Fishing is a selective agent and has the potential to change populations. Many fishers target a particular species or group of species and target large adults as opposed to juveniles. It is well known that different fishing methods catch different size fish in the same area, e.g. the trawl and longline fisheries of the North Atlantic Ocean (ICNAF 1963) and the seine, gillnet and troll fisheries for Pacific salmon off British Columbia (Ricker et al. 1978). In spite of this potential for selection, evidence for genetic changes due to fishing, especially in marine populations, has been limited. In an early review Miller (1957) concluded that there was little evidence for any heritable changes due to exploitation or introductions in freshwater fishes. Since that period there have been several reports of genetic changes in fish populations associated with heavy fishing pressure. Most reports are based on a change in life history characters over time, and these characters may change non-genetically in response to environmental change.
The Atlantic cod has been exploited throughout the coastal waters of the North Atlantic Ocean for centuries, but biological sampling of catches has been carried out only for the past 50 years. The discovery and excavation of a fishing schooner lost of the coast of Nova Scotia in the 1750s provided a unique opportunity to compare cod lengths, based on measurements of cleithra bones from the pectoral girdle, between the 18th and 20th centuries (Kenchington and Kenchington 1993). Modern hook caught cod off Nova Scotia are much smaller than cod caught during the 1750s, although cod of similar maximum size are occasionally caught (Kenchington and Kenchington 1993).
On the Scotian shelf in the Northwest Atlantic catches peaked at over 80 000 tonnes in 1968 and were accompanied by a marked decline in biomass and corresponding decline in mean size and age of cod (Beacham 1983a). In addition the median length and age at maturity declined: the median age at sexual maturity declined from >5 years in the 1960s to < 3 years in 1978 (Fig. 1) for both males and females (Beacham 1983a). Cod that matured at smaller or younger sizes would have a selective advantage under heavy fishing pressure as the larger and older maturing cod would be captured before the onset of sexual maturity (Beacham 1983a). Similar observations have been made for Atlantic cod at West Greenland, where the average age at first maturity declined from 9.9 years in 1917 to 6.4 years in 1936 in the northern fishery and from 9.3 years in 1922 to 7.6 years in 1936 in the southern fishery (Hansen 1949).
In the Arcto-Norwegian stocks of Atlantic cod the majority of the fish matured at 8–10 years in the 1930s but by the 1970s the majority of cod matured at 6 years (Borisov 1979). This decline has been attributed to a selective removal of late maturing cod from the population (Borisov 1979). No trend in length at age was noted over a 40-year period suggesting that growth was not density dependent, although a decline in length at age in the last half of the 1980s coincided with a decline in relative abundance of capelin, the major prey item for large cod (Jorgensen 1992). Law and Grey (1989) have suggested that fisheries on the younger 4-year old “feeding stocks” in the Barents Sea and on the Spitzbergen Shelf should be restricted and fishing concentrated on the larger spawning fish on the Norwegian coast. This fishing pattern would select for late maturing individuals and give an increase in yield.
Figure 1. The median age at sexual maturity for Atlantic cod caught in Fishery Division 4W on the Scotain Shelf, Canada between 1959–79. The traingles joined by a solid line are female data and the circles joined by a dashed line are male data, the dotted line indicates no data available and the vertical bars indicate 95 % confidence limits. (Data from Beacham 1983a).
The haddock has been heavily exploited in the demersal fisheries of the northwest Atlantic. Catches of haddock from the St Pierre Bank off Newfoundland rose rapidly to a peak of 58 000 tonnes in 1955, based largely on an exceptional year class in 1949. This large catch was followed by a decline to around 6 000 tonnes in 1957 and further reductions to less than 1 000 tonnes in the 1970s (Templeman and Bishop 1979a). Biological data collected between 1948–51 and 1969–75 show a decline in the mean age at 50% maturity (Fig.2.) from 4.6 to 3.3 years in males and from 5.9 to 4.3 years in females (Templeman and Bishop 1979b). Beacham (1983b) has pointed out that this decline occurred over a period of both increasing and decreasing growth rates, so that the change in age at maturity is not simply related to changes in growth rate due to the compensatory effect of reduced biomass, but is most likely genetic.
