|
|
EFAN Report 4-2000 |
Guidelines for validation studies
By
Beatriz
Morales-Nin
CSIC/UIB-IMEDEA
Miguel
Marqués 21
07190
Esporles, Spain
|
European Fish Ageing Network (EFAN) |
|
Phone: (47) 37 05 90 00; Fax: (47) 37 05 90 01;
Email: bente.lundin@imr.no Office address: Flødevigvn. 49, Hisøy (Arendal),
Norway. Coordinator: Erlend Moksness, Phone (direct): (47)
37 05 90 41; E-mail: moksness@imr.no |
Introduction
The
aim of EFAN is to ensure that age determination becomes a reliable element of
the assessments underlying the scientific management advice on fisheries and
environmental resources. Thus, the determination of the accuracy of the ages is
of paramount importance for EFAN. This document reflects the guidelines
elaborated by the Cell and also includes the results of a workshop held on 1999
with the participation of Cell 5 members.
According to Beamish et al. (1983) age
validation is a process of estimating the accuracy of an age estimation method.
Validation of an age procedure indicates that the method is sound and based on
fact (Kalish 1995).
The term
validation has two meanings; in a more restricted onenarrower sense the term is used,
that is to determine the temporal meaning of the growth increment
used in ageing and in
a wider one sense, that the term is usedis
to prove that the whole
age determination procedure is accurate.
FFollowing
Francis (1995) the definition of accuracyaccuracy is a matter of degree, which
measures how close an estimated age is likely to be to the true age. This wider sense certainly should be preferred,
because each age
reader will reach a different level of
accuracy and especially
because an
individual age reader will reach different levels of accuracy for each age group (older specimen
are more difficult to age). Therefore, for each age group the accuracy should be
measured from
age readings of individual age readers
by estimating
how close the estimated ages are to the true ages.
Thus, to interpret age we must have a standard that
should be meet. For
example, aAn ageing procedure might be considered
accurate when
the majority of
the true age group will be assigned to the correct year class yr. It should be taken
into account that the estimated relative bias (difference between estimated age and
e.g. modal age) does not indicate accuracy, but only the agreement to the age reading method of the group of age
readers (all age readers could apply a completely wrong age reading method).
The standard would depend of the degree of accuracy
acceptable for a particular ageing pbocedure and for the use to which the
estimated ages will be put.
Three levels of validation are possible (Francis 1995): 1)
The first n increments are annual, but there are insufficient data to determine
the periodicity of the latter increments and to make a quantitative estimate of
the accuracy. 2) All the increments are effectively annual but there are insufficient
data to make a quantitative estimate of the accuracy. 3) The increments are
effectively annual and quantitative estimate of the accuracy is provided.
A revision of the validation papers carried out as part of
the Cell activities has underlined the lack of complete validated ageing for
most species, only in some fresh water studies the degree of validation reached
the highest level.
The
validation methods can be roughly divided into indirect and direct ones,
depending if they give support to the growth rates determined in the population
(like modal length progression analysis) or they assess the increment
periodicity of individual fish.
Indirect validation methods
Indirect
validations are verification methods that support the growth rates determined
by the age reading. They cannot be considered as validation in the sensus of
Francis (1995), but frequently they are the only available methodology to
support the age determinations.
Length based methods
The
length-based methods might be of two kinds, like following a single cohort along
time or analyse the length composition of the landings to identify
the different
age classesgroups present, based on the assumption that
each age group has a normally distributed length distribution and have a
differential modal length. Petersen (1891) was the first to identify the modes
on the length composition as corresponding to age groups. The age class 0
correspond to the smallest mode present in the sample obtained after the
spawning period, the following modes correspond to age 1 etc.
Since
then many methods have been developed to identify the age classes in a length
composition. There are: a) Graphical methods in which the groups of points
obtained by mathematical transformation of the length frequencies, corresponding
to an age class are identified (Cassie 1954, Bhattacharya 1967). b)
Computational methods, using the maximum likelihood (Hasselblad 1966), with
previous information of the age classes present and their mean length
(Macdonald and Pitcher 1979) or incorporating biological information (Schnute
& Fournier 1980) (inter alt.).
If
there are age length keys associated with the original length frequencies, these could be
used to compare predicted length at age distributions against observed
distributions. These predicted length at age distributions can then be compared
with the observed distributions using a chi-square test.
Some
species have a relatively long spawning period, high overlapping of lengths
between age groups and growth variability, which increases with age. Thus,
statistical methods for the disaggregationdesegregation of age
groups by length frequencies often are only reliable for juvenile fish.
