Field Document 6
October 1990


A report prepared for the project
Fisheries Development in Qinghai Province


James F Muir
Aquaculture Specialist

This report was prepared during the course of the project identified on the title page. The conclusions and recommendations given in the report are those considered appropriate at the time of its preparation. They may be modified in the light of further knowledge gained at subsequent stages of the project.

The designations employed and the presentation of the material in this document do not imply the expression of any opinion whatsoever on the part of the United Nations or the Food and Agriculture Organization of the United Nations concerning the legal or constitutional status of any country, territory or sea area, or concerning the delimitation of frontiers.

Rome, 1990

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.




2.1 Background biological data
2.2 Existing hatchery facilities
2.3 Design options
2.4 Hatchery specifications

2.4.1 Layout and design
2.4.2 Heat balance
2.4.3 Pumps and pipes
2.4.4 Filtration and storage
2.4.5 Aeration
2.4.6 Control systems
2.4.7 Backup

2.5 Fry Rearing
2.6 Implementation


3.1 Background
3.2 The Biology of the Fish

3.2.1 Spawning
3.2.2 Early lifecycle stages
3.2.3 Recruitment

3.3 Characteristics of the Qinghai catchment

3.3.1 The Qinghai catchment
3.3.2 Climate and streamflow data
3.3.3 Soils and landuse

3.4 Assessment of conditions

3.4.1 Background
3.4.2 Present state of feeder streams
3.4.3 Conclusions

3.5 Recommendations


4.1 Trout farm development
4.2 Water supply and pipe sizes
4.3 Recommendations


People met


Table 1 Spawning and hatching characteristics of naked carp
Table 2 Requirements for the naked carp hatchery
Table 3 Estimated heat inputs
Table 4 Heat balance calculations
Table 5 Pump and pipework specification
Table 6 Aeration requirements
Table 7 Safety and backup details
Table 8 The Qinghai catchment
Table 9 Climatological data
Table 10 Stream data
Table 11 Streamflow data


Figure 1 Layout of proposed hatchery system
Figure 2 Recycle system and filter layout
Figure 3 The Qinghai catchment


The Government of the Peoples Republic of China, assisted by the United Nations Development Programme and the Food and Agriculture Organisation of the United Nations are engaged in project CPR/88/077, Fisheries Development in Qinghai Province. In support of the project, FAO assigned Dr Laszlo Varadi as aquaculture and drainage engineering specialist, from 24 May to 21 June 1989, with the following terms of reference (Varadi, 1989) :

Due to unforeseen circumstances, Dr Varadi's consultancy had to be curtailed, and its work amended to focus on the planning of the naked carp hatchery. Varadi's report - Field Document 3 - reviews the background to the biology of the naked carp and proposes outline specifications and equipment lists for hatchery facilities. There is also a brief comment on the hydrology of the area.

Subsequently, the project team has been able to clarify some of the issues affecting fish production in Qinghai. Some preliminary aquaculture facilities have been installed, and initial monitoring of the water bodies, their stocks and their management is helping to identify the main technical priorities. The present consultancy was therefore initiated, to follow on from the uncompleted earlier assignment, with terms of reference correspondingly modified to:

The consultant was assigned to the project from 18 September to 12 October. Details of the mission are provided in Appendix 1.


2.1 Background biological data

A brief review of the data relevant to hatchery requirements was provided by Varadi, 1989, based on earlier studies carried out by the national project team. To quote Dunn (1990, pers. comm):

Gymnocypris has evolved into a “salmonid analogue”, probably responding to the typical salmonid environment that it has exploited. The species therefore has a typical anadromous “run” every spring and early summer (approximately May till July) and lays atypical cyprinid non-sticky eggs in gravel substrates. The adults return to the lake in the summer, leaving the young to develop. It appears that the females are serial spawners, as underlined by the difficulty in stripping more than a small proportion of the eggs from mature migrating females.”

Preliminary trials have been made at the project and elsewhere in the country - at the Fisheries Research Institutes of Hei Longjian and Inner Mongolia - in the use of artificial spawning and hatching methods. These confirm that the injection of pituitary hormones using typical methods for Chinese carp will stimulate a more complete and controllable release of eggs, permitting conventional artificial fertilisation. Although there would appear to be little shortage of potential broodstock in season, this approach is more easily managed by holding broodstock for conditioning prior to spawning. This process also offers greater control of the timing and quantity of egg supply.

Initial trials have also confirmed that eggs can be hatched -possibly using several conventional techniques, and the larvae successfully first fed and developed to fry or fingerling stage. The techniques for this have not as yet been fully established, and will need to be refined during the course of the project. Due to the need to control spawning, the initial concepts as described by Varadi, of stripping the broodstock at river sites and transferring fertilised eggs to a separate hatchery, may not be appropriate for routine use. A more complete hatchery with combined broodstock, hatching and fry rearing facilities, may be more suitable.

Varadi (op cit) has described conventional carp culture facilities which may be suitable for naked carp hatching. There may also be a need to experiment with other facilities, and to examine alternative approaches to early fry and fingerling rearing. Thus the initial approaches to hatchery design and operation will have to be reasonably flexible. Table 1, based on the data of Varadi and others, summarises the currently defined spawning and hatchery characteristics of the species.

Table 1: Spawning and hatchery characteristics of Naked Carp,

Gymnocypris przewalskii przewalskii (Kessler)

Time needed for maturationminimum 5*, typical 7 years
Size of mature fish35–40cm, min approx 25cm*
Weight of mature fishtypical 200 to 500g*
Spawning seasonMarch - July
Main hatchery spawning targetJune - July*
Water temp range during spawning6 – 17.5°C
Sex ratio (F:M)1:2 to 1:3 (up to 1:4 at nest*)
Spawning areaShallow (0.1 – 1.1m) transparent water with sandy-gravel bottom, slow but constant water flow,
Spawning site25–30cm hollows dug out for eggs, typically 1 per 10m2*
Spawning habitFish of different size and age spawn following upstream migration - to 50km or more*
Hatchery spawning methodStripping at river sites, or hypophysation of held stock*
Eggs/breeder, averageapprox 16,000
Size of eggs2.5mm dry, 4.5mm swollen
Avge weight of swollen eggsapprox 35mg
Incubation time from fertilis-112 hours (72 day-degrees)
ation to hatching, at 15.5°C 
Target hatching temperature13–15°C *
Target fry size1.5 to 2.5cm*
Target time required7 to 10 days*

Note: all data from Varadi/Biological Institute of Qinghai, except that marked *, from current local information

2.2 The existing hatchery facilities

Facilities are to be set up at the existing Chinese carp hatchery at the Fisheries Research Station, Xining. The complex offers a series of traditional earth ponds, (deeper than normal to allow overwintering during the severe local conditions), supplied by pump with well water and by gravity with rather poorer quality irrigation canal water. The ponds are operated in conventional manner for the holding of carp broodstock, spawning, and fry and fingerling rearing. Provision is also made for combined duck and pig production, and one pond, which also acts as a settling pond for incoming canal water, is given over completely to duck rearing.

The pumped water supply is drawn from a 40m well, at a water level of some 27m below site level. This is reportedly a good quality supply, at an approximately constant 8°C year-round temperature. The water is currently delivered at a flow-rate of about 70m3/hour for two periods of 3 hours per day, via a single electrically powered tubewell pump. The system has been tested continuously over a 72 hour period at this flowrate, without significant reduction of well level. The canal water supply is warmed to ambient conditions, but is of significantly poorer quality, particularly after it has passed through the duck rearing area.