Figure 2. Trends in the mean age at 50% sexual maturity for male and female haddock on St Pierre Bank, Canada between 1948–75. The vertical bars represent the 95% confidence limits. (Data from Templeman and Bishop 1979b).
On the Scotian Shelf haddock catches peaked in 1965 at more than 50 000 tonnes and subsequently declined to less than 11 000 tonnes in 1967. Over the same time period the median length at sexual maturity was relatively constant but the median age at maturity declined between 1959–64 and 1975–79 from 3.9 to 2.7 years for males and from 4.4 to 3.0 for females (Beacham 1983c).
On the Grand Bank in the Northwest Atlantic the catch of plaice Hippoglossoides platessoides increased during the 1950s and peaked at around 90 000 tonnes per annum in the 1960s before declining to around 45 000 tonnes (Pitt 1975). Over the period 1959–72 the mean size at age increased gradually, while the age at sexual maturity declined (Pitt 1975). Average length was correlated with stock abundance, but not water temperature, leading Pitt (1975) to conclude that the changes in size at age were to a large extent due to a reduction in density. The cause of the decline in age at maturity is less clear as there are regional variations in the age at maturity; in some areas faster growing fish mature at a greater age than slow growing fish from other areas (Pitt 1975). Fishing is likely to remove a higher proportion of the faster growing fish, producing a decline in the age at maturity but which was not due to a density dependent increase in growth rate. Beacham (1983b) has presented evidence for a decline in the median length and age at sexual maturity for plaice (Fig. 3) and yellowtail flounder Limanda ferruginea in several fisheries in the northwest Atlantic. This decline in both size and age at maturity is not easily accounted for by an increase in growth rate due to the compensatory effect of reduced biomass (Beacham 1983b), but is likely to reflect a genetic change in the population.
In the yellowfin sole Limanda espersa fishery off the Russian Federation the size and age at which 50% of the females reached maturity declined from 29.9 cms and 8.5 years to 27.2 cms and 7.2 years over the period 1961–69 (Tikhonov 1977). An increase in fecundity in fish of the same size was recorded over the same period.
In contrast in the North Sea fishery an increase in growth rate in the sole Solea solea was related to food availability from trawling activities rather than a reduction in stock density (de Veen 1976). Size at maturity remained constant for a long period and then increased, while age at maturity did not change over the same period. Likewise Rijinsdrop et al. (1991) found no evidence for density dependent growth in adult sole and plaice Pleuronectes platessa over a 30-year period of high exploitation. In plaice the age at maturity, but not the size at maturity, varied over the period 1958–88 and was positively correlated with growth rate in juveniles. In sole there was no relationship between juvenile growth rate and age at maturation, and for both plaice and sole there was no consistent trend in growth rate or size and age at maturity (Rijinsdrop 1991).
Figure 3. The median age at sexual maturity for Atlantic plaice caught in Fishery Division 4W on the Scotian Shelf, Canada between 1964–80. The squares joined by a solid line are female data and the circles joined by a solid line are male data, the dotted lines indicate no data available ant the vertical bars indicate 95 % confidence limits. (Data from Beacham 1983c).
Observations on natural populations of rock lobsters suggest that the onset of sexual maturity is age and not size related. When the breeding stock of Panulirus longipes cygnus from the west coast of Australia is at low density the size at first breeding and upper size range of females is higher (Chittleborough 1979). Under crowded (unexploited) conditions the growth rate is retarded and the onset of sexual maturity occurs at smaller size (Chittleborough 1979). In Jasus lalandii in the Benguela ecosystem off South Africa the growth rates and the size at sexual maturity are greatest where food availability is high (Byers and Goosen 1987). A reduction in size at onset of egg production and an increase in asymptotic length was reported in the spiny lobster Panulirus marginatus from Hawaii following heavy fishing pressure on a virgin stock (Polovina 1989). This decline in size at onset of egg production in P. marginatus is in the opposite direction to that predicted from density relationships and may indicate a genetic change in the population.
