The behaviour of humpback dolphins Sousa chinensis at the Richards Bay shark nets: efficacy of acoustic deterrents.

 

Cetaceans and fishing nets

Worldwide, a variety of small cetaceans are caught and killed in fishing nets (Jefferson and Curry 1994, Cockcroft 1994, Goodson 1997 and Julian 1997). Although some fisheries directly target cetaceans, either as markers (indicating the presence of the target commercial fish) or as a convenient source of bait (Goodson 1997), usually the cetacean catches are incidental. However, the incidental capture of small cetaceans by fisheries was identified by the International Whaling Commission (1994) as the single greatest cause of cetacean mortality (Cockcroft 1994). Cetaceans in South African waters also face the threats of indiscriminate fishing gear set by commercial fisheries (Ross and Best 1995).

In KwaZulu-Natal, small cetaceans are also caught in non-commercial fishing nets set at a number of beaches to protect bathers from sharks. These shark nets have effectively diminished the probability of shark attack by substantially reducing the number of sharks residing in the netted areas (Davis et al. 1995). The shark nets are permanent and hang stationary in the water about 500 m from the shore and trap any animal that cannot pass through the 25 cm mesh. Thus the shark nets are not shark-specific and, besides the 21 853 sharks caught between 1981 and 1993, 1 059 dolphins were caught, as well as 916 turtles and 4 841 batoids [(rays and guitarfish); Davis et al. 1995]. The inshore population sizes of most of these different animals are unknown, making it difficult to put the number of mortalities into perspective. However, conservation measures should be taken so that the local populations of these animals are not further depleted.

Various species of cetaceans have been caught in the shark nets. Indian Ocean bottlenose dolphins Tursiops truncatus, Indo-Pacific humpback dolphins Sousa chinensis and common dolphins Delphinus delphis are caught frequently (Cockcroft 1990). Other species are caught less frequently, e.g. spinner dolphins Stenella longirostris, striped dolphins S. coeruleoalba, Fraser’s dolphins Lagenodelphis hosei and false killer whales Pseudorca crassidens (Cockcroft 1990). The infrequently-caught species inhabit the offshore and pelagic zones (Carwardine 1995) and rarely enter the inshore region, reducing the chances of capture. Although the common dolphin is an offshore species, during the austral winter these animals follow the annual migration of sardines Sardinops ocellatus northwards along the coast (Armstrong and Thomas 1995, Ross and Best 1995), potentially increasing the chances of capture. Approximately 70% of common dolphin captures between 1980 and 1988 occurred in winter (June, July and August; Cockcroft 1990).

The 316 common dolphins caught between 1980 and 1988 probably represent an annual loss of less than 1% of the total population (Cockcroft 1994). The situation is somewhat different for bottlenose and humpback dolphins. These two species both use the inshore region throughout the year (Durham 1994, Peddemors 1995, Karczmarksi 1996). This is confirmed by the catch data which show that they are caught in the shark nets in all months of the year (Cockcroft 1990). The capture and death of a minimum 67 humpback and 279 bottlenose dolphins between 1980 and 1988 represented an annual decrease of 4% and 3.5%, respectively, of their estimated populations (Cockcroft 1990). The International Whaling Commission (1994) idenrified the humpback and bottlenose dolphins of KwaZulu-Natal as populations experiencing levels of mortality that are probably not sustainable.

In an attempt to understand the causes of dolphin capture in shark nets, Cockcroft (1994) examined biological, environmental and physiographic parameters (table 1) pertaining to the capture of the frequently-caught species (common, bottlenose and humpback dolphins). Most of these parameters were not correlated with capture, and of those parameters that appeared to be correlated, none was common to all three species. Since the shark nets are now lifted while the sardine migration traverses the KwaZulu-Natal coast, and since the common dolphin population does not appear to be threatened by the nets, I will consider the information pertaining to bottlenose and humpback dolphin captures only. The only parameter common to capture in these two species is that current direction was significantly different to normal on days of capture (Cockcroft 1994). Other parameters (listed below) were in direct contrast.