Another
approach is to calculate the growth parameters from the length frequencies
(Pauly & David 1981) and compare them with the growth parameters obtained
from the age-length keys.
Age based methods
There is a strong
indication that the age reading method is accurate, if exceptional good or weak year classes can be followed in age compositions over a long time series. The strength or weakness of a year class would
be lost rapidly, if the age reading method would not be accurate, because of
assignment of ages to wrong year classes (Eltink & Kuiter 1999). However, reader bias, which is defined as the subjective assignment of ages caused by the
knowledge of existing strong or weak year classes, is likely to have an effect on the age reading results. This age reader bias
can be excluded by age reading a set of calcified structures of this exceptional strong
year class collected in the different years (approximately an equal number in each year collected). Analysis of these age
readings determine whether such an otolith set of an extremely strong year
class can be used in otolith reading workshops to estimate the precision and the accuracy and the relative/absolute bias in age reading (an example for horse
mackerel is presented in ICES 1999) and to estimate the effect of age reading errors on the
assessment (Addendum of ICES 1999).
Marginal otolith structure development
The
evolution over time of the otolith marginal structure shows the period of
increment formation. The most commonly used validation method is to determine
the percentage of translucent and opaque increments in the edge with monthly periodicity.
As sexual dimorphism has been reported (Morales-Nin et al. 1998) and
differences depending on maturation could be found (Beckman & Wilson 1995),
it is recommended to consider these factors when using these methods.
The
marginal increment evolution can be quantified measuring the marginal increment
in formation against the previous increment laid down in the otolith. Samame’s
(1977) index could be used:
I=(R-rn)(
rn- rn-1)-1
Where:
I marginal increment index, R otolith radius, rn n increment.
One
of the objectives of EC Study Contract 98/096 (Distribution and Biology of
anglerfish and megrim in the waters to the west of Scotland) is to indirectly
validate the age the ‘northern stock’ of the anglerfish (Lophius piscatorius). This is being achieved by verification of the
first annulus from primary (presumed daily) otolith increments and marginal
increment analysis.
Backcalculation methods and chronology of annual
growth zones
Back
calculation can be considered a complementary method to corroborate the age
estimated. The relation between linear size of the structure and that of the
body, which is usually determined empirically, can provide specific and useful
information concerning individual growth. The procedure of back calculation of
annuli and growth zones in calcified structures has been used to reconstruct
the previous size at age. The technique needs a large sample representing a
broad range of sizes, including fish almost exactly one year old and annuli
must be located very precisely and correctly. This technique includes various
procedures, assumptions and limitations (Carlander 1981, Francis 1995).
In
several fish species, especially those that live in waters where year-to-year
variation in thermal conditions is considerable, temperature affects growth.
The variation in growth, partly related to thermal conditions, is observed in
the annual growth zones of bones, scales, and to a lesser extent also in
otoliths. Exceptionally warm or cold growing seasons in the past are often
identifiable in the calcified structures of long-lived fish species. Thus, if
an exceptionally wide (or thin) annual zone in a calcified structure is located
as many rings back from the catching time as an exceptional warm (or cold) year
has been in years before the catching, this growth zone, possibly with some
other exceptional growth zones, support the assessment of age.
Chemical methods
Three
major assumptions underpin the application of otolith microchemistry to age
validation, that:
1.
Otolith composition
reflects seasonal temperature variations
2.
Temperature
variations correspond to visible features
3.
Variations
in composition validate the seasonal frequency of visible features
Elemental
composition
The concentrations of the main
microconstitutents of the otoliths, as Ca, Sr, Na, K, Fe, Mn and Mg can be
measured in the otoliths using a Wavelength Dispersive Spectrometer (WDS or
electron microprobe) to validate the seasonality of visible opaque and
translucent (hyaline) bands used for age estimation. Elemental ratios in the otolith calcium carbonate, especially the
Sr/Ca ratio, have been hypothesised to vary in response to seasonal changes in
water temperature.
