YE Shao-Wen， ZENG Wen-Tao， CHEN Yu-Shun， ZHANG Tang-Lin， LIU Jia-Shou and LI Zhong-Jie

(1. State Key Laboratory of Freshwater Ecology and Biotechnology， Institute of Hydrobiology， Chinese Academy of Sciences，Wuhan 430072， China; 2. Shaoxing University， Shaoxing 312000， China)

Abstract: Knowledge of the vulnerability of different species and size groups to various types of fishing methods is vital to the development of optimal harvesting strategies and rational exploitation of fish resources. In this study， the selectivity of four types of fishing methods (weir nets， gill nets， electrofishing， and cormorants) used simultaneously in a typical multi-species shallow lake (Lake Niushan， China) within the Yangtze River basin， was compared for catch composition， efficiency， and size selection. Species selectivity was evaluated with the index of relative importance (IRI). Potential fishing effects on the trophic structure of fish community were analyzed by calculating the mean trophic level (MTL) of the catches and the weight ratio between prey and predator fishes. Results indicated that the use of weir nets and electrofishing should be reduced or eliminated， because they were relatively unselective and caught many species， including juvenile individuals. Fishing efforts of cormorants should be restricted due to their high fishing efficiency， especially on piscivorous fishes. Gill nets with appropriate mesh size are likely to be a more sustainable fishing method since they were more selective in capturing larger individuals and could balance the prey-predator relationship. We suggest shifting from single-species approaches toward multi-species management， and taking trophic interactions into account to better manage fishery resources.

Key words: Multi-species fisheries; Fishing selectivity; Trophic structure; Trained cormorants; Fishery management

Along the middle and lower reaches of the Yangtze River， there are numerous freshwater lakes that together represent about 30% of the total lake area in China[1]. These lakes are generally shallow(without thermal stratification)， and usually have abundant macrophytes cover， high biological productivity and a diverse community of freshwater fish(40—70 species or even more in some lakes)[2，3]. As these lakes are located in densely populated areas，most of them are intensively used for fisheries. They have long been one of the most important commercial fishing bases， with a fish yield constituting around 2/3 of the total landing of freshwater fish in China[4].

However， fish resources in the Yangtze lakes have declined severely during the past half century，mainly due to intensive human activities such as habitat fragmentation/modification， deterioration of the aquatic environment， and unsustainable commercial exploitation[3，5]. As a matter of fact， reckless overfish-ing is common in these lakes， and results mainly from poor management of the multi-species and multigears lake fisheries[2].

Extensive studies have indicated that fishing gears select catches by fish species and size (e.g.[6—8]， and can directly or indirectly affect structure and biomass of fish communities， ecological interactions among species， and productivity and functioning of ecosystems[9—11]. Therefore， knowledge of the vulnerability of different species and size groups to various types of fishing methods is vital to the development of optimal harvesting strategies and rational use of fish resources[12，13].

Given the current concern to achieve sustainability of lake fisheries in the Yangtze River basin， there are only a few studies on descriptions of catch composition and population biology of fish landed by commercial fishing[14—16]. Limited information is available on quantitative evaluation of the impacts of various fishing methods on fish composition， population structure and trophic relationship in the lakes. In the present study， we compared and evaluated selectivity by four types of commonly-used fishing methods (i.e. weir nets， gill nets， electrofishing， and cormorants) in Lake Niushan， a typical shallow macrophytic lake in the Yangtze River basin. Our main purposes were to: (1) examine catch composition and diversity by the conventional fishing methods， (2)compare species and size selectivity of the four different methods as well as their potential impacts on the trophic structure of fish community， and (3) provide basic information for an improved management of the multi-gears fishery in multi-species shallow lakes.

1 Materials and Methods

1.1 Study area

Lake Niushan (30°16′—30°22′ N， 27′—114°38′E) is a 38 km2shallow lake located on the south bank of the middle reach of the Yangtze River， within Hubei province， China (Fig. 1). This lake was originally one part of a larger lake， Lake Liangzi， which is connected to the Yangtze River by a channel. It has been separated from Lake Liangzi by a dyke since 1978. The dyke is kept closed for most time of each year (usually from March to October)， to prevent the flooding of the Yangtze River into the lake. During this study， Lake Niushan was heavily covered with two dominant submerged macrophytes:Potamogeton maackianusandMyriophyllum spicatum. Its water depth averages between 2.0 and 3.5 m， with a maximum depth of 5.0 m. Most of the lake shoreline is surrounded by villages and the land use in the drainage basin is primarily agriculture.