The difference in the catch statistics of the four northerly installations may reflect the difference in distribution of the two species, even though they overlap over much of their range (Saayman and Taylor 1979, Karczmarski 1996, Cockeron 1990 in Durham 1994). The difference in "captured" age-sex classes of the two species may indicate that there is sexual segregation in habitat use (Cockcroft 1994). The "stomach-fullness results" are rather confusing unless bottlenose dolphins are usually caught during feeding, whereas the humpback dolphins are caught at the onset of feeding. It is possible that the stomach-fullness measurement is crude (Cockcroft 1994) and therefore misleading. Perhaps the high proportion of young (incompletely weaned) bottlenose dolphins confounds this measure. The only unequivocal conclusion that Cockcroft (1994) could draw about factors affecting capture is that capture occurs where the presence of the dolphins coincides with the presence of the nets.

 

Warning devices

Attempts have been made to alert dolphins to the presence of nets and various devices have been set into experimental nets as visual, auditory and echolocatory stimuli. Peddemors et al. (1990) incorporated three types of visual and/or echolocatory devices into the shark nets: plasticised aluminium foil; aluminium discs and stainless steel wire. These devices should act as reflectors making it easier to see, or echolocate, the extent of the net. Unfortunately, they were not hardy enough: the aluminium foil and disks corroded and they caused net entanglement; the stainless steel wire broke soon after installation (Peddemors et al. 1990). Clangers, rattles and bell buoys were devices designed to emit sound signals from the nets to stimulate hearing. Since the energy source of these devices was wave action, they did not function continually. While the rattles and bell buoys were deployed, dolphins were observed in the vicinity of the nets but no reaction to the devices was apparent. Subsequently, a net with rattles caught a young female bottlenose dolphin and a net with the bell buoy caught a young female humpback dolphin. After a while, the bell buoy was lost and the clangers corroded. Any device to be set permanently in the nets will have to be sufficiently hardy to withstand the harsh underwater environment; in addition, a more reliable source of energy should be used to produce a more effective sound signal.

To stimulate echolocation, Peddemors (1995) set small, air-filled sonar reflectors into experimental shark nets which dissected a well-used path of bottlenose dolphins. The results showed that the bottlenose dolphins could detect both the experimental (reflector enhanced) and control (ordinary) shark nets. Surfacing patterns and sound emission by the dolphins suggested that the dolphins could detect and recognise the experimental net more readily and from a greater distance than the control nets. On encountering control nets, they exhibited stress behaviour (decreased swimming speed towards the nets and increased breathing rate) which was not exhibited when the experimental nets crossed their path. Due to the success of this project, the reflectors have been deployed at two installations in southern KwaZulu-Natal (Margate and Southbroom). However, the reflectors make net maintenance difficult and bottlenose dolphins have been caught in these nets (VM Peddemors personal communication).

Similar sonar reflectors have been investigated in nets in Europe. Harbour porpoises Phocoena phocoena avoided fishing nets with reflectors, first changing direction of travel by 180o, then giving the nets a wide berth (Goodson 1997). In the "no nets" control, they traveled along a straight path. Unfortunately, when experiments were conducted out at sea, Goodson (1997) found that sonar signals at 100 kHz were strongly scattered by the aeration effects of breaking wave tops, thus reducing the effectiveness of the reflectors in the nets. Goodson (1997) suggested a double warning system that exploits the animal’s hearing to attract attention whilst reflectors allow its echolocation sense to ‘see’ the extent of the barrier.

Acoustic devices have been tested on harbour porpoises (Kraus et al. 1997, Kraus and Brault 1997). These devices are sometimes referred to as "acoustic warning/deterrent devices;" however, I prefer to call them pingers, which is a neutral label based on the sound that they produce. The pingers used by Kraus et al. (1997) emitted a loud broadband signal which lasted 300 ms and was repeated every 4 seconds. Two harbour porpoises were captured in nets with activated pingers and 25 were captured in nets with similar but silent devices. Thus the pingers effectively reduced the number of mortalities. However, exactly how this device worked and its effect on the porpoises is unknown. It was suggested that the pinger’s signal affected herring Clupea harengus which was the harbour porpoise’s main food item at that time (Kraus et al. 1997). Because the herring possibly avoided the sound signal and thus the nets, the porpoises did not get close to the nets and the number of porpoises captured was reduced. To test this hypothesis, replicates of the experiment were carried out in spring when there were no herring in the area (Kraus and Brault 1997). The number of harbour porpoise mortalities was still significantly smaller in nets with pingers than in nets with similar but silent devices (Kraus and Brault 1997).

