makhotin et al., 2001; DickeyCollas et al., 2003). Thus, to assume a constant mortality rate, through all egg developmental stages could not be always appropriate. Daily egg production methods were developed for the Atlantic mackerel (Priede and Watson, 1993). Likewise, mortality estimates generated from the abundance of stage IA eggs, were already successfully integrated in the TAEP method of 1996 for this species (Anon, 1996); this is consistent with the concept of the number of eggs spawned at sea is closer to the abundances of stage IA. However, neither stage-specific mortality estimates, nor curves of population decay from laboratory controlled conditions, have ever been described for the NEA mackerel egg stage. The aim of this study was to check the influence of temperature on development and mortality of artificially fertilised eggs of NEA mackerel, from stages I to V. The hatching time was included for comparison with previous studies. Moreover, we examined the suitability of the traditional exponential models and we explored the event analysis, to estimate embryonic mortality. 2. Materials and methods 2.1. Incubation neous stage identifications may have affected their results on mortality. In turn, the Northwest Atlantic mackerel embryonic period was studied from temperature controlled conditions by Worley (1933) and Lanctot (1980), who used eggs of mature individuals from fish populations inhabiting close to Woods Hole, Massachusetts and St. George’s Bay, Nova Scotia areas, respectively. The development of mackerel eggs, between both localities, was shown to be very similar; however, the temperature range over which development took place was greater in the study of Worley (1933). Since 1977, independent estimates of the spawner biomass of NEA mackerel have been based on direct eggs surveys. These estimates of biomass are derived from the total annual egg production method (TAEP) (Gunderson, 1993), and then, they are used as inputs for the assessment with the integrated catch at age analysis (ICA) (Patterson and Melvin, 1996). An important requirement of this method is that spawned eggs of the target species can be aged (Stauffer and Picquelle, 1985). Age determination requires precise information on the effect of temperature on the rate of development of a large number of egg stages (Moser and Ahlstrom, 1985). Lockwood et al. (1977) modelled the temperature dependent development rate for each of the egg stages of NEA mackerel. Nowadays, such rates are being used to apply the method of assessing abundance of mackerel fishes in the Bay of Biscay. Adult mackerel, S. scombrus, were captured in March 2004 during the spawning season in the Biscay Bay area at 45◦ 47′ N 2◦ 25′ W; sperm and eggs were collected from 84 ripe female and 60 running male mackerels. Only females that yielded eggs freely (i.e. with little or no pressure on the vent) were used in these experiments. Eggs and sperm were combined in filtered seawater at ambient temperature. After 1 h of incubation the temperature was lowered to 5 ◦ C to keep the gametes at the basal metabolic activity level, during transportation to culture tanks (see, Table 1). After approximately 6 h from fertilisation, at the laboratory, 300 eggs were sampled and individual diameters measured to the nearest 0.1 mm, using a dissecting microscope. Fertilised eggs were selected and organised as follows: around 40 000 eggs were separated and incubated into five black cylindroconical 200 l tanks (35–40 ind l−1 ); 250 eggs were placed in five 1 l jars (50 ind l−1 ); 150 viable eggs were separated into individual 35 ml sterile vials. Thirty of these vials were transferred to each of five tanks, maintained at 8, 11, 13, 15 and 17 ◦ C water baths (Table 1). Seawater in each vial was replaced every day, throughout the experiment. Every bath was filled with UV-sterilised filtered (1 ␮m) seawater and temperature conditions were controlled by the combined function of an air-conditioning device and flow-through water 160 D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 Table 1 Details concerning handling, transportation and incubation of Atlantic mackerel eggs together with technical data of the different thermal groups used in the experiment Parameter Conditions Time from stripping to fertilisation (h) Temperature upon stripping and fertilisation (◦ C) Time from fertilisation to arrival (h) Temperature from fertilisation to arrival (◦ C) Quantify of eggs at arrival 4 12.4 ± 0.6 6 5 ± 0.8 about 40 500 Parameter starta (◦ C) Incubation temperature at Incubator tank (l) Initial density in tank (ind l−1 ) Incubator vial (l) No. of vials Initial density in vial (ind l−1 ) Incubator jar (l) Initial density in jar (ind l−1 ) a Group 1 Group 2 Group 3 Group 4 Group 5 8.5 200 40 0.035 30 28.6 1 50 11 200 38 0.035 30 28.6 1 50 13.1 200 35 0.035 30 28.6 1 50 15 200 35 0.035 30 28.6 1 50 17.4 200 38 0.035 30 28.6 1 50 All groups were previously acclimated through 1 h of water bath to each corresponding temperature. cooling (Aqua-Medic, Titan 500). The selected temperature range was greater than the range spawning mackerel would be expected to encounter, between February and March, over the area the adult fishes were caught (Anon, 1967; Valencia et al., 1988, 2004). The whole experiment was performed in a controlled temperature room, at 13 ◦ C. Eggs were not treated with any antibiotic solution. Vials and eggs incubated in tanks were checked daily, to determine mortality and development rates, respectively. Eggs incubated in jars were used to validate the hatching