Fig. 1 Geographic location and sketch map of Lake Niushan in the middle Yangtze River， China

Like many other shallow lakes in the Yangtze River basin， Lake Niushan was also used intensively for stocking fisheries. In the recent two decades， there has been a shift in the stocking species from domestic carps (e.g. grass carpCtenopharyngodon idellus， bighead carpAristichthys nobilis， silver carpHypophthalmichthys molitrix， and black carpMylopharyngodon piceus) to high-quality pisivorous fishes (e.g.mandarin fishSiniperca chuatsiand topmouth culterCulter alburnus)， with the aim of maintaining the macrophyte cover and utilizing excess small fish resources[4，17]. In autumn and winter， fish populations in the lake are usually harvested by the following fishing methods.

1.2 Passive fishing methods: weir nets and gill nets

Weir nets， a kind of entrapment gear， are one of the most common and traditional fishing methods in China’s inland waters[18]. They are made of polyethylene net and bamboo in recent years， instead of wicker， shrub or reed used in ancient China. They generally consisted of three parts: fence net， catching chamber， and hoop net. Fish are intercepted and guided by the fence into the catching chamber， and then into the hoop net where fish escape is difficult or even impossible. About 80 weir nets with stretched mesh size of 5 cm were set up in Lake Niushan; the length of each fence net was more than 50 m.

The gillnets used in Lake Niushan were made of monofilament nets with stretched mesh size from 8 to 12 cm. Usually， for one couple of fishermen， about 50 gill nets (25 m long and 2 m high per net) were connected from end to end， and were spread into the lake vertically by weights and floats， setting overnight from dusk to dawn.

1.3 Active fishing methods: electrofishing and cormorants

Wooden boats (about 4 m long and 1.5 m wide)，equipped with 7.5 kilowatt diesel generator that pro-duced direct current to energize electrodes， were used by fishermen for electrofishing. The cathode was a linear array of iron poles suspending in the water， and the anode cable was tied at the end of a scoop net with hand-activated switch. One operator standing on the front of the boat held the net and carried out electrofishing by controlling the switch. Four electrofishing boats were used in the lake mainly at the daytime.

The cormorantPhalacrocorax carbohas long been trained to capture fish in China. During our study， a total of 98 trained 3—5 year-old cormorants were used by fishermen. Before the birds were released for fishing， a noose was tied around the base of the neck to stop them swallowing any fish they might catch. The cormorant team worked in the lake usually from 9:00 am to 4:00 pm with a short break at noon.

1.4 Data collection and treatment

Fish catches by the four fishing methods were surveyed randomly at landings sites of the lake in the autumn and winter of 2005. The catches were identified to species， counted and measured (TL， total length to the nearest 0.1 cm). Total weight of each species was recorded to the nearest 50 g with an electronic balance or to the nearest 1 g whenever possible.During this study， 12， 14， 8 and 13 times of surveys were conducted for weir nets， gill nets， electrofishing and cormorants， respectively.

Species diversity for each catch was presented as: (1) species richness， i.e. the number of species present in the catch， and (2) Shannon-Wiener index(H′)， using the formulawherepiis the proportion of individuals of speciesito the total number of individuals in the catch.

A measure of the relative importance of species i in the catches was estimated using an index of relative importance[5](IRI):Where %Niand %Wiis number and weight percentages of speciesiin the catches; %Fis percentage frequency of occurrence of speciesi. This index combines percentage-by-number， percentage-by-weight and frequency-of-occurrence data， and to some extent can avert the undue emphasis on number or weight of some species.

One-way ANOVA comparison was used to test for the differences in species richness， Shannon-Wiener index， and mean total length of major fish species among the four fishing methods. When overall significance was found， pair-wise comparisons were then conducted using Tukey’s honest significant differences (HSD) test to determine which fishing methods were different.

The mean trophic level (MTL) of the catches was calculated for each fishing method with the formulawhereTLiis the fractional trophic level of speciesiand %Wiis weight percentage of speciesiin the catches. The fractional trophic levels of fish species calculated by Zhang[19]were adopted here.

