Fluctuations in numbers of great white pelicans at Lake St Lucia in response to changing water-levels


*E-mail: downs@ukzn.ac.za


In January 2004, Lake St Lucia, a major part of the Greater St Lucia Wetland Park, was reduced to a fraction of its normal capacity as a result of a severe drought in this region of KwaZulu-Natal, South Africa. After rains in the area, the lake level rose and then fluctuated considerably over the next 42 months. During the first 38 months the mouth of the estuary into the sea was closed. The area entered a second severe period of prolonged drought from mid 2005 through to the spring of 2006. Great white pelican (Pelecanus onocrotalus Linnaeus, 1758) numbers and lake levels were monitored during these 42 months. Pelican numbers were highly variable ranging from 0 to 6000. When lake levels were very low or too high, no pelicans were present. Pelican numbers appeared more indicative of food availability. Implications of these trends to the management of the lake and the conservation of the avifauna are discussed. It is suggested that the great white pelican could be used as an indicator species for the fish dynamics of the lake at medium to low lake levels.


En janvier 2004, le lac Ste Lucia qui est une partie majeure du Greater St Lucia Wetland Park fut réduit à une fraction de sa capacité normale suite à une grave sécheresse dans la région du KwaZulu-Natal, en Afrique du Sud. Après des pluies dans la région, le niveau du lac est remonté et a considérablement fluctué au cours des 42 mois qui ont suivi; pendant les 38 premiers de ces mois, la bouche de l’estuaire vers la mer est restée fermée. La région a connu une nouvelle période de sécheresse prolongée entre le milieu de 2005 et le printemps 2006. Le nombre de grands pélicans blancs (Pelecanus onocrotalus Linnaeus, 1758) et le niveau du lac furent suivis pendant ces 42 mois. Le nombre de pélicans était très variable et il a fluctué entre 0 et 6000. Lorsque le niveau du lac était trop bas ou trop haut, il n’y avait aucun pélican. Le nombre de pélicans semble indiquer davantage la disponibilité de la nourriture. L’on discute des implications de ces tendances pour la gestion du lac et la conservation de l’avifaune. On suggère d’utiliser le grand pélican comme indicateur pour la dynamique des poissons du lac lorsque le niveau de ce dernier est moyen à bas.


A major part of the Greater St Lucia Wetland Park, a world heritage site, is Lake St Lucia in KwaZulu-Natal (KZN) (Fig. 1). This lake is a highly dynamic water body. The lake is fed by several rivers and seepage zones and is temporarily closed to the sea (M. Bowker, personal observations). It is a large expanse of water being in excess of 70 km long and up to 18 km wide, with a maximum surface area of about 417 km2 (Johnson, Barnes & Taylor, 1998; Fig. 1). The lake is relatively shallow with mean depth ranging from 1 to 2 m and depending on the depth of the lake, it experiences high evaporation rates (Taylor, 1982). Salinities in the lake can vary enormously ranging from 110 ppm in October 1970 to 0 ppm in August 1991 [Ezemvelo KZN Wildlife (EKZNW), unpublished data] and this in turn influences the natural resources and dynamics of the system. When the level of Lake St Lucia drops below a critical point, the lake separates into discrete bodies and these units respond differently to prolonged periods of drought (Fig. 2). The Narrows (Fig. 1) which is deeper and has a small surface area is subject to less evaporation than the rest of the lake which is shallower and has a high surface area to volume ratio. Consequently, the Narrows never dries up even in the height of droughts. Lake St Lucia can therefore provide a range of habitats for a variety of birds and the lake levels and resulting salinities influence which species are able to use this resource (Whitfield & Cyrus, 1978).

Figure 1.

 Lake St Lucia, at its maximum surface area and its catchments in north eastern KZN, South Africa

Figure 2.