success, at each temperature. At all temperatures, the sampling of five eggs was undertaken, to identify the development stages initially every 2 h. This interval was increased as the stages of development advanced, reaching a maximum of 4 h over the period of hatching. At the same time, 30 eggs per temperature treatment were checked every 2 h, to record the time of death of each individual. The eggs for the staging were taken from those stocked in tanks; they were not used again, being immediately registered, photographed and preserved in 4% formalin (neutralised with borax), after examination. Stages were assigned to each egg sample using stereoscopic binocular microscope. The stages IA, IB, II, III, IV, V, Hatch 50% and Hatch total were classified, following the morphological keys reported by Lockwood et al. (1977). These categories were not designed to represent equal proportions of development time. The eggs for the mortality study were those stocked in vials. Negative buoyancy, white colour and deposition over the vial bottom were observation keys, to decide when the eggs were likely to be dead. Mortality was checked through the glass, avoiding any possible contact with the egg and allowing it use for successive samplings. Eggs were removed only to confirm a mortality event; then, they were observed under a dissecting microscope, registered, photographed and preserved. 2.2. Analysis The effect of temperature on the process of embryonic development was studied using the power function, as in Lockwood et al. (1977): Ii = ai T bi (1) where Ii is the incubation time to reach the end of each stage of development i, at incubation temperature T (◦ C), and ai and bi are the scale and temperature parameters, respectively. The model was fitted using ordinary least-squares for the logtransformed equation: ln Ii = ln ai + bi T (2) Stages duration was compared using a one-way ANOVA between temperature trials; when differences were significant, a Tukey–HSD multiple comparison post hoc test was performed, to differentiate statistically mean values between temperature levels. The log-transformation employed in this study, to model the rate of development with temperature, has been used in other studies of mackerel. Herein, results were compared with those reported by Worley (1933) and Lanctot (1980). Furthermore, total development rates and duration of each stage of development, were compared with the temperature-dependent models reported by Lockwood et al. (1977). ANCOVA analysis was used to compare slopes and intercepts at 95 and 99% confidence intervals. Likewise, percentages of difference, between the IB and 50 hatch stages of both studies, were also estimated from the fitted models. In order to study mortality, the traditional models of exponential decay (Ahlstrom, 1954) were used initially; abundances were fitted on age, for the calculation of the daily mortality rates. Event analysis was then performed, fitting age-at-death series to contrast between the best suitability of the Weibull (1951) and exponential distributions. 161 D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 At each temperature, the traditional exponential decay model (Eq. (3)) was used for the calculation of mortality rates, according to the method reported by Houde (2002): Nt = N0 exp(−Zt) (3) where Nt is the number of survivors at age t (d−1 ), N0 the abundance at 6 h post-fertilisation for eggs, and Z (d−1 ) is the daily mortality coefficient. The model was fitted using ordinary least-squares for the log-transformed equation of survivors on age: ln Nt = ln N0 − Zt where x′ = (x1 , . . ., xk ) is the vector of explanatory variables and ␤ is the vector of parameters to be estimated. The covariates considered in this case were temperature and stage. The model was fitted using maximum likelihood, from which the Pth percentiles of the cumulative distribution functions of survival were derived. The suitability of the Weibull distribution was compared with the exponential, log–normal and log–logistic distributions. 3. Results (4) The effect of temperature was evaluated through ANCOVA analysis. Moreover, cumulative mortality rate Zt (d−1 ), survival rate (Eq. (5)) and the proportion of individuals dying in time (Eq. (6)) were also calculated for each temperature, according to S(t) = e−Zt (5) A(t) = 1 − e−Zt (6) Mortality was also studied using survival analysis. The mortality events were assumed to follow a Weibull distribution (Eq. (7)) (Chambers and Legget, 1989; Pepin et al., 1997). The corresponding survival and hazard functions are given by S(t) = exp(−(t/α)γ ) and h(t) = (γ/α)(t/α)γ−1 (7) where α is the scale parameter in units of event observations (time) and γ is the shape parameter, which is dimensionless. Type I, II or III survivors curves are derived (Deevey, 1947), when γ > 1, =1 or <1, respectively. In particular, when γ = 1, the Weibull distribution is identical to the exponential distribution. The effect of additional covariates on the mortality events, such as environmental variables, can be studied by incorporating them into the Weibull models. Then, the survival and hazard functions of the Weibull model are S(t) = exp(−(t/α)γ exp(x′ ␤)) and h(t) = (γ/α)(t/α)γ−1 exp(x′ ␤) (8) 3.1. Effect of temperature on development Ripe eggs and sperm were obtained from females and males of 35.91 ± 3.37 cm total length (mean ± S.D.). The sea surface temperature at the position of capture was around 12 ◦ C. The mean egg diameter of viable eggs used in this study was 1.24 ± 0.04 mm (mean ± S.D.) (n = 300). The time (h) from