THA:75:008:WP/80/19 SELECTION AND INBREEDING OF CULTIVATED MACROBRACHIUM ROSENBERGII IN THAILAND CONTENTS

by

Roger W. Doyle

Consultant to
the
Programme for the Expansion of Freshwater Prawn Farming
(FAO/UNDP/THA/75/008)

Bangpagong, Chacheongsao
Thailand
1980

Hyperlinks to non-FAO Internet sites do not imply any official endorsement of or responsibility for the opinions, ideas, data or products presented at these locations, or guarantee the validity of the information provided. The sole purpose of links to non-FAO sites is to indicate further information available on related topics.

This electronic document has been scanned using optical character recognition (OCR) software. FAO declines all responsibility for any discrepancies that may exist between the present document and its original printed version.

CONTENTS

1.1 SUMMARY

2.1 INTRODUCTION

3.1 TERMS OF REFERENCE

4.1 INBREEDING

4.2 Definitions

4.3 Relationship between inbreeding coefficient and inbreeding depression in Macrobrachium

4.4 Inbreeding in Thai stocks

5 SELECTION

5.1 Principles

5.2 Note on the numerical data

5.3 Note on the selection objectives

5.4 Definitions

(a) selection differential

(b) intensity of selection

(c) response to selection

(d) heritability

(e) fitness

(f) fitness function

6 SELECTION PLANS

6.1 PLAN A: controlled breeding

(a) summary

(b) flowchart

(c) rate of progress

(d) when to select

(e) requirements for implementation

6.2 PLAN B: artificial selection of females

(a) summary

(b) selection intensities

(c) rate of progress

(d) requirements for implementation

6.3 PLAN C: enhanced natural selection

(a) procedure

(b) rate of progress

6.4 PLAN D: current practices in Thailand

7 REFERENCES CITED

8 ACKNOWLEDGEMENTS

9 APPENDIX A: outlines of research and training program in Macrobrachium quantitative genetics

10 APPENDIX B: Itinerary

List of Tables

1. Inbreeding rate, sex ratio 1:1

2. Inbreeding rate, sex ratio 1:4

3. Some examples of inbreeding depression

4. Inbreeding in various Thai situations

5. Selection intensities and sample statistics, PLAN A (3 months)

6. Selection intensities and sample statistics, PLAN A (6 months)

7. Selection intensities and sample statistics, PLAN B (3 months)

8. Selection intensities and sample statistics, PLAN B (6 months)

9. Selection intensities and sample statistics, PLAN C (3 months)

10. Selection intensities and sample statistics, PLAN C (6 months)

List of Figures

1. Egg volume vs body length (female fitness function)

2. Weight-length relation, both sexes

3. Claw thickness - body length relationship, immatures and males

4. Male fitness function at 6 months

SELECTION AND INBREEDING OF CULTIVATED MACROBRACHIUM ROSENBERGII
IN THAILAND

Roger W. Doyle^a

1.1 SUMMARY

This consultant's report was written in Bangkok under the sponsorship of FAO/UNDP project THA: 75/008, “Programme for the Expansion of Freshwater Prawn Farming in Thailand”. The report is based on biological data collected during November and December of 1980, on published material and on discussions with the staff of Bangpakong Fisheries Station, Chachoengsao province.

Conclusions on inbreeding: In theory, a small hatchery-farm combination in Thailand could suffer a decrease in productivity of up to 34% after about 7 years of inbreeding, but this is easily avoidable if such hatcheries occasionally exchange spawners and post-larvae. Hatcheries producing 1 million post-larvae or more per year will have no inbreeding problems.

Conclusions on selection: Chachoengsao hatchery has been selecting large females as broodstock when practicable, but because spawners are often taken after the first harvest this selection has not been maximally efficient. Three alternative plans are suggested for increasing the intensity of growthrate selection. With an assumed heritability of 0.15 to 0.35, the most efficient controlled-breeding procedure might increase growth rate by up to 55% per year (if growth-rate variability is not unduly affected by behavioural interactions). Other plans are less efficient but easier to implement. A selection program should be initiated because the potential rewards appear to be great.

Matters for attention and decision:

1. Wild animals should never be introduced into the cultivated Thai stocks or the process of domestication will be set back.
2. Small hatcheries should exchange stock occasionally to prevent inbreeding.
3. If spawners are taken from a farm which practices continuous restocking, selection for increased growth rate will not be possible.
4. Selection can be greatly increased by taking large spawners from the first harvest or earlier.
5. The optimal age to select spawners according to the easy-to-implement PLAN C should be precisely identified.
6. An efficient selection program will require at least one resident staff member who has some quantitative genetics training (APPENDIX A).
7. A hatchery undertaking any of the selection plans must take its broodstock only from those farmers it has supplied with post-larvae.

^a Department of Biology, Dalhousie University, Halifax, Nova Scotia, Canada B3H 4J1.

2.1 INTRODUCTION

Cultured Macrobrachium rosenbergii in Thailand live out their entire life cycle in a human-controlled environment. This environment is quite unlike the natural one as regards water quality, food, predation and competition, and it is inevitable that the species will begin to adapt - by a process analagous to natural selection. This evolutionary change can be called “domestication” (Doyle & Hunte, in press). The rate with which domestication takes place depends on the intensity of “natural” selection in the artificial environment, on the amount of heritable variation in life-history traits such as growth and development rates, and to some extent on population size and sex ratio. Malecha, Peebles and Sarver (1979) have written a very interesting account of domestication in the Anuenue (Hawaiian) strain of M. rosenbergii. Their paper makes it unnecessary to include a general literature review in this report. The analytical portions of the Malecha, Peebles and Sarver paper deal with biometrical and behavioral changes which may have taken place in cultivated Macrobrachium. To my knowledge, no attempt has yet been made to obtain quantitative estimates of either the intensity of domestication selection or of inbreeding rates. The purpose of this consultants report is to try to make these calculations and also to make some suggestions as to how the rate of domestication might be speeded up in Thailand.