 Selley’s Lakes section of Lake St Lucia during a dry down period in July 2004 showing how the lake separates into discrete bodies. Pelicans in the foreground congregate near a traditional breeding site

In areas with high biodiversity and conservation importance, such as Lake St Lucia, focal-species approaches have shown potential for conservation initiatives and management (Landres, Verner & Thomas, 1997; Simberloff, 1998). A variety of these focal-species approaches and terminology have been developed. For example, an indicator species is a species whose characteristics are used as an index of attributes for other species or environmental conditions of interest (Landres et al., 1997). However, the indicator species concept is problematic if there is no agreement on what it is supposed to indicate and which species best suits the criteria (Landres et al., 1997; Simberloff, 1998). To be a flagship species, the species need only operate in the public relations or fundraising spheres. It can do this by capturing the imagination of the public and induce people to support conservation efforts. They play a socio-economic role rather than an ecological one (Walpole & Leader-Williams, 2002).

Great white pelicans (Pelecanus onocrotalus Linnaeus, 1758) are large, white, highly visible birds (Maclean, 1985) that are reliant on healthy water bodies for their food and Lake St Lucia is one of these water bodies (Whitfield & Blaber, 1979; Berruti, 1983). Pelicans could be developed as a flagship species for the lake. Furthermore, because of their size and their reliance primarily on aquatic food, pelicans might qualify as an ecological indicator species.

Great white pelicans are not able to dive, and generally feed socially. They swim in a ‘C’-shaped formation, which is thought to drive the fish forward. They then all thrust their heads into the water simultaneously in an attempt to capture their prey. When prey is abundant this synchronized fishing method breaks down and the birds forage individually within a loose group (Din & Eltringham, 1974). Prey consists mainly of fish that are <100 mm long and weigh <100 g, although they are able to feed on fish over 2 kg in mass (Brown & Urban, 1969; Whitfield & Blaber, 1979).

Between January 2004 and June 2007, the study area experienced two extremely dry cycles. The first of these was at its height when the study started in January 2004 and ended in the latter half of that same month. The second started in summer 2005 and continued till the end of 2006 (Bowker, 2006). In March 2007, a combination of unusually high tides and storms at sea resulted in the breaching of the mouth of the estuary, and sea water entered the lake.

It was expected that lake levels would influence the number of great white pelicans on Lake St Lucia. The aim of the study was to determine the effects that changing lake levels have on the numbers of pelicans present at Lake St Lucia, through investigating water-levels and corresponding pelican numbers over a three and a half year period from January 2004 to June 2007. The study also considers whether the great white pelican numbers can be used as an ecological indicator for the system.

Materials and methods

Monthly from January 2004 to June 2007, 42 aerial counts of great white pelicans were conducted at Lake St Lucia. Aerial counts were made from the EKZNW Cessna 182 Skylane around the end of the third week of each month. The entire lake north of the Narrows was surveyed from a height of about 80 m (Fig. 1). The same aircraft, pilot and observer was used for each flight, and all flights were carried out between 09.00 and 11.00 hours. Visibility on all flights was good, allowing a total count of all pelicans present at the lake. It was possible to see large groups of pelicans from considerable distances as well as spot individual birds along the flight path.

All pelicans observed were photographed digitally with a Fujifilm FinepixTM S5000® (Nikon Corp., Tokyo, Japan) digital camera synchronized with a Garmin eTrex LegendTM Personal Navigator® (Garmin International Inc., Olathe, KS, USA) so that exact geographical locality was recorded. All groups of pelicans photographed were counted subsequently.

The recorded track from the Garmin eTrex LegendTM Personal Navigator® and the digital photographic images were downloaded to a desktop computer for analysis. The digital images were opened in Adobe® Photo-3 Deluxe Home Edition 4 (Adobe Systems Inc., San Jose, CA, USA) and enlarged where necessary. For each picture a new transparent layer was created over the original image on which each individual pelican was marked with a dot. For large groups of birds the image was divided into smaller areas and each area was dealt with separately. The dots were counted; the number of birds for each site recorded and subsequently entered on a map using the location given by the Garmin navigation aid track. A total monthly pelican count was then determined by the summation of all pelicans observed.

The levels of Lake St Lucia were recorded automatically by the Department of Water Affairs (DWAF) by a recorder (W3R002a04) at the bridge to St Lucia town in the Narrows area of the lake, and by a recorder (W3R002a02) at Charter’s Creek (Fig. 1). Using the daily water-levels for the respective sites, monthly averages were calculated and used when relating water-levels to the number of great white pelicans in the lake.

