6. BIOLOGY OF TUNA (Contd.)

6.1.3. Growth

6.1.3.1. Introduction

Growth is an aspect of fish biology essential for the use of numerous population dynamics models and thus also for the management of stocks. After a brief review of the principal methods used to determine the age of the three tuna species that interest us (yellowfin, skipjack and bigeye), we will show the results; when a wide choice of publications is presented, we will retain the more recent works and/or those that use a large sample of fish (number and size range of individuals concerned).

6.1.3.2. Determination and expression methods of results

Three different techniques are generally used to determine growth of fish: the progressions of modes observable in size frequency distributions of individuals sampled, the direct reading of age from different hard parts (bones, scales, otoliths…) and tagging. Each of these methods will be summarized.

6.1.3.2.1. Modal progressions or the Petersen (1895) method

This method follows the change in time and thus the growth of modal sizes that appear in length-frequency distributions. The size frequency data of fish come from samples generally collected with a regular periodicity (e.g. monthly). It is supposed that individuals are born in different successive groups and that these groups will be identifiable by their average size (or mode) in size frequency samples collected out periodically. It is considered that the method will be most applicable when, as Postel indicates (1955), “the spawning period is short and the population homogeneous”.

Based on this principal, there are numerous techniques to decompose plurimodal distributions (in which the modes may overlap more or less), into different characteristic modes the changes in which one follows over time (Cassie, 1954; Tanaka, 1962; Hasselblad, 1960; Bhattacharya, 1967; Gheno and Le Guen, 1968; Pauly and David, 1981…).

6.1.3.2.2. Direct reading of age

This method is based on the observation that growth of a fish is not regular and that halts or even slowing of growth for any cause (depletion of food, temperature of environment, reproduction, migration, diseases…) will be materialized by a visible phenomenon in certain calcified tissues (skeletal or not), such as: vertebrae, maxilla, scales, otoliths, fin rays … The appearance of growth stoppage marks or “checks” can even depend on an internal biological rhythm; the periodicity of their formation is in this case very precise, around 24 hours, and one then speaks of “daily growth rings”.

Once the periodicity of these different checks in the tissues is established and known, it is possible to determine the age of an individual fish by simply counting the checks. To establish a growth curve, either these observations can be repeated on several fish of different sizes, or if the mathematical relation tying the number of marks to the size or age of the species concerned is known, the sizes or ages corresponding to each check may be “back calculated” from a limited number of individuals.

It should be emphasized that as well as the often delicate preparation and recording techniques, the principal difficulty of the method resides in the “validation” of results, that is the precise determination of the periodicity of appearance of growth checks.

For fish such as tuna that are difficult to raise, the most widely used validation technique consists of injecting substances such as tetracycline into fish which have been previously measured and identified by a tag. These substances have the property of fixing themselves almost instantly into the calcified tissues during growth and leave a very visible and specific mark; when the fish is recaptured, possible growth checks formed between dates of tagging and recapture can be clearly identified.

6.1.3.2.3. Tagging

This technique consists of placing tags on live previously measured fish and then releasing them. The subsequent recapture of these individuals which are identifiable by their tags, permits the determination of their growth from the relation between the observed increase in size and time passed between tagging and recapture. In order to be effective, this method requires that the number of fish recaptured is sufficient; the magnitude of the number of fish to tag will depend on the area in which these fish are likely to move (migrations) and their rate of exploitation. However accuracy in size measurements at tagging and recapture is indispensable for the success of this technique.

6.1.3.2.4. Expression of results

Results obtained by application of the three methods carried out below, are generally expressed more often in the three following ways:

• growth rate: the increase in length (expressed in millimeters or centimeters) or of weight per unit of time (generally month or year). Growth rate is often given by length or weight range of the fish.

• the length (or weight)-age key: it is a table of correspondence between actual or relative age and the length or weight of the fish.

• the von Bertalanffy (1938) growth function; this function is expressed according to the classical equation:

L_t = L∞ (1 - e^-k(t-t_o))

with L_t = length at age t,

L∞ = asymptotic length,

k = growth rate, and

t₀ = theoretical age at which length is zero.

The exponential curve that corresponds to this equation, tends to an asymptotic length in which the value is symbolized by L∞. It is necessary to note that this length L∞ is essentially a mathematical parameter, characteristic of the von Bertalanffy equation. The value of this parameter depends mainly on the size composition of fish in the sample used to calculate the growth equation; it is only if all the individuals (including larger fish) are well represented in samples used, that L∞ will correspond fairly well with the average size of the largest individuals sampled. One can not therefore a priori associate L∞ with the maximum size that the species considered can reach in nature.

We will indicate maximum sizes that have actually been observed of each species independently of values of L∞. It must be emphasize that as a general rule, it is dangerous to extrapolate growth results obtained from a sample of fish covering a given size range, outside of this size range.

6.1.3.3. Results

The results obtained by each of the different growth study methods will be given for the three species, yellowfin, skipjack and bigeye; finally, we will indicate the more probable synthetic global results concerning the growth of each species.