The pandalid shrimps are protandrous hermaphrodites (reproducing first as males) in which the age of first breeding and subsequent sex reversal is related to individual size (Charnov 1979). In the shrimp, P. borealis, fishery in the Skaggerak a reduction in size of females occurred over a period of heavy exploitation (Jensen 1965, 1967). The combined Danish and Swedish shrimp catch more than doubled between 1956 and 1960 while the percentage of large shrimps, predominately females, declined from 44% in 1944 to 14% in 1961. More surprising was the appearance of a small size class of females (Table 1), approximately the same size as the one-year old males. Small females (< 75mm) were not present in 1953, but by 1961–62 formed 21–30% of the catch (Jensen 1965). Charnov (1981) has suggested that increased fishing mortality on the large females has selected for individuals that mature as female at the age of first breeding.
For species with an extended spawning period and associated seasonal patterns of behaviour and movement, fishing activities may inflict heavier mortality on one component of the resource. In seasonal and unrestricted fisheries there is a temptation for commercial operators to take as much of the catch as soon as the season opens. This may put greater pressure on early returning individuals and favour late returners. The first arrival and subsequent spawning of Atlantic herring C. harengus on the Norwegian coast has changed over a 60-year period (Devold 1963). Around the turn of the century herring first returned to the spawning grounds in September–October, but this return date has been delayed progressively so that by the 1950s herring did not appear on the spawning grounds until January (Devold 1963). Mathisen (1989) has suggested that this change is due to the effects of fishing which has selectively harvested the first returning sub groups of herring, so that with time these contribute less to the fishery and are replaced serially by later returning sub-groups.
|average length (mm)|
|Males and transition||74||69||68||67|
|Males and transition||65||81||85||83|
This interpretation of Devold's observations must be balanced against non genetic changes which may have occurred over the period. Changes in the spawning time of other groups of Atlantic herring have been explained by changes in environmental conditions. The herring is a phenotypically plastic species which has been subdivided into a large number of taxonomic groups based on morphometric characters and spawning periods and localities (Parrish and Saville 1965). The heritability of spawning time in herring is unknown, but there is evidence for a strong environmental component. In the Baltic, herring spawn between the spring and autumn (Aneer 1985) with the less fecund spawning in the spring and the more fecund in the autumn (Anokhina 1971). The disappearance of autumn spawners has been linked to improved feeding conditions, due to eutrophication, whereby adults have sufficient food reserves to spawn in the spring (Aneer 1985). In the northwest Atlantic Ocean, autumn- and spring- born herring, identified by otolith characteristics, have been found spawning in the opposite season (Messieh 1972).
In an extension of this selective spawning-fishery hypothesis Mathisen (1989) has suggested that the decline of the Peruvian anchoveta Engraulis ringens was accelerated by disruptive selective fishing on the “best reproductive units” leaving only the marginal groups to rebuild the stocks.
In the Pacific salmon fisheries of British Columbia most fish are caught as mature adults when they return to rivers for spawning. In this respect the fisheries are less likely to demonstrate the ‘fishing up’ effect seen in trawl fisheries for marine teleosts. A detailed study of five species of Pacific salmon concluded that at least three species exhibited a decrease in mean size at age following years of size selective fishing (Ricker 1981). The gillnet and troll fisheries for coho Oncorynchus kisutch and pink salmon O. gorbuscha tend to catch the larger fish and these species exhibited the greatest declines in mean sizes (Ricker 1981). The chinook salmon O. tshawytscha has decreased in both mean size (Fig.4) and age between 1951 and 1975 (Ricker 1980). Ricker (1981) suggested that this decrease in size and age was due to the troll fishery capturing both maturing and nonmaturing individuals which selects against late maturing individuals. However, the data are complex as the fishing methods and mesh sizes have changed over the sampling period and for O. nerka there was an increase in growth rate over the period which was related to oceanic cooling (Ricker 1981).
In the coho O. kisutch males mature after six months in the ocean as jacks or after 18 months as hooknose. These two alternative life history strategies are maintained by disruptive (natural) selection favouring small “sneaky” males and large “fighting” males (Gross 1985). The fishery has selectively harvested the larger fish (Fig.5) increasing the relative frequency of jacks (Gross 1991). However, Gross (1991) has shown that changes at other stages of the life cycle can influence the ratio of jacks to hooknose males. Stream clearance may reduce the available refuges for jacks, thereby favouring hooknose fish on the spawning grounds. Other environmental changes, such as eutrophication, may increase fry growth rate leading to an increase in the proportion of juveniles maturing as jacks (Gross 1991).
