4. BIOLOGY OF NAKED CARP

4.1 Introduction

The naked carp Gymnocypris przewalskii (Kessler) is endemic to Qinghai Lake, where it has been isolated since tectonic movements severed the connection between the lake and the upper Yellow River (Huanghe) some 130,000 years ago (Wu 1987). G. przewalskii has diverged in isolation, but retains a close relationship to G. eckloni of the Yellow River (Wu 1984). Like other Schizothoracinae, these species lack scales except in two columns around the anus, forming a split behind the ventral fin (G. przewalskii may also have small post-opercular scales).

Apart from G. przewalskii, the fish fauna of Qinghai Lake includes four species of loach (Cobitidae: Nemacheilinae), namely Triplophysa alticeps, T. microps (syn. T. dorsonatus: Wu in press), T. scleropterus and T. stoliczkae. These appear to be related to species in the Yellow River, but their biology is unknown and they are not considered further here.

This section draws upon a review of the biology of G. przewalskii, prepared as part of the present investigation (Yang & Walker 1993a). It considers earlier studies of the biology of G. przewalskii, and incorporates new data.

4.2 Evolutionary History

During the Quaternary, as the Himalayas and other mountain ranges were uplifted, the low-altitude, tropical or subtropical conditions prevailing on the Qinghai-Xizang Plateau were replaced by the present high-altitude, cold, arid continental climate (Wu 1980). There were many extinctions of fish and other biota, leaving a small relict fauna. At this time the Schizothoracinae probably arose from surviving species of Barbinae (Cuvier 1840, Hora 1953, Chao et al. 1962, Das 1963, Mirza 1975), through forms represented by the Pliocene fossil Plesioschizothorax macrocephalus (Wu et al. 1980). The case to regard the primitive barbines as evolutionary progenitors of the Schizothoracinae rests upon apparent similarities in morphology, palaeobiology and zoogeography (Chao 1981, Wu 1984). The two groups have at least five common characters: (1) caudal fin without hard ray, (2) olfactory organ with large hollow, (3) quadrate absent, (4) vertebrae 2–3 fused, with compound neural spine, and (5) basioccipital not flat behind pharynx.

The Schizothoracinae includes 11 genera in three evolutionary groups (Wu 1987). The most primitive group, represented by the Schizothorax subgenera Racoma and Schizothorax, is distributed mainly at lower altitudes (1200–2500 m), in peripheral areas of the Qinghai-Xizang Plateau. These species retain uroneurals, which are present in barbines but absent in the more specialized forms of Schizothoracinae (Wu 1984). An intermediate group, including Ptychobarbus, occurs at middle altitudes (2750–3700 m) and a third more specialized group, including Gymnocypris and Schizopygopsis, is found in high-altitude areas (4000–4750 m) near the centre of the plateau. Members of the third group are distinguished by the absence of barbels and the presence of a canalis preoperculo-mandibularis.

4.3 Taxonomy And Distribution

The Schizothoracinae are distinguished from other Cyprinidae (Cypriniformes) by lacking scales apart from two columns on either side of the anus, forming a split behind the ventral fin. Wu (1984) listed 25 diagnostic characters: (1) scales minute; (2) two pairs of barbels; (3) lower lip bi-lobed only at corners of mouth; (4) lower jaw without horny sheath; (5) upper jaw over-reaches lower jaw and mouth, which is subterminal; (6) ventral fin originates opposite to or slightly behind vertical line of dorsal fin; (7) posterior ray of anal fin in mature males soft and not hooked; (8) three rows of pharyngeal teeth; (9) pharyngeal teeth hooked; (10) prepalatine present; (11) second pre-ethmoid of layers of connective tissue; (12) dorsal rim of maxilla broad and flat, with premaxillary process directed forward and down and ethmoid process clear; (13) detary triangular with slender anterior rim, similar bill, or pre-lateral part curves; (14) long, narrow canalis preoperculo-mandibularis; (15) parasphenoid horizontal slightly prominent in middle and directed downward; (16) entopterygoid on upper forward part of quadrate, upwardly oblique and fan-shaped; (17) supratemporal present; (18) opisthotic on ventral face of skull, not joined with post-temporal; (19) post-subtemporal fossa absent; (20) third neural spine broad with straight dorsal margin; (21) at least 42 vertebrae; (22) five suborbital ossicles; (23) cleithrum with narrow dorsal rim; (24) uroneuralia present; (25) skull short and broad.