3.1 TERMS OF REFERENCE

Several aspects of Macrobrachium genetics are included in the terms of reference of this study. The first and most important is the rate of inbreeding. Inbreeding presents dagers to all domestic animals, and it is imperative to estimate the magnitude of this threat in Macrobrachium cultivation. The second aspect is to find out what kind of animals (relative to the rest of the population) are being chosen as breeders and whether this choice of broodstock by farmers or hatcheries is likely to be beneficial, neutral or harmful to the long range objective of domestication. The third aspect is to evaluate the possibility of introducing a deliberate program of stock improvement through selective breeding. It was understood from the outset that very intense selection does not necessarily lead to stock improvement and very rapid inbreeding does not necessarily lead to deterioration. Certainly these results would follow in most other domesticated animals but the genetics of the response to selection and inbreeding is still unknown for Macrobrachium. No doubt the genetics program at the Anuenue Fisheries Research Center in Hawaii will someday give us the answer to these important questions. If a scientifically-monitored selection program is begun in Thailand it will also contribute to our scientific understanding of the genetics of aquaculture production. The original motivation for the consultancy was partly based on prudence and partly hope; prudence, because for safety reasons it seemed best to avoid inbreeding, hope, because a program of selective breeding might well succeed spectacularly.

The practical difficulty of setting up a breeding program in Thailand was a major consideration in preparing the report. The Department of Fisheries hatchery at Chachoengsao, which is now or soon will be the largest single supplier of post-larvae (New, Singholka and Vorasayan, 1980) might be willing to invest effort in selective breeding if it can be shown that (a) the effort is not greater than the resources available will support and (b) the program makes scientific sense. It will doubtless be difficult to interest commercial hatcheries in undertaking such a program because any increase in production would be cumulative, and might not even be noticeable for several years. The long-term perspective should belong to government agencies which are interested in the possibility of a steady, cumulative increase in the productivity of Thai Macrobrachium stocks. I have attempted to consider both the scientific and the practical problem of genetically improving the Thai strain of Macrobrachium. Several selection programs of varying complexity are outlined, and Appendix A consists of an outline Research & Training proposal to provide the expert Thai manpower required for the success of such a program.

4.1 INBREEDING

Inbreeding can be informally defined as mating among relatives -- brothers and sisters, parents and offspring, or to a lesser extent, among cousins and more distant relatives. It occurs in randomly mating populations of animals whenever the population size becomes small. In humans, close inbreeding has unfortunate medical and sociological consequences. In animals, too, inbreeding can have strongly deleterious effects on survival, growth rate, disease resistance and fertility. An important objective of this consultancy is to estimate the level of inbreeding in Thailand and, if this proves to be significant, to suggest ways of reducing it.

4.2 Definitions:

The analysis of inbreeding in a random-breeding population is based on two key concepts, the “inbreeding coefficient” and the “effective population size”.

The inbreeding coefficient, F, expresses the average degree of relatedness between the parents of each individual in the population. It is a comparative measure which relates back to an ancestral population in which F is defined to be zero. In Thailand the ancestral, or base population is not well defined because cultivation started in the mid sixties, with wild-caught animals being added to the breeding population at least until the Chachoengsao hatchery began large-scale production in 1977. It is reasonable to take 1977 as the date at which inbreeding may have started to accumulate in the Thai Macrobrachium stocks.

At any time t (measured in generations) the level of inbreeding in the population is:

1. F_t = 1-(1-ΔF)^t × 100

   Where F_t is the current level of inbreeding in generation t expressed as a % of the maximum value of 1.00, and ΔF is the rate of inbreeding per generation.

   The expression for the rate of inbreeding is:

2. ΔF = ½N_e

   Where N_e is the effective population size.

   The effective population size N_e is a function of the number of successfully breeding animals in the population at any given time. When the number of breeders is constant and the sex ratio is l:l, male:female, the expression for effective population size is:

3. N_e = N + ½ (approximately, for constant N)

   Where N is the actual number of breeding animals of both sexes.

   If there are unequal numbers of male and female parents, as would be the case in a Macrobrachium breeding program the appropriate expression is:

4. approximately.

   With 20 females and 5 males, for example, N_e = 16 and F = 3.13% per generation. N_m and N_f are the number of male and female breeders, respectively. When the number of breeders fluctuates from generation to generation the expressions relating N_e to the actual population size is a little more complicated and when the generations overlap they are more complicated still. The reader should refer to standard texts such as Falconer (1960) and Turner and Young (1969) for more information. These complications are not dealt with here, however. The Thai population is treated as if all animals born within about 3 months of each other belonged to the same generation, and the generation length is approximately 9 months.

4.3 Relationship between inbreeding coefficient and inbreeding depression in Macrobrachium

The problem of estimating rate of inbreeding in Macrobrachium reduces to the problem of finding out what the effective population size is in the hatcheries, or in groups of hatcheries interconnected by the occasional exchange of stock. Note that the number of spawners determines N_f, not the number of gravid females in the pond. The fact that large numbers of animals mature and breed in the farm ponds is irrelevant to this calculation because the resulting offspring do not make it through to the next generation.

I have calculated ΔF and F_t for a number of different population sizes and listed the results in Tables 1 and 2. Table 1 refers to a randomly-breeding population in which it is safe to assume that each spawning female has been fertilized by a different male (1:1 sex ratio) and Table 2 to a population in which the male:female sex ration is 1:4, eg. a controlled-breeding population set up according to the recommendations of Ling (1969). The depressive effect of various levels of inbreeding on domestic animals is illustrated in Table 3 with data from Falconer (1960) and Turner & Young (1969), the two references available to me as this report is being written. Wild animals at the beginning of their adaptation to an artificial environment would certainly react to inbreeding faster than the animals in Table 3. In speculating on the inbreeding question in Macrobrachium I have assumed that the depression due to inbreeding is about three times as great as Table 3, which accords with what information ia available from my own (unpublished) observations on the amphipod Gammarus lawrencianus. This is a convenient working hypothesis because it makes the level of inbreeding depression (as a reduction from the “normal” growth rates, viability etc.) numerically about the same as the current value of F_t. In other words, where Table 2 indicates that 12 spawners every generation yields an inbreeding coefficient after 10 generations (F₁₀) of 41.4%, the growth rate and viability of the stock may be as much as 44% less than its original value. This is probably a maximal estimate of the effects of inbreeding depression. The true value of the relationship between F_t and inbreeding depression production would be an urgent research objective if the levels of F were expected to be significant (which they are not, as will be shown below).