In
a recent study in cod and hake (DGXIV Study Project 96-075) it was shown that
the opaque and translucent zones were generally different in composition during
the early stages of development although the variation declined toward the edge
of the otolith. Sr/Ca ratios were generally higher in translucent zones than in
opaque zones. Na/Ca ratios were
inversely related to Sr/Ca ratios. The decreasing variations in Sr/Ca ratios
between translucent and opaque bands towards the otolith edge could be a result
of either the decreasing width of the bands or an ontogenetic effect. Because
there was such a close coupling of the visual pattern of otolith zone formation
and the chemical composition, it seemed unlikely that simple cyclic seasonal
temperature fluctuations were responsible for all of the variation in the Sr/Ca
signal. Therefore, it was not possible to use the Sr/Ca variations to validate
which zones corresponded to annual otolith increments in many of the hake
otoliths. There is increasing evidence in the literature that the Sr/Ca ratio
in fish otoliths responds only indirectly to ambient temperatures. The elemental
ratio may be more of a reflection of physiological processes, and these may
simultaneously induce visual changes in the otolith. Thus, the Sr/Ca ratio is not independent of otolith growth and
cannot be used as an independent validation tool.for this species.
Isotopes
Isotopes have been used to validate the age
estimates of fishes. They have mostly been used for long-lived species where
the intention has been to ‘validate’ either the younger ages estimated from
whole otoliths or older ages estimated from sectioned otoliths.
The
literature n the use of radiometric techniques for age validation or age
estimation in fish was reviewed for EFAN Cell 5 ( Gordon, EFAN Report 1/98;
EFAN Home page: 30 June 1998). 210Pb/226Ra disequilibria
in otoliths has been the most useful technique because it is suitable for ages
of up to about 100 years. 228Th/228Ra disequilibria has also been used but is only
useful for fish up to about 10 years. The requirement for relatively large
quantities of material for analysis (c. 1 g) necessitates the pooling of
otoliths or parts of otoliths. Whole otoliths are less useful than otolith
cores because of the need to model otolith mass growth rates. Non-proportional
uptake of parent or daughter isotopes can invalidate the results. The many
assumptions associated with the method means that the radiometric technique is
best used to validate age estimates obtained by other methods. Bomb radiocarbon
can also be used to validate age estimates of long-lived species or of species
where otoliths have been archived.
The
210Pb/226Ra disequilibria method has been the most widely
used technique and there has been considerable debate over the validity of the
results. An important assumption is that there are no losses or gains of 226Ra
or 210Pb after uptake, other than by radioactive decay or in growth.
One of the decay products is 222Rn and it has been argued by Gauldie
(EFAN web site; Cell 5: 30 June 1998) that this can escape from the otoliths
and hence invalidate the results. Other
authors argue that otoliths are similar to aragonitic corals where no losses of
222 Rn have been reported. Some recent publications relevant to the
debate are Gauldie (1998), Gauldie and Cremer (1998), Gauldie et al. (1998) and
Tracey and Horn (1999).
A
new method of measuring 226Ra using ion exchange and thermal
ionisation mass spectrometry has been described by Andrews et al. (1999). The
technique has been applied to validate the longevity of the Pacific grenadier (Coryphaenoides acrolepis) (Andrews et
al. 1999).
In
a similar way to the Sr/Ca ratios, the stable oxygen isotope (d 18O) signals show a seasonal variation
coincident with the otolith’s visible growth increments, generally supporting
increment-count age estimates (Weidman & Millner 2000).
Direct methods
Direct validation takes into account a precise
temporal reference mark on a calcified part relatively to the other growth
marks, it is an individual based technique often based on observations of
growth in captivity or in mark-recapture experiments.
Tagging and
release
When the
fish is measured and externally tagged before release and again at recapture, the length increase by time
is determined, which makes the determination of the growth rate possible
(Kirkwood 1983). When
the marked fish is released at the age of 0+, the number of annual marks should
be identical to the age in years at recapture. The marking method may not
affect the growth of the marked fish.
Several marking techniques are
available, from the insertion of external plastic coded tags, to coded visible
implants under the skin, tattoo-ink and cold and hot-branding. These techniques
have to be applied depending of he fish size and are not suitable for larvae.
The direct
methods are based on observations of growth in captivity or in mark-recapture
experiments. The mark can be only of fish which is measured and at recapture
the length increase by time is determined. This allows to determine growth
rates but if the experiment includes otolith marking, it allows to validate the
periodicity of the otolith increments.
Calcified structures marking
If the tagging
experiment includes calcified
structure marking
by the incorporation of a chemical marker, it allows validating
the periodicity of the otolith increments. Otoliths as well as other
calcified structures can be marked in reared or wild fish using different
markers that are incorporated into the otolith composition. Various forms of
tetracycline (Beamish et al. 1983), alizarin (Villanueva&Molí 1997),
calcein (Thomas et al. 1995) and strontium (Clear et al. 2000) (either injected
or dissolved in seawater for ambient exposure), have been used successfully for
producing a mark on the otolith. The three first markers appear as a
fluorescent mark under ultraviolet light. Strontium has to be detected through
elemental analysis. Another way of marking otoliths is through thermal shock
producing distinct increment patterns (Volk et al. 1995).
