A study to assess the status of the large scale commercial fisheries operating in southern Lake Malawi was conducted using generalised linear modelling and Principles of Precautionary Approach. The generalised linear modelling analysis was used to standardise catch per unit of effort data from 1976 to 2009 and to determine the effects of fishing vessel characteristics such as engine size and hydro-acoustic devices on fish catch rates. The Precautionary Approach, through reference points like Bmax (reference point corresponding to period of relatively high fish catch rate), Bpa (reference point at which precautionary management action should be undertaken), Blim (limit reference point beyond which a fishery is considered seriously depleted) and Bcur (reference point corresponding to current fish catch rate), was used to assess the status of fish stocks using standardised fish catch rate which approximated fish biomass or stock size.

Results of the assessment indicated that the current fish biomass level in all three categories, bottom pair trawl, stern bottom trawl and stern mid-water trawl fisheries, is well above Blim, but is 9–12% above Bpa, which is very close suggesting that the fishery is fully exploited and precautionary management measures are required. Suggested measures included introduction of fishing closed season for the large scale commercial fisheries which will be dependent on the biology of the most important target fish stocks, relocation or reduction of fishing effort in the current fishing grounds and intensification of patrols to reduce incidences of illegal, unreported and unregulated fishing activities.

Introduction

Lake Malawi, situated in the African rift valley between 9°30”S and 14°30”S and bordered by Malawi, Mozambique and Tanzania, is the third largest lake in Africa with an average depth of 292 m (Patterson and Kachinjika, 1995). The shallow areas are the most productive and vast expanses exist in the southeast and southwest arms of the lake (Figure 1).

The lake supports a highly diverse capture fishery that can be categorised as large-scale commercial, small-scale commercial and subsistence (Banda et al., 2001). The large-scale commercial fishery is a mechanised fishery that operates trawls, purse seines or lift nets and it started in 1968 (Tarbit, 1972a). The fishery is now well established in areas D, E and F of the south west arm and areas A, B and C of the southeast arm of Lake Malawi. There are currently over 14 registered fishing units of which 43% are stern bottom trawlers while 28% are stern semi-pelagic trawlers with the rest being stern bottom pair trawlers (GoM, 2010).

Total annual yield from all species in the lake is estimated at over 90,000 tonnes of which the large scale and small-scale fisheries account for about 5% and 95%, respectively. For the last five years fish landings in the large scale commercial fisheries averaged 4,192 tonnes annually. The most dominant species groups were the cichlid genera Diplotaxodon (Ndunduma), Copadichromis (Utaka) and Rhamphochromis (Ncheni) with 47%, 31% and 5%, respectively. Catfishes of the genera Bathyclarias and Bagrus and other non-catfish species groups accounted for 17% altogether.

Malawian fish stocks especially those within shallow waters have declined in recent years. The main reason for this decline is probably excessive fishing effort. Another relevant cause can be sought in destructive fishing practices such as use of small meshed gears and fishing for juveniles (Palsson et al., 1999). The decline of the fisheries has stressed the need for efficient fisheries management, based on scientific knowledge. The aim of this report is to present this knowledge base as far as possible under the present conditions. In addition, our interpretation of this knowledge base is used to make a number of specific recommendations to act on the analysis of that knowledge, i.e. a translation of scientific knowledge into fisheries management actions.

Materials and methods

Data collection

Catch and effort data from fishing companies operating in southern Lake Malawi are available from 1976 to date. Over ten groups of fish species including Chambo, Chisawasawa, Ndunduma, Ncheni, Bombe, Kampango, Utaka, Usipa and Mbaba among others are usually caught using purse seines, semi-pelagic and demersal trawl nets. The catches of fish are usually recorded on a daily basis. Many other quantities are also recorded, including: effort levels (pulls or days), engine horse-power, vessel tonnage, date, vessel name, type of gear and area fished.

Methodology

The precautionary approach

In agreement with current practices in fisheries management an attempt is made to apply the principles of the precautionary approach. The FAO Code of Conduct for Responsible Fisheries, adopted in 1995, stipulates in article 7.5: “States should apply the precautionary approach widely to conservation, management and exploitation of living aquatic resources in order to protect them and preserve the aquatic environment. The absence of adequate scientific information should not be used as a reason for postponing or failing to take conservation and management measures.” And in Paragraph 12.13 of Article 12: “States should promote the use of research results as a basis for the setting of management objectives, reference points and performance criteria as well as for ensuring adequate linkage between applied research and fisheries management.”