fertilisation, to reach the end of each stage of development, together with the relationship between age and temperature at a given developmental stage, are summarised in Table 2. As expected, a negative relationship was found between temperature and time to hatch (Fig. 1A). The result reported here, in which the exact time of fertilisation is known, has indicated that the embryonic period lasted at most 251.6 h at 8.6 ◦ C and 77.4 h at 17.8 ◦ C; the incubation time variability Ii , ranged between 86% and 99% explained by the temperature model. The observed duration of egg stages changed uniformly between temperatures. At all temperatures, the cumulative rate of development, from stage to stage, showed a pronounced retardation in the intermediate (III and IV) stages. A slight increase in the development rate then took place providing, as result, a sigmoid-shaped curve (Fig. 1B). The development rates of eggs were significantly higher at high temperatures than at low; likewise at extreme stages (IA, IB, II, FH, 50H) than at intermediate stages (III and IV) of development (ANOVA, p < 0.001; Tukey, p < 0.05). The inter-specific duration of stages III and IV varied the most, over the ranges of temperatures studied (Fig. 2). For most stages, the duration ranged roughly between Table 2 Estimated age (hours from fertilisation), to the end of each stage of development, and parameters a, b and determination coefficient (R2 ) for the relationship between the temperature and the observed age for each stage Mean temperature Sd Tem IA IB II III IV First hatch 50% hatch Total hatch 8.6 11.1 13.2 15.1 17.8 0.253 0.238 0.129 0.163 0.420 31.3 25.9 22.8 20.6 18.3 59.0 42.1 33.5 28.1 22.6 85.7 59.0 45.9 37.7 29.7 159.8 108.4 83.4 68.1 53.1 197.4 134.6 103.9 85.0 66.5 226.2 153.6 118.2 96.5 75.2 239.6 159.8 121.5 98.3 75.8 251.6 166.2 125.6 101.0 77.4 −0.74 5.04 0.86 −1.31 6.90 0.95 −1.45 7.58 0.97 −1.51 8.33 0.98 −1.49 8.50 0.98 −1.51 8.67 0.99 −1.57 8.87 0.99 −1.62 9.01 0.99 b a R2 The regression coefficients are for the fitted equation: ln Ii = ln ai + bi T. 162 D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 Table 3 Coefficients of Ii = ai T bi , where Ii is the “average” age of Atlantic mackerel eggs at hatching, at temperature (◦ C), from different localities Locality Reference a b Woods Hole (USA) Nova Scotia (Canada) Bay of Biscay (Europe) Bay of Biscay (Europe) Worley (1933) Lanctot (1980) Lockwood et al. (1977) This study 9.7151 10.2280 9.3797 8.8767 −1.8932 −2.0318 −1.7647 −1.5790 3.2. 2004 development data versus 1977 development data and NWA mackerel Fig. 1. (A) Functional relationship of the age from fertilisation to the age at the end of each stage of NEA mackerel egg development, in relation to incubation temperature. (B) Fitted age from fertilisation to the age at the end of each stage of development, between stage IA and total hatching. 1 h at high (17.8 ◦ C) and 20 h at low (8.6 ◦ C) temperatures. However, for stages III and IV such duration ranged from 13 to 73 h, across the whole range of temperatures utilised in the experiment. Below 11.1 ◦ C, stages IA, IB and II showed similar durations, but warmer than that threshold, their duration differed by more than 10 h. Fig. 2. Total duration of the stages of mackerel egg development, for a range of five different temperatures, derived from the fitted models. The effect of temperature, on the incubation time to reach 50% of hatching, was compared with previous studies undertaken on Atlantic mackerel (Table 3). There were significant differences between such studies (ANCOVA, d.f. = 3, f = 38.7, p < 0.05). The stages duration obtained during this study, together with those reported by Lockwood et al. (1977), were also compared. No significant differences (ANCOVA, p > 0.05) (Table 4) were found at any of the II–V and total hatch stages. However, for the rates of development within stages IA, IB and 50% Hatch, significant differences between studies (ANCOVA, p < 0.05) were identified. In the present study, stages IA and IB were significantly more rapid than those reported by Lockwood et al. (1977), at low temperature ranges and slower at high temperature ranges, with a midpoint of coincidence at 12 ◦ C (Fig. 3). The percentage difference in the incubation time to reach 50% of hatching, between the two studies, ranged from 16.5% at 7.4 ◦ C to −2.4% at 17.8 ◦ C. Likewise, such difference in stage IB ranged from 25.3% at 7.4 ◦ C to −0.6% at 17.8 ◦ C. The negative values indicate a more rapid development in the Lockwood’s study (Table 5). 3.3. Effect of temperature on mortality Estimates of daily mortality rates were calculated from exponential models of decay; these were represented as a mean value, from fertilisation to hatch (Table 6A). The cumulative mortality at hatch undertaken at 11–17.8 ◦ C ranged from 67 to 80%; the daily mortality rate ranged from 0.17 to 0.38 d−1 between these two specific experimental temperatures (Table 6A). Subsequently, stage-specific mortality, for each temperature, was calculated as a percentage of population dying at stage of development. Temperature did not show any significant influence on the daily mortality rate of mackerel fish eggs (ANCOVA, d.f. = 4, f = 2.16, p > 0.05), or on the stagespecific mortality rate (ANOVA, p > 0.05). Thus, a mean mortality value was calculated for each stage of development (Table 6B), covering the temperature range 8.6–17.8 ◦ C; these ranged from 5.0 ± 4.34% at IA and to 20.0 ± 8.3% at stage V. The distribution of the observed mortality events is plotted in Fig. 4A and B, for each temperature and stage, respectively. 