6.1.3.3.1. Yellowfin

Direct reading of age

Scales

The first important work on the direct reading of age of Atlantic yellowfin from scale reading has been published by Yang, Nose and Hiyama (1969) from a sample of 296 yellowfin fished by longline; these authors have carried out reading of scales taken under the fourth dorsal spine. Their results based on 144 individuals are:

• Growth rate: 34 cm/year (i.e. 2.8 cm/month) for size (determined by back calculation) between 66 and 130 cm.

• Von Bertalanffy equation: Under the hypothesis advanced by Yang et al., according to which the check marks appear 2 times per year (in May and September), the growth parameters calculated by these authors (L∞ = 222.8 cm; k = 0.139) correspond therefore to a six month time scale because 6 months separate the formation of two successive check marks; converted to an annual scale, these parameters become: L∞ = 222.8 cm; K(annual) = 0.278.

• Length-age key: the age given is relative age, as the authors point out that they can not determine the periodicity of appearance of rings (2 per year for the months of March and September) except by analogy with hypotheses made in the Pacific, and that they do not know the age of the appearance of the first ring; the relative age (expressed in semesters) corresponding to each mark of growth arrest would depend therefor on this double hypothesis:

- First dorsal fin ray

Draganick and Pelczarski (1984), have attempted to determine the growth of Atlantic yellowfin from the recording of 171 transverse sections of the first dorsal fin of yellowfin caught by longline. The authors point out the hypothetical side of their results by emphasizing the following observations:

• No actual validation of the periodicity of the appearance of check marks (estimated at 2 per year).

• Small sample size and consisting of mainly large individuals (over 120 cm).

We will add to these observations that, as experience has shown, the larger the individuals, as is the case here, the more the sections of the first fin become difficult, even impossible, to read because of the bony redeposition that occurs in the center of this ray rendering this very important central part unreadable.

Taking into account these observations, the parameters of the Von Bertalanffy equation calculated by these authors are: L∞ = 192.4 cm; k(annual) = 0.37; t₀= -0.003 years

Modal progressions (Petersen method)

After the work of Le Guen et al., (1969), the most important synthetic work on yellowfin growth, determined from size frequency distributions was been carried out in 1973 (Le Guen and Sakagawa, 1973). Size frequencies established from length measurements of yellowfin caught between Senegal and Angola from 1966 to 1970 by French and American tuna fleets (pole and line boats and purse seiners) were collected and analyzed by these two authors.

The size range covered by this numerically very large sample includes all sizes of yellowfin from 35 to 180 cm in length, that were and are still exploited in the eastern Atlantic. However, considering the more or less stationary aspect of modes between 40 and 55 Le Guen and Sakagawa only included individuals between 60 and 171 cm in their calculations.

The parameters of the best growth estimation made by these authors for all eastern Atlantic yellowfin, and according to their hypothesis of a fixed birth date of the first of March, are: L∞ = 194.8 cm; k(annual) = 0.420; t₀ = 0.6233 years.

The value of t₀ has been fixed according to hypothesis that fish recruited at 60 cm have an age of 18 months and considering the existence of two birth dates fixed by these authors on March 1 and July 1. The length-age key and the growth curve established from this equation (table 6.4, figure 6.17) indicate a close enough agreement with results calculated from parameters of Yang et al., (1969), at least for sizes corresponding to ages between 2 and 5 years.

This growth curve of Le Guen and Sakagawa has been used extensively as a reference for all population dynamics work for Atlantic yellowfin.

From the observation according that in the recent period following the Le Guen and Sakagawa study, catches of small yellowfin less than 60 cm., as well as those of large yellowfin more than 140 cm. have greatly increased, Fonteneau (1980) stresses that it is indispensable conduct another study in order to clarify the growth of these individuals which were poorly represented in the samples used by Le Guen and Sakagawa.

The samples collected from 1971 to 1977 from FIS pole and line and purse seine catches, as well as those coming from Ghana, Japanese and Korean fleets that exploit mainly small yellowfin, skipjack and bigeye in the Gulf of Guinea, have been used by Fonteneau (1980) to determine the growth of yellowfin. We should point out here that this sample covers all of the Atlantic yellowfin fishery better than the one available for Le Guen and Sakagawa; the sample used by them was in fact a juxtaposition of seasonal samples of diverse geographic origins.

In the Fonteneau analysis, three size scales (fork length) are considered:

• small yellowfin from 35 to 70 cm

• average size yellowfin (70 to 130 cm)

• large yellowfin (over 130 cm)

Modal progressions give the following results:

Small yellowfin (FL from 35 to 70 cm, and W from 0.8 to 6.7 kg)

These individuals seem to grow according to a slow growth law (figure 6.18), with an average growth rate of 1.56 cm/month (18.7 cm/yr), well under the rate calculated for the same size range in by the Le Guen and Sakagawa equation (4.9 cm/month).