4.4 Inbreeding in Thai Macrobrachium stocks

Estimation of inbreeding rates for several situations relevant to Macrobrachium cultivation in Thailand are presented in Table 4. There is at least one combination hatchery/farm in the Bangkok area which can in theory be described by the first row in this table (personal communication from Mr. Paiboon Vorasayan, Chief Extension Officer, Chachoengsao Station). If this farmer always uses animals from his own ponds and always seeds his pond from his own hatchery he could encounter serious inbreeding problems. The estimated decrease in growth and survival may be as high as 19% after 10 generations, giving a decrease in productivity of about 34%. This seems to be the worst possible situation which is at all likely to occur in Thailand. In practice, however, the farmer in question sometimes uses spawners from other farms and buys post-larvae from other hatcheries, thus breaking the inbreeding cycle and resetting at F = O.

According to Ling (1969) a medium-sized hatchery producing 1 million post-larvae per year would need to maintain a breeding population of approximately 60 females to provide 12 spawners per month. If we consider all of these females to spawn in the same “generation” (actually their offspring will differ in age by up to 4 months), then N_f = 60. The number of males required to fertilize these females is either 60 if the population is a hypothetical one with gravid females collected directly from the farm ponds, or about 15 if the population is a real one maintained at the hatchery for breeding purposes. The inbreeding rates for both sex ratios are shown in row 2 of Table 4. The effect of inbreeding may still be considered significant because the decrease in both viability and growth is 9% after 10 generations, for a potential loss of productivity of 17% if a special breeding population is maintained. If the size of the “Ling-type” hatchery is increased so that it produces 20 million post larvae per year, which is the objective of the Chachoengsao hatchery, the inbreeding coefficient is reduced by at least an order of magnitude and becomes too small to worry about (row 3). In fact the inbreeding situaction in Thailand for hatcheries producing a million larvae or more is even better than rows 2 and 3 indicate, because the number of females used as spawners in each generation certainly exceeds 60. Mr. Paiboon Vorasayan suggests that a hatchery tank producing 20,000 post-larvae per month requires approximately 20 gravid females per month, which works out to a standing population of approximately 400 females for a production of 1 million post-larvae per year. (This assumes that 12/60 of the females are ready to spawn each month.) Hatcheries producing on this schedule would have negligible inbreeding rates (rows 4 and 5, Table 4).

Table 1. Inbreeding rate per generation (ΔF) and accumulated inbreeding after 10 and 20 generations (F_t). Ne = effective population size. Male:Female sex ratio = 1:1.

Table 2. Inbreeding rate per generation as in Table 1, but with a male: female sex ratio of 1:4.

Table 3. Some examples of Inbreeding Depression (% decrease in trait)

Data from Falconer (1960) p. 249 except sheep viability from Turner and Young (1969) p. 292.

Table 4. Inbreeding depression in situations of interest to Macrobrachium farming in Thailand.

If all 32.4 million post-larvae produced in Thailand (New, Singholka and Vorasayan (1980) came from a single population of females the value of N_f would be about 13,000. The inbreeding in such a large population would be essentially zero (row 6, Table 4). Because hatcheries occasionally mix spawners from different farms, the idea that all the animals under cultivation in Thailand belong to a single population is probably not far from the truth. This population is very large, and so inbreeding is not a problem in Macrobrachium farming in Thailand.

5. SELECTION

5.1 Principles:

Individual M. rosenbergii, like most wild aquatic organisms are highly variable, even when reared together in the same environment. In principle, the simplest program of genetic stock improvement is to choose superior animals as breeders so that as generation succeeds generation the variation in the original population is translated into improved production. This straightforeward approach can be guaranteed to work only if certain conditions are met: (1) the variation must be heritable so that the superior qualities of the parents are passed on to their offspring, (2) the qualities designated “superior” must be easy to recognize so that large numbers of animals can be classified quickly, (3) traits under selection must not be correlated in a way which is counterproductive (in M. rosenbergii, for example there would be serious problems if selection for decreased time-to-metamorphosis resulted in slower post larval growth), (4) it must be physically convenient to induce the selected individuals to mate and to keep the selected offspring separate from the rest of the population, (5) the progress of the selection program must be carefully monitored to maintain the integrity of the experimental design over many generations.

These are stringent requirements even the laboratory using wellunderstood animals like Drosophila. In M. rosenbergii under conditions of commercial cultivation in Thailand not one of the requirements can be fully met at the moment. Therefore the various selection programs I have outlined in this report are all compromises and they come without a guarantee.

The present procedure for propagating Macrobrachium at Chachoengsao involves selection of the larger and younger of the available females whenever possible; this has been done deliberately with the goal of stock improvement in mind. Therefore the present system is analysed, in this report, as a “plan” under the assumption that it is competing with other possible plans and will be continued if it is not replaced by some other. The “no guarantee” warning applies to this plan as well.

There can be no doubt that some sort of selection program is well worth a try. Based on analogies with other organisms the potential payoff is enormous (a potential gain of 55% per year for (PLAN A, below).

The present method of selection (PLAN D) could be made more intense with little change in procedure. The five conditions for guaranteed success will be discussed further in relation to the various selection plans.

5.2 Note on the numerical data

Morphometric measurements were made on animals collected at three locations near Bangkok:

1. Thai Prawn Farm (Mr. Peng). Thai Prawn farm is one of the few farms in Thailand which continuously harvests and restocks its ponds on the “Hawaiian plan”. Although this is doubtless very efficient from the production viewpoint, the animals in the ponds are a mixture of ages and so cannot yield growth rate information or be used for size-dependent selection. Morphometric measurements were made on animals of both sexes over a wide size range with the help of the farm manager, Mr. Sawat, whose assistance is gratefully acknowledged. The measurements were (a) standard length (orbit to telson), (b) diameter of the sub-terminal segment of the 2nd thoracic leg (the segment just behind the pincer of the main claw), and (c) volume of the egg mass of pregnant females, estimated by sedimenting the eggs under gravity in a graduated centrifuge tube.
2. Three-month-old animals were collected by the staff of the Chachoengsao fishery station at a nearby farm. The collections were made by cast net (entangling the animals as it fell) and were probably quite random in respect to body size. The measurements made on these animals were the same as those at Thai Prawn Farm.
3. Six-month-old animals (not previously harvested) were also collected and measured, the collections again being made by the Chachoengsao staff. On this occasion the cast-net collection was supplemented by about 5 animals collected by hand, which may have biassed the results slightly.