“A precautionary reference point is an estimated value derived through an agreed scientific procedure which corresponds to the state of the resource and of the fishery, and which can be used as a guide for fisheries management” (FAO, 1997). Precautionary reference points are often based on time series of age dependent models, which are presently not applicable to Malawian fish stocks due to limited amount of age-based data, but can also be based on age independent data. Examples of reference points are various levels of fishing mortality (F) such as F0.1, FMSY, Fmed, Fhigh or different levels of biomass (B).

However, in situations of limited data alternative biomass estimates can be used, such as CPUE (Catch Per Unit of Effort), as an estimator of stock size or biomass. A simple approach in this case is to select a given CPUE value as a reference point. In the absence of virgin biomass this can be the maximum CPUE observed during past years or, preferably, the mean CPUE over a period of relatively high CPUE. This value is referred to as Bmax in this report. The next step is to define a biomass limit (Blim) below which the stock would be considered seriously depleted and even in danger of a collapse and, therefore, which should be avoided with very high probability. The choice of this point is rather subjective and a value of 20% of Bmax is suggested. Since there should be a high probability of staying away from the Blim level, another value is needed at which a precautionary management approach must be undertaken in order to avoid the stock dropping below Blim. This value, Bpa (pa = precautionary approach) can be calculated as, Bpa = Blim exp (2*σ) where σ is a measure of uncertainty in the biomass estimate and the constant 2 reflects the approximate 95% confidence. The value of σ is usually taken as 0.2–0.3. However, in view of the high uncertainty in stock size estimates for Malawian fisheries a value of 0.4 is suggested. This results in a Bpa = 0.45, i.e. 45% of Bmax.

Thus, when the stock appears to drop below Bpa a recommendation should be made to reduce fishing effort. The extent of the effort reduction should be in accordance with how close the point estimate of biomass is to Bpa and Blim. The closer the point estimate is to Blim the more reduction in effort should be recommended. If the point estimate is at or even below Blim a closure of the fishery would seem the only realistic action. The point estimate of a stock, Bcur (%), is defined as the current level of the biomass (CPUE) relative to Bmax. In this report the reference years for calculation of a mean Bcur are the last 3 years available in the data series.

In view of the actual values chosen for Blim and Bpa various criteria should be considered, such as uncertainties in size and productivity of the stocks. This uncertainty, although difficult to quantify, must be regarded as high for Malawian stocks due to limited and unreliable data. Furthermore, most Malawian fish stocks are characterised by low fecundity and therefore more vulnerable to excessive fishing pressure than more fecund stocks. In view of this a Bpa less than 45% of Bmax would be regarded as inappropriate. (For further details of this subject a reference is made to: ICES CM 1997/Assess:7. Report of the study group on the precautionary approach to fisheries management. International Council for the Exploration of the Sea, Copenhagen.)

Standardisation of catch rate data

The use of catch rate data as an index of abundance depends on being able to adjust for the impact on catch rates of changes over time of factors other than abundance (Maunder and Punt, 2004). Generalised linear models (GLMs) are general and powerful statistical techniques that are commonly used for standardising catch and effort data (GLMs; Nelder and Wedderburn, 1972). They are defined by the statistical distribution for the response variable (usually catch rate) and how some linear combination of a set of explanatory variables relate to the expected value of the response variable (Maunder and Punt, 2004). To apply a GLM it is necessary to: (a) choose a response variable, (b) select a sampling distribution for the response variable, (c) choose a link function appropriate to the distribution and (d) select a set of explanatory variables (Maunder and Punt, 2004).