The size-age relation, given by Fonteneau, for small yellowfin is based on the observed displacements of modes and on the hypothesis of a fixed birth date of 15 January:

Figure 6.17 Growth curves of yellowfin based on parameters K and L∞ of the Von Bertalanffy equation calculated by different authors.

Table 6.4 Relation between age and size in yellowfin (Thunnus albacares) from the Atlantic calculated from parameters proposed by different authors and determined by various methods. When the growth model proposed by an author is composed of several distinct parts, these are separated by a dotted line. The domain of strict theoretic application of these different age-length relations is indicated by figures marked by an asterisk (*).

Figure 6.18 Pattern of cumulative size-frequency distributions of yellowfin (from 1969 to 1977), by fishing area (Dakar, Abidjan, Pointe-Noire) and by month, derived from measurements made on yellowfin caught by purse seine and baitboat by the FIS tuna fleet (after Fonteneau, 1981). Note the similarity between observations made at Dakar, Abidjan and Pointe-Noire, the slow apparent growth of fish of size less than 70 cm, and the dispersion of observations from yellowfin of size greater than 125 cm.

This relation is included in Fonteneau's general size-age key (table 6.4).

Average size yellowfin (70 to 130 cm; 6.7 to 42.1 kg)

The parameters of the Von Bertalanffy equation calculated by Fonteneau for individuals of the size scale are: L∞ = 166.4 cm; k(annual) = 0.864; t₀ = 1.2917 years.

These parameters and the length-age key that permits calculations for yellowfin included in this size range (table 6.4) are similar to those proposed by Le Guen and Sakagawa (1972).

Large size yellowfin (over 130 cm; weight over 42 kg) For these individuals, Fonteneau points out that “modes are sometimes observed in the size range; but, when they exist analysis of modal progressions is most often impossible” (figure 6.18).

Various reasons could be devised to explain this difficulty:

• a strong inter-individual variability in growth

• the existence of differential growth in males and females

• the mixture of several groups of yellowfin born at different dates

• lack of precision in fork length measurements of these individuals which are in fact calculated from predorsal length measurements in accordance with a mathematical relation (Caveriviere, 1976).

At this point, we will conclude however that for average size yellowfin (70 cm to 140 cm in length), the modal progressions analyses of Le Guen and Sakagawa (1972) and of Fonteneau (1980) give similar results that agree with the conclusions of Yang et al., (1969) obtained from scale reading. On the other hand, Fonteneau's results, that indicate slow growth for small yellowfin (35 to 70 cm) diverge completely from those obtained in the Atlantic by other authors which regardless of method, seem to indicate a “rapid” growth of these yellowfin. However, the sample used by Fonteneau is the only one that actually contains a significant number of fish of this size range (35 to 70 cm).

Given the uncertainties that can exist in results obtained by the Petersen method applied to small yellowfin (selectivity of fishing gear, variability of growth linked to zone and times of catch, continuous recruitment without a fixed date, incorporation of young bigeye with yellowfin…) we will wait to examine the results of tagging experiments in order to discuss more precisely these two divergent hypotheses on growth of young yellowfin.

Tagging

The first calculation of growth rate observed on tagged and recaptured yellowfin was made by Fonteneau (1980) based on 82 individuals for which the time passed between tagging and recapture was over 1 month. In this analysis the two size classes (previously mentioned in the growth analysis by the Petersen method) are considered separately. The results (table below), similar to those obtained by this author by analysis of size frequencies (paragraph 6.1.3.1.), seems to confirm the hypothesis of slow growth of individuals under 70 cm.

After 1975, tagging experiments have been pursued and the number of usable recaptures for estimation of growth has increased. Bard (1984) gathered all data concerning recaptures of tagged yellowfin in the Eastern Atlantic by different countries (Korea, Côte d'Ivoire, France, Japan, Senegal) and retained 243 individuals for growth analysis. After having stated that the growth rate of individuals under 70 cm. was noticeably slower than that of larger yellowfin, this author, by different successive adjustments, determined that that the growth rate was modified starting at 65 cm. Similar to Fonteneau, he describes the growth of yellowfin, by separately considering two size ranges of yellowfin, small (under or equal to 65 cm) and large:

Small yellowfin (35 to 65 cm. or 0.8 to 5.4 kg.)

Growth rate: 17.7 cm/yr (1.47 cm/mo)
Linear growth equation: L = 17.71t–35.0
with L =fork length in centimeters; t = relative age in years, 35 cm being considered as the size corresponding to age 0 at recruitment.

Table 6.5 Maximum longevity of the three major species of tropical tuna (yellowfin, skipjack, bigeye) with the maximum observed sizes and weights.

*: estimation

Large yellowfin (from 65 to 180 cm, or 5.4 to 110.8 kg)

Von Bertalanffy growth equation with
L∞ = 196.55 cm; k(annual) = 0.474; t₀ = 0.847 years

Bard's results concerning yellowfin of less than 65 cm confirm the slow growth rate of these individuals brought to attention by Fonteneau.