2 Results

2.1 Catch composition and diversity

A total of 22 fish species from six families (listed in Tab. 1) were found in the catches of the four fishing methods， among which 17 species are of commercial value and seven species (i.e.， bighead carpA.nobilis， crucian carpCarassius auratus， grass carpC.idellus， topmouth culterC. alburnus， common carpCyprinus carpio， Wuchang breamMegalobrama amblycephala， and mandarin fishS. chuatsi) were targeted by all the methods.

The total number of species caught by weir nets，gill nets， electrofishing and cormorants was 21， 10， 18 and 11 respectively (Tab. 1). Species richness of the catches was significantly higher (P＜0.05) for weir nets and electrofishing than the other two fishing methods (Fig. 2A). The Shannon-Wiener index of the catches was significantly lowest (P＜0.05) for cormorants， and was similar among the other three methods(Fig. 2B).

2.2 Selectivity on fish species

Species selectivity of the four fishing methods were evaluated and compared with the index of relative importance (IRI). According to the values ofIRIand a criterion of 10% (Tab. 2)， the selectivity of the fishing methods on the 17 commercial fish species were ranked as follows. (1) Weir nets: crucian carp ＞redfin culter ＞ sharpbelly ＞ grass carp ＞ common carp ＞yellow catfish; (2) Gill nets: crucian carp ＞ grass carp ＞mandarin fish ＞ common carp ＞ Wuchang bream; (3)Electrofishing: grass carp ＞ common carp ＞ crucian carp ＞ redfin culter; (4) Cormorants: mandarin fish ＞topmouth culter ＞ common carp ＞ crucian carp. It should be noted that cormorants were especially good at capturing mandarin fish and topmouth culter and electrofishing was efficient in catching grass carp. All the fishing methods seemed to have high selectivity on crucian carp and common carp.

2.3 Selectivity on fish size

Size selectivity of the four types of fishing methods on major commercial species were described and compared in Tab. 3.

(1) Mandarin fish.Weir nets caught a narrow length range (TL＜280 mm) of the fish. The other three fishing methods covered broader ranges， and cormorants especially captured the fish from 114 to 677 mm. Pair-wise comparisons also showed that the meanTLwas smallest for weir nets (167 mm).

Tab. 1 List of the 22 fish species captured by the four types of fishing methods in Lake Niushan， showing their presence (+) in the catches of each method

Fig. 2 Boxplots of (A) species richness and (B) Shannon-Wiener index of the fish catches calculated for the four fishing methodsThe central box covers 50% of data values and the horizontal line in the box indicates the median. Means with different letters are significantly different from each other at the 0.05 level

(2) Common carp.Weir nets and gill nets caught a wider length range than electrofishing and cormorants. The meanTLof the catches was largest for cormorants (551 mm)， with no significant differences among the other fishing methods.

(3) Crucian carp.Weir nets captured the widest length range (47—298 mm) with the smallest meanTL(123 mm). In contrast， gill nets caught the narrowest range (137—286 mm) with the largest meanTL(219 cm).

(4) Wuchang bream.Weir nets captured the widest length range (98—459 mm). The meanTLof the catches was largest for gill nets (346 mm)， with no significant differences among the other three fishing methods.

(5) Grass carp.Cormorants targeted a narrow length range (527—962 mm)， but with the largest meanTL(681 mm). The other three fishing methods covered broader ranges with similar mean TLs.

(6) Topmouth culter.Cormorants caught a wide length range with the largest individual ＞ 1 mTL.Weir nets also caught a wide length range (239—886 mm)， but mainly targeted young individuals， r-esulting in the smallest meanTL(385 mm).

2.4 Impacts on trophic structure

The 17 commercial fish species were classified into five trophic groups (Tab. 2) according to their feeding habits. It was evident that the weight percentages of different trophic groups in the catches varied greatly among the four fishing methods (Fig. 3). On the other hand， as shown in Fig. 4A， the mean trophic levels (MTL) of the catches was highest for the cormorants (3.15)， mainly due to the high proportion of carnivorous fishes (61.9%) in their catches (Fig. 3).The value of MTL was lowest for electrofishing(2.36)， as a result of dominance of herbivorous fishes(47.8%) in the catches (Fig. 3).