In this study the process of standardising catch rate data was done using generalized linear modelling assuming a continuous distribution and included year as one of the explanatory variables to detect trends in abundance over time (Xiao, 2004; Maunder and Punt, 2004) but also to understand what determines the catching power of individual fishing vessels (Hilborn and Walters, 1992). Taking fishing vessels as discrete entities, their catch rate can be obtained as the product of the abundance in a particular year times the efficiency of the vessel class (Hilborn and Walters, 1992) such that:
formula
(1)
where U is the catch rate, A is the abundance of fish, and q is the efficiency. The subscript t refers to time and i refers to vessel class. As a statistical model it can be written as follows:
formula
(2)
where U11 is the catch rate obtained by the first vessel class in the first time period, αt is a factor that is the abundance in year t relative to year 1, βi is the efficiency of vessel class i relative to vessel class 1, and ti is a factor that accounts for the deviation between the observed Uti and the expected value t and i.
The linear statistical model (equation 3) is obtained by taking logarithms of both sides of equation 2 such that:
formula
(3)
and can be used through the GLM approach to estimate log(U11), α and β.
To incorporate factors like vessel tonnage, horsepower and the presence or absence of a fish finder (colour hydroacoustic sounder), model 3 above was modified as follows (Hilborn and Walters, 1992):
formula
(4)
where now Uti is the catch rate of the ith vessel in the tth year, αt as before is a scale factor for the catch rate in year t relative to year 1, β1 is the log of the increase in catch rate expected for each unit of tonnage, Ti is the tonnage of the ith vessel, β2 is the log of the increase in catch rate expected for each unit of horsepower, Hi is the horsepower of the ith vessel, γ is the log of the increase in catch rate expected if the vessel is equipped with a hydro-acoustic device or fish finder, and Εi takes the value 0 if the vessel does not have the device and 1 if it has. ti is the residual for the ith vessel in the tth year.

Input parameters for vessel characteristics

Input parameters for the GLM analysis are shown in Table 1 and included engine horse power (Hp) of the fishing vessels included in the data series and presence or absence of a fish finder or hydro-acoustic device. Values for engine horse power are mean values but there were wide variations especially for stern pelagic and stern bottom trawlers as observed from the standard deviation values in Table 1.

Results

Pair trawl fishery

Results of analysis for the pair trawl fishery are shown in Figure 2 and Table 2. There is a general declining trend in cpue especially between 1976 and 2009 with highest cpue values occurring between 1976 and 1983. The cpue declined from an average of 1,398 kg day−1 between 1976 and 1983 to just around 685 kg day−1 between 2007 and 2009. The detailed analysis as per principles of Precautionary Approach are shown in Table 2. Bmax was estimated at 1,398 kg day−1 and corresponds to the highest average cpue attained by the fishery which was between 1976 and 1983 while Bcur was estimated at 685 kg day−1 and corresponds to the current average catch rate of between 2007 and 2009. The point at which precautionary management measures must be undertaken (Bpa) was estimated at 629 kg day−1. Blim which is the point at which the stock should be considered seriously depleted and in danger of collapse was estimated at 279 kg day−1.

Comparison of Bcur against Blim and Bpa indicates that the current fish biomass level in the pair trawl fishery is well above Blim but is just 9% greater than Bpa suggesting that the fishery is fully exploited and precautionary management measures are required for the fishery. The suggested measures include:

  • Institute fishing closed season for major species harvested by the sub-sector.

  • Relocate/reduce fishing effort.

  • Intensify patrols to reduce incidences of illegal, unreported and unregulated fishing activities including usage of under-meshed cod ends.

Stern bottom trawl fishery

Results of analysis for the stern bottom trawl fishery are shown in Figure 3 and Table 3. There is a general declining trend in cpue especially between 1976 and 2009 although in the early 1990 s the fishery experienced a sharp drop in catch rates before picking up again in the mid 1990 s. However highest cpue values averaging 2,552 kg day−1 occurred between 1976 and 1984 and declined to an average of 1,285 kg day−1 between 2007 and 2009. Detailed results of the analysis are as shown in Table 3.

Bmax was estimated at 2552 kg day−1 and corresponds to the highest average cpue attained by the fishery which was between 1976 and 1984 while Bcur was estimated at 1285 kg day−1 and corresponds to the current average catch rate of between 2007 and 2009.

The point at which precautionary management measures must be undertaken (Bpa) was estimated at 1,148 kg day−1. Blim which is the point at which the stock should be considered seriously depleted and in danger of collapse was estimated at 510 kg day−1. Comparison of Bcur against Blim and Bpa indicates that the current fish biomass level in the stern bottom trawl fishery is well above Blim but is just 12% greater than Bpa suggesting that the fishery is fully exploited and precautionary management measures are required for the fishery (Table 3). The suggested measures are the same as those of the pair trawl fishery and include:

  • Institute fishing closed season for major species harvested by the sub-sector.