At the furthest end of the site, there is a separate hatchery building, comprising a small group of work rooms, a roof-mounted header tank, and a greenhouse-covered tank area (approx 600m2), holding spawning tanks, warming ponds and jar hatching facilities. The 15m3 header tank is supplied at present from the settling (duck) pond via a small centrifugal pump. At the opposite end from the header tank, the greenhouse area is supplied by the well water through a 4cm steel pipe and valve. Adjacent to the hatchery unit is a fingerling rearing area and a small pond suitable for holding broodstock. Both these areas can be covered with plastic greenhouses if required.

The hatchery and associated pond facilities can be made available to the project once the main carp spawning season is over - from approximately the end of May. This will fit in conveniently with the normal spawning period of the naked carp - from June onwards. Although the overall quantities of water are likely to be adequate throughout the year, one of the most critical aspects of its use for the naked carp will be to ensure that - assuming similar conditions to those found in the spawning streams - water temperature and water quality are suitable.

2.3 Design options

A summary of the requirements for the hatchery is given in Table 2. Water supplies can be obtained either from the warmer canal (duck pond) source - perhaps 20 to 25°C during this part of the year, or from the colder, higher quality groundwater, at 8°C.

Table 2 Requirements for the naked carp hatchery

Capacity106 over season
Water temperature13 – 15°C
Egg hatching system5 × zug jars, 150litres
8 × zug jars, 200 litres
3 × trough incubators
(7 × 40cm × 40cm trays)
Hatching/cycle time5 to 6 days
Fry rearing volume4 × (2m × 2m × 0.3m) tanks
Rearing/cycle time7 to 10 days
Water flow, hatchingContinuous, 13 × 180l/min = 2.3m3/hr
Water flow, tankContinuous, 4 × 3m3/hr
Total water flowContinuous, 15m3/hour

Given the need for a reasonably constant temperature in the 13–15°C range, the following could be considered:

The main risk of a recycle system lies in the possible transfer of diseases between stock units, and the difficulties of controlling them if established. However, this must be balanced against the major advantages of being able to maintain stock healthily in optimum envronments, and of avoiding external contamination, eg from the pond water. If eggs are routinely treated prior to introduction, and uncontaminated spring water is used, the recycle system should prove more than adequate for producing healthy stocks. If found to be necessary, provision should (and can) be made to treat for disease without excessively disrupting the system.

A further risk with this type of system concerns the possibility of breakdown of one or more of its components - eg pumps, control systems, etc. In this case the risk is no greater than that for the alternative systems, but care will have to be taken to ensure that the system is as breakdown-resistant as possible, and that adequate backup options are available in the worst case. Given that these risks can be handled, a simple recycle system as described would be the preferred choice.

2.4 Hatchery specifications

2.4.1 Layout and design features

The layout of the proposed system is shown in Figure 1. The hatchery jars are arranged in an ‘L’ shape around the circular tank, while the tanks are sited along the edge of the adjacent rectangular tank. This tank serves as the main storage, filtration and aeration unit, from which water is pumped, using a simple submersible electric pump, to the header tank as needed. From the header tank, water flows by gravity (using the existing pipe system) to the hatchery jars and tanks, from which it normally drains to the storage/treatment tank. Both units - jars and tanks, can also be drained externally if needed - eg after disease treatment, and the complete system can be drained by gravity via an outflow from the main tank. Fresh spring water is supplied to the tank via an extension of the existing pipe. Its use is defined by the heat balance of the system (next section). Aeration is supplied with a blower or compressor fed diffuser system.

2.4.2 Heat balance

The fundamental design requirement is to ensure that the system can operate and easily be maintained at the required temperatures. As fresh water enters the system at 8°C, it will have to be warmed up to the operating levels of 13–15°C. The system must not be allowed to heat up excessively during the daytime, and may have to be cooled. Equally it should not cool down too much overnight. It was not possible to get accurate data on temperature and heat conditions, but water in the greenhouse system during this time of year is said to heat up from 8°C to 22°C in approximately one week. This corresponds to the following heat input levels - Table 3.

Table 3 Estimated heat input

Assumed system volume/mass:150m3/150,000kg
Assumed system area:150m2
Heat input for 14°C temp rise:2,100,000kCal
Average temperature rise/day:2°C
Average heat input per day:300,000kCal
Average heat input/hour (24hrs):12,500kCal/hr
Average heat input/hour (14hrs):21,500kCal/hr
Average/maximum equivalent energy:14.5/25.1kW
Average heat input/m2 per day:2,000kCal/m2/day
Average heat input/m2/hr (24hrs)83.3 kCal/m2/hr
Average heat input/m2 hr (14hrs):142.9kCal/m2/hr
Avge/max equivalent energy/m2:0.097/0.167kW/m2

In addition, the circulating pumps of the system will heat the water, depending on their operating efficiency (much of the energy not used for pumping goes to heat - affecting both the process water and the outside air). However, on the basis of 1kw rated pump power, perhaps 200W is likely to enter the system - in this case relatively insignificant.

In the initial stages, the system would be filled and allowed to warm up. For the complete system, based on an operating temperature of say 14°C, this would take about 3 to 4 days. If faster startup were needed, the system could be operated at reduced volume, reaching working temperature in 1 to 2 days, then filled up over the next few days. Alternatively, another pond in the greenhouse could be used for additional heating either before or while the system is started, and its warmed water pumped or syphoned into the storage/treatment pond.

The simplest approach to heat management is to allow the system to warm up slightly during the daytime, and to rely on heat storage in the system to maintain (slightly falling) temperatures overnight, allowing the temperature to pick up gradually during the daytime (eg Table 4) During the daytime, some additional cooling water would be needed to control temperatures to the required levels. In practice, these could be adjusted according to actual conditions.

Table 4 Heat balance calculation

a) ‘Steady-state’ conditions: number of hours of top-up flow required to maintain temperature at stated final levels System mass, kg: 150000 Initial temperature, °C: 8

Daily heat input, '000 kCal:

flowrate, 1pm    

b) Example of daily temperature fluctuations in typical temperature regime: Maximum heat input, kCal/hr: 30000

% of max
heat input
Temperatures:% of max
heat input

If individual tanks have to be drained, their volume is insignificant compared to that of the main system, which can continue to operate as normal, refilling the tank once its water supply is turned on again. When the system is topped up with spring water, its temperature will drop marginally, but this will easily be recovered during the daytime. If the complete system has to be drained, water can be stored in an adjacent pond to heat up for refilling.

2.4.3 Pumps and pipes

The system can be set up so that jars and tanks can drain by gravity into the main storage/treatment pond. All that is then required is to return the storage pond water to the header tank, from which it flows by gravity to the hatchery units. A single submersible continuously rated electrical pump is recommended, with if possible a similarly sized standby/backup pump - of submersible or conventional centrifugal type. If possible (see later) the pump would be operated intermittently, controlled by water levels in the header tank. If this cannot be done, the pump would operate continuously, regulated by an outlet valve, with an overflow from the header tank to the storage pond.