In the genus Gymnocypris the ventral fin originates posterior to the dorsal fin, and there are seven suborbital ossicles. G. przewalskii is entirely free of scales apart from the two distinctive columns on either side of the anus. Since 1876 there have been seven recorded descriptions of the species in Qinghai Lake, namely Schizopygopsis przewalskii (Kessler 1876), Gymnocypris przewalskii, G. roborowskii and G. leptocephalus (Herzenstein 1891) and G. chengi, G. depressus, G. chinghainensis and G. convexaventris (Zhang & Zhang 1963a, 1963b). These forms were synonymized as G. przewalskii (Kessler) by Zhu & Wu (1975), who further distinguished two subspecies, G. p. ganzihonensis and G. p. przewalskii. The former occurs only in the Ganzhi River, isolated from Qinghai Lake by the receding lake level in the 1960s, and is distinguished from G. p. przewalskii on electrophoretic and morphological grounds (Zhu 1975). The main morphological distinction is the smaller number of gill lamellae: the gill arch of G. p. ganzihonensis has 28 inner lamellae and 18 outer lamellae, whereas G. p. przewalskii has 46 inner and 30 outer lamellae. The respective ratios are 1.56 and 1.68, whereas Wu (1984) proposed 1.28 as a threshold for subspecific status (cf. Meil 1965).

The Schizothoracinae is confined to the rivers and lakes of the Qinghai-Xizang Plateau and adjacent areas of central and southern Asia, and more than 80% of known species are from China (Wu 1987). Gymnocypris occurs only on the plateau and, as noted, G. przewalskii occurs only in Qinghai Lake and affluent streams. In the following, unqualified references to G. przewalskii should be taken to refer to G. p. przewalskii.

4.4 Salinity Tolerance

As part of the Qinghai Lake fishery investigation, a series of trials was undertaken to estimate the upper limit of salinity tolerance of naked carp. This was of interest in view of the trend towards increasing salinity in Qinghai Lake, and the potential for translocations of naked carp to high pH saline lakes elsewhere in China. Juveniles are likely to be more sensitive than adults, and thereby a better indicator of the effect of salinity on populations.

Fingerlings of naked carp were obtained from the hatchery ponds maintained by the Bureau of Aquatic Products, Xining, in May 1992. A stock solution of 22 g L^-1 was prepared by slow evaporation of water from Qinghai Lake (12.5 g L^-1), and test solutions were made by dilution with distilled water. Groups of 10 fingerlings were assigned to treatments of 16,18, 20 and 22 g L^-1 (determined by a pilot experiment), and to two controls, one with salinity corresponding to that of the lake and the other containing water from the hatchery pond. Both treatment and control solutions were replicated three times, with 4 L of solution allocated to each group of 10 fingerlings. The trials were conducted in a laboratory at water temperatures from 13–14.5°C. Fingerling survival was monitored at frequent intervals over 130 h. After probit transformation (Finney 1971), the data indicated an LD₅₀ (Lethal Dose to kill 50% of the population) of 18.4 g L^-1, with 95% confidence limits 15.9 and 21.2 g L^-1.

The figure of 18.4 g L^-1 is not far beyond the present lake salinity, and if this increases as a result of the falling level, the survival of the species in Qinghai Lake could be threatened. The LD50 result takes no account of the ability of the species to acclimate to steadily rising salinity, however, and it should be regarded as conservative.

Two incidental notes pertaining to salinity tolerance are that naked carp from Qinghai Lake have been successfully translocated to Huangqi Lake in Inner Mongolia, where the salinity is about 18 g L^-1 (section 4.10), and that the species has disappeared from Gahai, isolated from Qinghai Lake by the falling lake level in the 1960s. Gahai now has a salinity of about 40 g L^-1.