Data were analysed using statistica (Tulsa, OK, U.S.A.). Descriptive statistics was obtained, and Generalized Linear Models (GLIM) regression analysis and correlation analyses were conducted.


General considerations

Throughout the study period, the water-levels at Charter’s Creek and the Narrows tracked each other. However, from April 2004 to September 2005 when the lake levels were relatively high, the recordings at the Narrows and at Charter’s Creek were more similar. As the levels dropped, as they did after September 2005, the difference between water-levels at each site became markedly different as the level at Charter’s Creek dropped more rapidly. This was most obvious in January 2004 and December 2005. The difference between the mean monthly levels at these two sites decreased from 0.56 m in the drought of January 2004 to a minimum of 0.08 m in September 2004 and then increased to 0.65 m in February 2006 (Fig. 3). The recorder at Charter’s Creek provides a better indication of the levels of the main body of the lake than does the recorder at the Narrows (M. Bowker, personal observations).

Figure 3.

 The mean monthly water-levels (m) at the St Lucia Estuary and at Charter’s Creek and numbers of great white pelicans counted in the Lake St Lucia system from January 2004 to June 2007. Mean sea level is c. 1 m above the mean lake bottom level. Sea water did not enter the lake until April 2007 as the mouth was closed by a sand bar (Missing data for Charter’s Creek as a result of water-level recorder failure)

On the 24th January 2004, the Umfolozi River spilled over and fresh water from this system entered Lake St Lucia (R.Taylor, personal communications). This raised the mean daily maximum water-level in the Narrows (DWAF, unpublished data), but the level remained below the mean sea level. Consequently, the mouth of the estuary was not breached and the lake remained separated from the sea. The level at the recorder at the St Lucia bridge in the Narrows then dropped quite rapidly as the water pushed northwards into the lake, topping over into these areas and raising the level at the recorder at Charter’s Creek. Consequently, the lake returned to being a single body of water and remained as such until spring 2005.

Water-levels and pelican numbers

In the first half of January 2004, the lake was at its lowest level (Fig. 3). Only 366 great white pelicans were recorded on the lake during the January 2004 count. The number of pelicans dropped to 123 in February 2004 and then increased steadily over the next 21 months, rising to a maximum of 6558 in September 2005 (Fig. 3). From February 2004 until November 2004 the lake level fluctuated, showing a slight declining trend. Over the same period, great white pelican numbers increased steadily to a maximum of 2693 in October 2004. The general trend was a steady decline in the lake levels (Fig. 3). The rains of the 2004–2005 summer season caused the water-levels to rise until they peaked during March 2005 at a mean monthly water-level of 0.64 m in the Narrows and 0.26 m at Charter’s Creek. However by August 2005, these levels fell to 0.48 m in the Narrows and 0.18 m at Charter’s Creek (Fig. 3) and by September the lake was again divided into separate compartments (M. Bowker, personal observations). This period of slowly declining water-levels was again associated with a general increase in pelican numbers from 1528 in March 2005 to 6558 in September 2005. The level of the northern section of the lake then dropped rapidly again as it had done in the 2003–2004 drought, while the level in the Narrows dropped more slowly. There was no water in the lake north of Fanies Island, except for some seepage along the eastern shore and water-levels continued to decrease until December 2005, coincidental with a rapid decline in pelican numbers from 6558 in September 2005 to 283 in January 2006. The northern part of the lake then experienced a long period of very low levels (−0.14 to 0.14 m with reference to mean lake bottom level) extending from January 2006 till November 2006 (Fig. 3). During this period there were only isolated, shallow bodies of water in the area between the Narrows and Fanies Island. Although there was a steady but small increase in the lake levels during these months, this merely produced large expanses of very shallow water. The difference between the level in the Narrows and the rest of the lake was high (c. 0.65 m) as the Narrows was again separated from the main body of the lake. The pelican numbers dropped from 1769 in December 2005 to 283 in January 2006. No pelicans were seen at the lake from September 2006 to the end of that year (Fig. 3). Summer rains raised the levels of the Narrows in December 2006 to 0.75 m above mean lake bottom level and this allowed for an influx of water from the Narrows into the main body of the lake. This combined with the summer rainfall in the area allowed the water-level in the lake to rise and the discrete bodies of water to join (M. Bowker, personal observations). Then a combination of unusually high tides and extreme weather conditions in March 2007 resulted in the breaching of the mouth of the St Lucia estuary and sea water flowed into the system. This raised the water in the lake to the highest levels experienced throughout the study period. Following the increase in the water-levels throughout 2006, there was an increase in pelican numbers rising from 63 in January 2007 to 3177 in March 2007. Exceptional rains in June 2007 also added to these increased levels and added fresh water to the lake. Pelican numbers remained fairly constant from February to June 2007, averaging 2230 individuals (Fig. 3).