Similar work, conducted also on yellowfin tag-recapture data, has permitted Miyabe (1984) to also confirm Fonteneau's hypothesis (1980) of slower growth of yellowfin of less than 60 cm.

In a general manner the size-age key (table 6.4) resulting from Bard's analysis agrees rather well with the one established in 1980 by Fonteneau, but only up to a size of around 150 cm (an age of around 4 years).

The results of both Fonteneau and Bard diverge strongly from those of Le Guen and Sakagawa for yellowfin less than 1 meter, agree well for those between 100 and 165 cm, and then diverge again. It is necessary to emphasize that the strict application interval of Bard's results, is limited to yellowfin sizes actually included in his analysis; in consequence, the equation and the length-age key for yellowfin more than 150 cm remains very hypothetical, in the same way as those for other authors.

Longevity and maximum size

The largest recorded yellowfin caught in the Atlantic was taken by a sport fisherman and weighed 176 kg with a length of around 2.10 meters (IGFA, 1985). At present, one can only estimate roughly that the maximum life span of yellowfin would be 10 to 15 years (table 6.5).

Discussion

Given that the growth results obtained by direct reading of age from scales (Yang et al., 1969) or from sections of the first dorsal fin (Draganick and Pelczarski, 1984) must remain very hypothetical since the have not been validated (paragraph 6.1.3.3.1), only growth rates or equations deduced from the analysis of modal progressions or of recaptured tags will be considered. Considering that the calculations carried out by Le Guen and Sakagawa (1973) do not take into account individuals less than 60 cm, it is a priori less than rigorous to extrapolate the growth equation proposed by these authors for yellowfin to this size category.

The low growth rate (1.4 cm/month) of young yellowfin (40 to 70 cm) that Fonteneau (1980) has calculated from the observation of size frequencies could in fact be due to bias linked to the sampling itself: selectivity of fishing gear, insufficient number of individuals, seasonal sampling, mixture of young bigeye.

This interpretation seems unlikely given the size of the sample, its good spatio-temporal distribution and the similarity of results when the samples from different fishing gears (figure 6.18), when pole and line or purse seine (Fonteneau, 1980) are considered separately. Further, the confusion of young bigeye with yellowfin of the same size in the samples, is possible given the close resemblance of these two species at these sizes, can only introduce a minor bias in the variability of observations because of the large numerical dominance of yellowfin in eastern Atlantic tuna catches.

The weak displacement of modes could also be explained by a more or less permanent recruitment of young yellowfin in the fishery. Yellowfin seem to reproduce more or less all year (chapter 6). However in the Gulf of Guinea reproduction seems to pass a maximum at the beginning of the year, that which appears to confirm the existence of well separated modes observed in measurements of the largest yellowfin (70 to 130 cm). The fate (recruitment) of yellowfin that are born at other times of the year, notably in the third quarter (paragraph 6.1.1.), is still poorly understood. These fish only appear in size frequency samples at lengths over 1 meter and progressively cause overlapping of different modes, preventing their separation.

The analysis of tagging results (Fonteneau, 1980; bard, 1984; Miyabe, 1984) confirms a low growth rate of young yellowfin.

At present, tagging appears to be the most direct and dependable method for determining growth. Biases linked to errors in length measurements of fish at tagging or at recapture, or further variations in length caused by the conservation mode of recaptured fish (freezing) seem negligible. A slowing of growth induced by stress caused by the tagging operation can hardly be involved when it is known that only individuals more than one month at liberty have been included (Bard, 1984), that this stress did not appear in slightly larger individuals (Fonteneau, 1980), and that very recently tagged fish have shown a feeding behavior that seems normal (Cayré, 1982).

Since there is good enough similarity in the growth curve of average size yellowfin (65 to 40 cm) proposed by different authors (figure 6.17), divergences that appear for sizes over 140 cm are a result of poor representation of these individuals in both size frequency samples and in tagging data. Although it has never been directly confirmed, the hypothesis according to which males and females follow different growth equations seems very probable (paragraph 6.1.2.) and could explain the difficulty of following the modal progressions for large individuals. Finally, the probable high individual variability in growth of these individuals and the progressive admixture of fish of different origins (cohorts, spawning areas…) renders the modal decomposition of size frequencies impossible and complicates enormously the analysis of their growth. The extrapolation of diverse growth curves (figure 6.17) for large yellowfin is therefore extremely hazardous, but remains the only means available for estimating the average growth of these individuals.

Conclusion and yellowfin age-size-weight table

The most recent analyses of yellowfin growth, regardless of method, seem to indicate that the growth of young individuals (less than 65 to 70 cm) occurs at a relatively slow rate on the order of 1.4 to 1.6 cm per month. The reasons for this slower growth are not yet known; the hypothesis that it is linked to the relative poverty of the Gulf of Guinea waters, where the young yellowfin are found. has been advanced (Fonteneau, 1980). Even the actual importance of the fraction of individuals that pass this phase of slowed growth can be questioned: do all eastern tropical Atlantic yellowfin follow this growth mode or rather only those born at the beginning of the year in the Gulf of Guinea? What importance do yellowfin born outside the Gulf of Guinea play in the exploitation of this species in the east Atlantic and which growth pattern do they follow?