4.5 Feeding

The naked carp has three modes of diet (Wang 1975):

1. Invertebrates including zooplankton (mainly copepods) and midge larvae (Tendipes), with few algae and little sediment

2. Algae, mainly diatoms, with large amounts of sediment and few zooplankton.

3. Mixed invertebrates and algae, with large amounts of sediment

Although the species is omnivorous its diet changes seasonally, apparently in response to changes in plankton biomass and food availability (Wang 1975). In May the algal content is maximal and invertebrate content is minimal. The proportion of algae falls steadily from the beginning of June, as copepods and midge larvae become more available, but rises again in February or March of the following year.

A list of planktonic and benthic organisms recorded in the diet is shown in Table 4.1. There are also occasional remains of macrophytes, loaches and naked carp fingerlings, and the prevalence of sediments in the gut suggests that the fish may derive some nutrition directly from sedimentary organic matter.

The diet varies also with the age and size of the fish. There are three ontogenetic stages, corresponding to morphological differences in the digestive organs:

1. Up to 15 mm body length, the main diet is small zooplankton, including rotifers, small chironomid instars and copepodites.

2. From 15–45 mm length, the fish feed upon copepods and larval and adult insects.

3. Beyond 45 mm length, the fish become omnivorous, as described above.

In the first two stages the main diet is small animal prey, although there may be small amounts of algae. With increasing body size the digestive organs change, particularly by increases in the length of the intestine and the number of intestinal loops. Fish in the first group have a straight intestine; those in the second group have two loops and adults have four loops. The mid-region of the intestine grows fastest, expanding its absorptive area.

Table 4.1
Organisms in the diet of naked carp in Qinghai Lake (mainly after Wang 1975).

Fish smaller than 35-mm body length have loose gill filaments which allow comparatively large prey to be taken. Gill arches form when the fish attain a body length of 35 mm, and at 45 mm the gills have developed a network of lamellae that acts like a filter, allowing more efficient feeding on plankton.

Feeding activity changes seasonally. Peak activity is in summer (June–August); there is little change through autumn and winter but a fall in spring, when the fish must increase their energy intake to sustain increased activity. This is commonly the case for omnivorous fish. The spawning population shows a contrary pattern in which feeding falls to a minimum during the spawning season (May–June) before recovering in July.

4.6 Reproduction

4.6.1 Sex Determination

The sex of naked carp is readily determined from physical characters, particularly the shape and appearance of the anal and dorsal fins (Hu 1975). The anal fin of the female is oval, smooth-sided, continuous and slightly pointed. In the mature female the ends of the rays of the anal fin become thickened (especially the second and third undivided rays). The anal fins of male fish have an obvious gap along the edge, and some have an inverse hook on the last divided hard ray. In males also there is less thickening of the anal fin rays with maturity. Finally, there is a breach in the edge of the dorsal fin of the male, but not the female.

In the Qinghai Lake population, females typically outnumber males by 3–4 to 1. For example, in an aggregate sample of 506 fish taken from the commercial trawler catch in 1990–92 (section 4.7.2), the proportion was 401 females and 105 males, or 3.8:1.

4.6.2 Gonad Development

Gonad development is in six stages:

4.6.3 Early Development

At a mean temperature of 15.5°C (range 13.5–18.0°C) and pH 8.0, the first cell division is completed 3.5 h after fertilization. The blastula is formed in 20 h, the gastrula in 34 h and the pluteus in 45 h, and the blastopore is closed by 48 h. The fry hatch after 138 h and begin feeding 11–15 days later. The hatchlings gradually move or drift out of the nest and congregate in small pools or along the stream edge, where there is shelter from the current.

Experiments by Guong (1979) showed that the longevity of sperm was 1 min 19s in distilled water, lmin 40s in Buha River water, 18min 20s in lake water, 20min 50s in 0.65% Ringer's solution and 16min 30s in 0.65% NaCl solution. The success of artificial reproduction obviously depends on sperm longevity, and Guong's data suggest that relatively saline solutions are appropriate for maintaining the highest possible fertilization rate under culture conditions. It is curious that sperm longevity is much greater in the lake water (12.5 g L^-1) than in water from the Buha (<0.5 g L^-1), where most spawning occurs. It is not known whether there is significant spawning within the main body of the lake.