The animals from these three farms are probably representative of those grown in central Thailand. Calculations based on them are therefore representative of the situation in Thailand but are not averages in any statistical sense. Obtaining true mean values (or standard errors including between-farm variance) is not an exercise which can be justified at the present time. The work required would be immense because of the ecological variation around the country, and the principal limitation on the accuracy of the predictions made in this report is not inadequate data, but uncertainty about the genetic effects of a given level of inbreeding and ignorance of heritabilities and genetic covariances. Research effort should be devoted to getting this kind of information.

5.3 Notes on the selection objective:

It is assumed in this report that if selection is stepped up on Macrobrachium in Thailand it will, in the first instance, be aimed at improving the gross productivity (per hectare, per year) of the selected stock. There are two aspects of productivity where improvement is potentially possible: survivorship and growth rate. The survivorship component is automatically put under strong selection whenever the life cycle is closed so that only those animals which survive in the aquaculture environment are chosen as parents of the next generation. The present cultivation technique is as good as any other that could be easily implemented, and Mr. Singholka's personal observation is that the survival of larvae and postlarvae from the time of hatching to metamorphosis is much better than that of larvae from wild females that he has tried from time to time. This evidence of domestication augers well for the success of a selection program but not much more can be said about it here. The present practice is adequate and the only way it could be set back would be by deliberate or accidental introduction of new “wild genes” into the domesticated stock. This should be rigorously avoided, and it would be as well for the extension workers to ask the hatcheries never to use wild gravid females.

It is the growth-rate for size-at-age component of productivity that creates the difficulties in designing selection programs in Macrobrachium. These difficulties are related to the second and third of the five general restrictions on success mentioned above. Individual differences in growth rate may have many possible causes, both metabolic and behavioral, and until one understands what causes the variation it is hard to predict what will happen to the stock when selection begins to succeed in changing its biological characteristics.

Assume for a moment that the growth rate variation is purely physiological (“anabolism” vs. “catabolism”). The question arises whether this implies variation in the rate of food consumption or in the efficiency of food utilization -- both possibilities being significant when profit rather than biomass is the objective of the farmer. Or does it imply variation in the way energy is partitioned between somatic growth and reproduction? This latter possibility is easy to envisage for femaes but the situations in males is more complicated. Do small physiological differences in growth rate become, in the males, amplified by behavioral interaction such as dominance hierchies and the hypothetical “bull male” effect? If so would size selection in the males have the undesirable effect of increasing the size variance which is already a major practical difficulty?

Obviously it is desirable to find the answer to these problems, and there are as many possible research programs as there are scientists interested in them. The most direct approach would be experimental -- the construction of a physiological energy budget for the different ages and sexes. This is the province of physiologists and nutritionists and it is probably the most viable approach providing it can be kept focused on the main problem, which is to explain how variation in growth rate can be translated into a rational selection scheme. My own approach (because of scientific background) would be a biostatistical study of what is actually going on in the farm ponds in terms of changes in the means and variances during rearing etc. This would involve the construction of formal hypotheses which would be tested against the data. The Thai cultivation scheme in which there are many farms, each of which has many replicate ponds is ideal for this sort of analysis -- a well-controlled experimental setup which could not be duplicated for many millions of dollars.

In fact the two approaches complement each other. Until some kind of research is done on the causes of growth rate variation it is only possible to make genetic calculations on size-at-age as a simple trait. In the pages which follow the rate of response to selection will be estimated without further comment about what aspect of the biology of Macrobrachium would actually change.

5.4 Definitions

As in the selection on inbreeding there are several technical terms which cannot be avoided and which are informally defined in this report for the convenience of the reader.

(a) Selection differential, S.

The selection differential is the difference between the mean of the animals which have been chosen as breeders, and the mean of the general population. In other words, a measure of the superiority of the broodstock over the other animals in the same generation. The magnitude of S depends on two factors, the proportion of the population which is being chosen as breeders and the amount of variation in the population. The figure below should make this clear.

In this figure approximately 30% of the population is shown as being selected. The selection differential is about 7 grams.

(b) Intensity of selection, I.

A frequently used informal measure of the intensity of selection is simply the proportion of the population selected as breeders -- about 30% in the figure above. Thus choosing 30% represents less intense selection than would choosing, say 5%. In general, of course, the more intense the selection the more rapid will be the evolutionary improvement in the population.

(c) Response to selection, R.

The response to selection is what one is aiming for, namely a change in the mean value of a trait from one generation to the next. Like the selection differential, R can be measured in units such as weight

(d) Heritability, h².

Heritability is a measure of the ability of a population to respond to a given intensity of selection. More specifically, it is the proportion of the variation in the population which can be ascribed to the kind of genetic variation that can respond to selection and as such, represents the extent to which variation is transmitted from one generation to another. Obviously, h² is an extremely important concept because if h² = 0 even the most intense selection will yield no progress, while if h² = 1.0 (its theoretical maximum) improvement will be observed just as quickly as one can select for it.

There are various ways of estimating h² of which all but one would require a very large commitment of research facilities and personnel on the scale of the Hawaii program. Pending the result of these studies h² will remain the great unknown quantity and a major source of uncertainty about the success of any selection program. There is a way of measuring h² during selection, however, which is highly relevant to the program in Thailand and where a Thai program could make an important contribution to the solution of an outstanding scientific puzzle. The reason why a production-oriented program can have this result as scientific “spin-off” is that heritability can be defined as the ratio of the two quantities R and S:

h² = R/S

Thus if the selection intensity and response to selection are both being monitored, the so-called realized heritability can be calculated. In fact this is the best way to do it because h² then applies to the environment in which the information will actually be used, namely the production environment. This is an important research objective which does not duplicate the work in Hawaii.

(e) Fitness, W.