Most of the existing pipework can be used directly (Table 5)

Table 5 Pump and pipework specifications

System pumpFlowrate 15m3/hr, max head 7m, allowing for pipe losses; 0.48kW (0.63HP) at 60% efficiency. Submersible continuous-rated electric pump with 2-inch (50mm) outlet.
Supply, store to headerFlow 15m3/hr, 50mm pressure pipe, at TEL = 50m gives 0.15m head loss. Note new pipe required, also control valve on pump outlet, and pump connector.
Supply, header to jarsFlow 3m3/hr, existing pipe sufficient extended with 50mm pipe to new incubators 12mm feeds, flow controls to each unit.
Supply, header to tanksFlow 12m3/hr, uses existing downpipe to circular tank, extended in flexible 75mm pipe to tank area, where joined to rigid 75mm pipe, then 25mm feeds to each tank, with flow control valves.
Return, jars to storage75–100mm pipe or gutter, receiving flow from tops of jars. Jars can be drained to floor, or via temporary flexible pipe to storage pond.
Return, tanks to storageVia existing below-tank drains, possibly attached to swinging arm standpipes to control internal levels. Alternative drainage to floor.
Overflow, header to storageVia 100mm (min 75mm) flexible hose downpipe - if possible change to more permanent plastic or steel pipe.

2.4.4 Filtration and storage

The rectangular pond has an area of approximately 150–170m2 (approx 24m × 7m), or at 1m depth a volume of 170m3. The waste water from the hatchery units would be filtered to remove solid materials, and also to remove metabolites (carbon and nitrogen-based wastes) via biological oxidation on the filter surface. For a maximum flowrate of 15m3/hour, and a conservative residence time in the filter of 2 hours, this requires a filter volume of 30m3. At this size, there will be more than ample capacity during the working season to operate without the need for cleaning. As sand and gravel is easily available locally, this can be used as the filtration material. The filter area can be separated from the main storage volume with a simple brick or concrete wall.

The remaining volume can be used for storage - as a heat reservoir, and as a zone for aeration. If preferred, one section of this volume - perhaps 40m3 of the remaining 140m3 can be separated with a brick or concrete dividing wall to make a reservoir of heated spring water for topping up the system - eg during the night. This will reduce the overall volume of the system, and hence its heat stability, and with careful operation should not be necessary. It may however give some useful extra flexibility if managed properly.

Figure 2 gives a suggested layout and shows how the filter would be constructed. To avoid excessive phytoplankton growth, the complete storage pond area should be covered with black polythene. This can conveniently be laid over curved plastic pipe hoops or a light metal or wood/bamboo frame, or alternatively suspended or supported over ropes attached to the greenhouse frame or the pond walls.

2.4.5 Aeration

The system requires aeration to oxygenate the incoming spring water and to make up for oxygen removed by the metabolism of the stock and by the treatment filter. These requirements are summarised in Table 6. Assuming a rather low efficiency - allowing for a generous safety margin - of 0.2kg oxygen per kWh applied, a unit of 0.6kW (0.8HP) would be required. A simple electrically powered blower or compressor would be suitable, supplying air to diffuser blocks or to airlift devices. On the basis of a 10% transfer of applied oxygen (20% by volume of air supplied), this would correspond to an air flowrate of about 10m3/hour. At the low pressures involved, a blower would normally be more efficient.

Table 6 Aeration/oxygen requirements

Incoming water:Assume 100lpm (6.0m3/hr) at 0.5mg/l, raising to 6.5mg/l = 0.036kg/hour
Stock metabolism:Assume 15m3/hr water flow with 2.5mg/l oxygen uptake by stock, = 0.038kg/hour
Filter oxidation:Assume 15m3/hr water flow with 3.0mg/l oxygen uptake by filter = 0.045kg/hour
Total requirement:0.119kg/hour

Aeration would be provided to the main tank, but could also be supplied to the hatchery tanks and to the header tank. For a blower system, main supply would be via 50–75mm low-pressure grade plastic pipe, feeding standard aquarium air tube, or for larger diffusers, 10–15mm plastic (garden) hose pipe. For a compressor system, main supply would be via reinforced wall garden hose pipe, or equivalent (10–15mm) plastic pipe, feeding standard aquarium air tube. These distribution pipes could normally be suspended from the greenhouse ceiling bars.

The aeration system would run continuously, though there is some safety margin in the stored oxygen in the system. Thus at the maximum demand level, assuming the storage tank was fully saturated, say at 6.5mg/litre, the system could run without aeration for about 3.5 hours before reaching 4.0mg/litre, perhaps a critical limit for the stock. Excluding fresh water supply, this could be extended to almost 5 hours, longer during the night, when the stock metabolism would be reduced.

Note that oxygen could be provided by photosynthesis - eg by growing phytoplankton in the storage pond, but algae would cause water quality problems elsewhere in the system, and would respire during the night, increasing oxygen demand. It is therefore not recommended in this instance.

2.4.6 Control and alarm systems

The system is designed to operate as simply as possible, and once set up should be able to run with only occasional adjustment. The main control suggested is for the pump supply to the header tank, switching the pump on at low header level, and off once the tank is filled. This allows the pump supply to adapt efficiently, and without manual adjustment, to a range of flowrates. A simpler but less flexible alternative is to use a single rocker switch. Note that the system pressure, and hence flowrate to the hatchery units, will fluctuate with the header tank water level. If there is too great a difference between the upper and lower switching levels, this may cause problems - but is unlikely to in this case.

If equipment is unavailable, the system can be operated on continuous flow, by controlling an outlet valve on the pump to match, as well as possible, pump delivery and tank flows. Normal surplus water would return via the hatchery tank supply pipe to the storage pond, and an additional high-level overflow, draining to the storage pond, would prevent the header tank overflowing. This method will be inefficient in terms of pump power, but over the operating season will not be prohibitive. An option may be to use two or three smaller pumps, switching on as many as are necessary to provide the flow required at the time.

Alarms could if necessary be provided for pump failure, aerator failure, for header tank high and low level, and for storage tank temperature. However, alarm systems should not be used as a substitute for routine inspection and monitoring, as they may themselves be unreliable.

2.4.7 Backup

Any system such as this depends on the successful operation of a number of sub-systems and components for its effectiveness. To make sure of its reliability, and its safety in the event of failure, it is important to make sure that there is adequate backup. Table 7 itemises the most important components of this system, and the suggested security measures needed.

Table 7 Safety and backup details

Component/aspectRisk/backup provided
Power failurePump, aerator, lights stop - see relevant sections; check via signal light, power alarm or stopping of individual items of equipment; needs backup generator, minimum 1kw, simple plug-in to system.
Pump stops/blocksHeader tank drains - 1 hour at maximum flow, could normally be extended to 2–3 hours with good aeration and quiet conditions; storage pond will fill approx 10cm; check via pump noise, indicator pipe, pressure gauge, pump power alarm, supply pipe pressure alarm, header low-level or storage high level alarm; needs easy pump access, backup pump, simple pump/pipe and pump/power connections.
Upper level switch failsPump does not switch off, header tank reaches overflow level, water drains via overflow pipe to storage pond, visible in hatchery. System can operate safely this way until switch is repaired or replaced.
Lower level failsPump does not switch on, header tank drains, switch conditions as for pump stopping.
Blower/compressor stopsAeration ceases - 3 or more hours if system well saturated; can pump/splash water in storage pond, increase water flow; check appearance of storage tank, pressure gauge, or via power or air pressure alarm on aerator; needs backup unit, simple power and pipe connections 
Header tank blocksFlow partially or completely ceases - 20–30min if complete stop, can be extended with aeration in jars or tanks, may connect temporary pump from storage pond to jar, tank supply pipes; header tank will overflow if pump continues, as upper overflow has limited capacity; pump may need to be stopped or flow reduced; check via flows to tanks/jars upper high level alarm, reduced water level in storage pond, sound of overflow; needs good access to header tank and downpipe outlets, possibly need to connect pump to downpipes to backflush.
Filter blocksWater level in return section of storage pond rises, other section drops - 2 to 3hours at normal flows. Filter can be bypassed by running return wastes directly to storage, to allow filter to be cleaned, etc. Care required to drain dirty filter material out of system. System can be operated for several days without filtration.
Spring water blocksTop-up supply to storage tank stops. System can pipe operate for several days, depending on temperature. Water can be pumped from pond supply canal as needed, and/or stored in adjacent tanks.