  • Relocate/reduce fishing effort.

  • Intensify patrols to reduce incidences of illegal, unreported and unregulated fishing activities including usage of under-meshed cod ends.

Stern semi-pelagic trawl fishery

Results of analysis for the stern bottom trawl fishery are shown in Figure 4 and Table 4. A general declining trend in cpue is evident especially between 1976 and 2009 although a sharp drop in catch rates occurred in the early 1990 s but picked up again in the mid 1990 s. Highest cpue values averaging 4,652 kg day−1 occurred between 1976 and 1984 and declined to an average of 2,342 kg day−1 between 2007 and 2009. As indicated in Table 4, Bmax was estimated at 4,652 kg day−1 and corresponds to the highest average cpue attained by the fishery in the early 1980 s while Bcur was estimated at 2,342 kg day−1 and corresponds to the current average catch rate of between 2007 and 2009. The point at which precautionary management measures must be undertaken (Bpa) was estimated at 2,093 kg day−1. Blim which is the point at which the stock should be considered seriously depleted and in danger of collapse was estimated at 930 kg day−1. Comparison of Bcur against Blim and Bpa indicates that the current fish biomass level in the stern semi-pelagic trawl fishery is well above Blim but is just 12% greater than Bpa suggesting that the fishery is fully exploited and precautionary management measures are required for the fishery (Table 4).

The suggested measures are the same as those of the previous fisheries and include:

  • Institute fishing closed season for major species harvested by the sub-sector.

  • Relocate/reduce fishing effort.

  • Intensify patrols to reduce incidences of illegal, unreported and unregulated fishing activities.

Status of the whole fishery

Results of analysis for the whole large scale trawl fisheries are shown in Figure 5 and Table 5. Relative biomass index fluctuated with a general declining trend between 1976 and 2009. Highest relative biomass index values averaging 0.84 occurred between 1976 and 1984 and declined to an average of 0.43 between 2007 and 2009.

As indicated in Table 5, Bmax was estimated at 0.84 and corresponds to the highest average relative biomass idex attained by the fishery in the late 1970 s and early 1980 s while Bcur was estimated at 0.43 and corresponds to the current average relative biomass index of between 2007 and 2009. The point at which precautionary management measures must be undertaken (Bpa) was estimated at 0.38 while Blim was estimated at 0.17.

Comparison of Bcur against Blim and Bpa indicates that the current fish biomass level in the large scale trawl fisheries as a whole is well above Blim but is just 13% greater than Bpa suggesting that the fishery is fully exploited and precautionary management measures are required for the fishery (Table 5). The suggested measures are the same as those of the previous fisheries and include:

  • Institute fishing closed season for major species harvested by the sub-sector.

  • Relocate/reduce fishing effort.

  • Intensify patrols to reduce incidences of illegal, unreported and unregulated fishing activities.

Effect of vessel characteristics on catch rates

Vessel characteristics that were assessed in the GLM analysis are Engine size in terms of horse power (Hp) and presence or absence of a Hydro-acoustic device or fish finder. The following are some of the findings.

Engine size (Hp) and make

Over five different makes of engines are operational in fishing vessels operating in southern Lake Malawi. Of the five brands Caterpillar engines are the most popular and are installed in 43% of the fishing vessels followed by Cummings (29%), Volvo Penta (14%), Lister Peter and Yamaha with 7% each (GoM, 2010). There is also great variation in terms of engine size among the various fishing units (Figure 6). Generally engine horse power varied from 42 to 380 Hp with an average of 192 ± 112.5 Hp. Engine power for pair trawlers averaged 44 ± 1.4 Hp while that of the larger fishing units ranged from 125 to 380 Hp with an average of 251.7 ± 67.7 Hp (GoM 2010). However results from the GLM analysis indicate that engine size had little effect on catch rates. In fact engine size contributed less than 0.00001% to increase in catch rates for all fishing crafts lumped together.