The Atlantic salmon S. salar exhibits wide variation in the mean age at first spawning between different river systems. It is believed that such variation is adaptive and maintained through natural selection (Schaffer and Elson 1975). Fishing pressure on returning adults has eliminated the larger and older fish from some runs leading to an early age of first returning fish (Schaffer and Elson 1975).
In species with a long spawning season unequal fishing pressure over the season may produce a selective mortality. In the Columbia River chinook salmon Oncorhynchus tsawytscha the spawning run lasted from April through to August with a peak in June-July when the fishery first developed last century (Thompson 1951). The fishery operated mostly over the summer when weather was suitable and the fish most abundant. Heavy fishing pressure on the peak run reduced catches so that by 1938 the overall numbers caught were reduced and the spawning run peaked early in May and in August (Fig. 6) with few fish caught in the original peak period of June-July (Thompson 1951). Mathisen (1989) has interpreted these observations as disruptive selection, selecting for both early and late spawners, which have not replaced the middle peak period spawners.
Figure 4. The median weights for troll-caught chinook salmon off British Columbia, Canada. (Data from Ricker 1981).
Figure 5. The effects of size-selective fishing on the relative fitness of hooknose and jack male4s of coho salmon. The selective removal of hooknose males increases the percentage of jacks in the fishery, pl* no fishing, p2* and p3* increasing levels of fishing. (Data from Gross 1991).
Figure 6. Seasonal changes in catches of chinook salmon in the Columbia River. A. Daily gillnet catches in 1876 and B. catches in 1938. (Data from Mathieson 1989).
Ricker (1982) has described a disruptive selection for size in the sockeye salmon Oncorhynchus nerka. The gillnet fishery has harvested fish from the mid size range favouring the survival of small 3-year ocean fish and large 4-year ocean fish. As a result the 3-year old fish have become smaller and the 4-year old fish larger, the difference between them increasing by about 500g (Ricker 1982).
Many lake fisheries use gillnets which selectively remove large heavy fish, producing a non random mortality with respect to size (Hamley 1975). In Canada whitefish are abundant and the lake whitefish Coregonus clupeaformis has supported fisheries in several lakes. A comparison of exploited and unexploited populations of lake whitefish showed the typical compensatory effects of fishing where growth rate increased with exploitation (Healey 1975). Growth rate in heavily exploited populations was similar to the most rapid growth rate in unexploited populations (Healey 1975). However, in heavily exploited populations fish matured at an younger age and smaller size than fish from unexploited populations.
In Lesser Slave Lake in Alberta the fish resources have been heavily exploited. The lake trout Salvelinus namaycush was fished out by the 1920s and other species have declined in abundance (Handford et al. 1977). Samples of C.clupeaformis were analyzed from commercial and research catches between 1941 and 1975. Following an initial increase in mean weight and length there was a decline in both length and weight at given age. By the 1970s fish of a given age were less than half the weight of fish of the same age in the 1940s, although mean length at age was about the same in the 1970s and 1940s (Handford et al. 1977). Similar observations have been made in other lakes (Handford et al. 1977). In addition the catches in Lesser Slave Lake showed an overall increase in mean age, a result to be expected under moderate levels of exploitation when selection favours slower growth rates and overcomes the counteractive effects of density dependent compensation (Handford et al. 1977). Lake fisheries for percids have shown similar patterns of an increase in growth rate and reduction in age of first spawning (Spangler et al. 1977).
The fish populations in Lake George, Uganda were studied between 1967 and 1972 where the cichlid Tilapia nilotica accounted for around 80% of the fish catch. Fishing has been the major source of adult mortality for this species which decreased in mean size from about 0.9 kg in 1950 to 0.4 kg in 1970. Over the same period the size at maturity decreased from 20–29cm to 18–24cm (Gwahaba 1973).
In Lake Femund in Norway the polymorphic whitefish consists of three morphs. A pelagic gillnet fishery started in the 1980s has selected against the pelagic morph which has decreased in proportion to the two other morphs and has shown a decrease in the proportion of large fish and a reduction in size at age (Sandlund and Naesje 1989).