Crucian carp and mandarin fish， both dominant in Lake Niushan with high biomass and occurrence(Tab. 2)， are a typical pair of prey and predator. Potential impacts of the four fishing methods on the prey-predator relationship were evaluated by calculating the weight ratio between the two species in the catches (Fig. 4B): the cormorants captured less crucian carp (prey) than mandarin fish (predator) with a weight ratio of 0.23 in their catches， while the weir nets and the electrofishing captured more prey than predator with high weight ratios (20.78 and 6.15 respectively). The weight proportions of the prey and the predator were almost equal (1.28) in the catches by gill nets.

3 Discussion

Tab. 2 Compositions of commercial fish species of different trophic groups in the catches by the four fishing methods

Quantification of the roles that different fishing methods play in the utilization of fish resources has long been recognized as an important basis for rational exploitation[4，12]. Many studies have compared different fishing techniques in a number of fisheries in terms of catch composition， efficiency and size selection (e.g.[6—8]). In most cases the studies focused on one or two species， with inadequate examples of a multi-species nature. In the present study， we compared and evaluated the selectivity of four fishing methods used simultaneously in a typical multi-species shallow lake. As the results indicated， weir nets and electrofishing caught a wide variety of fish species with small individuals not able to avoid capture，while gill nets and cormorants selected a relatively narrow variety of species with larger individuals.Generally speaking， the catch composition and size structure of fishery harvests depend on the attributes of fishing methods， fish morphology， behavior and distribution， as well as key environmental factors (e.g.macrophytic cover， water depth and temperature). Accordingly， specific reasons for the capture selectivity of the four fishing methods， implication for fishery research and management are considered below.

Tab. 3 Range and mean values of total length (TL) of six major species caught by the four fishing methods

Fig. 3 Weight percentages of different fish trophic groups in the catches by the four fishing methods

Fig. 4 (A) Mean trophic levels of the catches and (B) weight ratios of prey (crucian carp) to its major predator (mandarin fish) in the catches by the the four types of fishing methods

The weir nets are colloquially called ‘mazing nets’ in China， because fish intercepted by the fences are inescapably lost in the elaborate catching chamber and ultimately trapped into the hoop nets alive. In our study， the higher diversity of their catches can be ascribed to several reasons. (1) The weir nets were usually deployed in macrophyte environment where fish diversity and abundance are positively related with the complex habitats[17]. (2) The long extending fences covered the whole water column from lake bottom to surface， so that they can block the movement of fishes with different spatial niches. (3) The unrestricted fine mesh (only 3 cm) contributed to more small fish species and higher proportion of juvenile fish in the catches. On the other hand， as a passive fishing method， the efficiency and selectivity of weir nets can be influenced greatly by fish activity and behavior[20，21]. For instance， the nets had higher capture efficiency on active species (e.g. redfin culter，crucian carp and bighead carp) than less active species (e.g. black carp and dark sleeper). In particular，they were inefficient on the mandarin fish， a demersal stalking piscivorous species[22]. Thus， further studies in terms of fish behavior and relevant environ-mental factors are necessary for a better understanding of selectivity by the weir nets.

Species and size selectivity of gill nets are generally determined by mesh size， deployment of the nets， as well as fish morphology and activity[20，21]. In Lake Niushan， the bottom-set gill nets with medium mesh size were used by fishermen to target demersal commercial species (especially the valuable mandarin fish). Small mesh size was not considered there because small fishes are of low or no market values and time-consuming to pick off from the nets. As a consequence， the gill nets contained a lower diversity of fish catches. The gill nets were efficient in harvesting mandarin fish by tangling its teeth， maxillaries or spines， while they were incapable of capturing the bottom-dwelling snakehead fish perhaps due to their cylindrical and slippery body. On the other hand，young individuals of commercial species were not the most abundant size group in the catches. Size selectivity curves for various mesh sizes of gill nets have been estimated for many species in other areas[6—8]， but no similar knowledges are available for the Yangtze lakes until now. Experimental researches on size selectivity of gill nets in the Yangtze lake environment are therefore urgently needed， so that correct mesh sizes can be determined for the sustainable management of fish populations.