Fitness is defined here (operationally, for females) as the number of eggs or total volume of eggs she is carrying at any given age. The fitness concept is an essential component of the theory of evolution in both wild and artificial populations like Macrobrachium in farm ponds. What “fitness” really means, of course, is the relative ability of an individual to contribute offspring to the next generation. Whatever morphological or behavioral traits contribute to variation in fitness will, generally speaking, be under natural selection even in a highly artificial environment. The more complicated and speculative definition of fitness as it applies to male Macrobrachium will be dealt with in the discussion of selection Plan B.

(f) Fitness function.

A fitness function is a mathematical relationship between fitness and some morphological or behavioral trait which is thereby under selection. In Macrobrachium females fitness appears strongly related to body weight under farm-pond conditions, and the fitness function relating egg volume to body weight is an upward sloping line (Fig. 1).

6. SELECTION PLANS

6.1 Plan A, Controlled Breeding of Macrobrachium at the hatchery.

(a) Summary.

This is the most efficient and also the most complicated of the selection schemes. It includes both a production and a scientific component and there are major implications for facilities and manpower. It will be shown below that a not-unreasonable estimate of the rate of potential gain is 55% per year (under the assumption that some of the Macrobrachium growth variation is heritable).

The most important part of the program is of course the production aspect in which the major objective is to provide juveniles to the farmers. The scientific aspect is not required for either selection or production per se but would nevertheless be essential for long term success. The justification for the science is not so much to obtain realized heritabilities as to provide a “watchdog” over the nature of the biological response to intense, prolonged selection. The value of good h² estimates to worldwide planning of aquaculture programs should not be minimized, however.

PLAN A differs from present practice in taking spawning animals from a special breeding pond where they have been held for 1 or 2 months, rather than directly from the farmers. The breeding pond is regularly re-stocked with the best (eg. largest) animals from the farm ponds. These animals are selected either by a special harvest at about 3 months when only the fastest-growing animals are sexually mature, or during the first regular harvest at about 6 months. After being collected from the farm pond, but before being put in the breeding pond, the gravid females must be cleared of any eggs they may be carrying (the results of uncontrolled breeding) by keeping them in concrete tanks similar to the present hatchery tanks until they spawn. These larvae are discarded. The relative advantages and disadvantages of the 3 or 6 month selection will be discussed in a later section.

Fig. 1. Female fitness function

(b) Flowchart.

(c) Rate of Progress.

The only way to estimate rate of response to selection in the absence of heritability information is to multiply the observed selection differentials by h² values which seem reasonable in the light of other studies on other organisms. The range of h² values used in this report is 0.15 to 0.35, which is fairly conservative for growth rate in terrestrial animals but is too high for fish, according to Moav and Wohlfarth. My own work with the estuarine amphipod Gammarus lawrencianus is giving values in this range. The summary statistics for the samples of 3-month-old and 5-month-old animals are given in Tables 5 and 6. The relationship between female body length and egg volume (used in the fitness function for females) was established on the basis of measurements on animals at Thai Prawn Farm, (Fig. 1). Data on which the male fitness function is based were also collected at Thai Prawn Farm but are not used in connection with Plan A. The relationship between body length and body weight was established from data made available by Mr. Paiboon on animals of various ages (Fig.2).

The results of the calculations on Plan A are shown in Tables 5 and 6.

Fig. 2. Relationship between standard length and wet wight in Macrobrachium rosenbergii (both sexes combined). Data from Mr. Paiboon Vorasayan. Line fitted by eye.

Table 5. Selection statistics for selection at 3 months. Plan A.

Table 6. Selection statistics for selection at 6 months. Plan A.

The selection intensities (10%--20%) are usual for this sort of program, with much higher intensities unduly limiting the ultimate response level and, sometimes, giving rise to pathological effects. The sex ratio in the breeding pond is the recommended male:female ratio of 1:4.

The possible rate of progress with Plan A is considerable (Table 5), up to 55% per year depending on heritabilities and age at selection. Even with all the caveats and untested assumptions taken into account this is surely worth thinking about.

(d) When to select.

Selection at 3 months requires a special harvest for the purpose, but the actual choice of animals would be easier because the largest animals are easily recognized (only the largest females have egg masses or mature ovaries). Selection at 6 months can be done during the first regular harvest, but choosing the largest individuals might be rather difficult under field conditions. The workers would be up to their waists in water trying to sort the animals into males-for-breeding, females-forbreeding, marketable-size and rejects. However, in the 6-month sample I looked at many of the early-maturing females had already spawned once or twice and happened to be without eggs at the time of observation. If this is generally true of 6-month-old populations it would allow one to eliminate the clearing stage and perhaps reduce the duration of the breeding stage of the selection cycle.

Selection at any time after the first harvest is not worth considering, as the fastest-growing animals will have been removed from the population and sold.

(e) Requirements for implementation.

1. Clearing tanks of the type presently in use as hatchery tanks at Chachoengsao. Two or 3 should be sufficient unless it is necessary to keep the gravid females more than a week or so.

2. Breeding ponds capable of holding at least 400 females and 100 males with low mortality rates. Presumably earthen ponds of 1000 m² would be suitable.

3. Standard rearing ponds of 0.1 hectare or so, similar to small farm ponds but more carefully controlled. These ponds would be used for the scientific aspects of Plan A, and need not be at the hatchery where selection is taking place. Their numbers might be, say 20 in groups of 4, or 40 in groups of 8, distributed among 5 locations in Thailand. The grouping into 4's or 8's is for the purpose of experimental design, but the details of such a hypothetical design would seem to be beyond the terms of reference of this report (but see Appendix A).

4. Technicians, not labourers would be needed to select the animals properly from the farm ponds and to obtain data from the experimental ponds.

5. Scientists interested in the problem who would be able to interpret the results and feed them back into the production program.

6.2 Plan B, Artificial selection of females.

(a) Summary.

This plan differs from Plan A in that only the females are selected according to size at 3 or 6 months. The males are not artificially selected at all but are under rather strong “natural” selection for size-at-age. The selected, gravid females are collected from the farm ponds and are placed immediately into the hatching tanks as is presently the practice. There are no clearing tanks or breeding ponds so PLAN B is much less expensive than the previous one. Extra technicians would however be required to perform the selection properly. A scientific component is warranted by the importance of discovering the biological basis of growth rate variation. The possibility of estimating realized heritability has been lost, unfortunately, because of uncertainty in the estimate of S in males which are not selected directly.