An appropriate way to investigate the mechanism behind the reduced capture rate would be to investigate the behavioural reaction of cetaceans to the acoustic alarms. I suggest that the behaviour of dolphins near nets changes when pingers in the nets are actively emitting a sound signal.

 

Testing pinger efficacy

Using the data published in Cockcroft (1990), I calculated the catch per unit effort (CPUE) of each installation in KwaZulu-Natal, which is the number of incidental dolphin mortalities per net: CPUE = M/N,

where CPUE – Catch per unit effort

M – the number of mortalities at each installation

N – the number of shark nets at each installation.

The installation with the highest CPUE is Richards Bay (fig 1). This makes Richards Bay the appropriate installation at which to test the efficacy of the pingers. It is the northernmost beach in South Africa with shark nets (table 2) and is situated on the Tugela Bank where the continental shelf is wider than that of the rest of the coast. Since the Richards Bay cetacean by-catch is mainly of humpback dolphins (Cockcroft 1990, Durham 1994, Natal Sharks Board unpublished data) it is an appropriate study species. The Natal Sharks Board (NSB) has recorded 100 humpback dolphins caught in shark nets along the KwaZulu-Natal coast between 1981 and 1997. Of these, 57 have been caught at Richards Bay, which represents only 3.3% of the number of nets in KwaZulu-Natal. Thus shark nets at Richards Bay in particular appear to be a threat to the humpback dolphins. The KwaZulu-Natal humpback dolphin population was estimated in 1994 at about 161 to 166 individuals (95% confidence limits 134 to 229) and the annual mortality due to shark net captures is approximately 4.5% of this population (Durham 1994). Thus, there is an additional reason for studying this taxon and its association with the shark nets.

 

Humpback dolphins

Classification of the humpback dolphin is in dispute. Although there could be as many as five different species (Ross et al. 1994, Carwardine 1995), generally only two are accepted: the Indo-Pacific humpback dolphin Sousa chinensis and the Atlantic humpback dolphin S. teuszii. The Indo-Pacific humpback dolphin has a robust body (150 – 200 kg, 2 – 2.8 m in length) with a small triangular dorsal fin on the top of a fatty hump (Figure 1; Ross et al. 1994, Carwardine 1995). The beak is long and slender and is exposed during the distinctive surfacing pattern which is characteristic of humpback dolphins (Carwardine 1995). The beak breaks the water surface at an angle of 30o to 45o and the beak and sometimes entire head is exposed; a few seconds later the back is exposed and arched, usually quite steeply.

The distribution of the humpback dolphin is distinctly tropical, extending into higher latitudes in areas where waters are warm (Ross et al. 1994, Carwardine 1995). Within this tropical distribution, they are limited to coastal areas, usually in shallow water (less than 20 m deep; Saayman and Taylor 1979, Durham 1994, Ross et al. 1994, Karczmarksi 1996). In comparison to the rest of KwaZulu-Natal, the size of humpback dolphin population was larger in the shallow turbid waters of the Tugela Bank (Durham 1994). However, water turbidity does not appear to limit their distribution and they can be found in clear waters in KwaZulu-Natal and the Eastern Cape (Durham 1994, Saayman and Taylor 1979, Karczmarksi 1996).

Humpback dolphins feed on littoral and estuarine associated fish, and on demersal species primarily associated with reefs (Barros and Cockcroft 1991, Ross et al. 1994). Although humpback dolphins in South Africa appear to feed independently of others in the group (Durham 1994, Karczmarski 1996), in Mozambique they have been observed feeding co-operatively (Peddemors and Thompson 1994). The average group size in KwaZulu-Natal, Plettenberg Bay and Algoa Bay (Eastern Cape) was consistent at seven animals per group (Durham 1994, Saayman and Taylor 1979, Karczmarksi 1996). These groups are not stable, however, and are characterised by their temporary nature and fluctuating membership (Saayman and Taylor 1979). In addition, solitary individuals form a large proportion of sightings in KwaZulu-Natal and the Eastern Cape (Durham 1994 and Karczmarksi 1996).