The
form of marking depends of the fish size, larvae and juveniles are usually
marked by immersion in dilutions of the marker, while bigger fish are generally
marked by intraperitoneal injection. The oral administration of a marker
incorporated to the food allows marking large numbers of fish, fluorescent
markers are successfully provided in this way (Nordeide et al. 1992).
Multiple
marking can be achieved and used as group marks on sub-populations with
different characteristics (Tsukamoto 1989). Due to the different otolith growth
rates along the life span, the moment of marking has to be selected in suitable
periods. In multiple marking the interval between marks has to be sufficiently
long to avoid merging of two separate marks.
Because
the time elapsed between marking and release and capture is known, the marginal
structures formed after the mark can be compared with the time elapsed and
their periodicity of formation is determined.
Otolith microstructure
There is now considerable
evidence that microstructural observations can assist in the interpretation of
otolith macrostructure (see Arneri et al., EFAN report 1/98 for review). Counts
of microscopic daily increments have been used to directly verify that one
opaque and translucent zone represents an annulus (Victor & Brothers 1982;
Taubert & Tranquilli 1982). However, this direct verification approach is
limited by the problem in detecting fine microscopic increments in fish > 1
year old and the need for validation of the daily periodicity of increment
structures. Due to these problems daily increments have most usually been used
to validate just the first annulus. However, changes in microstructure may also
provide a means of distinguishing between seasonal zonations and secondary
structures. In temperate species daily deposition can be very compressed or
arrested in cold periods (Taubert & Coble 1977; Campana & Neilson 1985)
and using light microscopy an annulus often appears as a discontinuity preceded
by increasingly narrow increments (Victor & Brothers 1982). Seasonal
differences in increment width may also be associated with changes in micro-crystal
thickness and compaction (Morales-Nin 1987). As such there may be
microstructural differences between annuli and secondary structures. However,
the growth discontinuities and the thin increments in mature fish, limit the
application of the method in fish older than 1 year.
References
Andrews, A.H., Coale, K.H., Ngwicki,
J.L, Lundstrom, C., Palacz, Z., Burton, E.J. and Cailliet, G.M. (1999)
Application of an ion-exchange separation technique and thermal ionization mass
spectrometry to 226Ra determination in otoliths for radiometric age
determination of long lived fishes. Canadian Journal of Fisheries and Aquatic
Science, 56, 1329–1338.
Andrews, A.H., Cailliet, G.M. and
Coale, K.H. (1999) Age and growth of the Pacific grenadier (Coryphaenoides acrolepis) with age
estimate validation using an improved radiometric ageing technique. Canadian
Journal of Fisheries and Aquatic Science, 56, 1329–1338.
Arneri, E., H. Mosegaard, P.J. Wright and B. Morales-Nin, (1998).
Microstructural validation of annual increments. In: The present status of
otolith research and applications, ed. P.J. Wright. EFAN report 1/98.
Bhattacharya, C.G. (1967). A simple
method of resolution of a distribution into Gaussian components. Biometrics 23:
115-135
Beamish, R.J., McFarlane, G.A.,
Chilton, D.E. (1983). Use of oxytetracycline and other methods to validate a
method of age determination for sablefish (Anoplopoma
fimbria). In Proceedings of the International Sablefish Symposium.
Anchorage, Alaska, Alaska Sea Grant Report 83-3: 95-116.
Beckman, D.W., Wilson,C.A. (1995).
Seasonal timing of opaque zone formation in fish otoliths. Recent developments
in fish otolith research. D.H.Secor,J.M.Dean,S.E.Campana (eds)University of
South Carolina Press, 27-44.
Cassie, R.M. (1954). Some use of probability
paper in the analysis of size frequency distributions. Australian
Jour.Mar.Fresh.Res. 5: 513-522.
Clear,N.P., Gunn, J.S., Rees, A.J.
(2000). Direct validation of annual increments in the otoliths of juvenile
southern bluefin tuna, Thunnus maccoyii,
by means of a large-scale mark-recapture experiment with strontium clhoride.
Fisheries Bulletin 98: 25-40.
Eltink, A. and C.J. Kuiter, 1989. Validation of ageing
techniques on otoliths of horse mackerel (Trachurus trachurus L.). ICES CM
1989/H:43, 15pp.