163 D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 Table 4 Parameters of Eq. (2), for each stage of development reported by Lockwood et al. (1977) and the present study 2004 Stage IA IB II III IV V 50H Tot.H Lockwood et al. (1977) This study 2004 Analysis of covariance a b a b d.f. F intercept p F slope p 7.206 7.759 8.276 8.938 8.987 8.739 9.379 9.470 −1.599 −1.614 −1.588 −1.682 −1.647 −1.552 −1.764 −1.753 5.036 6.902 7.578 8.327 8.497 8.671 8.876 9.067 −0.740 −1.313 −1.454 −1.512 −1.494 −1.511 −1.579 −1.643 1 1 1 1 1 1 1 1 0.09 46.57 277.5 49.01 17.7 2.74 5.73 39.04 0.77 <0.01 <0.01 <0.01 <0.01 >0.05 <0.05 <0.01 32.92 46.57 2.61 2.82 2.78 0.22 0.032 2.25 <0.01** <0.01** >0.05 >0.05 >0.05 >0.05 <0.05* >0.05 ANCOVA was used to compare slopes and intercepts at the 95 and 99% confidence intervals. Fig. 3. Individual plots showing the time at the end of each egg developmental stage of Atlantic mackerel, from stage IA to total hatch, in relation with temperature. The plots compare the reported data by Lockwood et al. (1977) and those from this study (2004). The results showed that mortality occur later at lower temperatures than at higher ones. Likewise, mortality was found to be more probable during the latest stages of development. In particular, such mortality events, at stage IV of development are very rapid. The fitted coefficients of the Weibull model are given in Table 7. Survival and hazard functions of the Weibull distribution are showed in Fig. 5A and B for the different stages of development, at temperatures of 13, 15 and 18 ◦ C. The Weibull distribution, in terms of maximizing the likelihood, provided a better fit than the log–normal or log–logistic distributions. Furthermore, a likelihood ratio test Table 5 Incubation times of Atlantic mackerel eggs, at IB and 50H stages together with the percentage difference between Lockwood et al. (1977) and this study Temperature (◦ C) This study IB (h) 50H (h) IB (h) 50H (h) 7.4 8.4 9.4 10.4 11.5 12.6 13.4 14.4 15.1 16.1 17 17.8 71.7 60.7 52.4 45.9 40.2 35.7 32.9 29.9 28.1 25.8 24.1 22.6 303.6 248.2 207.8 177.1 151.1 130.8 118.7 105.9 98.3 88.8 81.5 75.8 96 74 61 54 45 40.5 36 31 29 26 25 22.5 363.3 280.4 221.8 182.8 158.7 134.8 118.5 106.2 96.0 86.8 82.2 77.3 Lockwood et al. (1977) IB (%) difference 50H (%) difference 25.3 17.9 14.1 15.1 10.7 12.0 8.7 3.5 3.1 0.6 3.8 −0.6 16.5 11.5 6.3 3.1 4.8 3.0 −0.2 0.3 −2.4 −2.3 0.9 2.0 164 D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 Fig. 4. Box-plots. Observed times at which the mortality process occurs for each temperature (A) and for each stage of development (B). The median values are represented by solid lines and the outliers by circles. Table 6A Instantaneous mortality rate (Z), cumulative mortality (Cum Z), cumulative survival (Cum S) and final mortality percentage (A%), at temperatures of 8.6, 11.1, 13.2, 15.1 and 17.8 ◦ C NEA mackerel T (◦ C) Z (d−1 ) Cum Z Cum S A% Embryos 8.6 11.1 13.2 15.1 17.8 0.32 0.17 0.22 0.24 0.38 3.31 1.17 1.14 0.99 1.23 0.03 0.33 0.27 0.30 0.20 96 67 73 70 80 showed that the Weibull distribution fitted better than the exponential distribution (likelihood ratio test χ2 = 372.72; d.f. = 1; p < 0.001). Age-specific mortality increased with age at stage, as documented by the γ = 10 > 1 value, and with increasing temperatures (χ2 = 359.31; d.f. = 8; p < 0.001) between 11.1 and 17.8 ◦ C. The percentile values, for 50% of the total population dead, ranged from 2.5 d−1 at 17.8 ◦ C, to 6.3 d−1 at 8.6 ◦ C. The resultant biological function corresponded to a Deevey’s (1947) type I survivorship curve, where mortality events did not occur at a constant, rate but increase during the late stages of development (Fig. 5B). Table 6B Mean stage-specific mortality values, given as % of population dying at a particular stage of development (8.6/11.1/13.2/15.1/17.8 ◦ C) Stage Mean A% stage-specific IA IB II III IV FH 50H Tot.H 5.0 4.1 4.1 10 17.5 20.0 2.5 13.3 ± ± ± ± ± ± ± ± 4.34 2.9 3.3 0.11 4.2 8.3 6.9 8.3 Table 7 Parameter estimates of the Weibull model, from stage IA to total hatch (Tot.H), with the corresponding standard errors (S.E.) and p-values Stage IA IB II III IV V 50H Tot.H Temperature Scale, α S.E. p-Value 1.186 1.826 1.998 2.541 2.912 3.040 3.070 3.118 0.061 0.064 0.064 0.053 0.058 0.050 0.054 0.050 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 −0.107 0.109 0.003 0.073 <0.001 <0.001 βi 4. Discussion 4.1. Development Most studies concerned with the assignation of ages to eggs collected at sea, are based on the time of spawning, the stage of development and the surface temperature at the time of collection. As expected, the results of this development study show that the incubation times vary between stages and temperatures, with the extreme stages developing more rapid than the intermediate ones. These observations were reported also for anchovy and sardine, where the incubation time decreased with increasing temperatures (e.g. Lo, 1985; Miranda et al., 1990; Motos, 1994; Bernal et al., 2001). Moreover, retardations in the development rate, at intermediate stages (e.g. Le Clus and Malan, 1995) were characteristic of the stage III and coincident with the closure of the blastopore. Such observations provide important biological information on the probability of capture of each mackerel egg-stage for field surveys; they outline the possibility of producing spe- D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 165 Fig. 5. (A) Estimated survival functions, S(t), for NEA mackerel eggs maintained at 8, 