The growth rate (from 2.8 to 3.3 cm/mo) and the growth parameters for average size yellowfin (70 to 140 cm) seems well enough defined if one believes the convergence of results calculated by different authors using different methods; the differences that can appear would be mainly due to methods (sampling, calculation, expression…) used to determine growth. However the relative growth acceleration of these individuals at around 65 cm is not explained, even given that this size corresponds roughly to the size at puberty.

Growth of large yellowfin over 150 cm and of juveniles from the larval stage up to 35 cm, as well as spatio-temporal and probable sexual variability in growth of all species are subjects that remain to be studied and may have important consequences for fisheries management.

Table 6.6 Table of correspondence between age, length (fork length, FL) and weight of yellowfin in the eastern tropical Atlantic. This table was calculated from a composite growth curve: First stanza: FL < 65 cm (age 1.7 years), linear growth of 17.7 cm/yr (Bard, 1983) Second stanza: FL > 65 cm, Von Bertalanffy growth equation with K (annual) = 0.42, L∞ = 194.8 mm (from Le Guen and Sakagawa, 1973) with the addition of t₀ = 0.967 yr. The length-weight relation used is that of Caverivière (1976), cf $ 6.3.

Despite these uncertainties, the description of yellowfin growth recently adopted by the ICCAT (ICCAT, 1984) is divided into two parts:

• For yellowfin less than 65 cm (age under 1.7 years) the growth equation is the one described by Bard (1984); that is linear growth with a rate of 177 mm/yr.

• For yellowfin more than 65 cm (corresponding age over 1.7 years), the growth equation adopted is the one proposed by Le Guen and Sakagawa (1973) with parameters: L∞ = 194.8 mm and k(annual) = 0.42 and the addition of the parameter t₀ = 0.967 years to connect this growth with that of yellowfin less than 65 cm.

The resulting length-age key (table 6.6) is tabulated in relative age from an age 0 which corresponds to that of yellowfin entering the fishery at 35 cm. This correspondence table (table 6.6) can at present be considered as the best for eastern tropical Atlantic yellowfin.

6.1.3.3.2. Skipjack

Direct reading of age

Skipjack scales appear unsuitable for age determination (Shabotiniets, 1968); it is mainly vertebrae, the first dorsal fin and otolith sections that are used for reading ages.

Due to the reduced dimensions of skipjack otoliths and the complexity of their preparation for age reading (Wild and Foreman, 1980), there were no observations of otoliths made on Atlantic skipjack. There are several works on age determination and growth from readings of thin sections (400 microns) of the first dorsal fin of Western Atlantic skipjack (Batts, 1972; Carles-Martin, 1975) and Eastern Atlantic skipjack (Cayré, 1979; Antoine et al., 1982, 1983d; Antoine and Mendoza, 1986). In regards to the Eastern Atlantic, the preliminary results of Cayré (1979) obtained for skipjack from 40 to 60 cm indicate a growth rate of 8.1 cm/yr and the following size-age key:

These age recordings, based on the non validated hypothesis of the appearance of one growth check per year, were taken up on a large scale during the International Program of Skipjack Research (Antoine et al., 1982 and 1983; Antoine and Mendoza, 1986). The results of this work, that involved many readings and observations per section, are very different from those obtained previously in the Atlantic (Batts, 1972; Carles Martin, 1975; Cayré, 1979) and indicate a low growth rate of 5 cm/yr. The length-age key resulting from these observations and the use of back calculation (table 6.7), suggests the existence of different growth rates from one zone to the other, and notably that the growth was more rapid in the north-east tropical region (Senegal) than in the Gulf of Guinea.

Table 6.7 Age-length (fork length, FL) of skipjack obtained by reading transverse sections of the first dorsal fin ray for two regions of the eastern Atlantic by two methods (after Antoine, Cayré and Mendoza, 1982).

Nevertheless, the authors, although their work was conducted rigorously on an unprecedented scale, point out the difficulty of these readings and the subjectivity of their interpretation. Further, no periodicity in the formation of growth checks, common to all samples or individuals, could be demonstrated despite the use of tetracycline tagging (paragraph 6.1.3.3.2.2.). The periodicity of the appearance of growth checks probably have heterogeneous and multiple causes which makes the determination of skipjack age from sections of the first dorsal fin arbitrary and not very reliable.