4.6.4 Spawning Behaviour

G. przewalskii is a migratory, anadromous spawner. The spawning stock, typically of individuals 150–250 g weight, congregates around the mouths of rivers in spring and moves upstream after floods that bring snowmelt. This usually begins in mid-late May (sometimes as early as April) and reaches a peak in June or July. The Buha River is the main spawning area, although the Shaliu, Haergai and other rivers are also significant. It is not known whether distinct stocks are associated with the different streams.

The Buha is unaffected by weirs like those on the Haergai, Qianji and Shaliu, and the effective spawning area extends 15–50 km upstream of the river mouth. In recent years the migration in the Buha has commenced on 19 May 1988, 10 May 1989, 15 May 1990, 26 May 1991 and 19 May 1992. Figure 4.1 suggests that there is no simple correlation between the onset of the migration and river discharge per se, although the flow data are monthly averages and must therefore obscure short-term fluctuations. Water temperature, photoperiod and salinity also may be implicated.

The migrating fish may cover distances of 40–50 km over several weeks, given adequate flows. At times when flow in the Buha has declined, large numbers of fish have been confined to isolated pools and fish kills have occurred. The fish are able to move against reasonably strong currents, and probably can swim at speeds of 50–100 cm s^-1 with burst speeds of perhaps 200 cm s^-1. The fish are reputedly able to leap small obstructions, perhaps 50 cm in height

Spawning occurs at water temperatures between 6–17.5°C. The eggs are non-adhesive (or very slightly adhesive), and released over a substratum of fine sand or gravel into water that is clear, slow-flowing (100 cm s^-1) and up to a metre deep. The eggs are laid into a 10-cm deep, 30-cm diameter depression. The density of these ‘nests’ in the stream bed may be as great as one per 5–10 m². The eggs are not covered, and both eggs and fry remain vulnerable to changes in the strength of the current over the following 1–2 weeks.

4.6.5 Fecundity

Although female naked carp come to maturation only once each year they are serial spawners, releasing their eggs over an extended period. There probably is some readsorption of un-shed eggs, judging from the large numbers remaining in the ovaries of some fish caught at the end of the season. Fecundity is related to body length and weight, and averages 16,242 or 28.75 eggs per gram of female body weight (Hu 1975). The fecundity of fish 48–51 cm in length is 6.3 times that of fish of 20–23 cm length, and fish weighing 1051–1250 g produce 4.2 times as many eggs as fish of 51–250 g. In either sex, maturity is attained at about 7 years, although there are indications of some fish having matured by 5 years.

4.7 Growth

4.7.1 Methods

In the course of this investigation population samples were obtained directly from the commercial trawler catch and using monofilament fleet nets at various locations, particularly near the Buha mouth. More data were obtained than are presented here (for example, records of gonad weights, gonadal states and the diameters of scale annuli); these will be useful in future analyses of the dynamics of the population.

The ages of individual fish may be determined from inspection of scales (from the anal area), vertebrae, hard rays and opercula. The scales show well-defined annuli (although they become indistinct in fish older than 10 years), and may be used to monitor the age structure of the trawler catch.

4.7.2 Length and Weight

Weight-length regressions for 506 fish sampled from the trawler catch are shown in Figure 4.2. The database represents 15 discrete samples during the trawling seasons of 1990–92. For females the regression equation is:

W = 0.01 L^3.03 (r² = 0.97, n = 401)

and for males:

W = 0.02 L^2.92 (r² = 0.91, n = 105),

where W is total fresh weight (g) and L is Standard Length (cm). Note that in Figure 4.2 the last two decimals in each coefficient and exponent are spurious because weight was measured to the nearest gram and length to the nearest 0.5 cm.' Note also that the disparity in numbers of females and males is a true reflection of their relative numbers in the trawler catch (section 4.6.1).

The slopes of the two equations are not significantly different (ANCOVA: F_l,502 = 1.921, P >0.05). If the data for males and females are pooled, the combined equation is:

W = 0.01 L^3.02 (r² = 0.93, n = 506).