13, and 18 ◦ C. (B) Age-specific instantaneous mortality rate (hazard function, h(t)) for the Weibull distributions in (A). The stages of development (IA, IB, II, III, IV, V, 50% hatch and total hatching) are represented by time events and percentile values. cific stage capturability indexes. The morphological patterns of development, observed during this study, were consistent with those reported by Russell (1976), based on wild eggs of S. scombrus L. Lockwood et al. (1977) estimated the total duration of the NEA mackerel embryonic period, as 222 h at 9 ◦ C and 77 h at 17 ◦ C, respectively. The total duration of the embryonic period is similar to that found by Lockwood et al. (1977). However, in the trials undertaken here, the stage duration time was generally more rapid. Particularly, the significant differences obtained between theoretical equations, of both studies at the IA, IB and 50% hatching levels at intermediate temperatures (10–15 ◦ C), suggest a different response. These differences might be due also to the experimental techniques. Blaxter (1981) explained the importance of not exceeding a density of 500 ind l−1 , to obtain an optimum development of eggs. Likewise, Hempel (1979) and Rombough (1988) reported the important role of the respiration, on egg survival. Heath (1992) included the oxygen concentration as an exogenous source of mortality, whereas Kjorsvik et al. (1984) explained that, in the later embryonic stages, a decrease in egg strength can be caused by infectious bacterial growth. In this way, three differences could be considered of importance, between this study and Lockwood’s study: the former used batches of 200 eggs placed in 50 ml of seawater volume; here, batches were sorted according to one egg per 35 ml. The frequency of examination reported by Lockwood was every 6 h, whilst it progressed here every 2 h. During the former study, the dead eggs, likely caused by over-density, hypoxia or bacterial growth, had to be removed twice daily from their tubes. In this way, our experimental procedure was previously designed to avoid any contamination, lack of information or handling of eggs. The differences at stages IA and IB, indicated that mackerel eggs herein developed more rapidly at low temperatures than those reported by Lockwood. In the presenting study, a temperature for cooling and transport of fertilised eggs was used prior to the experimentation in the laboratory. This suggests that this previous acclimation of eggs at low temperatures could have been beneficial for the development rates at such temperatures. Rates of development without retardations, together with maximum survival rates at these two levels, would support such a concept in the present experiments. Jobling (1994) reported that the thermal tolerance is dependent upon previous thermal history and that the compensatory changes in the physiological rates may occur more rapidly with increasing temperatures than with decreasing ones. Moreover, the convergence of the IA curves of both studies, at 12 ◦ C, is coincident with the natural thermal preferences of mackerel in the Bay of Biscay (Valencia et al., 1988; Bez and Rivoirard, 2000) and with the temperature of acclimation used in Lockwood et al. (1977). On the other hand, differences between the times to hatch, of all the experiments from different localities, suggest that the development rate could be dependent on the origin of the eggs. Fox et al. (2003) reported that the temperature- 166 D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 dependent egg development rates of plaice were influenced by egg size, however in spite of effects attributable to gametes and egg size (e.g. Rideout et al., 2004) most of the divergence in development are mostly attributed to differences in incubation temperature (e.g. Pepin, 1991; Pepin et al., 1997; Fuiman, 2002). Blaxter and Hempel (1963), have shown that the time from fertilisation to hatch, in sardine, does not depend on egg size; this suggests that, although the sizes of the eggs used in this study were different from those of Worley (1933), Lockwood et al. (1977) or Lanctot (1980), such a property might not be valid to explain the different results reported here. In this way, since Lockwood used eggs collected in the Celtic Sea and West of Ireland, the differences (between the total incubation times from this study and the earliest) could indicate that regional or seasonal differences in the NEA could have produce different genetics tolerance to temperature, or that slight differences in thermal history of the parent fishes prior to spawning could have affected the development rates of the eggs (which we can not confirm). 4.2. Mortality The survival of eggs of this specie is a probable source of variability in recruitment to adult stocks (Hjort, 1914). In this study, the daily mortality rate of the NEA mackerel eggs, without predation between 11 and 17.8 ◦ C, range from 0.17 to 0.38 d−1 ; as such, this would result in only 20–33% of the fertilised eggs surviving to produce larvae. These results lie close to the 0.44 d−1 reported by Ware and Lambert (1985), for NWA mackerel eggs under natural conditions; this would result in 20% of surviving eggs, after following the total incubation period. In the published literature, suggestions have been made of links between laboratory-derived observations and field estimates, in terms of development and mortality on fish eggs without predation (see, for example, reviews in Dickey-Collas et al. (2003)). During the present study, the analysis of the survival process