Similarly， two categories of factors are considered for fishing characteristics of the electrofishing in Lake Niushan. (1)Biological factors. The efficiency of electrofishing can be influenced by fish behavior， morphology， and distribution[21]. Specifically，the benthic fishes (Monopterus albus，Misgurnus anguillicadatus，Odontobutis potamophila) that sink to lake bottom when stunned， were difficult to capture by electrofishing. Larger fish individuals， such as the largest grass carp (TL=1025 mm) in our study， tended to be selected by electrofishing. The common explanation is that total body voltage increases with fish length at a given voltage gradient， resulting in a greater electroshock to larger fish[23]. Predatory fish may be more vulnerable to electrofishing， not only because of their larger size but also because their territorial behavior makes them less likely to avoid an electrofishing operation. (2)Environmental factors. As the electrofishing was carried out mainly in shallow macrophytic areas of the lake， the species that diurnally inhabit nearshore macrophyte covers (e.g. sneakhead fish， grass carp， and mandarin fish) were usually more vulnerable than pelagic species (e.g. topmouth carp and bighead fish). Relatively low water temperature in the late autumn and winter might decrease fish metabolism， leading to a decreasing ability to perceive and escape an electric field. Moreover，stunned fish may be easier to capture at low temperatures because they recover less quickly.

The Chinese people have practiced fishing with cormorants since ancient times. However， training cormorants has become too complicated and unprofitable in modern times to be carried out. It seems that today this method is used for fishing in limited inland waters of China[18]. For instance， the fishermen with their cormorants in this study all came from a little village in Hubei Province. The trained cormorants hunted bottom-dwelling fish by diving from lake surface， using their wings and webbed feet for underwater propulsion. Our investigation indicated that they were most efficient in capturing mandarin fish with great importance (IRI=97.2%) in the catches and secondly efficient on topmouth culter (IRI=25.7%)，another high-priced piscivorous fish. As we observed，the cormorants were so fierce that they were able to bring huge individuals of the two species (max 677 and 1044 cmTLrespectively) to the lake surface.However， they were inefficient on grass carp， perhaps due to visual and swimming barriers created by the submerged macrophyte cover. In addition， turbidity and bad weather caused by wind or rain could also influence catch composition and fishing efficiency， although these environmental factors were not explicitly considered in this study.

Pauly，et al.[24]argued that the mean trophic level of landed fish could be used as a simple index of sustainability in multi-species fisheries. Managing for sustainability and trophic level maintenance can be improved by determining the trophic levels by various fishing methods and adjusting the mix of methods to maintain the full trophic composition of the fishery. In this study， fishing effects on the mean trophic level of the catches were compared and the cormorants appeared to have the most distinct selectivity. They selectively removed piscivorous species at the highest trophic levels and might have the highest potential ecosystem effects through direct and indirect relationship among the biological components. On the other hand， Friedlander and DeMartini[25]suggest that harvesting top predators may have reduced consequences if lower trophic levels are extracted in equal proportions. By calculating the weight ratios between the crucial carp and the mandarin fish in the catches， we evaluated the possible impacts of the four fishing methods on prey-predator relationship. In our view， the gill nets were more likely to balance the prey-predator relationship， and the rational mixed use of weir nets and cormorants could also balance the relationship because of their opposite weight ratios.Furthermore， special attentions should be paid in the future to trophic network modeling approaches such as ECOPATH/ECOSIM[26，27]， which could provide insights into the food web dynamics under exploitation by various fishing techniques in the Yangtze lake ecosystems.

In summary， this study addressed selectivity by various fishing methods used in the Yangtze multispecies lake. From a fishery management perspective，the use of the fishing techniques and understanding the factors that influence their use are vital to improve the chances for achieving fisheries sustainability. According to the results， the use of weir nets and electrofishing should be reduced or eliminated， since these relatively unselective fishing methods caught diverse species， including juvenile individuals. Fishing efforts of the cormorants should be restricted due to their high fishing efficiency， especially on the top predators. Gill nets are likely to be a more sustainable fishing method because they were more selective in capturing larger individuals. By the use of the appropriate mesh size and a high degree of uniformity in the size of all meshes， the gill nets can be highly selective. Here we suggest shifting from single-species approaches toward multi-species management. We also advocate taking trophic interactions into account to better manage fishery resources.