2.5 Early fry rearing

This is discussed in Varadi's report, whose proposals are probably the most suitable for the conditions available. There may be some value however in experimenting with the remaining ponds in the greenhouse area as semi-intensive fry production units, possibly using the circular pond as an intensive live feed production unit, with partial volumes being transferred at regular intervals to the rectangular ponds. Further work would be required to establish the feeding requirements at this stage, however, and to define the optimum or most effective feeding strategy.

Following a semi-intensive initial stage, the part-grown fry could be transferred to outdoor ponds and/or released to natural or semi-natural stream or lake habitats as determined in the later stages of the project.

2.6 Implementation

Some materials - plastic pipe, some valves were available as part-orders for equipping the hatchery, or as surplus to immediate needs for the trout hatchery. The additional materials required were identified and selected locally. Unfortunately there was insufficient time available to install the complete system, though the major components - pump, tanks, jars, were put in place, ready for connection. By the end of the mission, there were still some unresolved questions about connectability of the different pipe and valve systems available, but it is hoped that these can be overcome. If necessary, a small number of additional fittings could be imported for early 1991, when the CTA returns to the project, and can supervise the final stages.

A series of design sketches was prepared, to assist the farm staff and technicians in completing the installation. By the end of the mission, the following tasks had to be completed:


3.1 Background:

The Naked Carp is anadromous, and has developed a typical salmonid pattern of a spawning “run” every spring and early summer, up the feeder rivers and streams of the Qinghai lake. There has been considerable concern that spawning success and hence recruitment to the Qinghai Lake fishery was increasingly being threatened by changes in stream and river habitats, caused in particular by artifically lowered water levels.

The objective of this component of the work was therefore to assess the current status of the feeder streams to the Lake, to attempt to assess the actual constraints to stock replenishment, and to propose, where feasible any means to resolve these, including where appropriate, the use of specific engineering measures. This involves an assessment both of the biological aspects of spawning, early life-cycle stages, and recruitment needs, and of the present and likely future hydrological characteristics of the system.

3.2 The Biology of the fish:

3.2.1 Spawning characteristics:

These have been reviewed in the previous section. The spawning naked carp migrate in spring and early summer (approximately May till July). The fish are reportedly serial spawners, as evidenced by the difficulty in stripping more than a small proportion of the eggs from mature migrating females. There is also probably a considerable readsorption of unshed eggs, as large numbers of eggs are found in the ovaries of some fish caught at the end of the season (Dunn, pers comm)

According to local workers, the mature fish, typically 150 to 500g in size (avge approx 250g), will migrate considerable distances, some 40 to 50 km for example in the Buha River, over a period of several weeks, during which adequate spawning flows are required. In cases where water levels have dropped during the spawning run, considerable numbers of fish have been seen confined in small pool sections of the streams, and fish kills have resulted. During the spawning runs, the fish will swim against reasonably strong currents and can do so for some time. Actual currents have not been measured, but based on a typical body length of 30cm, continuous swimming speeds of 50 to 100cm/sec might be attained, with burst speeds of up to 200cm/sec.

The fish are stated to be capable of clearing small weirs or obstructions - perhaps 50cm, but there are no reports of salmonid leaping behaviour. Given the fish's family characteristics, its morphometry, and its lacustrine adult habitat, it might well not be as vigorous as a salmonid in similar conditions.

Spawning is reported to occur in mixed gravel and sand banks, quite similar to salmonid ‘redds’. These are typically found in the braided sections and side-channels of the feeder streams, and have relatively low water velocities compared with the main stream section, or downstream passages. The female clears away a small circular hollow - typically 30cm diameter, perhaps 10cm deep at the centre, in which eggs are laid. An estimated maximum density of 1 ‘nest’ per 5 –10m2 was indicated. Although considerable upstream migration is said to be normal, there are local reports of some fish, possibly distinct families or even sub-species, spawning in sandy gravel beds adjacent to the lake itself. There is no information at present on distinct stocks associated with the separate feeder streams.

The eggs are not covered, as they are reportedly slightly adhesive (cf Dunn's remarks), and/or are left in place by the gentle water currents. If flows increase during this period, however, it is said that the eggs can be washed out of the beds. The eggs hatch over a 10-day period, and apparently remain in the nest area in the early stages.

3.2.2 Early lifecycle

The adults return to the lake in the summer, leaving the young to develop. Unlike, eg the Pacific salmon, there appears to be insignificant spawning mortality in normal conditions, and hence little opportunity for nutrient input from this source. Little appears to be known about the early feeding stages of the fish, but it was suggested that the first feeding fish gradually move or drift out of the nest areas, and collect around low-velocity zones - eg small pools, perhaps seasonally isolated from the river, or stream edges, in which feeding can be found. Based again on bodylength estimates, a typical 2–3cm fish at this stage might be capable of a position-holding current velocity of up to 2–3cm/sec, a short-period velocity of 3 to 5cm/sec, and a burst speed of perhaps up to 10cm/sec. A sustained water velocity of more than 3 to 5cm/sec might be expected to drift the fish downstream, though as, eg with young salmonids, the fish may well be able to find lower current micro-zones.

Feeding is supposed to be largely benthic. By September all stages of young (year 0) fish are found from the hatchings that have occured throughout the season - during the present mission we observed small groups of 2–5cm fish moving in small pools and margins of several of the feeder streams. The water was relatively clear, and there was ample benthic cover, but little other feed material - eg insect larvae was immediately apparent. It is uncertain whether these young fish are able to overwinter in the streams, as they are locally reported to do. Although the area is extremely cold and extensively frozen in winter, the streams are said to continue to flow due to groundwater flows reaching the surface under the ice. Given the nature of the substrate, this may be possible, but it does not seem to have been corroborated.

3.2.3 Recruitment:

It must be presumed at this stage that the overwintered year 1 fish will drop, actively or passively (perhaps moved by the spring rise in streamflow), to the lake. Predation is reportedly very low, but there is a sizable (and protected) bird colony in the lake, on Bird Island, near the main feeder outlet on the Buha River. Otherwise, there is apparently no record of cannibalism, there are no piscivorous fish present, and the first real predators are the fishermen, catching 5+ year fish in pair trawls or gill nets. Thus the main mortality appears to occur at around the spawning size; catch sizes are typically 200g+, and gill nets are being used, legally or otherwise, at the mouths of rivers to trap migrating fish. Dunn (pers comm.) notes that ‘the fish enter the trawl fishery at 5, 6, and 7 years old (25 –30 cm standard length)’, but also considers it ‘likely that the spawning adults of which the larger individuals are in the order of 50 – 70 cm are probably in the region of 15 – 20 years old’. Depending on the relative contribution of the older stock the effective pre-spawning mortality may clearly be much higher.