was undertaken by studying, simultaneously, development and mortality rates. The results obtained on cumulative mortality of mackerel eggs at hatch (67–96%), fluctuated between temperatures similar to those reported by Danielssen and Iversen (1977) (75–90%) and Lanctot (1980) (62–90%); furthermore, the minimal mortality found between 11 and 13 ◦ C, correspond to the water temperature at the time of capture of the parent fish. This suggests that acclimation of adult fish might influence the thermal tolerance of the embryos and subsequent survival. Bunn et al. (2000) also suggested that, the absolute temperature change that most eggs can tolerate is about 6 ◦ C and that susceptibility to mechanical pressure is dependent on the stage of development. Therefore, it must to be considered that factors like cooling, transportation or handling could have affected the results of survival. Other endogenous factor, such us quality of eggs and sperm and the artificial fertilisation are also suggested as a possible cause of early mortality (see reviews by Kjorsvik et al. (1990), Bromage (1995) and Brooks et al. (1997)). However, our results show that, at all tempera- tures, stages I and II represent the most significant period of survival for mackerel eggs and that mortality increases as hatching approaches. Moreover, the values of mortality found herein are to be lower than those previously reported for this species, from eggs spawned and fertilised in nature (Ware and Lambert, 1985). The mortality events observed during the presenting study were more pronounced at the extremes of the experimental temperature range, than at intermediate values; this was observed also by Worley (1933), Lockwood et al. (1977) and Lanctot (1980). This trend would simply indicate that mortality is lower at temperature ranges where the developing mackerel eggs are usually encountered naturally. Likewise, the trend of the daily mortality rates found herein would support the popular thinking of that natural daily mortality rates increase with temperature (see Pepin, 1991). However, the high mortality experienced at 8.6 ◦ C, would also suggest that this temperature becomes suboptimal for the NEA mackerel embryonic development. Field studies on egg production often incorporate estimates of the rate of natural mortality into their models. Such estimates take into account the mortality suffered from the time when the egg is released into the water, to the time at which it is sampled, through the calculation of an instantaneous daily mortality rate (Z); this, in turn, is derived from an exponential model of decay (Bunn et al., 2000). Exponential models imply that the mortality risk occurs at a constant rate, which might not always be appropriate. During the present study, event analysis has indicated that variations in temperature, per se, have significant effects on the time of survival of eggs of NEA mackerel; however the exponential model of decay did not detect such differences. As expected, the eggs did not die at a constant rate, throughout all the stages of development; this was documented by the instantaneous stage mortality rates and the probability values of the Weibull likelihood ratio tests. These observations suggest that, although the exponential models can be useful methods to rate the magnitude of the whole mortality process during the egg stage, the biological meaning of the different stage strengths (Thompson and Riley, 1981; Iversen and Danielssen, 1984; Rombough, 1996), the combination of the temperature-stage related effects, or the endogenous factors (Kjorsvik et al., 1990), are aspects that should be taken into account to produce appropriate mortality estimations. In this way, Thompson (1989), using the exponential model of decay to fit mortality curves to mackerel natural egg counts, reported that several problems in the calculation of the stage-duration could have produced erroneous estimations. Likewise, Dickey-Collas et al. (2003) found that ageing imprecision can lead to a bias in the estimation of survivorship of wild plaice eggs from spawning to hatch. Such problems can now be avoided using the temperature-stage related Weibull distribution, which has shown to fit the age at death series better than the exponential model. Such an approach covered the entire distribution of known-age mortality events and it rigorously compared, D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 between groups, the huge mortality events that occur during the latest embryonic stages of development. The results herein show that the information measured from individuals, provides more accuracy and allows more complete and less biased use of data than when measuring from populations. The current study adds more evidence of the differential daily mortality with stage and proposes a useful mechanism to deal with this aspect. In conclusion, event analysis has shown that a strong age-dependent relationship of stage-specific mortality rate exists, during the egg period of the NEA mackerel; this has been reported also by Irvin (1974), Blaxter (1988) and Pepin (1991), for other species. This pattern would suggest that, within the optimal thermal range for development, temperature affects on mortality, when reductions or increases in any development time are proposed by unexpected changes of temperature. Such differences in development and stagespecific mortality rates could have consequences, both when calculating egg mortality in the field and when applied to estimating the spawning stock biomass, through egg production