In the 1980s a new fishery developed for orange roughy off the coasts of New Zealand and during the first ten years of exploitation the biomass was reduced by 60–70%. Genetic diversity, measured by gel electrophoresis, showed a significant decrease on three geographically isolated fishing grounds (Table 2) between 1982–83 and 1988 (Smith et al. 1991). The recent finding of new spawning grounds provides a further opportunity to test for genetic effects of exploitation on virgin stocks. Preliminary results from allozyme surveys indicate levels of genetic diversity similar to the pre-exploitation levels reported in 1982–83 (Smith unpub.). In a similar situation with the spiny lobster Panulirus marginatus no genetic changes were detected over a nine-year period following heavy exploitation of a virgin stock (Seeb et al. 1990).
|Fishing ground||Year of sampling||Mean heterozygosity||Probability|
|Chatham Rise||1982||0.20||< 0.02|
|East coast||1983||0.25||< 0.02|
|Challenger Plateau||1983||0.22||< 0.01|
In sockeye salmon populations spawning in isolated lakes on the Kamchatka Peninsula the young spend one to three years in the lakes and then migrate to sea for a further one to four years, before returning as sexually mature adults. The males mature at two different size: as large males after two or more years at sea, and as small jacks or “kayiriki” after one year at sea (Altukhov 1990). In addition some lake residual males mature as “dwarfs” (Altukhov 1990). This sexual dimorphism is similar to that reported in coho salmon O. kisutch (Gross 1985). Large sockeye males have an advantage over small males on the spawning grounds, but small males are more successful in years of low water levels and in shallow spawning areas that the large males cannot penetrate (Altukhov 1990).
Over a 35-year period between 1935–76 the numbers of sockeye returning to Lake Dalneye, on the Kamchatka Peninsula, declined dramatically (Altukhov 1990), while the proportion of jacks increased in the returning males and the proportion of resident dwarf males in the population increased dramatically (Table 3). More recent allozyme studies of sockeye populations in three Kamchatkan lakes (Table 4) have shown that heterozygosity measured at two enzyme loci is greater in small males (jacks) than in large males (Altukhov 1993). Allele frequencies are similar between the two groups but small males are characterised by a significant excess of heterozygotes at the phosphoglucomutase (PGM) locus and large males a significant excess of homozygotes in two lakes. Heterozygosity in females is intermediate between the large and small males (Altukhov 1990). Altukhov (1990) suggests that these changes in heterozygosity are due to the selective effects of fishing. While fishing may have favoured the survival of small males, as may have changes in the freshwater environment, there are no historical data available on allozymes for the two size classes of males. In three Kamchatkan lakes the level of heterozygosity at PGM increases with the proportion of small males in the lake (Altukhov 1990).
Kirpichnikov et al. (1990) have shown that during the fry stages slow growing individuals are less heterozygous than fast growing individuals. If these fast growing individuals mature early as dwarfs or as jacks, and in salmonids fast growth is associated with early maturity, then this would account for the observations of Altukhov (1990). The overall population heterozygosity has increased due to the indirect effects of the fishery favouring the survival of small males.
O. mossambicus is widely used in tropical pond aquaculture. It is a mouth breeder and is easily reared in tanks. To measure the impact of selective fishing two populations were established and, after 39 months, harvested at two monthly intervals (Silliman 1975). Harvesting removed approximately 10% of the fish: one population was selectively fished by removing only fish that could not pass through a grid placed in the tank. The second control population was unselectively fished by removing a similar number of fish but from all size classes. After 63 months the fishing pressure was increased to 20% and continued for a further 12 months. To test for a genetic impact of fishing groups of 46 mature fish were removed from the selected and unselected populations after 77 months and ongrown for a further period. Growth was measured at 55–56, 118–119, and 150–151 days after the start of the second experiment. In both the selected and unselected populations males grew faster than females, but the males in the unselected population grew more rapidly than males in the selected population (Silliman 1975). Thus selective fishing resulted in a decline in growth rate in males in only three generations. It is likely that females were less affected by the size selection due to their slower growth rate than males.
|lake (10 000s)|
|Percentage of jacks|
|in migrating males||0.2||0.6||4.3||37.5|
|dwarfs in male||26||49||74||89|
* significant genetic imbalance
Life history differences between populations of guppies are associated with predation: some predators (cichlids)prey on large sexually mature guppies and other predators (killifish) prey on small immature guppies. A population of guppies was transplanted from a river with a cichlid predator to a tributary site with no guppies but killifish predators. Over a period of 11 years the transplanted population produced larger offspring and larger females (Fig. 7) which first reproduced at a larger size than females in the control population (Reznick et al. 1990). Rearing groups of descendants from the two populations under similar laboratory conditions showed that the observed life history differences had a genetic basis (Reznick et al. 1990).