The behaviour of humpback dolphins off the Tugela Bank differs from that of these dolphins elsewhere in KwaZulu-Natal. Although the proportion of feeding behaviour was the same, resting and socialising were observed more frequently and travelling less frequently on the Tugela Bank (Durham 1994). In a study of the behaviour of dolphins around the shark nets, Peddemors (1995) found that the social behaviour of bottlenose dolphins was significantly reduced.

 

Hypotheses and aims

There are four main hypotheses in this study.

    1. The behaviour of the humpback dolphins is changed when acoustic warning devices are actively emitting a signal.

      2. a) Humpback dolphins will spend less time in the netted area when the acoustic warning devices are active.

      b) Average surfacing distance of the humpback dolphins from the nets will decrease.

    2. The behaviour of humpback dolphins at the Richards Bay shark nets is different to the behaviour of this species elsewhere in the Richards Bay area.
    1. There will be less social behaviour at the nets than elsewhere in the study area.
    2. The proportion of feeding behaviour will be greater in the netted area than elsewhere in Richards Bay.
    1. The behaviour of the humpback dolphins is related to certain environmental conditions (e.g., water depth, turbidity, reef presence).
    2. The capture of humpback dolphins is related to current speed and direction.

The study has five aims.

    1. To examine the reactions of the humpback dolphins to the signal emitted by the pingers.
    2. To quantify the behaviour of the humpback dolphins at and away from the nets in the Richards Bay area.
    3. To ascertain whether water characteristics affect both humpback dolphin behaviour and capture.
    4. To analyse the spatial pattern of capture within the Richards Bay shark net installation.
    5. To make recommendations with respect to the use of the pingers in the shark nets of KwaZulu-Natal.

METHODS AND MATERIALS

Historical capture data

A c 2 test will be used to determine if historical catch data (1980 – 1997; NSB unpublished data) are evenly distributed for all nets within the Richards Bay installation. The length of each net will have to be accounted for because all the nets are not equal in length. In an effort to understand what environmental conditions are associated with capture, historical environmental data, including wind direction, surface water temperature, swell height and current direction (NSB unpublished data, CSRI unpublished data) will be correlated with catch data using Spearman’s Rank correlation.

Field procedure

Weather permitting, a boat-based search for humpback dolphins will be initiated at daybreak from the Richards Bay harbour. The time of search initiation will be noted. The search area is limited to the area between Durnford Point and the Richards Bay Lighthouse (16 km in length). Although the width delimitation of the study area is 4 nautical miles from the shore, searches only take place within 1 km of the shore line since Karczmarski (1996) always found the dolphins within 150 – 350 m from the shore. Two transect lines will be followed in search of the humpback dolphins, one approximately 250 m from the shore, the second approximately 750 m from the shore, this allows a 250 m wide search area on each side of the boat. Search speed will be approximately 10 km per hour.

When humpback dolphins are seen, time, geographic position, behaviour, minimum number of dolphins, shark net number, photograph number and any comments will be noted.

    1. Time: Real time.
    2. Geographic position: Latitude and longitude recorded by a Garmin II-Plus Global Positioning System (GPS).
    3. Behaviour: Travelling, Feeding, Socialising, Resting or Undetermined (see below).
    4. Minimum number of animals.
    5. Net number: If any animal is within 100 m of a shark net.
    6. Photograph number: Obtained from the concurrent photo-identification study by Mark Keith. This will allow later confirmation of animal identity, number of animals and group composition.