Francis,R.I.C.C. (1995) The analysis
of otolith data-A mathematician’s perspective. Recent developments in fish
otolith research. D.H.Secor,J.M.Dean,S.E.Campana (eds)University of South
Carolina Press, 81-96.
Gauldie, R.W (1998) Winding back the
clock on orange roughy ages: what does it do for allowable catch? Seafood New
Zealand, August 1998, 42-43.
Gauldie, R.W. and Cremer, M.D. (1998)
Loss of 222Rn from otoliths of orange roughy, Hoplostethus atlanticus, invalidates old ages. Fisheries Science,
64, 543-546.
Gauldie, R.W., Thacker, C.E., West,
I.F. and Wang, L. (1998) Movement of water in fish otoliths. Comparative
Biochemistry and Physiology, Part A, 120, 551-556.
Gordon,
J.D.M. (1998) Radiometric Ageing. In: The present status of otolith research
and applications. Ed. P.J. Wright. EFAN Report 1/98.
ICES, 1999. Report of the Horse Mackerel Otolith
Workshop, Lowestoft, UK, 15-19 January 1999. ICES CM 1999/G:16
Hasselblad V. (1966) Estimation of parameters for a mixture of normal
distributions. Technometrics 8(3): 434-444.
Kalish, J.M. (1995). Radiocarbon and
fish biology. Recent developments in fish otolith research.
D.H.Secor,J.M.Dean,S.E.Campana (eds)University of South Carolina Press,
637-654.
McDonald, P.D.M., Pitcher, T.J.(1979)
Age groups from size-frequency data: a versatile and efficient method of
analyzing distribution mixtures. Jour.Fish.Res. Bd.Canada 36: 987-1001.
Morales-Nin,
B., Torres,G.J., Lombarte,A., Recasens, L. (1998). Otolith growth and age estimation in the European
hake. Journal Fish Biology 53: 1155-1168.
Nordeide,J.T., Holm, J.C., Otteraa,
H., Blom, G. Borge, A. (1992) The use of oxytetracycline as a marker for
juvenile cod (Gadus morhua L.).
Journal Fish Biology 41, 21-30.
Pauly, D., David, N. (1981) ELEFAN I,
a BASIC program for the objetive extraction of growth parameters from
length-frequency data. Meeresforsh. 28:205-211.
Petersen, C.G. (1891) Eine methode
zur bestimmung des alters und wuchses der fishe. Mitteilungen.Deutscher
Seefisherei-Verein 11: 226-235.
Thomas, L.M., Holt,S.A:, Arnold,C.R.
(1995). Chemical marking techniques of larval and juvenile red drum (Scienops ocellatus) otoliths using
different fluorescent markers. Recent developments in fish otolith research.
D.H.Secor,J.M.Dean,S.E.Campana (eds)University of South Carolina Press,
703-717.
Tracey, D.M. and Horn, P.L. (1999)
Background and review of ageing of orange roughy (Hoplostethus atlanticus, Trachichthyidae) from New Zealand and
elsewhere. New Zealand Journal of Marine and Freshwater Research, 33, 67-86.
Tsukamoto K. (1989). Otolith daily increments in the Japanese eel. Nippon
Suisan Gakkaishi 55, 1017-1021.
Schnute,J., Fournier, D:(1980) A new
approach to length-frequency analysis:growth structure.
Can.Jour.Fish.Aquat.Sci. 37:1337-1351.
Victor, B.C and E.D. Bbothers,
(1982). Age and growth of the fallfish Semotilus
corporalis with daily otolith increments as a method of annulus
verification. Can. J. Zool., 60: 2543
Villanueva,
R., and Molí, B. 1997. Validation of the
otolith increment deposition ratio using alizaring marks in juveniles of the
sparid fishes, Diplodus vulgaris and D.puntazzo. Fisheries Research 30:
257-260.
Volk, E.C.,Mortensen,D.G.,
Wertheimer, A.C. (1995). Nondaily otolith increments and seasonal changes in
growth of pink salmon (Oncorhynchus gorbuscha)
otoliths. Recent developments in fish otolith research.
D.H.Secor,J.M.Dean,S.E.Campana (eds)University of South Carolina Press,
211-226.
Weidman,C.R., Millner, R. (2000).
High-resolution stable isotope records from North Atlantic cod. Fisheries Research
46: 327-342.
Wilson,C.A. (1995). Chemical marking
in otoliths. Recent developments in fish otolith research.
D.H.Secor,J.M.Dean,S.E.Campana (eds)University of South Carolina Press, 719.