The Petersen Method

The application of this method of determining skipjack growth has been attempted twice on large samples from the Eastern Atlantic (Bour, 1976; Cayré et al., 1986). In addition to the biases or errors inherent in the method itself (absence of certain age classes in the samples, subjectivity of the choices of the modal progressions) Cayré, Diouf and Fonteneau (1986) specify that they could not determine skipjack growth by this method because of the observed stability of the modes (often a single mode). This stability may be explained in this case by the combination of different phenomenon:

• seasonal and geographic variability of growth (Bard and Antoine, 1986; Cayré et al., 1986).

• opportunist mode of reproduction of the species, permanent spawning without either well defined period or zone (Cayré and Farrugio, 1986).

• more or less permanent recruitment of skipjack into the fisheries.

• sudden and frequent migrations (immigrations and emigrations) in the fishing zones.

Analogous conclusions have been reached (Josse et al., 1979) in the Pacific based on an analysis of a large quantity of data collected over more than 20 years; these authors point out more that the apparent progression of modal sizes can lead to rapid, slow, or zero growth, depending on regions and years studied and that the very subjective aspect of the method in its application to skipjack renders results, when they exist, extremely doubtful.

Tagging

Until recently (1983), tagging and recapture of skipjack in the Atlantic Ocean were numerically insufficient for a serious growth analysis. Thus in 1976, the growth estimation of Atlantic skipjack was based on only 12 recaptures (ISRA-ORSTOM, 1976); the results were reduced to the simple estimate of the mean growth rate of 11.5 cm/year.

Since this date, numerous skipjack (nearly 30,000) have been tagged in the Eastern Atlantic during the “International Atlantic Skipjack Research Program” coordinated by the ICCAT. Several growth analyses from data on tagged and recaptured fish have been made (Bard et al., 1983; Bard and Antoine, 1986; Cayré et al., 1986).

Growth rates

One of the first results of these analyses has been to make it clear that there is variability of growth rate for skipjack linked to tagging and recapture zone:

• Growth rates observed by Cayré et al., (1986) on skipjack 40 to 55 cm in length, tagged in the north tropical zone (Senegal - Cape Verde) vary from 13 to 20 cm/yr depending on tagging period and time at liberty; globally, the growth rates observed for the 177 tagged skipjack recaptured more than one month after tagging is 18.9 cm/yr in the north tropical zone.

• The growth rate of skipjack from the Gulf of Guinea, calculated from different growth equations proposed by Bard and Antoine (1986) varies, depending on the parameters and time at liberty actually used, from 7.1 to 9.8 cm/yr.

In comparing the growth rates observed on tagged skipjack in the north tropical zone at different times, Cayré et al., 1986, were able to show that the high average growth rate observed for this zone (18.9 cm/yr) was linked to a seasonal component of the environment. The growth rates observed on tagged skipjack in this zone in the beginning of the fishing season (June) and at the end of the fishing season (October) are respectively 20 cm/yr and 15 cm/yr. The difference between these two values indicates a strong seasonal variability in growth which is maximal during summer at the moment where ecological conditions are more favorable and when individuals seem to stay in this region.

In conclusion, the growth rates of skipjack during their period of seasonal presence (3rd and 4th quarter) in the north-east tropical zone (18.1 cm/yr) appeared generally much faster than that observed in the equatorial zone (8.3 cm/yr).

After taking into account this seasonal growth acceleration, and although no data are available for the south-east tropical zone or for the eastern Atlantic, an average growth rate of 12 cm/yr is currently adopted for Atlantic skipjack.

A very high variability of average annual growth rates (9 to 32 cm/yr) appears also in works carried out on Pacific skipjack. As in the Atlantic, these growth rates seem to depend on the study zones and probably also reflect the strong spatio-temporal variability of the growth of the species.

Parameters and growth curve

Given the existence of a geographical and seasonal growth variability, the parameters K and L∞ of the Von Bertalanffy growth equation, have been estimated separately for the equatorial zone between 5°N and 5°S (Bard and Antoine, 1986) and for the north-east tropical zone 10°N to 2°N, from the coast to 20°W (Cayré et al., 1986b). For each one of these zones, skipjack recaptured after less than one month at liberty, have been eliminated from calculations in order to avoid an eventual bias linked to trauma that may be caused by the tagging operation.

- Equatorial zone (Bard and Antoine, 1986)

The few individuals (n = 28) at liberty more than one year, and therefore of large size, have been deliberately excluded from calculations by these authors because of modifications, assumed by them, that occur in the behavior (emigration) and the physiology of these large skipjack. Estimation of the parameters K and L∞ have therefore been made from 369 recaptures corresponding to time at liberty between 30 and 365 days: L∞ = 741.5 mm; K(annual) = 0.3758; t₀ = 0.

Figure 6.19 Growth curves and parameters for skipjack calculated from tagging data and corresponding to the tropical northeast and equatorial east zones of the Atlantic (after Cayré, 1985).

The authors, deciding that a value of L∞ equal to 80 cm permitted a better schematisation of the growth (this value corresponds to the larger skipjack captured), have calculated the value of K corresponding: L∞ = 80.0 cm; K (annual) = 0.32; t₀ = 0.