At this stage however there are insufficient data to define the state of the stocks and the year classes relative to the actual fishing effort and recruitment potential, though there are as yet no signs of dramatic collapse of the fishery. However Dunn also comments that ‘within the time span of the present spawning population there have been massive changes in the environment of the streams’, and considers that ‘the condition of the fishery is probably more a function of the recruitment from the streams, rather than any form of fishing activity’

Although it is difficult to make accurate comments on the relative importance of the contributing factors, it may be useful, at least in ‘order of magnitude’ terms, to attempt to quantify the possible role of stream capacity in sustaining recruitment to the fishery. The current catch from the system is around 4,000t. Assuming an average weight of perhaps 250g, this represents 16 million individuals. Assuming a (perhaps high) mortality of 50% within the lake, and 80% in the feeder streams, this would require a recruitment to the lake of 32 million fry, and the successful hatching of 160 million eggs.

Assuming 50% of the migrating females actually spawn, and an average fecundity of 16,000 per female. This would require 10,000 spawning females, 20,000 in total, and with an average F:M ratio of 1:2.5, 50,000 males. In the streams themselves, assuming all females make nest areas, 20,000 nests at 1 per 10 m2 requires 200,000m2 of suitable substrate. Assuming perhaps 5% of the actual streambed is potentially suitable, and assuming spawning sites are used only once in each season, this would require 4 x 106m2 (400ha) of total stream area, corresponding eg to a single stream area of 20km length and 200m width

Simplistic calculations such as these are clearly very sensitive to the assumptions made; mortality for example is likely to be related to population density and to feed availablity, and may decrease, providing a feedback mechanism to increase fry survival if hatched production declines. Some tentative conclusions might however be drawn. Unless river-based gill netting devastates migrating stocks (as it might do), the numbers do not suggest too critical a shortage of spawning adults. More seriously, the system does appear to require substantial areas of spawning ground, and this would be extremely sensitive to water flow conditions. Stream environments may also be critical for fry survival and lake recruitement. These are examined in the next section.

3.3 Characteristics of the Qinghai catchment

3.3.1 The Qinghai catchment

An outline of the Qinghai Lake area, and its major features is shown in Figure 3. Unfortunately, it was not possible to obtain supplementary information, such as local climatology, topographical detail, geology and soils, or land-use. Such information as was available is summarised in Tables 8 to 11, and in the following comments.

Table 8 The Qinghai catchment (based on Kelts et al, 1989, and williams, 1990)

Altitude, m.a.s.13,196
Lake area, km24,635
Catchment area, km234,950
Catchment/lake area ratio7.54
Lake volume, 106m385,450
Maximum depth, m27
Mean depth, m17.5
Maximum length, km106
Maximum width, km63
Perimeter, km306
Water renewal time, yrs60.4
Salinity, g/l12.5 – 14.15
Approx. mean rainfall, cm35
Approx mean evaporation, cm55

Along the south shore there are a number of small inflows and in the SW corner of the lake a larger stream, the Hima river. At the western end is the Buha river, which is described as the most important inflow contributing 80% of the input to the lake. The Buha river enters the lake through an extensive delta, and 20 – 30 km upstream from the mouth the river course appears to break up into a small inland delta. Along the north shore there are a number of small streams and three larger inflows of which the most important, the Gongcha river enters by means of a broad delta. There are two freshwater lakes (effectively lagoons) adjacent to the lake, one in the NE corner and one at the eastern end.

According to Williams (1990), lakes such as Qinghai result from major tectonic changes in the Tertiary period, and increasing aridity, resulting in the development of the typical endorheic drainage basins, and increasing salinity, in the case of Qinghai lake developing, through climatic change, over the last 40,000 years. Williams also notes that the lake, in common with others of the region, falls well within the theoretical limits for salt lakes. Based on a crude interpretation of the mean rainfall and evaporation data associated with these limits (much would depend on catchment land use, and hence net evapotranspiration), the catchment would appear to be subject to continuing desiccation. This would be supported by local reports of a net reduction in lake water level of some 10cm per year (Williams quotes a drop of ca 2m in 15 years), and is in turn supported by an apparent rise in salinity from 12.5g/l recorded prior to 1979, to 14.15g/l in 1989 (Kelts et al, 1989). However, medium-term trends, as evidenced by a recent rise in level through the relatively wet year of 1989, may not be so severe.

3.3.2 Climate and streamflow data

According to Williams (1990), the climate can be described as semi-arid, continental, with cold winters, hot summers, and limited precipitation, mostly in summer. There was no information available on the actual rainfall patterns over recent years, as meteorological data, though collected locally, was not readily available. Local accounts relate a higher than average rainfall during 1989, and a consequent increase in lake water level, while 1990 has seen average to low levels.

Table 9 Climatological data (Qinghai Hydrological Institute)

Average air
Mid S.-0.4-
Hima R.-0.5-0.4n/a2.16.910.012.412.58.63.0-0.2-0.518

Note n* is number of years recorded

Table 10 Stream data (Qinghai Hydrological Institute)

Yearly average:Flow range,m3/secn*
flow,rainfallmax (year)min (year)
Buha R.(1)784021.484.739.0(1964)10.0(1973)19
Buha R.(2)1433725.957.152.7(1967)6.3(1973)21
Jiang R.304811.1115.118.5(1967)5.7(1960)8
Xialio R.14427.4169.711.0(1972)3.3(1979)21
Hima R.14253.985.96.0(1959)2.9(1960)5

Table 11 Stream flow data (Qinghai Hydrological Institute)

Average monthly
Water flow
Buha R.(1)2.031.913.2411.821.735.463.655.336.514.56.032.9219
Buha R.(2)2.542.452.634.1715.035.684.875.854.622.36.232.8821
Jiang R.0.500.510.763.0110.417.537.125.122.910.53.360.958
Xialio R.
Hima R.0.900.852.293.093.684.5110.

Notes: Buha River: (1) is measured at one of the outlet channels,

(2) is measured further upstream. It is not clear however whether these represent total flows. Given the size of the system compared with the others, the total may be much more than this.

From the material available, it is clear that streamflows vary significantly throughout the year, the major flows occurring over the normal spawning period of the fish. If the young fish overwinter in the streams, they are probably also likely to move lakewards during this period. The relative contribution of rain and snowmelt to streamflow is not clear from the records available; it is probable however that snowmelt might provide much of the ‘base flow’ to the streams, rising in the springtime as the melting increases, and supplemented by short-period increases due to localised rainfall. There is little information on the characteristics of the stored ice and snow, but this would clearly be important for spring and early summer streamflows, and the realtively cold and clear water may act as a migration stimulus for the fish.

Local reports suggest that streams rise quite quickly during rainy periods, suggesting a relatively high ‘runoff coefficient’ -perhaps as much of the watershed may be saturated during the spring to summer season. Once rainfall ceases, the stream levels are likely to fall equally rapidly. This information is probably collected by the Hydrology Institute, but was not made available at the time. The pattern of short term fluctuation over a rising spring base flow would be characteristic of a ‘moving and waiting’ migration strategy for the spawning fish.

There is also, over the record available, a substantial year to year variation in average streamflow. The commercial fishery of the lake, and its associated records are however only recent, and cannot be easily correlated with this data to ascertain whether specific year classes had been affected by this variation.

3.3.3 Soils and Landuse

Topsoil material appears to be sedimentary in origin, with little humic organic content or root binding material, and of typical coarse silt to sand particle size. Base material contains substantial sand, gravel and boulder material, and is extracted, at least on a small scale for local construction work. As such, the material is appears to be relatively easily drained, and moderately to highly susceptible to hydraulic and aeolic erosion forces. Overall gradients are moderate to steep in the outer areas of the catchment, falling to very slight, almost horizontal, in the plains surrounding the lake itself. Gradients are generally steeper in the southern basin. Around the catchment, the valley of the Buha is unique in its elongated lower plain.