The estimated maximum response to this sort of selection is 26% per year, with the usual caveats on heritability.

(b) Selection intensities.

The selection on females is performed just as in Plan A and the estimated selection intensities are the same.

The only selection acting on the male population is natural selection on traits which affect mating success. It is obvious that at 3 months, and to a lesser extent at 6 months, all the males cannot be sexually active or equally successful at reproduction. The question is how to quantify this difference in fitness and relate it to variation in the trait we are interested in, namely body size. The most convincing calculation would be based on the proportion of males that have functional sperm, and the “electroejaculation” technique of Sandifer and Lynn is an apparently ideal methodology for this.

In this report I have attempted to use the development of secondary sexual characteristics to make the connection between size and fitness in males. The most immediately obvious difference between mature male and mature female Macrobrachium (aside from a visible egg mass) is body size and the diameter (thickness) of the main claw. In mature males the subterminal segment is much larger relative to body length than it is in females. The ratio of claw thickness (in cm) to body length (in cm) averages about 0.65 in mature males and about 0.35 in mature females. Surprisingly enough these differences are not in themselves sex-specific, as both sexes fall on the same allometric line (Fig. 3). In immature animals of both sexes the relative size of the dactyl is the same as it is in animals which are too young to classify, approximately 0.22. As the males mature the allometrically claw thickness increases with body length, at least over the range of sizes encompassed by the 3 to 6-month-old population (Fig. 3). The data in this figure came from all three collections.

The argument is as follows:

In every population which contains gravid females there must be some males which are engaged in sexual activity. Let us suppose that the males most probably sexually active are those with the best developed secondary sexual characteristics, especially characteristics which, like exaggerated claw thickness, are involved in combat and display. Then the males with the highest claw:body length ratio in a population containing pregnant females are assigned a probability of reporduction of 1.0, which is their “fitness”. Males which have the lowest claw: body length ratio (immature or subordinate, or both) are assigned a fitness of 0.0. These two assumptions limit the scale by which any given set of claw: body ratios can be transformed into relative fitness values. To complete the specification of the fitness function we make the assumption that the transformation is linear. With these transformations the claw: body length ratio shown in Fig. 3 can be presented as the fitness functions of Fig. 4 and Fig. 5 for 3 and 6 months, respectively.

Fig. 3. Relationship between the ratio of claw thickness to body length and body length. Males and animals too small to classify. CST in mm, standard length in cm.

Table 7. Selection statistics for selection at 3 months, Plan B.

(Note: X = length in cm
Wt = weight in grams
R = assumes h² = 0.15–0.35)

Table 8. Selection statistics for selection at 6 months, Plan B.

Table 9. Selection statistics for selection at 3 months, Plan C.

Table 10. Selection statistics for selection at 6 months, Plan C.

It can be proven (Doyle and Myers, in press) that when a fitness function relating reproductive success to any trait X is linear, the rate of change of X during natural selection (the selection differential) is:

where S_x is the selection differential, r_w,x is the product-moment correlation coefficient between X and W, ♂_x and ♂_w are the standard deviations of X and W, and W is the mean fitness.

This equation has been used in Tables 5 and 6 to provide a “natural selection” for fecundity differences among the selected females. In females the correlation r_w,x was the correlation between egg volume and body size, and w was the mean egg volume of the largest 16% of the females. For males the correlation r_w,x is the correlation between the probability of reproduction and body size shown graphically in Figs. 4 and 5.

(c) Rate of progress.

Calculations on the rate of progress under Plan B are given in Tables 7 and 8 along with the summary statistics on claw: body length. The same assumptions about heritability are made as in Plan A. The rate of progress is of course much less than in Plan A. The rate of progress has a reasonable maximum value of 26% or 21% per year, depending on whether selection is at 3 or 6 months.

(d) Requirements for implementation.

1. Technicians and all the other support required to make the selection of females as accurate as possible.

2. Standard rearing ponds and scientific staff as in Plan A if the progress of selection is to be followed scientifically.

6.3 Plan C, Enhanced natural selection.

(a) Procedure.

This plan is a simple modification of current hatchery practice at Chachoengsao. Spawning females are not selected by size but are collected as young as possible, either at the first regular harvest at about 6 months or in a special harvest at about 3 months after stocking. Careful timing enhances the intensity of natural selection on growth rate for three reasons: (1) only the larger females have eggs, (2) these females have been fertilized by the larger males in the population, (3) larger animals of both sexes have not yet been harvested and are still available as broodstock. The maximum rate of progress which could reasonably be anticipated is about 12% per year with the assumptions made here.

Fig. 4. Male fitness function (6-month population).

(b) Rate of progress.

Calculations on the rate of progress for selection at 3 months and 6 months under Plan C are shown in Tables 9 and 10, respectively. At three months the maximal rate of response is about 12%. At 6 months the rate is lower, about 7%, and it is noteworthy that the selection differential on females is actually negative at 6 months because the gravid and full-ovary females are smaller than the average female at this time. This presumably indicates that the larger females have already spawned at least once and are not yet gravid again. Six months appears to be too late to select unless some effort is made to choose large females, not just gravid females. Plan C is the easiest to implement because it does not require judgements about the relative sizes of individuals under trying field conditions. However, it is clear that the optimal time to take spawners has not yet been identified, and furthermore it probably varies from farm to farm and also with the season. The optimal time is easy to identify -- just find the age (probably between 3 and 6 months) at which the difference between the size of the gravid females and the mean size of the whole female population is maximal.

6.4 Plan D, Current practice in Thailand.

Procedure and rate of progress. The techniques used to cultivate Macrobrachium rosenbergii in Thailand have been admirably described in a series of working papers published by the National Freshwater Prawn Research and Training Center, Department of Fisheries with the support of FAO/UNDP (THA/75/008). A summary has been published by New, Singholka and Vorasayan (1980).

From a geneticist's point of view the situation appears to have two main features. Firstly, the effective population size is probably large as previously pointed out in the section on inbreeding. Selection of part of the population will lose much of its effectiveness when the selected animals are genetically mixed in with the rest. If selection of a “Chachoengsao strain” is to be effective contamination of the broodstock with unselected animals will have to be avoided. For the same reason, the importation of wild animals into the broodstock will have to be avoided.