The growth curve corresponding to these parameters, (figure 6.19) is the one actually used in different skipjack population dynamics models, after however arbitrarily fixing the size at age 1, either at the size of the smallest skipjack entering the fishery (around 35 cm, Bard and Antoine, 1986), or at the probable size of the species at age 1 year (38 cm; Cayré, 1985).

- North-east tropical zone (Cayré et al., 1986b)

The parameters of the Von Bertalanffy equation have been calculated (according to the same method as the one used in the east equatorial zone) from recapture data of 170 skipjack having time at liberty between 30 and 200 days: L∞ = 620.0 mm; K (annual) = 2.0805; t₀ = 0

The growth curve corresponding to these parameters (figure 6.19) gives the impression of, as we have seen previously, strong regional and seasonal variability in skipjack growth.

Age-length-weight key of skipjack

Taking into account the fact that tagging is at this time the most dependable technique for determining the growth equation of skipjack, that, from direct age reading, a skipjack of 1 year measures probably less than 40 centimeters, and finally that skipjack frequent only seasonally the tropical region, a length-age key (table 6.8) based on the growth parameters of Bard and Antoine (1986) for the equatorial region and adopting a size of 38 cm at the age of 1 year, has been established (Cayré, 1985).

Longevity and maximum size

The largest skipjack caught in the Atlantic occasionally reach lengths between 90 cm and 1 meter; however such individuals are practically never caught in the study zone; the weight that corresponds to these sizes is between 17 and 25 kg. In spite of all the uncertainties of age determination of this species, its maximum longevity can be estimated to be around 10 years (table 6.5).

Table 6.8 Table of correspondence between age, length (fork length) and weight of skipjack in the eastern tropical Atlantic. (from Cayré, 1985).
- growth calculated from by the Von Bertalanffy equation (Bard and Antoine, 1986) with: K (annual) = 0.332, L∞ = 80.0 cm
- weight calculated by the length-weight relationship of Cayré and Laloë (1986).

Table 6.9 Table of correspondence between age, length (fork length: FL) and weight of bigeye in the eastern tropical Atlantic. This table was established from the parameters of Von Bertalanffy growth curve calculated by Cayré and Diouf (1984): K (annual) = 0.1127, L∞ = 285.37 cm and the addition of t₀ = -0.5 yr.
The length-weight relation used is that of Parks et al. (1982), cf $ 6.3

Conclusion

Of the different techniques used to determine growth, tagging remains the currently most reliable. What is known now on Atlantic skipjack growth remains, in spite of everything, quite fragmentary. Generalization of growth observed in only one part of the spatio-temporal distribution of the species (east equatorial zone) and mainly from the size range of individuals most frequent in the catches (35–60 cm), is a source of serious potential bias, if significant fractions of the population follow very different growth patterns during a significant period of their exploitation (Cayré, 1985). The demonstration of a seasonal variability in growth should lead to a supplementary analysis in different locations and periods of abundance of the species.

6.1.3.3.3. Bigeye

Direct age reading

Age determination of Atlantic bigeye from scale reading is recognized as impossible (Gaikov et al., 1980). Only the first dorsal fin has been used (Gaikov et al., 1980); Draganick and Pelczarski, 1984).

The reading of transverse sections of the first dorsal fin of bigeye caught by longline has permitted Gaikov et al., (1980) to establish a length-age key (table 6.9) calculated from the following Von Bertalanffy growth parameters: L∞ = 253.75 cm; K(annual) = 0.173; t₀= -0.15 years

The size range of individuals sampled by Gaikov et al., was from 30 to 200 cm, but neither the number nor the size frequency distribution of individuals used for age estimated are given. Further, the hypothesis used by these authors to determine age and according to which there were two growth checks per year is not verified or validated. This hypothesis is all the less verifiable since, as the authors point out, their sample has only been collected at a single period of the year.

A more recent study (Draganick and Pelczarski, 1984) in accordance with the same hypothesis and from a sample of bigeye of sizes mainly between 110 and 165 cm in length, results in the following growth parameters in size and weight:

L∞ = 281.8 cm; K(annual) = 0.23; t₀= -0.02 yr.

W∞ = 206.4 kg; K(annual) = 0.24; t₀= -0.03 yr.

The length-age key calculated from these parameters (table 6.9) is similar to that proposed by Gaikov et al., (1980), but here again the periodicity of the appearance of growth checks in the dorsal fins is not validated.

It is necessary, in addition to the major obstacle represented by the absence of validation of the periodicity of growth checks, to keep in mind the often subjective character of age reading, a phenomenon well demonstrated for skipjack. Also, experience has shown that the larger the individual, the more difficult or impossible the sections of the first fin become to read because of the calcareous deposition that occurs in the centre of this fin render this very important central part unreadable.