The dolphin(s) will then be followed. In an attempt not to influence dolphin behaviour, the boat will be driven parallel to the direction of movement at a distance of no less than 20m. If the behaviour or group size changes, the time and geographic position will be noted. In addition, geographic position will be recorded every five minutes. A watch (Casio DB-34H) with a countdown timer with automatic reset function marks the five-minute interval with a 10-second alarm. If the dolphin or dolphin group is not seen for more than 10 min, it will be considered ‘lost’ and time and geographic position of the last sighting will be considered the end of that follow, the search will then be resumed. Searches will be abandoned in Sea States >3 when the occurrence of white caps and increased swell height decrease the probability of sighting dolphins. This is in accordance with international survey techniques (Leatherwood and Show 1980). However, if a follow is in progress and the sea state changes, it will not be terminated until the dolphins are lost or the Sea State increases to a rating of 5. The time of termination of searching will be noted. The start and stop times of search can be used to calculate "observer effort."

Behavioural studies

 

Five types of behaviour will be scored, namely travelling, feeding, socialising, resting and undetermined. These categories were derived from Peddemors (1995), Karczmarski (1996) and personal observation, and represent behavioural states (Martin and Bateson 1993). I will use a combination of six cues to assess behaviour: a) directional versus localised movement, b) speed of movement, c) amount of time spent at the surface, d) regularity of breathing (surfacing pattern), e) dive angle, and f) group geometry (spatial arrangement).

Travelling: Movement is persistent and directional with a regular pattern of surfacing and diving. Dive angles are shallow. Animals are not underwater for extended length of time. The boat is easily driven parallel to the animal(s). If the speed of the boat is less than 7 km/h, travel is slow. If speed is greater than 7 km/h, travel is fast.

Feeding: Localised movement, extended submersion times (two minutes or more). Dives are frequent and steep, dive direction is irregular, varying both within and between individuals. Swimming occurs at high speed bursts along erratic courses. Some aerial behaviour (overshooting the surface, somersaults). Fish are sometimes seen. If there is more than one animal, they are usually widespread (more than 15 m between them). It is difficult to follow feeding animals because of erratic courses and long submersion times. Searching for food, i.e. foraging is included in this category; movement is more directional than feeding, group geometry is more clumped, no aerial behaviour but submersion times are extended. Because of extended submersion times, following is difficult, but it is easier to anticipate where they will surface because of directionality of movement.

Socialising: Movement is localised and much time is spent at the surface. Dive direction is unpredictable. If levels of activity are high, with aerial behaviour (jumps, somersaults, lobtailing, spyhopping) then behaviour is classed as play, especially if it is socially directed (i.e. animals interacting with one another). Presumably it is possible for a single animal to play; if there was some aerial behaviour and the animal spent most of the time near the surface I would class it as solitary play. Because of erratic changes in direction, it is not always easy to follow playing animals, although it is relatively easier than following feeding animals because playing animals spend more time near the surface. Not all socialising is as energetic as play. When activity levels are low, and much time is spent at the surface and animals are clumped and interact with one another, the behaviour is scored as socialising.

Resting: Movement may appear directional but often the animals are circling over the same area. Much time is spent at the water’s surface and levels of activity are low. Dive angles are shallow. No social interaction even though animals may be clumped. It is easy to follow resting animals because they spend much time in view at the surface.

Undetermined: Behaviour does not fall into any of the above categories.

Once dolphin(s) have been spotted, focal sampling will be used as the sampling rule (Martin and Bateson 1993) and attempts will be made to remain with the same animal or group, even if there are other (subgroups) in the near vicinity of the animal or group being followed. The follow will be continued for as long as possible, i.e. until animals are lost, the weather conditions change or the sun sets. The behaviour of the majority (>50%) of the group will be recorded although other activities (and the proportion of animals engaged in these activities) will be noted. There are a few exceptions to the above sampling rule. If only one animal is followed, all its behaviour will be recorded. If equal numbers of animals are doing different activities, I will record all behaviour that occurs. Travelling animals will be given following priority since it is necessary to study broad-scale movement patterns and to ascertain all places used or visited by the humpback dolphins. In addition, travelling animals are easier to photograph than animals moving locally and erratically - this is important for photo-identification.