Pelican numbers showed conflicting correlations with water-levels in Lake St Lucia from January 2004 to June 2007 (Fig. 4) showing that pelican numbers are not directly related to water-levels. The relationship was marginally negative at the Narrows (r = −0.098, F(1, 40) = 0.384, P = 0.539), but positive at Charter’s Creek (r = 0.311, F(1, 35) = 3.740, P = 0.061). Water-levels in the Narrows were higher than those at Charter’s Creek (Figs 3 and 4).

Figure 4.

 Correlation of pelican numbers and water levels at two sites; (a) the Narrows and (b) Charter’s Creek, Lake St Lucia from January 2004 to June 2007 (Dotted line shows the 95% confidence level)

Pelicans were present in large numbers (>500) from April 2004 to December 2005 and from February 2007 to June 2007. The lake was used mainly for feeding, roosting and loafing (Bowker, 2006). There was limited breeding at the lake in the winter of 2004.


At the height of the drought at the beginning of 2004, the surface area of Lake St Lucia was reduced to an estimated 25% of its maximum size (Whitfield et al., 2006) and the remaining area was extremely shallow (this study; Bowker, 2006). The lake became unsuitable for many aquatic biota and many birds moved away from the lake (M. Bowker, personal observations). The number of great white pelicans visiting the lake was reduced dramatically as they moved to more suitable foraging sites outside of KZN, possibly in Mocambique (G. Nanni, personal communications). Subsequent rains over the lake and in the catchments raised the lake levels and the system became progressively more suitable for pelicans. The conditions at the lake and in the catchments varied over the next 2 years, but generally over this time the lake levels were low and dry conditions prevailed. From January 2004 to March 2007, the lake was cut off from the sea as a management practice to stop the influx of sea water and thus limit the salinity of the lake (R.Taylor, personal communications). This impacted on the water-level and resulting salinities within the system (EKZNW, unpublished data). Mouth closure also prevented the movement of prey species into and out of the lake (Cyrus and Vivier, 2006a,b). Limited fresh water enters the lake, coming mainly from the east flowing rivers, as well as through seepage which occurs mainly along the eastern shore (Johnson et al., 1998).

Pelicans are extremely mobile birds and in Africa may cover great distances, presumably related to foraging or breeding (Brown, Urban & Newman, 1982; Crawford, 2005). When the lake area and depth decreased to the low levels in January 2004, foraging opportunities decreased and birds left the lake.

The fish populations in Lake St Lucia vary considerably at different times of the year, especially if the estuary is open to the sea (Blaber, 1982). The closure of the mouth of the lake to the sea impacts negatively on this potential source of food for pelicans and other piscivorous avian species, as the migrations of fish and prawns into the estuary are not possible (Blaber & Cyrus, 1982; Cyrus & Vivier, 2006a,b). Furthermore with lower lake levels, the biomass and number of fish species decline (Cyrus & Vivier, 2006a,b). In addition, mass mortality of fish occurs in the lake during periods of low lake water-levels or extremely high salinities (Cyrus & Vivier, 2006a).

In late January 2004, fresh water flowed into the lake, but only filled it to about half its maximum depth. This was enough however to encourage the birds to start returning and by July 2004 the count reached 2028. Pelican food stocks had obviously recovered as well, as these birds remained in the system and were joined by additional pelicans increasing their numbers to a peak of 6558 pelicans in September 2005. It appears that the increase in pelican numbers was not a direct result of the rise in lake levels, but rather the result of the change that this rise manifested in the system. A newly filled water body does not provide suitable foraging for pelicans immediately as prey have not yet had the opportunity to breed. When the level starts dropping after a period of being high, then conditions become ideal for feeding, as prey are now more concentrated and within reach of these nondiving birds. Increased water levels create the habitat needed for the production of prey not only in the lake system but also in the wetlands of the whole of north eastern KZN.