Figure 7. Size of offspring and average size of sexually mature females in transplanted and control samples of guppies over an eleven-year period on the Aripo River in Trinidad. Circles joined by a solid line are the introduced site and triangles joined by a dotted line are the control site. At the control site the main predator preyed predominantly on large guppies and at the transplant site the main predator preyed on small guppies. (Data from Reznick et al. 1990).
A laboratory experiment was carried out with the water flea D.magna to measure the impact of two contrasting harvesting regimes, removing small individuals and removing large individuals (Edley and Law 1988). At the first harvest 50% of the stock was removed and there after 40% removed every eight days. Yields declined under both harvesting regimes, but were greatest in populations harvested for large individuals. Under the harvesting regime of removing small individuals the mean size at age increased as did the size at first reproduction, while under the harvesting regime removing large individuals, the reverse occurred with a decline in size at age and decline in size at first reproduction (Edley and Law 1988).
Menshutkin et al. (1989) produced a simulation model of the sockeye salmon Oncorhyncus nerka fishery in Kamchatka. When growth was controlled by a single gene with two alleles the model population became unstable with loss of one allele, supporting observations that growth is under polygenic control. When growth was polygenic and multiallelic then the percentage of “fast growth” (which equate to early maturity) alleles increased in the selected population and the proportion of non migratory fish increased. These results would be expected from observations in the fishery where there has been an increase in the proportion of jacks and lake-resident fish (Altukhov 1990). Favro et al. (1979, 1982) used a simulation model to estimate the magnitude of genetic effects in a trout fishery subject to minimum size limits and in which growth was controlled by a small number of major genes. Results showed that mean size and total numbers decreased with moderate levels of fishing pressure and were in agreement with observations in a brown trout Salmo trutta fishery in Michigan. Extending the model to select for fish from a specific size range, whereby a minimum and a maximum size limit was set in the fishery, then the model showed that the double-size limit produced a similar decrease in larger fish as did conventional minimum size limits (Favro et al. 1980).
Law and Edley (1989) have used an age specific model to describe exploitation which acts as a selective force on genetic variation in life history characters. Selective fishing will lead to changes in life histories that in turn produce evolutionary changes in yield. In principle the fishery manager may opt for the harvest pattern that will select for an optimum life history which will produce the greatest yield, which the authors refer to as the “evolutionary stable optimal harvesting strategy” (Law and Edley 1989). Applying this model to the Arcto-Norwegian cod Gadus morhua, which has shown a decline in age at maturity, then an increase in yield could be obtained by selecting for late maturing fish. In practice this could be achieved by restricting fishing based on immature fish in the northern feeding grounds and focusing effort on the spawning fisheries (Law and Edley 1989).
A large-scale experimental management approach to test the heritability of growth rate and the effects of size selective fishing on a natural population of pink salmon Oncorhynchus gorbuscha has been proposed by McAllister and Peterman (1992) with the overall aim of increasing catch biomass. This species, as with other west coast salmonids, has shown a reduction in mean size over the past twenty years (see section g above and Ricker et al. 1978) and there is uncertainty if the changes are due to selective removal of larger fish, to selective depletion of stocks with large body size, or to changes in oceanographic conditions (McAllister and Peterman 1992). Controlled selective harvesting of small fish would permit a test of the heritability of growth rate and, provided that growth rate is inherited, lead to an increase in mean size in the fishery. Monte Carlo simulations showed that block designs with few replicates and short duration times generated high statistical power for determining the importance of size selective fishing. In an extension of the experimental design McAllister and Peterman (1992) used a decision analysis to demonstrate that the experimental approach to management of the pink salmon stocks was likely to produce a higher harvest value than the current management practice.