Secondly, “large body size” is not exactly equivalent to “fast growing” in the current procedure, because animals graded by size are frequently of different ages (if taken from a continuously restocked pond) or are the residue after the fast-growing animals have been removed (batch-stocked pond). It can be concluded that the intensity of selection on growth rate, and consequently the rate of response is rather low with the current system.

There will, of course, be very effective selection for survival under the present system. As previously mentioned, there are signs of improvement in this important component of production.

7. REFERENCES CITED

Doyle, R.W. and W. Hunte, 1981 in press. Genetic changes in “fitness” and yield of a crustacean population in a controlled environment. J. Exp. Mar. Biol. Ecol.

Doyle, R.W. and R.A. Myers, 1981 in press. The measurement of the direct and indirect intensities of natural selection. In H.Dingle and J. Hegman, Eds. Variation in Life Histories: Genetics and Evolutionary Processes. Springer-Verlag.

Falconer, D.S., 1960. Introduction to quantitative genetics. The Ronald Press. New York. 365 pp.

Ling, S.W., 1969. Methods of rearing and culturing Macrobrachium rosenbergii (de Man). F.A.O. Fish. Rep. 57(3); 607–619.

Malecha, S.R.,J.B. Peebles and D. Sarver, 1979. Approaches to the study of domestication in aquaculture: the development and characterization of genetic stocks and their hybrids in the cultured freshwater prawn, Macrobrachium rosenbergii. Paper presented at the statutory meeting of the International Council for the Exploration of the Sea. Warsaw, Poland October 1–10, 1979.

Moav, R., 1976. Genetics and genetic improvement of Fish. Chapter 10 in Pillay, T.V.R. and W.A. Dill, Eds. Advances in Aquaculture. Fishing Book News. Farnham, England. p. 610–622.

New, M.B., Singholka, S. and P. Vorasayan, 1980 in press. Current status of freshwater prawn farming in Thailand. Proceedings World Mariculture Society 11.

Turner, H.N. and S.S.Y. Young, 1969. Quantitative genetics in sheep breeding. Cornell Univ. Press. Ithaca, N.Y. 332 pp.

8.1 ACKNOWLEDGEMENTS

This consultant's report was written in Bangkok under the sponsorship of FAO/UNDP project THA: 75/008, “Programme for the Expansion of Freshwater Prawn Farming in Thailand”. I am grateful to Mr. Somsuk Singholka, director of the Bangpakong Fisheries Station, Chachoengsao Province and to Mr. Michael New, FAO project officer and Senior Fisheries Biologist (Aquaculture) for their courtesy and friendly cooperation throughout my 6-week stay in Thailand. With their aid, obtaining the numerical data and other information upon which this report are based proved to be easy. More than this, and on a personal level, they made my first exposure to Asia a thoroughly pleasant and rewarding experience. Mr. Peng and Mr. Sa-Ngar Meknavin, respectively general manager and associate General Manager of Thai Prawn Farm provided me with overnight accomodation and biological samples for which I am very grateful.

9.1 APPENDIX A

TO: Mr. Somsuk Singholka, Mr. Michael New, Mr. Alex Fedoruk

FROM: R. W. Doyle, December 5, 1980

CONCERNING: Outline of a research and training program in Macrobrachium quantitative genetics.

OBJECTIVE

The objective would be to provide research and training support for a stock-improvement program undertaken by the Department of Fisheries in Thailand. If such a stock-improvement program were begun it would of course have increased production as its primary aim, and several suggestions as to how this might be accomplished are included in my consultant's report. It is desirable -- and perhaps in the long run essential -- to support the practical goal with a quantitative genetics research program. The scientific activity outlined in this memo has two aspects: (1) an analysis of the genetic changes which will take place as selection begins to have an effect, and (2) training of Thai personnel who can exploit this genetic information by feeding ideas back into the selection and production programs. A stockimprovement program would be unlikely to succeed without resident staff members at the various fisheries stations who are familiar with the objectives and techniques of selective breeding. This research and training proposal deals with the problem of finding such people.

JUSTIFICATION

The technique used in Thailand to cultivate Macrobrachium has the potential to bring about profound and rapid genetic changes in the productivity of the stock. These changes could well be beneficial, but this depends on two factors, (1) the direction and intensity of selection of superior broodstock, and (2) the underlying genetic structure of those traits which affect survivals of the major components of productivity.

A discussion of various sorts of selection are included in my consultant's report. Data analysed in that report indicate that the potential gain in growth rate under intense selection could be as high as 55% per year, assuming a heritability of growth rate of 0.35. Less efficient (but easier to implement) programs or more conservative heritability estimates would of course reduce this figure. Even so, there is reason to seriously consider some sort of stock-improvement program based on artifical selection.

There are several lines of evidence suggesting that selection would bring about beneficial results. Selection for larval survival has been intense since the commercial-scale production of juveniles began in Thailand, and it is Mr. Singholka's impression that the survival of the cultivated strain of larvae is now better than that of larvae taken from wild-caught females and tested under the same conditions. Malecha, Peebles and Sarver (1979; ICES document CM 1979 F:48) report differences in the size distribution of the cultivated Anuenue strain, relative to wild strains, which they attribute to behaviour changes associated with domestication. Experiments on the amphipod Gammarus lawrencianus have induced substantial changes in survival and growth rate as the result of relatively weak and indirect selection (Doyle and others, several references). While it may not be fair to depend on an analogy between Macrobrachium and an amphipod, the yield of the latter has increased by 300% under a Macrobrachium-like model aquaculture system, which shows that with some crustaceans at least it can be done.

If we suppose that some form of enhanced selection were to be incorporated into the production cycle at Chachoengsao or elsewhere, we should consider the case for a research and training program to go along with it. One of the primary objectives, as already mentioned, would be to train resident staff members so that they could select efficiently, monitor progress and solve problems in an informed way as they arise. The research, as distinct from training, objectives are also important. At the moment virtually nothing is known about the biological basis of growth rate variation in Macrobrachium which could be caused by differences in metabolism, behaviour, or both. Some sources of variation will respond more quickly to selection than others, and changes could conceivably be deleterious. (If, for example, selection for large males increased social dominance and hence population variance the change would probably be considered deleterious even if the gross harvest increased.) For reasons of efficiency and caution, then, intense selection should be accompanied by a research program.