Modal progressions (Petersen method)

The first estimation of bigeye growth by the Petersen method (Champagnat and Pianet, 1974), was based on size-frequency samples of individuals caught by eastern Atlantic surface fisheries (pole and line boats and purse seiners) from the Congo to Senegal. The Von Bertalanffy growth equation parameters determinated in this way are: L∞ = 338.53 cm; K(annual) = 0.104,097; t₀ = -0.5425 yrs. This growth equation and resulting length-age key (table 6.9), is applied theoretically to a size range of bigeye from 60 to 140 cm in fork length.

Subsequently Marcille et al., (1978), have extended this analysis based on measurements of predorsal length (from the extremity of the head to the base of the first dorsal fin) of bigeye caught by the FIS surface tuna fleet (France, Côte d'Ivoire, Senegal) from 1969 to 1977. This sample, larger than that used by Champagnat and Pianet (1974), permits them to extend the region of strict application of the growth equation that they propose to a size range of bigeye between 45 and 150 cm in fork length. After the measurements of predorsal length had been converted to fork length by the relation established by Champagnat and Pianet (1974), the age-length key (table 6.9) can be calculated from the growth equation parameters: L∞= 259.6 cm; K(annual) = 0.1488; t₀ = -0.3983 yrs. This key is very similar to the one proposed by Pianet and Champagnat (1974).

The same method applied later to more larger samples, including measurements of bigeye caught by longline (Weber, 1980); Pereira, 1984), has permitted the calculation of age-size relations (table 6.9) similar to those previously established. The parameters calculated by Weber (1980) on a sample including individuals from 40 to 190 cm long are: L∞ = 491.6 cm; K(annual) = 0.054; t₀ = -0.952 yrs.

The very high value of L∞, comes from the fact that large fish (> 150cm) are quite poorly represented in the sample, as well as from the subjectivity of modal progressions for large bigeye. Pereira (1984), who has updated these measurements to 1982, emphasizes the irregular representation of large individuals in the samples and the difficulty to localize the modes in the size frequencies for individuals over 150 cm. The length-age key deduced from Pereira's growth parameters (L∞ = 381.47 cm; K(annual) = 0.08508503) and addition to these parameters of a value of t₀ equal to -0.4 yrs, is therefore very similar to the previous ones (table 6.9) for bigeye from 40 to 150 cm in fork length.

Figure 6.20 Growth curves for bigeye in the eastern tropical Atlantic determined by different authors and different methods (see text).

Tagging

An analysis of bigeye growth has been done by Cayré and Diouf, 1984 from bigeye tagging in the eastern Atlantic by the Côte d'Ivoire, France, Japan and Senegal. The 130 recaptures used by these authors have permitted them to show that, contrary to yellowfin, young bigeye (length under 60 cm) do not seem to have slow growth. The age-length key (table 6.9) calculated from growth parameters determined by Cayré and Diouf: L∞ = 285.37 cm; K(annual) = 0.1127, with the addition of the parameter t₀ = -0.5 yrs., is only applied theoretically to individuals where the size is between 38 and 110 cm. These results are similar to those obtained by other methods in a size range between 40 and 150 cm. Also, in spite of the relatively conservative number of recaptures, the analysis by Cayré and Diouf (1984) shows that there does not seem to be a difference in growth between bigeye from north and south tropical regions of the eastern Atlantic.

Longevity and maximum size

The maximum size that bigeye can reach is between 2 meters and 2.5 meters but individuals over 180 cm are very rarely caught in the fisheries; the maximum longevity of the species, in spite of uncertainties of growth of large individuals, would be in the order of around 15 years (table 6.5).

Conclusion and age-size-weight table

Although all of the study methods on growth, regardless of reliability, result in similar growth curves (figure 6.20), all stumble on the problem of growth of individuals of more than 150 cm in fork length. The reasons for this limitation of the study methods are linked, jointly or not, to the method itself (direct recording), to the poor representation in the samples of large individuals (Petersen method, tagging), and finally even to the biology of the species that can introduce major variability in individual growth (spatio-temporal variability of growth, migration…) and render otherwise proven methods (modal progressions) inoperable. As for yellowfin, the hypothesis of different growth in males and females from a size around 140 cm seems probable and could also explain the difficulty in following the modal progressions for bigeye of this size.

Table 6.10 Age-length relation of bigeye in the Atlantic calculated from the parameters proposed by different authors and determined by various methods. The domain of strict theoretic application of these different age-length relations is indicated by figures marked by an asterisk (*).

The extrapolation of growth curves outside of their strict limits of application and the imprecision of various methods (with the exclusion of tagging) that the conclusion that growth of young individuals (less than 40 cm) follows the same growth equation as large bigeye (more than 150 cm) remain hypothetical. The best key relating age, size and weight (table 6.10), is, for the moment, the one derived from the growth equation established from tagging data (Cayré and Diouf, 1984). However, tagging of large bigeye (FL> 100 cm) by a method that remains to be found and the utilization of vital markers (tetracycline), prerequisite to any new tests of direct age reading, seem at this time, the steps to take in order to remove the remaining uncertainties in the knowledge of bigeye growth.