Two behavioural recording rules will be used, continuous and one-zero (Martin and Bateson 1993). Using the continuous recording rule, changes in behaviour will be recorded as they occur in real time. Thus frequency and duration of each behaviour can be calculated. In addition, information about the sequence of activities is preserved. However, the change in behaviour is not always instantly recognisable and there is often a lag between the change in behaviour and my realisation of the change. Therefore, every five minutes I will record the behaviour in 5 min intervals using one-zero sampling. Positional data are recorded using an instantaneous sampling rule, although position is recorded with changes in behaviour. The GPS has a track facility which records the exact path followed on a minute-by-minute basis and effectively continuous positional data are preserved.

Any activity that occurs within 100m of any net will be regarded as occurring within the netted area. When the dolphins enter and leave this area, the time and geographic position will be recorded. In addition, since the geographic position recorded by the GPS is that of the boat, the distance of the (closest) dolphin from the nets will be judged. This will enable a more fine scale recording and analysis of position in the netted area.

The frequency and duration of each behaviour in the vicinity of the shark nets will be compared to the frequency and duration of behaviour elsewhere in the study area using a two-way ANOVA.

The study area will be divided into 1 km zones. In each zone a Coefficient of Area Utilisation (AU; Karczmarski 1996) will be calculated. The AU, which ranges from 0.0 to 1.0 and represents the time spent by the dolphins in a particular zone as a proportion of the total observation time will be calculated as: AU = D/T,

where AU – Coefficient of Area Utilisation

D – time spent in a particular zone

T – total observation time.

A Kruskal-Wallis ANOVA could be used to test for a difference in the amount of time spent in each zone over the entire study period.

In addition to the AU, an Activity Index (IA; Karczmarski 1996) will be calculated for each zone. This index, which ranges between 0.0 and 1.0, will be used to represent the amount of time the animals are engaged in each of five behaviours in a particular area as a proportion of the total time spent by the dolphins in that area during any one day: IA = B/S,

Where IA – Index of Activity

B – time dolphins were engaged in a particular activity in a zone

S – time spent by the dolphins in this zone

The IA, and subsequently the mean IA, will be calculated separately for each of the five behavioural categories for each zone. A Kruskal-Wallis ANOVA could be used to compare IAs between zones.

Environmental conditions

Environmental data (cloud cover, wind direction, wind speed, sea state, average swell height and maximum swell height) will be recorded at the start of each search, at the end of each follow and when noticeable changes in the weather occur. These data (excluding cloud cover and sea state) will be confirmed by Port Control (the department controlling harbour activities) which measures these factors continuously. In addition, wave period and wave direction, current direction and speed will be obtained from Port Control. Depth data will be read off a bathymetric chart of the study area. I have scanned this chart into the computer and a grid of depths has been drawn up. Once the geographic position is specified, depth data can be easily accessed.

Water characteristics (surface and subsurface temperatures and water visibility) will also be measured at six set sites, all within 500 m from the shore (Figure 2). A 20-litre container will be used to sample surface water. If this container is warm it will be left in the water for a minute to equilibrate. Temperature will be measured with a thermometer to one decimal place. A subsurface water sampler (a cylinder that can be sealed underwater) will be used to retrieve water from 4.5m depth for similar temperature measurements. Water visibility will be measured with a Secchi Disk. These data will be collected at the end of each follow at the position where the dolphins last seen.

Pinger experiment

At the beginning of June, experimental work will begin with the introduction of the pingers into the nets. Pingers will be deployed 100 m apart in one of two nets, net 5 or net 99. There are two pinger states: on/off. The flip of a coin will determine which net has pingers each week and pinger state will depend on the previous state of pingers in that net. Both the deployment of pingers and the "coin-flip" procedure will be carried out by the NSB staff so that I will not know pinger state (on/off) during observations. This will remove any possible observer bias. Observations will continue as before.

The coefficient of area utilisation will be calculated and compared (using Kruskal-Wallis ANOVA) for days with pingers on and days with dummy pingers, comparisons may also be made between the AU before pingers were deployed and subsequent to pinger deployment. In a similar way, IAs will be calculated and compared.