In the absence of specific data it is difficult to be certain of the factors involved, but it would appear that feeder streams are in a typical depositional phase, evidenced by development of alluvial fans, braiding, etc, in the lower zones. This would of course be modified by seasonal variation, and there would still appear to be substantial localised movement and transport of soils, particularly where human activities had changed profiles, or exposed unstable materials.

Vegetation is of typical steppe grassland and tussock, traditionally used for extensive pastoral animal rearing by nomadic herders. There is little recorded data on changes in land use on the catchment, as information was again not easily available. Circumstantial accounts however describe a positive effort from the mid 1960's and into the 1970's to settle the area, build up small townships, and develop the Qinghai catchment from traditional seasonal pasture towards more intensive agriculture eg summer wheat production, based on irrigation of the broad areas of flat land around the lake itself. The lower areas are still used for seasonal herding of sheep, cattle and yaks. During the autumn, when animals are brought down from higher summer pastures, there would appear to substantial grazing pressure and probable physical damage to the lower soils.

Information on upland land use was even more difficult to establish, though it is unlikely that climatic conditions would favour substantial change in agricultural practice. This does not exclude, however the possibilities of mining, mineral exploitation or other forms of land movement.

3.4 Assessment of conditions

3.4.1 Introduction

The primary problem identified in the initial stages of the project was the possible threat to spawning success posed by artifically lowered water levels in major spawning locations such as the Buha river. An additional problem may arise if increased silt loads block spawning areas, change bank and water flow characteristics, and/or cover egg-laying zones, thereby restricting ventilation and reducing larval survival or quality. The following comments are based on the observations of Dunn and on information obtained in travelling round the lake and visiting feeder streams with Fisheries staff. They are however greatly limited by the quality of evidence available, the time available to visit the sites involved, and the opportunity to corroborate field observations.

3.4.2 Present state of feeder streams

(a) The Harge river

This is a small to medium sized stream on the northeast of the lake; the first of any substantial size in this area (a smaller, more easterly stream is reported to flow into one of the lagoons skirting the lake, and is apparently isolated from the main system, though fish (naked carp?) are said to spawn.

The Harge river has been known as a spawning site for naked carp, but has been substantially changed by an irrigation scheme -reportedly developed in 1965, with a diversion barrier a few hundred metres upstream of the road crossing, which is itself some 20–30km from the lake. At the time of visiting, very little water was running down the riverbed, but flows down the irrigation canal were quite substantial. During higher flow conditions, additional water tops the diversion weir and runs to the stream, and this is apparently still enough to stimulate some migration. Fish were reported to spawn in areas downstream of the bridge, and also in the past to have moved up beyond the site of the present dam. Fish (possibly not naked carp) were also reported in the irrigation canal.

The river itself has a moderate gradient at the bridge area, and appears to be much flatter in its lower reaches. There are numerous, though not extensive areas of sandy gravel of the type reported to be suitable for spawning, and relatively little silting in this area, though this may deposit further downstream. The stream bed appears to be relatively stable, as might be expected if much of its flow is diverted. There appears to be little attempt to regulate the stream by providing compensatory flows during the spawning season, and it must be concluded that the Harge river's capacity as a spawning source is severely limited by its use for irrigation.

(b) The Goncha (Sha Liu, or Ike Wula) river

This is a medium to large river, the main river on the north shore of the lake. Just downstream of the bridge crossing, the river is also controlled by a weir, reportedly from the mid 1960's, to supply irrigation water for a large agricultural development and to feed a power station. The development of the area also involved the establishment of a large township (Goncha) which still appears to be growing. There is some industry associated with this township, but there is no information on any effects on the river. At the time of visiting, however, during relatively low flows, there was a slight sulphide smell from the river, suggesting moderate organic loading, and some deoxygenation.

Although most of the flow appears to be diverted from the river, there is apparently still a sufficient quantity for some migration and spawning. The capacity for spawning is reported to vary greatly from year to year. Clearly if a fixed quantity of flow is being extracted continuously from the system, capacity will be very sensitive to fluctuations in flow. At the crossing point, the stream has a moderate gradient, and downstream sections appear to be reasonably stable, though there was some evidence of movement. Apparently trees bordering the stream were cut at some time, causing some instability of the banks. There is a substantial delta ground downstream of the main crossing point, and evident presence of considerable areas of sandy gravel beds.

Spawning is reported to occur from the lower-middle reaches of the river up to the mountains. The diversion weir, approx. 100m × 1m high, is reportedly passable by the fish during moderate flows. There is apparently a substantial gill net fishery on the river. The fish are reported to be slightly smaller than those in the Buha river, possibly a separate reproductive group? From a brief observation, Dunn thought this river might well be a major breeding area for the naked carp, as although overall streamflows are not as substantial as those of the Buha, there appear to be extensive areas of suitable spawning grounds. However, there is no quantitative data on the comparative areas of gravel sites etc.

Dunn notes information that the scheme was completed in the late 1960s to early 1970s, and comments that massive changes to the form and the flow of the river will have occured within the lifetime of the present population of older adult fish. As with the Harge river, there is apparently no system of compensatory flow. It would appear that the spawning capacity of the river, and particularly its year-to-year consistency, though not as badly affected as the Harge, must be seriously reduced by its present management.

(c) The Chang Ji river.

This is a small to medium sized stream on the northern shore of the lake. It also has small weir and irrigation canal, sited upstream of the bridge crossing, reportedly built in the 1970's. Fish have been reported to spawn in the river, though from its appearance its seems unlikely to contribute substantial quantities. The weir, approx 1.5m high, is apparently quite regularly washed through by floods and has to be rebuilt almost yearly. This might inadvertently give better opportunities for spawning than in some of the more controlled streams.

(d) The Buha river

The Buha river, feeding to the western edge of the lake, is considered by the Fisheries Institute staff to be the major resource for naked carp production, and is clearly the largest river system in the catchment, with the greatest flowrates and the most extensive areas of low-lying depositional substrate. Spawning fish are reported to migrate as much as 80km upstream on the system, which appears to have widely braided gravel zones along much of its lower length.

The lower road bridge crossing over the Buha river, about 15km upstream of the lake, has been used as a sampling point for a number of years by staff of the Fisheries Institute, who report large numbers of fish moving upstream every year. Dunn comments on the possible importance of the widely branched zone 20–30km upstream of this crossing; this may well be the most important spawning ground. At the upper bridge crossing, just above this zone, its extent could be observed; by comparison with the other areas visited, it would be by far the most extensive substrate area available.

There was less evidence of heavy abstraction of water on this system than with other visited, though two weir/diversion structures were seen. These have apparently not been much used in recent years, however. It is not known whether other such structures exist or are used. What may prove more important on this river is the apparently quite heavy gill-netting, pursued illegally during the spawning season. As the fish will clearly be concentrated at several points along their migration, they will be highly vulnerable.

Dunn comments on the changes in Bird Island, shown as an island in maps from some 20 years ago, now joined to the mainland by a shallow causeway across the mud flats of the Buha river delta. It was not possible to the examine causes for these changes, or whether they usefully indicate the changing stream processes over this period. The accretion of material forming the causeway could quite easily be explained by the normal depositional delta formation processes, possibly affected by localised wind-driven currents, together with a lowering of lake level. It is clear in any case that the Buha, with a large catchment area and a low gradient over its first 100km, is a major site for transport and deposition of alluvial material, and it is probable that this very characteristic gives it its importance as a spawning site.

(e) The Hima river.