There is another reason as well. The Thai selection program would provide a unique opportunity to get information on “realized heritability” of traits related to production. Heritability, a genetic parameter, can be efficiently estimated by comparing the rate of response to selection, generation after generation, to the intensity of selection that brought it about. Both the selection and the response have to be carefully measured, which is where the research comes in. What makes the Thai program unique is that, unlike the system of continous restocking being developed in Hawaii, the Thai style of cultivation is for the most part based on batch culture. This keeps the generations distinct, allows accurate estimation of selection intensity and response, and raises the possibility of getting heritability information which would be of immense usefulness wherever crustacean aquaculture is being contemplated. The inter-related hatcheries and farm ponds represent a country-wide “experiment” which would cost many millions of dollars to duplicate. It seems a good idea to try to take scientific advantage of this experiment. The admirable genetics program at the Anuenue Fisheries Research Center in Hawaii is based on a different sort of estimation procedure, and there would be little duplication of effort.

PROGRAM OUTLINE

Most of the training would be done at Dalhousie University and other institutions in eastern North America, while the research would take place in Thailand at fisheries stations located in Chachoengsao, Phayao and elsewhere. The laboratory and computer facilities at Asian Institute of Technology and NIFI would be used as well. If some of the research is to be done by graduate students as part of an M.Sc. program, the faculty of the degreegranting institution should be involved. However, the first priority is to train people from the staff of the fisheries stations (or seconded to the stations) where the Macrobrachium selection is to be implemented and the results monitored. If stock improvement is the responsibility of Chachoengsao station, as it probably will be, the scientific and monitoring responsibilities might be distributed among other stations in Thailand.

The suggested program has four components which repeat on an annual cycle:

1. Training in my laboratory during May, June and July. This is the best time for the Canadian part of the program for several reasons: it falls outside the regular university teaching semesters so there would be ample space in student laboratories, the research laboratories are at their peak of activity, the climate is suitable for travel and field work (1 or 2 days might even be rather pleasant), space is available in the otherwise crowded student residences, and winter clothing need not be purchased.

   During the training period the Thai students would learn some of the elements of quantitative genetic analysis and apply these techniques to the “model” aquaculture organisms in my laboratory -- amphipods -- as well as to the lobster Homarus americanus. They would spend some time at other laboratories in the region, in particular the fisheries biological station at St. Andrew's, New Brunswick, where there are programmes underway on lobster breeding biology and salmon genetics. If the timing is right it would be worthwhile to send the students to the 1-week Scientific Aquaculture course conducted annually by the Marine Biological Laboratory at Woods Hole, Massachusetts. I teach part of this course and could take the Thai students to Woods Hole and back.

   The most important objective of the training period would be for the Thai students to plan on an appropriate Macrobrachium research project to be followed up in Thailand. The project would be designed in as much detail as possible and exercises in data analysis would be conducted using data simulation. The detailed written research proposal would constitute the student “output” from the training program.

2. During August I would return to Thailand with the Thai students to help them set up the research projects. These projects would of course fall under the overall supervision of the chiefs of the stations where they are being carried out. At the same time, I would review the progress of projects already underway and discuss possible new ones with the station chiefs.

3. It is probably wise to plan a return trip to Thailand in December for a short period of one or two weeks. While this might seem an unnecessary expense, one review every year is hardly enough to keep beginning researchers motivated and on the right track.

4. If the Thai students are not already familiar with Macrobrachium hatchery and rearing practices before they come to Canada, they should become so through a brief apprenticeship at Chachoengsao or elsewhere.

10.1 APPENDIX B

Itinerary, R. W. Doyle from Nov. 9 - Dec. 20, 1980.

Nov. 9 Left Canada

Nov. 10 Arrived Bangkok, met at airport by Mr. Somsuk Singholka. Discussed program with Mr. Singholka.

Nov. 11 Visited Department of Fisheries with Mr. Michael New, met Mr. Ariya Sidthimunka and Dr. Piroj. Visited UNDP/FAO, met Mr. Rattana.

Nov. 12 Visited Chachoengsao hatchery, discussed program with Mr. New and Mr. Singholka.

Nov. 13 Visited Thai Prawn Farm to observe harvesting procedure, re-visited Chachoengsao hatchery in company with Mr. King To Chang.

Nov. 14 Visited Farm of Mr. Chob in company with Mr. Peng and Mr. Sa-Nagar, Thai Prawn Farm, to observe harvesting procedure.

Nov. 15–18 Detailed plan of program. Trial calculation on “made-up-data” (Bangkok).

Nov. 18–19 Stayed at Thai Prawn Farm to make morphometric measurements on Macrobrachium.

Nov. 20–24 Data reduction and selection calculations (Bangkok).

Nov. 25 Visited Chachoengsao to interview Mr. Paiboon Vorasayan on inbreeding questions.

Nov. 25–27 Inbreeding calculations (Bangkok).

Nov. 27 Collected 3-month-old animals in company with Mr. Vorasayan and Mr. Chamnan. Made morphometric observations at Chachoengsao.

Dec. 3 Collected 6-month-old animals with Chachoengsao technician Mr. Saksit. Morphometric observations.

Dec. 3–8 Data reduction and selection calculations (Bangkok).

Dec. 4 At NIFI discussed possible Macrobrachium genetics program with Mr. Alex Fedoruk, Mr. Wanit, Mr. Chen Foo Yan and Mr. Michael B. New.

Dec. 8 Submitted memorandum on possible Research/Training Program to Mr. Fedoruk. Visited Asian Institute of Technology to talk with Dr. Kanchit Malaivongs, Assoc. Dir. of the Regional Computer Center about possibilities of speeding up data reduction etc. and also about computer requirements for the genetics follow-up program.

Dec. 14 Submitted draft of final report, began work on research papers resulting from the consultancy.

Dec. 15–17 Holiday

Dec. 17–19 Work on research papers and revision of report.

Dec. 19 Seminar at NIFI on Selection and Stock Improvement in Macrobrachium in Thailand.

Dec. 20 Departed Bangkok.

Dec. 21 Arrived Canada.