Another quantifiable variable which will be used to analyse the effect of the pingers on dolphin behaviour is the minimum surfacing distance from the net. In other words, does net-avoidance occur when pingers are on? Minimum distance of the dolphins from the nets with active pingers will be compared to those with dummy pingers. In addition, the minimum distance from the nets with active pingers will be compared with data obtained prior to pinger deployment.

REFERENCES

Armstrong MJ and Thomas RM. 1995. Clupeoids. In: Oceans of Life. (2nd edition). AIL Payne and RJM Crawford (eds). Vlaeberg Publishers. Halfway House.

Carwardine M. 1995. Whales, dolphins and porpoises. Dorling Kindersley. London.

Cockcroft V.G. 1990. Dolphin catches in the Natal shark nets, 1980 to 1988. S. Afr. J Wildl. Res. 20: 44-51.

Cockcroft V.G. 1994. Is there common cause for dolphin capture in gillnets? A review of dolphin catches in the sharks nets off Natal, South Africa. Rep. Int. Whal. Commn. (Special Issue 15): 541-547.

Corkeron PJ. 1990. Aspects of the behavioural ecology of inshore dolphins Tursiops truncatus and Sousa chinensis in Moreton Bay, Australia. In: The bottlenose dolphin. S Leatherwood and RR Reeves (eds). Academic Press. San Diego.

Davis B, Cliff G and Dudley SFJ. 1995. The Natal Sharks Board. In: Oceans of Life. (2nd edition). AIL Payne and RJM Crawford (eds). Vlaeberg Publishers. Halfway House.

Durham B. 1994. The distribution of the humpback dolphin (Sousa chinensis) along the Natal Coast, South Africa. Unpublished Masters of Science, University of Natal.

Goodson AD. 1997. Developing deterrent devices designed to reduce the mortality of small cetaceans in commercial fishing nets. Mar. Fresh. Behav. Physiol. 29: 211-236.

International Whaling Commission. 1994. Report of the Workshop on Mortality of cetaceans in passive fishing nets and traps. La Jolla, California. 22-25 October 1990.

Jefferson TA and Curry BE. 1994. A global review of porpoise (Cetacea: Phocoenidae) mortality in gillnets. Biological Conservation 67: 167-183.

Julian F. 1997. Cetacean mortality in California gill net fisheries. Unpublished manuscript presented to the annual scientific meeting of the International Whaling Commission, Bournemouth, United Kingdom. September 1997. SC/49/SM02.

Karczmarski L. Ecological studies of humpback dolphins Sousa chinensis in the Algoa Bay region, Eastern Cape, South Africa. Unpublished PhD thesis. University of Port Elizabeth.

Kraus SD, Read AJ, Solow A, Baldwin K, Spradlin T, Anderson E, and Williamson J. 1997. Acoustic alarms reduce porpoise mortality. Nature 388: 525.

Kraus S and Brault S. 1997. A field test of the use of pingers to reduce incidental mortality of harbour porpoises in gill nets. Unpublished manuscript presented to the annual scientific meeting of the International Whaling Commission. Bournemouth, United Kingdom. September 1997. SC/49/SM42.

Martin P and Bateson P. 1993. Measuring behaviour: An introductory guide. 2nd edition. Cambridge University Press. Cambridge.

Peddemors VM. 1995. Aetiology of bottlenose dolphin capture in shark nets off Natal, South Africa. Unpublished Doctor of Philosophy, Faculty of Science, Department of Zoology. University of Port Elizabeth.

Ross GJB, Heinsohn GE and Cockcroft VG. 1994. The humpback dolphin. In: Handbook of marine mammals, Vol 5 (Eds. SH Ridgeway and R Harrison). Academic Press. New York.

Ross GJB and Best PB. 1995. Smaller whales and dolphins. In: Oceans of Life. (2nd edition). AIL Payne and RJM Crawford (eds). Vlaeberg Publishers. Halfway House.

 

Saayman GS and Tayler CK. 1979. The socio-ecology of humpback dolphins (Sousa sp.). In: Behavior of Marine Animals. Win, H.E. and B.L. Olla (eds), Vol 3. Plenum, New York. 438pp.