This is a small stream running past the rear of a small township. It carries a considerable silt load and probably receives high nutrient drainage from the town. It appears that in September, when the migrant herders carry out their winter cull, the river is used as general discharge for slaughtering activities. In September this stream was found to support a considerable (unquantified) population of Gammarus, suggesting that there is a considerable organic food supply for young naked carp. At the time of visiting, however, there was very little water in the stream bed

(f) Other rivers, sites.

Several smaller streams on the south side of the lake were seen. At the crossing points, gradients were moderate, but rising fairly quickly upstream. There was usually some substrate suitable for spawning. Most of these streams were modified at the bridges, resulting in channelling and some localised erosion. There may have been more extensive gravel beds downstream, but although uncontrolled for water extraction, there was unlikely to be enough flow in these for a significant contribution the the lake's needs.

The freshwater lagoon at the eastern end of the lake once had a population of naked carp but these were reported to have been either fished out or killed off by herders washing their sheep with chemicals in the water.

3.4.3 Conclusions

Most of the streams on the northern side of the lake have been affected by damming for irrigation, power generation and/or water supply, and there is clearly a substantial reduction in flow. Quite apart from possible effects on the overall water balance of the catchment, the effects on spawning opportunity for the naked carp must be marked, as the streams appear to receive little more than residual flows. Those streams which still support some reproductive populations will thus be extremely sensitive to rainfall and streamflow variation, particularly as there do not appear to be any provisions for compensation flow during the critical spawning season. In many cases, gravel beds will simply not be available as streams drop down into their central dry weather flow channels. Reportedly much of the irrigation demand also occurs in the early part of the year, and so the prospects for substantial compensation flow may be limited. However, of these streams, probably only the Goncha (Sha Liu) is likely to be a significant contributor to lake recruitment.

The most important of the systems in terms of water flow and substrate area, the Buha, shows relatively little evidence of water abstraction, at least within the lower-lying zones where much of the irrigation supply appears to directed in other systems. This may be at least partly explained by the apparently lower scale of development and human settlement than in the areas to the north and east. In terms of overall spawning area, there would appear to a sufficient quantity. The upper area of the Buha, for example, may conservatively be estimated to cover some 30km × 200m, or 6 × 106 m2 surface area, considerably more than that proposed for sustaining the fishery. The Goncha would easily provide 20km × 50m, or 1 × 106m2. While the area of the Buha, even partially covered, should continue to sustain the fishery, that of the Goncha, being more heavily regulated, will be far more uncertain. Given the apparent prominence of the Buha, however, its fishery practice must come under closer scrutiny. If substantial numbers are being gillnetted in the stream channels, this may become a greater risk than water flow. This might be particularly important during good water flows, when migrating fish would have a better than average chance of spawning success.

It is difficult to state whether the streams are markedly at risk from siltation. Overall it would appear not to be the case. The carriage of eroded material is a normal process; and sand and gravel bars and beds will be continuously formed, removed and reformed as the hydraulic energy of the streams changes with water flow. On balance it would appear that material is continuing to be moved down the streams, though where sizable flows are being diverted, there may be sediment accumulating upstream of barrages. Downstream by contrast, reduced water flows may create more stable environments, though not necessarily more suitable for the fish.

One other factor which may affect the downstream stretches of the feeder streams is the changing lake level; in general if the lake level falls, the stream energy increases, and more erosion will initially be seen at the stream mouth, balanced later by more transport from upstream, and deposition downstream. In severe instances it may be difficult for migrating fish to pass the gradient barrier at the mouth. A rising lake level will reduce stream energy, and will tend to increase deposition further upstream of the mouth. However, given the topography of the area, and the recent rise in lake levels, this is not likely to be important.

3.5 General conclusions and recommendations

The two major factors affecting the fishery are the recruitment from the rivers, which in some cases have probably been considerably modified within the lifetime of the present fish population, and the fishing activity. Comments on the rivers are given above. The fishing effort on the lake is badly documented and the statistics questionable. The factory in the SE corner of the lake has run two sets of wooden pair trawlers (150 hp) for the last 30 years or so. These have now been replaced by steel trawlers with 250 hp engine (2 in 1988; 2 in 1990). There are in addition two other steel pair trawlers landing at Shatos on the north shore, for which there appears to next to no information. There is also considerable ad hoc fishing going on with monofilament gillnets, apparently aimed at taking migrating adults in the spawning season, though it is officially only permitted from September onwards. There is also some small fishing activity by people living close to the lake, but this is not likely to be significant.

There is thus little reliable information on which to clarify the relative effects of the river conditions and of the fishing activity, and it is apparent that both have changed substantially in recent years. However, the most important stream systems, particularly those of the Buha River, still appear to support substantial spawning activity, and broad estimates suggest that currently available spawning habitat may not restrict recruitment. That said, many of the smaller streams, and significant parts of the larger ones have lost habitat, and with continuing change in land use, and additional fishing of migrating stocks, there is no cause for complacency.

In the case of salmonids, there is a considerable scientific basis on which to establish streamflow and habitat requirements (Fraser, 1975). Thus according to one of simpler methods, a minimum discharge of 30% of mean annual flow is adequate for suporting fisheries, and 60% for supporting spawning and reproduction activities in a resident population. More sophisticated analyses provide inventory methods for usable gravel areas, based on streamflow, and its relationship with depth of water cover over the gravel, velocity at 30cm depth, and gravel particle size characteristics. In this way it is possible to obtain reasonably accurate estimates of the potential spawning area under any specified flow regime, and to define, for multi-use management, realistic compromises between permitted discharge and spawning opportunity. Similar approaches can be used to define potential egg and larval rearing habitat.

While it is unrealistic to expect to approach the naked carp fishery in as comprehensive a manner, there is much that can be examined further. There are also some practical implications. Firstly, if flow is reduced so that water recedes significantly within the gravel banks, opportunities for spawning will virtually cease, though migration may still occur. In Scottish salmon rivers, Baxter (1961, in Fraser, op cit), quotes a flow of 1/8 of mean annual flow for this, somewhat less for larger rivers. At flowrates of 1/4 to 1/2 mean annual flow, Scottish rivers flowed across the full width of their banks, and much of the spawning area was accessible. Further, as it appears (Table 1) that the naked carp is quite tolerant of water depth, spawning in depths of 0.1 to 1.1m, gravel beds may not have to be extensively covered to provide spawning sites.

In some streams, though water flows may drop, their cross-section may actually cause localised increases in velocity, either preventing spawning or in extreme cases, restricting upstream movement. Such problems can be identified, and easily remedied. A further aspect of this is that if low flows have to be accepted, higher velocity areas can be broadened out artificially to create conditions suitable for spawning, to help compensate for those lost elsewhere. If compensation flows are possible, it may be feasible to improve spawning opportunities considerably with very little change in water regime, if specific river sections are properly inventoried. If it proves too difficult to adjust flows, it may be possible to create sections of the stream in which water is held up sufficiently (albeit with sufficient velocity), to provide an artificially enhanced spawning habitat. To do so effectively, however, would require a greater knowledge of the species than is currently available.

A seperate interest concerns the possibilities of fish spawning in lake edge areas. This has been reported anecdotally, and deserves careful examination. If indeed this occurs, and the relevant habitats and their controlling factors defined, this may substantially alter our view of the dependence of the fishery on on river habitats, and may allow river management, at least from the point of view of recruitment, to be somewhat relaxed.