methods. However, additional research should quantify the impact of the development differences on the egg production estimates based on field surveys. Special attention must to be paid in thermal acclimation and compensatory response of the physiological rates. The development model herein could be used to determine the age of mackerel eggs, collected at a range of natural temperatures. Acknowledgements We are grateful to the crews of the Divino Jesus de Praga and Gure Aita Jose, on board of which several fertilisation surveys were carried out; likewise, to the Aquarium personnel for their advice on the design of culture systems. This work was supported by a PhD grant from the Education, Universities and Research Department of the Basque Government. We wish to thank Leire Ibaibarriaga (AZTI-Tecnalia, Spain) for her assistance with statistical analysis, and similarly Enrique Portilla (FRS, Marine Laboratory, UK) for his useful comments throughout the preparation of this paper. Finally we appreciate the constructive comments of X. Irigoien (AZTITecnalia, Spain) and M. Collins (University of Southampton, UK) who helped to improve this paper. References Ahlstrom, E.H., 1954. Distribution and abundance of egg and larval populations of the Pacific sardine. Fish. Bull. 56, 83–140. Anon, 1967. Oceanographic atlas of the North Atlantic Ocean. Section II. Physical Properties, Publication No. 700. US Naval Oceanographic Office, Washington, DC. Anon, 1996. Report of the working group on the assessment of mackerel, horse mackerel, sardine and anchovy. ICES CM 1996/Assess 7. Bernal, M., Borchers, D.L., Valdes, L., Lago de Lanzos, A., Buckland, S.T., 2001. A new ageing method for eggs of fish species with 167 daily spawning synchronicity. Can. J. Fish. Aquat. Sci. 58, 2330– 2340. Bez, N., Rivoirard, J., 2000. On the role of sea surface temperature on the spatial distribution of early stages of mackerel using inertiograms. ICES J. Mar. Sci. 57, 383–392. Blaxter, J.H.S., 1974. The Early Life History of Fish. Springer-Verlag, New York. Blaxter, J.H.S., 1981. The rearing of larval fish. In: Hawkins, A. (Ed.), Aquarium Systems. Academic Press, New York, pp. 303–323. Blaxter, J.H.S., 1988. Pattern and variety in development. In: Hoar, W.S., Randall, D.J. (Eds.), Fish Physiology, vol. 11a. Academic Press, London, pp. 1–58. Blaxter, J.H.S., Hempel, G., 1963. The influence of egg size on herring larvae. (Clupea harengus L.). J. Cons. Perm. Int. Explor. Mer. 28, 211–240. Bromage, N., 1995. Broodstock management and seed quality: general considerations. In: Bromage, N.R., Roberts, R.J. (Eds.), Broodstock Management and Egg and Larvae quality. Blackwell Science, London, pp. 1–23. Brooks, S., Tyler, C.R., Sumpter, J.P., 1997. Egg quality in fish: what makes a good egg? Rev. Fish Biol. Fish. 7, 387–416. Bunn, N.A., Fox, C.J., Webb, T., 2000. A literature review of studies on fish egg mortality: Implications for the estimation of spawning stock biomass by the annual egg production method. Science Series Technical Report 11. CEFAS, Lowestoft, UK. Cameron, P., Berg, J., Dethlefsen, V., Von Westerhagen, H., 1992. Developmental defects in pelagic embryos of several flatfish species in the southern North Sea. Neth. J. Sea Res. 29, 239–256. Chambers, R.C., Legget, W.C., 1989. Event analysis applied to timing in marine fish ontogeny. Can. J. Fish. Aquat. Sci. 46, 1633–1641. Chambers, R.C., Witting, D.A., Lewis, S.J., 2001. Detecting critical periods in larval flatfish populations. J. Sea Res. 45, 231–242. Danielssen, D.S., Iversen, S.A., 1977. The development and mortality of mackerel eggs (Scomber scombrus L.) in different temperatures, ICES CM 1977/L19. Deevey, E.S., 1947. Life tables for natural populations of animals. Quart. Rev. Biol. 22, 283–314. Dickey-Collas, M., Fox, C.J., Nash, R.D.M., O’Brien, C.M., 2003. Plaice egg mortality: can we determine survivorship? J. Sea Res. 50 (2/3), 211–225 (Special issue). Fox, C.J., Geffen, A.J., Blyth, R., Nash, R.D.M., 2003. Temperaturedependent development rates of plaice (Pleuronectes platessa L.) eggs from the Irish Sea. J. Plankton Res. 25, 1319–1329. Fuiman, L.A., 2002. Special considerations of fish eggs and larvae. In: Fuiman, L.A., Werner, R.G. (Eds.), Fishery Science. The Unique Contributions of Early Life Stages. Blackwell Publishing, Oxford, UK, pp. 1–32. Gunderson, D.R., 1993. Surveys of Fisheries Resources. John Wiley and Sons, New York. Heath, M.R., 1992. Field investigations of the early life stages of marine fish. Adv. Mar. Biol. 28, 1–130. Hempel, G., 1979. Early Life History of Marine Fish: The Egg Stage. Division of Marine Resource, University of Washington, Washington. Hjort, J., 1914. Fluctuations in the great Fisheries of northern Europe. Rapp. P.-V. Cons. Perm. Int. Explor. Mer. 20, 1–228. Houde, E.D., 1987. Fish life dynamics and recruitment variability. Am. Fish. Soc. Symp. 2, 17–29. Houde, E.D., 2002. Mortality. In: Fuiman, L., Werner, R. (Eds.), Fishery Science. The Unique Contributions of Early Life Stages. Blackwell Publishing, Oxford, UK, pp. 64–87. Irvin, D.N., 1974. Temperature tolerance of early developmental stages of Dover sole, Solea solea (L.). In: Blaxter, J.H.S. (Ed.), The Early Life History of Fish. Springer-Verlag, Heidelberg, pp. 449– 463. Iversen, S.A., Danielssen, D.S., 1984. Development and mortality of cod (Gadus morhua L.) eggs and larvae in different temperatures. Flodevigen Rapportser. 1, 49–65. 