University of Groningen The biology and impacts of Oreochromis niloticus and Limnothrissa miodon introduced in Lake Kariba Chifamba, Chiyedza Portia

(1)

The biology and impacts of Oreochromis niloticus and Limnothrissa miodon introduced in

Lake Kariba

Chifamba, Chiyedza Portia

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2019

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Chifamba, C. P. (2019). The biology and impacts of Oreochromis niloticus and Limnothrissa miodon introduced in Lake Kariba. Rijksuniversiteit Groningen.

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

528794-L-bw-Chifamba 528794-L-bw-Chifamba 528794-L-bw-Chifamba 528794-L-bw-Chifamba Processed on: 6-2-2019 Processed on: 6-2-2019 Processed on: 6-2-2019

Processed on: 6-2-2019 PDF page: 57PDF page: 57PDF page: 57PDF page: 57

Submitted for publication

African Journal of Ecology

tilapia species in the Southern‐African Lake Kariba

Portia Chiyedza Chifamba

Britas Klemens Eriksson

Chapter

4

(3)

Nile tilapia (Oreochromis niloticus L.) is a widely used aquaculture species in African lakes, where it often has established and displaced native species. However, it is not clear if the ecological impact of the Nile tilapia is triggered by species competition or changed food-web interactions. In Lake Kariba, escaped Nile tilapia has displaced the native Kariba tilapia (Oreochromis mortimeri Trewavas 1966), but the ecological interactions between the species are not well known. To investigate if food competition is a potential driver of the displace-ment, the extent of diet overlap of the two tilapia species was studied. Benthic pennate diatoms were the dominating food source for all individuals studied, contributing 83% to the volume of the diet for Nile tilapia and 78% for the Kariba tilapia. Most tilapia individuals consumed both detritus (ca 80%) and sand (ca 70 %). Similarity indices demonstrated a strong diet overlap between the two tilapia species, and their relative protein digestion efficiency was also similar. Thus, the different tilapia species almost had an identical food niche, indicating that there is a significant potential for food competition being a contributing factor for the decline of the native Kariba tilapia after the introduction of Nile tilapia.

(4)

59

4

Introduction

Oreochromis niloticus L. (Nile tilapia) is a cichlid fish endemic to the Nile River, which is widely used in aquaculture all over the world due to its superior growth rate and adaptation to aquaculture conditions compared to other cichlids (Phillippart & Ruwet 1982; Blow & Leonard 2007). Today, Nile tilapia has been established into many African lakes, either by deliberate introduction or by escaping from fish farms (Ogutu-Ohwayo 1990; Weyl 2008; Chifamba 1998). In connection with establish-ment and subsequent increases in population sizes of the Nile tilapia, many native species have declined or totally disappeared (Ogutu-Ohwayo 1990; de Vos et al. 1990). The cause of the decrease in native fish fauna has been attributed to competitive exclusion by the introduced species, or to overfishing from a new and more intensive fisheries that have developed around the newly established species (Ogutu-Ohwayo 1990); Canonico et al. 2005; Njiru et al. 2006a). In this paper, we investigate if increased competition from the introduced Nile tilapia may have contributed to the decline of the native Kariba tilapia (Oreochromis mortimeri) in Lake Kariba by documenting the extent of their diet overlap.

Lake Kariba, on the Zambezi River, was created in 1958 to provide hydro-electricity, though other economic activities that include fisheries were anticipated. To facilitate the development of a fishery, some tracts of land were cleared of trees before inundation (van der Lingen 1973). Between 1964 and 1972, three fish species dominated the catches, a predator, Hydrocynus vittatus (Castelnau 1861) and two cichlids, Oreochromis mortimeri (Trewavas 1966) and Serranochromis condring-toni (Karenge & Kolding 1995), both endemic to the Middle Zambezi River. In 1962, O. mortimeri contributed 35% to the artisanal fishery catches from Sinazongwe area. Later in 1974, it formed 38% of catches from Sanyati to Sengwa basins (Kenmuir 1984).

Aquaculture activity in Kariba grew from small farms in 1980s to include a comer-cial scale cage culture production with a fish processing factory in 1990s. Kariba fish farmers preferred growing O. niloticus and soon escapees from the farms on the shores entered the lake. By 1993 the species had become established (Chifamba, 1998). With the introduction of O. niloticus, the once abundant O. mortimeri declined and disappeared from many parts of the Lake Kariba (Chifamba 2006; Zengeya & Marshall 2008). A similar effect on indigenous fish was observed in other waterbodies where O. niloticus has been introduced (Balirwa 1992; de Vos et al. 1990). In Lake Victoria, it displaced Oreochromis esculentus (Graham) and Oreochromis variabilis (Boulenger) and native cyprinids in Lake Luhondo (Balirwa 1992; de Vos et al. 1990). This negative effect has been attributed to competition for resources such as food.