Finally, concerning the overall strategy for resolving these issues; at this stage in our understanding of the system, the problem is too broad to be resolved by simple and separate technical measures. Rather, it concerns the issue of sound watershed management, based on an overall view of the priorities for the use of the Qinghai basin, and a better understanding of the relationships linking the components of the system. This relates to the long term water quality status of the lake itself as much as to the specific conditions of the rivers feeding it. The issues also concern decisions over the relative values of agricultural production, drinking water and sanitation, power supply, conservation, and tourism, as well as fisheries. Given the rather special characteristics of the area, and its recognised habitat value, a broadly-based conservation approach should be given serious consideration.

With the present limitations on available data, it is not realistic to propose a specific management policy. It is imperative however that relevant data be collected or made available, with a view to outlining such a policy. Without such a rational approach, there is little point in meddling with the fishery, as it may become profoundly affected by circumstances which are not being measured or controlled. Within a conservation policy, there would of course be the opportunity to restrain current abuses and control future development, allowing time for wider system information to be obtained and collated, pending a more comprehensive management approach. Amongst the necessary requirements for further work are:

On the basis of this information, the main technical priorities for support of the fisheries could be defined, which could include some or all of the following, depending on their relative effectiveness:

Any or all of these approaches should be made with a clear concept of where the limitations actually occur - thus if substrate is not limited, there is little point developing it significantly; if spawning stock are being overfished from the river, there is little point in operating expensive water management procedures. Priorities for action should be made strictly where there appear to be critical bottlenecks.


4.1 Trout farm development

The trout farm at Namenxia reservoir is currently under construction, due for initial operation in the spring of next year, with the introduction and hatching of the first batch of trout eggs. Work is going ahead reasonably well, and should be on schedule as required, provided current activities can be completed before the cold weather starts. Limits for normal concrete work are already being reached. However, most of the major building work has now been completed - very substantially by typical fishfarm standards.

An initial target capacity of 1 million eggs is planned, with fry rearing to a size of approximately 1g. There is ample capacity within the hatchery building to expand this level of egg intake. A spring water supply at the base of the dam is being tapped to deliver water, via a buried pipe, to the main hatchery and a set of external 8m fry tanks. The spring was originally estimated to supply 200 litres/sec, but repair works currently being completed to the interior wall of the dam, may reduce seepage, and hence supply to the spring.

In spite of earlier recommendations by project staff, the supply to the hatchery is being laid in 300mm concrete pipe, and there is considerable concern for its potential in supplying the required water flows. The options for water supply were therefore examined in further detail, as there was still the opportunity to amend specifications.

4.2 Water supply and pipe sizes

On the basis of the current hatchery plans, and its use of tanks and troughs, a total flow of approximately 30 – 40 1/sec is needed, while the external tanks would require up to 200 1/sec. Based on the planned header tank level of +3.0 m (hatchery floor datum) and the spring surface level of +3.6 m, the 0.6 m head loss would allow the pipe to deliver perhaps 55–60 1/sec (depending on the actual roughness of the concrete pipe). In similar conditions, a 400mm pipe would supply about 130 1/sec and a 500mm pipe 235 1/sec.

The internal pipework for the hatchery is to be run in 100mm plastic pipe, with 50mm and 25mm take-offs to the tanks and troughs. At the relative heads between header tank level and minimum supply level for the tanks and troughs, say + 1.4m, this pipework would carry much in excess of normal requirements. Thus by lowering the proposed header tank level to, eg + 2.6 m, a better balance can be obtained between delivery from the header and supply to it, which for the 300mm pipe could be raised to about 80 1/sec, which would just meet the short term needs of the hatchery and the initial stages of the external tank use.

This arrangement would solve the immediate problem of getting the hatchery operational, but would not for effective use of the outside tanks, which are necessary for taking the stock up to the intended production size. Nor would there be any potential for expansion inside the hatchery. The best approach would be to replace the 300mm pipe, even at this late stage, with 500mm pipe, which would accommodate the 200 1/sec planned flow, allowing for siltation and other incidental frictional losses. According to estimates, the cost of locally produced standard grade concrete pipe would actually be less than that of the higher-grade 300mm supplied to the project, which is considerably over-specified for the pressures involved.

There remains some doubt about the actual potential of the spring, once the dam has been refilled. If this fails to deliver the specified flow, some means will have to be found to capture other seepage springs in the area.

4.3 Recommendations

The effective operation of any aquaculture system is fundamentally dependent on its water supply. Without an adequate water supply there is little point in considering the production of fish. Although the initial stages of growth may proceed acceptably, shortage of water and poor environments will steadily cause loss through disease, poor growth, and poor overall condition of the stock.

The 300mm pipe is clearly inadequate for the needs of the hatchery. To avoid disturbance to the hatchery once it is operational, the recommended approach would be to replace this immediately with a larger pipe, preferably 500mm, to allow for siltation, etc. However, much of the trenching work has now been done, and the pipe laying is now almost complete. Moreover, there is some doubt about the final capacity of the spring itself.

The approach should therefore be as follows:


Varadi, L 1989

Appendix 1: Mission details


Mon 17/9Depart to Edinburgh
Tue 18/9Arrive London, collect visa am, travel to Rome
Wed 19/9Rome, ticket, administrative action
Thu 20/9Travel to Beijing
Fri 21/9Arrive Beijing, briefing FAO Beijing
Sat 22/9Travel to Lanzhou, Xining
Sun 23/9Discuss project with CTA and Fisheries Consultant
Mon 24/9Project office, meet project staff, further technical briefing, background data, request hydrology data
Tue 25/9Naked carp hatchery, collect data, develop and discuss initial design possibilities
Wed 26/9Trout farm site, assess water supplies, discuss pipe sizing with authorities, check alternative supplies
Thu 27/9Qinghai lake
Fri 28/9Qinghai lake, return to Xining
Sat 29/9Xining, visit carp hatchery, agree final design details
Sun 30/9Work on report/design details
Mon 1/10Local holiday; work on report, design
Tue 2/10Local holiday; finalise draft report
Wed 3/10Local holiday; discuss carp hatchery and trout pipe sizing, check pipe supplies for carp hatchery, materials for control systems
Thu 4/10Carp hatchery, locate materials for construction
Fri 5/10Carp hatchery, move equipment, order pumps, pipes, fittings, arrange for welding, etc.
Sat 6/10Position equipment, pipe runs, prepare instructions and sketches for installation, brief CTA and counterparts on completion needs.
Sun 7/10Travel to Lanzhou
Mon 8/10Depart Lanzhou for Beijing, debrief FAOR Beijing
Tue 9/10FAOR office, final report. Depart Beijing.
Wed 10/10Arrive Rome, debriefing Rome, discuss report
Thu 11/10Rome, finalise report
Fri 12/10Rome, report, administration
Sat 13/10Travel to UK


Hans A H Dall, FAO Representative, Beijing

Torsten Malmdorf, Programme Officer, FAO Beijing

Ad Spijkers, Senior Programme Officer, FAO Beijing

Liu Xueming, Programme Officer, Dept of International Co-operation, Ministry of Agriculture, Beijing

David Edwards, CTA, CPR/88/077, Xining

Ian Dunn, Fisheries consultant, CPR/88/077, Xining

Qiu Benchen, NPD, CPR/88/077, Xining

Ying Baicai, Deputy NPD, CPR/88/077, Xining

Wong Jilin, Leader, Lake Study Group, Fisheries Institute, Xining

Yang Hongzhi, Deputy Leader, Lake Study Group, Xining

Zhao Yimin, Leader, Fish Culture Group, Xining

Top of Page