Effect of hydro-acoustic devices

Hydro-acoustic devices or fish finders are electronic devices that are fitted on fishing vessels to help in locating fish. These devices have varying capabilities. Some are a combination of Global Positioning System (GPS) and an echo sounder/fish finder (Figure 7). There are some which are fish finders only. They can indicate direction of movement, size or magnitude of fish schools and depth at which the fish are found. They are therefore highly efficient tools for hunting fish.

Of the three categories of fishing units operating on Lake Malawi, fish finders are found only Stern bottom and Stern semi-pelagic trawlers. None have so far been found on pair trawlers. Results from the GLM analysis indicate that fish finders have great influence on catch rates. Generally catch rates increased by 7% on fishing vessels equipped with such gadgets.

Discussion

Scientific advice on fisheries management is generally based on the results of the application of some form of stock assessment technique. This usually involves estimating the parameters of some form of population dynamics model by fitting it to research and monitoring data and using the results of the fitting process to estimate quantities that are of interest to the decision makers (Maunder and Punt, 2004; Hilborn and Walters, 1992). In this report Generalised linear models (GLMs) were used to standardise catch per unit of effort data from which precautionary reference points such as Bmax, Bpa and Blim were derived.

Current catch rates for all three categories of large scale commercial fisheries assessed in this study Pair-trawl, Stern bottom trawl and Stern semi-pelagic trawl were 9.0%, 11.90% and 11.91% slightly greater than Bpa – the precautionary reference point at which precautionary fisheries management measures must be undertaken. In addition the calculated overall current relative biomass index for all the three categories was 13% greater than Bpa which is very close to Bpa suggesting that the fish stocks exploited by the large scale commercial fisheries operating in southern Lake Malawi are fully exploited and precautionary management measures are therefore required to address the situation.

The findings from this study are in agreement with those from previous studies done by Bell et al. (2012), Kanyerere et al. (2005), Bulirani et al. (1999) and Turner (1995) among others who observed that in the last decade, the annual catch landings especially of large fish species have declined significantly in Lake Malawi. This decline has been attributed to overfishing especially of inshore fish stocks in Lake Malawi which has been accompanied by changes in the fisheries where large fish species like Chambo (Oreochromis species) among others have been replaced by smaller fish species (Turner et al, 1995). In the 1990s Chambo which consists of three endemic Oreochromis (Nyasalapia) species O. karongae (Trewavas), O. squampinnis (Gunther) and O. lidole (Trewavas) was the mainstay of both the traditional and large scale commercial fisheries in southern Lake Malawi but has since been severely over-fished to almost extinction (Bell et al., 2012; Bulirani et al., 1999). There has also been no significant compensatory increase in abundance of smaller species in response to a decline in the larger species (Turner et al., 1995). The need for urgent management action to prevent the fishery from declining further cannot therefore be over-emphasized.

As such this study recommends that: (i) The Government of Malawi through the Department of Fisheries and in collaboration with other stakeholders should strengthen the current co-management framework that advocates for a more consultative and participatory approach to fisheries resource management (Njaya and Donda, 2007) including aspects of participatory monitoring, control and surveillance, (ii) Institute fishing closed season for both the mechanised and the traditional fisheries to protect major fish species harvested by both sectors. (iii) further biological studies are recommended so as to fully understand the population dynamics of major fish species targeted by both sectors. This would allow far more advanced age-structured population dynamics models such as multi-species yield per recruit (YPR) and multi-species virtual population analysis (VPA) to be used in obtaining refined management advice for the fisheries. (iv) Reduce or relocate fishing effort to other lightly fished areas such as central and northern parts of Lake Malawi. (v) Intensify patrols to reduce incidences of illegal, unreported and unregulated fishing activities.

Although there are opportunities of expanding and relocating some of the fishing units to central and northern parts of Lake Malawi as other operators such as Maldeco Fisheries Ltd and Linga Fishing Company have done, there are some technical and logistical challenges that have to be overcome. Technical challenges include lack of landing and maintenance facilities for trawlers especially starting from Salima going northwards. With the exception of Maldeco Fisheries Ltd most of the operators rely on dockyards and service centres that are concentrated in southern Lake Malawi particularly in Mangochi district. In terms of logistics, most of the smaller operators do not have trucks that can distribute their fish catch to distant markets such as Blantyre, Lilongwe and Mzuzu. Relocating such operators to fishing grounds other than those in Mangochi district will not be easy.