Further investigation is needed to ascertain where these large numbers of pelicans (>5000) moved to and how long fish populations take to recover after periods of severe drought.

By December 2005, the conditions on the lake resembled those of January 2004, with the whole of False Bay dry and the lake reduced to a few isolated bodies of water. This dry phase which started in November 2005 and lasted until November 2006 was accompanied by a decrease in the numbers of pelicans from 6558 in September 2005 to 283 in January 2006 and then to zero for the period September to December 2006. This was indicative of poor feeding conditions in the lake for these pelicans. The lack of water and consequently fish was the primary reason for the decline in numbers of pelicans during this period. It appears that a dynamic estuarine system such as Lake St Lucia cannot sustain such high levels of biomass production if closed off from the sea even when water-levels remain high (Cyrus & Vivier, 2006a,b). Pelican numbers decreased at the end of the dry winter months as carrying capacity was exceeded and competition increased. Most fish species breed after rains (Kok, 1980) and fish stocks were probably not replenished during this dry time.

There appears to be a lag effect between the rising water-levels and the increase in pelican numbers. This lag occurred in the first 6 months of 2004 and from November 2006 to March 2007. This would be anticipated as increased water-levels encourage a higher productivity in the lake, the effect of which would take time to benefit consumers higher up the food chains.

Great white pelican numbers varied greatly at Lake St Lucia during the study period, at times exceeding 6000 (Crawford & Taylor, 2000; present study). Under conditions of severe drought when lake levels were low, the pelicans abandoned Lake St Lucia as a feeding site. However, when lake levels rose sufficiently pelicans returned and their numbers kept increasing until September 2005, when the next dry cycle started. It appears that the lake contains an extremely rich and sustainable source of food and that the conditions experienced at the lake during 2004 and 2005 were ideally suited to the foraging of the great white pelican. Newly filled and dry pans contain no food and are of no benefit to foraging pelicans. Water bodies that are drying down provide the best feeding conditions as the prey is concentrated (Berruti, 1983). The shallow water may also have provided ideal fishing conditions for these pelicans as the fish could not escape by diving out of reach of the birds’ long bill and neck. The low water-levels do however discourage nesting. There are no large islands or areas protected by swamp and vegetation at the lake that offer the great white pelican a suitable area for breeding (Bowker, 2006). Although adults might be in breeding condition, no breeding takes place.

If one pelican feeding from the lake for 1 day is equivalent to one ‘pelican-day’, and the number of pelicans counted on each monthly count was the average number of birds for each day of that month, then the period 1 January 2004 till 31 December 2005 represents 1.76 million pelican-days for Lake St Lucia. It is estimated that the great white pelican consumes about 1.2 kg of food per day (Din & Eltringham, 1974) or about 10% of its body mass (average 11.5 kg) (Brown & Urban, 1969). This translated into a mass of c. 2110 t of food removed by this species alone. The period June–October 2005 was equivalent to 713,006 pelican-days, representing 855 t of food taken in 153 days or an estimated 5.59 t of food removed per day.

Throughout 2004 and 2005, the lake was cut off from the sea, so there could have been no replenishment of the food resources from the sea. The rivers supplying water for the system also dried up soon after the summer rains each year and could not have acted as a reliable reserve from which to source food. This implies that this biomass of food was trapped or produced within the lake and that this was available even when the number of birds on the lake was steadily increasing and the removal of fish was continuous.