This is similar to that reported by Zhao (1975), literally:

W = 0.012811 L^3.0285.

The relationships of age to length and weight are shown in Figures 4.3a-b and 4.4a-b. The degree of scatter and the nature of the data (multiple values of L and W for a given age) preclude the use of simple linear regression equations to describe the relationships (although more sophisticated models could be applied). In general, growth is relatively rapid for the first 4 years and thereafter slow and gradual, without distinctive stages, so that it is difficult to distinguish year classes. The variability and indistinctness of the cohorts are surprising in view of the strongly seasonal nature of the environment, but presumably are related, at least in part, to serial spawning. Individual fish take about seven years to attain a marketable size of 300 g and 30 cm, and about 10 years to attain 500 g and 36 cm. For fish over 4 years the average annual length increase is about 15 mm for males and 14 mm for females, and the corresponding weight increments are 55 g for males and 60 g for females.

4.7.3 Growth Curve

The data do not allow for precise definition of a growth curve, but it is of interest to attempt to fit the length vs age data to a von Bertalanffy curve, as follows:

l_t = L_∞(1-exp(-k(t-t₀)))

where l_t is the length at time t years, and L_∞, k and t_o are model parameters. Figure 4.5 shows a plot of length vs age (pooled data), with the von Bertalanffy model (solid line) fitted by least-squares. The parameters for the fitted equation are as follows:

l_t = 96.02(1-exp(-0.02(t+9.92))) (r² = 0.995, n = 448).

The broken line in Figure 4.5 represents a von Bertalanffy curve determined by Zhang & Chen (1980) for naked carp in Qinghai Lake. Their model is:

l_t = 59(1-exp(-0.07(t-0.14))),

where the parameter values k and t₀ were assumed and the asymptotic length of 59 cm was adopted from the largest fish recorded in earlier work by Zhao (1975). Zhang & Chen incorporated this equation in an application of the Beverton-Holt Dynamic Pool Model to estimate the maximum sustainable yield of the Qinghai Lake fishery. The main difference between the two von Bertalanffy models is in the value of the t₀ parameter, which influences the horizontal displacement of the equation. Clearly, the model of Zhang & Chen is not a good approximation to the present data (their yield estimates are considered in section 5).

The maximum length of 59 cm reported by Zhao (1975) was for a 21-year old female weighing 2700 g. This is considerably larger than any fish recorded during the present investigation, but consistent with the early removal of large fish that is typical of newly-opened fisheries. The largest individuals in samples from the present study are described in Table 4.2. It appears that males may have a shorter longevity than females (14 cf. 21 years). The maximum attainable size under the present fishery regime appears to be a length of about 45 cm and a weight of about 1 kg for males and 1.3 kg for females. Thus, the asymptotic length of 96 cm implied by the von Bertalanffy model fitted to the present data (Fig. 4.5) is unrealistically high.

Table 4.2 The largest and oldest fish of known age in samples (366 females, 82 males) from the trawler catch in 1990–92. The largest male captured (not listed here) was an individual of indeterminate age, 42 cm long and weighing 1080 g.

4.7.4 Cohort Analysis

Monofilament fleet nets were set at various locations in 1991, and generally provided samples of 300–400 fish. Conversion factors were used to correct the numbers of fish caught in each panel to an equivalent 500 m². These samples represent a truer cross-section of the population than the commercial catch, whose composition is influenced by the mesh size of the trawl nets. Ideally, these kinds of samples help to identify year classes, and so demonstrate changes in population structure associated with recruitment

Figure 4.6 shows the composition of samples taken in May, June, July and August 1991. The definition of cohorts (and thereby year classes) is blurred, evidently because the fish are serial spawners, and because individual growth rates are highly variable. Several computer programs (e.g. ELEFAN: Dr D. Pauley, ICLARM, Manila) have been developed to help distinguish cohorts within mixed populations, and these may be useful in further analysis to derive estimates of annual recruitment and mortality. Otherwise, the principal value of these data is as a benchmark for interpretation of future population trends.