Conclusions

To reduce over capacity in southern Lake Malawi, it is recommended that there should be deliberate efforts by Government or private investors to construct landing and maintenance facilities within central Lake Malawi. This would ensure that fishing vessels allocated to operate in central and northern Lake Malawi have easy access to fish handling and vessel maintenance facilities. This would also help reduce down time for fishing vessels.

Considering the multi-species nature of the fishery and complex interaction between the artisanal and mechanised fisheries, this study strongly recommends further biological and socio-economic studies to better understand and manage this complex fishery.

Acknowledgements

This report would not have existed without the support and dedication to duty of all FRU researchers and technicians. We also thank all owners of fishing units who timely submit their catch returns to the Department of Fisheries.

References

References
Banda, M.C., Jambo, C.M., Kachinjika, O., Weyl, O.L.F.,
2001
. Fisheries resource user groups in Malawi.
Short Communication 3
.
National Aquatic Resource Management Programme (NARMAP)
,
Monkey Bay
.
Bell, R.J., Collie, J.S., Jamu, D., Banda, M.,
2012
.
Changes in the Biomass of Chambo in the Southeast Arm of Lake Malawi: A Stock Assessment of Oreochromis spp
.
Journal of Great Lakes Research
38
,
720
729
. doi:.
Bulirani, A.E., Banda, M.C., Palsson, K.P., Weyl, O.L.F., Kanyerere, G.Z., Manase, M..M, Sipawe, R.D.,
1999
.
Fish Stocks and Fisheries of Malawian Waters
.
Resource Report 1999
.
Government of Malawi, Fisheries Department
,
Lilongwe, Malawi
.
FAO
,
1997
.
FAO Technical Guidelines for Responsible Fisheries. No. 4
.
FAO
,
Rome
.
GoM
,
2010
.
Annual Frame Survey for Large Scale Commercial Fisheries 2010
.
Fisheries Department
,
Lilongwe, Malawi
.
Hilborn, R., Walters, C.J.,
1992
.
Quantitative Fisheries Stock Assessment
.
Chapman and Hall
,
London
.
Kanyerere, G.Z., Booth, T., Weyl, O.,
2005
.
Management advice for the Diplotaxodon limnothrissa (Teleostei: Cichlidae) resource in the south-east arm of Lake Malawi, based on per-recruit analysis
.
African Journal of Aquatic Science
30
(
2
),
141
145
(5)
.
Maunder, M.N., Punt, A.E.,
2004
.
Standardising catch and effort data: A review of recent approaches
.
Fish. Res.
70
,
141
159
.
Njaya, F., Donda, S.,
2007
.
Fisheries Co-management in Malawi: An Analysis of the Underlying Policy Process
.
World Fish Centre
,
Cairo, Egypt
.
Nelder, J.A., Wedderburn, R.W.M.,
1972
.
Generalised linear models
.
J.R. Statist. Soc. A
137
,
370
384
.
Palsson, O.K., Banda, M.C., Bulirani, A.,
1999
. Review of Demersal Monitoring Surveys in Southern Lake Malawi.
Fisheries Bulletin No. 40
.
Fisheries Department
,
Lilongwe
.
Patterson, G., Kachinjika, O.,
1995
. Limnology and Phytoplankton Ecology. In: A. Menz (Ed.),
The Fishery Potential and Productivity of the Pelagic Zone of Lake Malawi/Niassa
, pp.
1
67
.
Natural Resources Institute
,
Chatham
.
Tarbit, J.,
1972
. Lake Malawi trawling survey: interim report 1969–1971.
Fisheries Bulletin 2
, pp.
1
16
.
Malawi
.
Turner, G.F., Tweddle, D., Makwinja, R.D.,
1995
. Changes in demersal cichlid communities as a result of trawling in southern Lake Malawi. In: T.J. Pitcher, P.J.B. Hart (Eds.),
The Impact of Species Changes in African Lakes
,
397
412
.
Chapman and Hall
,
London
.
Xiao, Y.,
2004
.
Use of individual types of fishing effort in analyzing catch and effort data by use of a generalized linear model
.
Fish. Res.
70
,
311
318
.

Note

Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/uaem.