At Lake St Lucia, great white pelicans eat mainly the freshwater cichlids Mozambique tilapia (Oreochromis mossambicus) and southern mouthbreeder (Pseudocrenilabrus philander) as well as barbs (Barbus spp.), thornfish (Terapon jarbua), elf (Pomadasys commersonnii), barracuda (Sphyraena spp.) and flathead mullet (Mugil cephalus) (Feely, 1962; Whitfield & Blaber, 1979). Normally, the diet of the great white pelican follows three distinct phases: a preincubation phase when migrating M. cephalus is targeted; an incubation and postincubation phase when adults obtain cichlids from Lake St Lucia and the Phongolo pans; and a postfledgling stage when the adults and juveniles feed on marine species in the lake before dispersing to other areas (Feely, 1962; Whitfield & Blaber, 1979). Oreochromis mossambicus was found to be the most common prey item in the regurgitations of c. 200 flightless young pelicans during the postincubation phase (Whitfield & Blaber, 1979).

Oreochromis mossambicus, T. jarbua, P. commersonnii and M. cephalus can all tolerate salinities of 0‰ to over 70‰ (Whitfield et al., 2006). Cyrus & Vivier (2006a) showed that in December 2004, the month before this study started, when the lake had become partitioned into four isolated compartments and salinities were high, catches per unit effort were generally low except for O. mossambicus, the grooved mullet (Liza dumerili) and Commerson’s glassy (Ambassis ambassis). Further they showed that O. mossambicus made up 45% of the total catch. Juveniles of at least twelve species including nine estuarine-associated marine species had spawned after mouth closure (Cyrus & Vivier, 2006b).

The pelican breeding cycle failed completely in 2005 and 2006, possibly as a result of the failure of the M. cephalus migrations as water levels were low and the lake was cut off from the sea. As a result adults were simply foraging for themselves and evidence suggests that they were preying mainly on O. mossambicus which seemed to be the most successful fish species in the lake during these dry, saline cycles.

Great white pelicans are gregarious birds and on Lake St Lucia form very large feeding groups. Many feeding groups were observed during the aerial surveys (Bowker, 2006), of which eight groups numbered in excess of 1000 actively feeding individuals, with a maximum group size of over 4000. This differs from the observations of these birds in Uganda made by Din & Eltringham (1974), where the largest feeding group numbered 129 and the average size was 8.5 birds per flock. Din & Eltringham (1974) recorded that the great white pelicans were often seen feeding solitarily or socially in groups that averaged 8.5 in number, but very seldom in groups >50.

Water level does impact on the suitability of Lake St Lucia as a feeding site for the great white pelican, but it appears that this is a secondary effect. The production and availability of fish under these changing water-levels appears to be the primary parameter. This in turn depends on the production of aquatic macrophytes and zoobenthos that are less able to tolerate hypersaline conditions (Whitfield et al., 2006). Consequently, it is suggested that water-level is not the causative factor, but an indicator of the suitability of Lake St Lucia and the surrounding wetlands for the great white pelican population. The levels of Lake St Lucia are not solely an indication of the state of the lake, but also reflect the prevailing conditions in the catchments and therefore most other wetlands in north eastern KZN. Water-levels may be regarded as proxy for all these factors and their effects (Bowker, 2006). Lake St Lucia represents the last sizeable body of water that remains in the area during a severe drought and consequently the last source of food for pelicans. With increased anthropogenic and development pressure in the north eastern region of KZN (Nanni, 1982) and the advent of climate change, it will become increasingly important to identify indicator species for the systems in the region.

The great white pelican could make an ideal and useful indicator species in Lake St Lucia at medium to low lake levels for the abundance of those species of fish that are part of its diet. A study of print media and websites advertising services available in the town of St Lucia shows that the great white pelican is already commonly used as a logo or an object of attraction and already partly qualifies as a flagship species for the area (M. Bowker, personal observations). Lake conditions that favour the great white pelican however will differ from those that favour other species and management of water-levels at Lake St Lucia levels will largely be governed by conservation goals of the controlling body. It appears that except for the management of the mouth of the estuary, there is little that can currently be undertaken to govern the water-levels. However, improved management of the catchments is required especially as rainfall in this region is the major factor determining water-levels in the lake (Bowker, 2006).


We thank the following for their input into this study: EKZNW for making their data available to us and for their assistance with the aerial surveys; Greg Nanni for his piloting expertise and assistance; The Greater St Lucia Wetland Park Authority for granting access to Lake St Lucia; Dr Ricky Taylor for guidance and assistance; and the Department of Water Affairs for the water-level data.