4.7.5 Tagging and Marking Investigations

Large numbers of fish were tagged during 1991, and relevant records are held by the Lake Investigation Team. If sufficient numbers of tagged fish are recovered in future, they will provide valuable data, on growth rates.

4.7.6 Fingerling Growth Experiments

Methods for artificial induction of spawning were successfully developed in part of the present project concerned with the feasibility of a naked carp hatchery (see FAO 1992). Fingerlings produced in this way were kept in ponds at the Bureau of Aquatic Products in Xining, and regular measurements were made of length and weight and pond temperature and other environmental conditions. The data have been recorded on a computer database and are included with the records of the project They showed that the growth rates of the fingerlings maintained in the ponds may be at least two times that of fingerlings in the lake, presumably because of the warmer conditions at lower altitude.

4.8 Mortality

The large seasonal bird population associated with Bird Island includes a number of piscivorous species, paticularly cormorants and gulls. The birds feed extensively upon fish, particularly juveniles, during the summer, and are a significant competitor for the fishery. The annual loss to bird predation has been estimated at 700 tonnes (Hu et al. 1975).

Five species of helminth parasites are known from the intestinal tracts of the fish (Yang 1991), and presumably have life histories involving birds as hosts. The species are Ligula sp., Diphyllobothrium sp. and Contracaecum sp., present in the fish as larval stages, and Neoechinorhynchus qinghaiensis and Echinorhynchus gymnocyprii, present as adults. Ligula occurs mainly in fish of length less than 200 mm, whereas the other four species are most frequent in fish of more than 200 mm length. The helminths occur also in at least one of the loach species, Triplophysa scleropterus (section 4.1). There is no evidence that parasitism is a significant agent of mortality among the fish.

Although external parasitic infestations are occasionally seen in trawled fish (especially in the eyes), the afflicted fish are processed as usual. There is no quantitative information about the incidence of parasites, nor about the reactions of consumers to parasitized fish.

4.9 Stock Recruitment

Experimental fishing in the Buha River before the break-up of the ice in 1991 provided evidence that at least some juveniles over-winter in the river. Although the winter ice is about 1-m thick, some flowing water must remain near the bed, perhaps through infiltration of groundwater.

It is not possible to draw conclusions in relation to a number of important issues. For example, spawning success is likely to vary in relation to the effects of annual variations in river discharge (including flow depletion caused by irrigation diversions) and siltation related to channel adjustments. Recruitment is likely to be influenced by bird predation (cf. section 4.8) and the inclusion of under-sized fish in the trawl catch (section 5.5).

Although incidental observations relating to spawning and recuitment are included in other sections of this report, many questions remain unanswered. This is arguably the signle most important target for future research.

4.10 Translocation

Naked carp from Qinghai Lake have been successfully translocated to Huangqi Lake in Inner Mongolia, where the salinity is about 18 g L^-1 (Qu 1988). Huangqi is a large lake (12,000 ha) at an altitude of 1266 m (Lian 1979). In the late 1950s several fish species were introduced to the lake, including bighead carp (Aristichthys nobilis), black carp (Mylopharyngodon piceus), “crucian carp” (Carassius auratus), grass carp (Ctenopharyngodon idella), silver carp (Hypophthalmichthys molitrix) and Wuchang bream (Megalobrama amblycephala). In the 1960s production was about 700–1000 tonnes (Lian 1979), but in 1970 the lake level began to recede as a result of diversions to a local irrigation scheme. In 2–3 years the surface area had decreased to 8000 ha, and the alkalinity and salinity increased markedly. The fish populations died out between winter 1972 and spring 1973, and a project was commenced to rehabilitate the fishery. In 1987, 38,000 naked carp fingerlings were artificially-spawned in ponds (in Inner Mongolia), and transferred to Huangqi Lake. After three years the body weight of the translocated fish had attained 300–400 g, indicating a growth rate approximately three times that in Qinghai Lake. There is no further information to indicate the status of the population.

Western China has many lakes of salinity <18 g L^-1, and there may be scope for establishing other populations of G. przewalskii. The relatively high tolerance of the species may make it an attractive option for establishing fisheries in lakes too saline to support other fish. The main constraint is likely to be that naked carp has low value as a fishery product (section 5.9).