168 D. Mendiola et al. / Fisheries Research 80 (2006) 158–168 Jobling, M., 1994. Fish Bioenergetics: Fish and Fisheries Series, vol. 13. Chapman and Hall, London. Kjorsvik, E., Stene, A., Lonning, S., 1984. Morphological, physiological and genetical studies of egg quality in cod (Gadus morhua L.). Flodevigen Rapportser. 1, 67–86. Kjorsvik, E., Mangor-Jesen, A., Holmefjord, I., 1990. Egg quality in fishes. Adv. Mar. Biol. A 26, 71–113. Lanctot, M., 1980. The development and early growth of embryos and larvae of the Atlantic mackerel, Scomber scombrus L., at different temperatures. Can. Tech. Rep. Fish. Aquat. Sci. 927. Government of Canada, Fisheries and Ocean, Dalhousie University, Halifax, NS. Lasker, R., 1964. An experimental study of the effect of temperature on the incubation time, development, and growth of Pacific sardine embryos and larvae. Copeia 2, 399–405. Le Clus, F., Malan, P.E., 1995. Models of temperature dependent rate of development of pilchard, Sardinops sagax eggs, to be used in routine procedures for estimating daily egg production. S. Afr. J. Mar. Sci. 16, 1–8. Legget, W.C., Deblois, E., 1994. Recruitment in marine fishes: is it regulated by starvation and predation in the egg and larval stages. Neth. J. Sea Res. 32, 119–134. Lo, N., 1985. A model for temperature-dependent northern anchovy egg development and an automated procedure for the assignment of age to staged eggs. In: Lasker, R. (Ed.), An egg production method for estimating spawning biomass of pelagic fish: application to the northern anchovy (Engraulis mordax). U.S. Dep. Commer. NOAA Tech. Rep NMFS.36. Lockwood, S. J., Nichols, J.H., Coombs S.H., 1977. The development rates of mackerel (Scomber scombrus L.) eggs over a range of temperatures. ICES CM 1977/J:13. Makhotin, V., Solemdal, P., Korsbrekke, K., Salthaug, A., 2001. Types and frequency of malformations and mortality in eggs of Arcto-Norwegian cod: a field study. ICES CM2001/N:12. Miranda, A., Cal, R.M., Iglesias, J., 1990. Effect of temperature on the development of eggs and larvae of sardine (Sardina pilchardus) Walbaum in captivity. J. Exp. Mar. Biol. Ecol. 140, 69–77. Moser, H.G., Ahlstrom, E.H., 1985. Staging anchovy eggs. NOAA Tech. Rep. NMFS 36, pp. 37–41. Motos, L., 1994. Estimación de la biomasa desovante de la población de anchoa del Golfo de Vizcaya, Engraulis encrasicolus, a partir de su produccion de huevos. Bases metodológicas y aplicación. PhD Thesis. University of Leioa, Spain. Patterson, K.R., Melvin, G.D., 1996. Integrated catch at age analysis version 1.2. Scott. Fish. Res. Rep. 56, Aberdeen. Pauly, D., Pullin, R.S.V., 1988. Hatching time in spherical, pelagic marine fish eggs in response to temperature and egg size. Environ. Biol. Fish. 22, 261–271. Pepin, P., 1991. Effect of temperature and size on development, mortality, and survival rates of the pelagic early life history stages of marine fishes. Can. J. Fish. Aquat. Sci. 48, 503–518. Pepin, P., Myers, R.A., 1991. Significance of eggs and larval size to recruitment variability of temperate marine fish. Can. J. Fish. Aquat. Sci. 48, 1820–1828. Pepin, P., Orr, D.C., Anderson, J.T., 1997. Time to hatch and larval size in relation to temperature and egg size in Atlantic cod. Can. J. Fish. Aquat. Sci. 54 (1), 2–10. Priede, I.G., Watson, J.J., 1993. An evaluation of the daily egg production method for estimating biomass of Atlantic mackerel (Scomber scombrus). Bull. Mar. Sci. 53 (2), 891–911. Rideout, R.M., Trippel, E.A., Litvak, M.K., 2004. Paternal effects on haddock early life history traits. J. Fish. Biol. 64, 695–701. Rombough, P.J., 1988. Respiratory gas exchange, aerobic metabolism, and effects of hypoxia during early life. In: Hoar, W.S., Randall, D.J. (Eds.), Fish Physiology, vol. 11a. Academic Press, San Diego, pp. 59–161. Rombough, P.J., 1996. The effects of temperature on embryonic and larval development. Soc. Exp. Biol. Sem. Ser. 61, 177–223. Russell, F.S., 1976. The Eggs and Planktonic Stages of British Marine Fishes. Academic Press, London. Stauffer, G., Picquelle, S., 1985. Procedures for sorting, staging, and ageing eggs. In: Lasker, R. (Ed.), An Egg Production Method for Estimating Spawning Biomass of Pelagic Fish: Application to the Northern Anchovy, Engraulis mordax. NOAA Tech. Rep. NMFS 36, pp. 33–35. Thompson, A.B., 1989. Mackerel (Scomber scombrus) egg mortality: the western mackerel stock in Biscay and the Western Approaches in 1977, 1980, 1983 and 1986. J. Plankton Res. 11, 1297– 1306. Thompson, B.M., Riley, J.D., 1981. Egg and larval development studies in the North Sea cod (Gadus morhua L.). Rapp. P.-V. Reun. Cons. Int. Explor. Mer. 178, 553–559. Valencia, V., Motos, L., Urrutia, J., 1988. Estudio de la variación temporal de la hidrografı́a y el plancton en la zona nerı́tica frente a San Sebastián. Resultados abril 1986–diciembre 1987. Informes técnicos Departamento Agricultura y Pesca, p. 20 (in Spanish). Valencia, V., Franco, J., Borja, A., Fontán, A., 2004. Hydrography of the southeastern Bay of Biscay. In: Borja, A., Collins, M. (Eds.), Oceanography and Marine Environment of the Basque Country. Elsevier, Amsterdam, pp. 159–194. Van der Land, M.A., 1991. Distribution of flatfish eggs in the 1989 egg surveys in the southern North Sea, and mortality of plaice and sole eggs. Neth. J. Sea Res. 27, 277–286. Ware, D.M., Lambert, T.C., 1985. Early life history of Atlantic Mackerel (Scomber scombrus) in the southern Gulf of St. Lawrence. Can. J. Fish. Aquat. Sci. 42, 577–592. Weibull, W., 1951. A statistical distribution function of wide applicability. J. Appl. Mech. 18, 293–297. Worley, L.G., 1933. Development of the egg of the mackerel at different constant temperatures. J. Gen. Physiol. 16, 841–857.