Earlier investigations of the diets of O. niloticus and that of O. mortimeri in Lake Kariba showed that the two species utilize similar food, but they did not quantify the

(5)

60

composition of different food items (Chifamba 1998; Mhlanga 2000). Oreochromis niloticus feeds on phytoplankton in Lake George and Turkana, though macrophytes are consumed in certain waters (Trewavas 1983). The percentage ‘Frequency of Occurrence’ (FO) used in these studies gave an indication of the food spectrum utilized by the two species. The method has the advantage that it is quick and requires minimum apparatus. However, the method does not give any information on the amount of the diet item consumed (Hyslop 1980), which is needed to calculate diet overlap.

This study estimates the quantity of consumed food items using the ‘Numerical Relative abundance method’ (NM) and ‘Volumetric Relative abundance method’ (VM) (Windell 1968; Hyslop 1980). The NM emphasises the importance of the small food item taken in large numbers, excludes categories that cannot be enume-rated due to mastication and the digestive process and those that do not occur in discrete units such as macroalgae and detritus (Hyslop 1980). It has been suggested that over-estimation may be compensated by faster digestion and evacuation of small prey compared to large ones (Ahlbeck et al. 2012). The volume method gives a better indication of the nutritional contribution of each food item to the diet and reveals any specialization whilst the VM emphasises the large rare food items. To minimize the bias of each method, we use the Index of Relative importance (IRI) that combines the FO, NM and VM (Pinka et al. 1971).

Diet overlap between O. niloticus and O. mortimeri, not before estimated, is then obtained using a modified percent similarity index, D (Schoener 1970) and the over-lap coefficient Q (Pianka 1973). The ability of different diet overover-lap indices to detect simulated distribution was compared by Linton et al. (1981) who found that only Schoener’s D index was accurate over a large range of diet overlap. Most studies of diet-overlap in fish employ Schoener’s index (Khallaf & Alne-ne-ei 1987; Bacheler et al. 2004, Kahilainen et al. 2005) but diets of cichlids fishes in Lake Chivero, Zimbabwe were calculated using Pianka’s Q index (Zengeya &Marshall 2007).

The aim of the present work is to evaluate the potential for food competition between the exotic O. niloticus and the indigenous O. mortimeri in Lake Kariba, by quantifying and comparing their diet composition, diet overlap, and digestion efficiency.

Materials and methods

Fish were caught from the littoral zone of Lake Kariba up to a depth of 13 m, using an electrofisher (Smith-Root Inc, Type VI-A) during daytime in July and August 2003 and at night using gillnets in October 2003, December 2004 and in February and June 2005 (Figure 4.1). The gill-nets were set perpendicular to the shoreline, at depths ranging from half a metre down to 13 m deep. Each gill-net fleet comprised of twelve 40-m panels of 38, 51, 64, 76, 89, 102, 114, 127, 140, 152, 165 and 178 mm

(6)

61

4

Figure 4.1 Skematic map of the Lake Kariba Sanyati basin showing the areas where fish for the

gut analysis were obtained (dark grey areas).

stretched mesh sizes. Fish used for estimation of the digestion efficiency were caught in the Nyaodza River bay by artisanal gill-net fishermen of Nyaodza Fishing Village (Figure 4.1). Standard length and body mass were recorded for each fish, after which the stomach was removed and stored in 70% alcohol, except for those to be used for deter-mination of digestion efficiency, which were stored on ice until further processing. Diet composition

In the laboratory, the stomach contents were removed and placed in a measuring cylinder and diluted with distilled water and then placed in a counting chamber. A good dilution was when individual food items were clearly visible and could be measured. The counting chambers were 24 mm in diameter and 4 or 10 ml in volume. The stomach contents were allowed to settle for 24 hours before counting on an Olympus CK40 inverted microscope, at a magnification of × 400. Material that could not be identified as any organism and consisting of what appeared to be organic matter was classified as detritus, and inorganic granules as sand. The proportion of sand and detritus in the sample was estimated visually and the sample then examined to identify the different taxa. All food items within a minimum of ten microscope fields of view (1.52 mm2_{) were identified to genera using various references}

(7)

62

(Needham & Needham 1962; Pesez & Pesez 1977; John 2000; Komárková-Legnerová & Cronberg 1994; Janse van Vuuren et al. 2006). Each food item was then scored against the list of taxa.

The length and width/diameter of each food item was measured using a calibrated eyepiece graticule. For calibration, the size of each eyepiece graticule unit was measured on a slide micro-meter. The volume of each food item was estimated using equations of nearest shape and using the relationships between biovolume and the length, width and diameter of size groups of different species found in literature (Hillebrand et al. 1999; Olenina et al. 2006) and unpublished (Ronald Bijkerk, personal communication). The detritus and sand were not included in the volumetric and numerical estimates. For the analysis of diets, food items were placed into food categories according to genera.

The stomach contents were analysed using three classic methods; percentage frequency of occurrence (FO), numerical relative abundance (NM) and the volumetric relative abundance methods (VM) (Hyslop 1980). The FO is the percentage of stomachs containing a particular food category out of the total number of stomachs analysed. The numerical method expresses the relative abundance of a particular food category of the total volume of all food categories. To prevent stomachs with many food items from influencing the estimate more than those with less food items, the mean proportion of each food category for each fish was calculated separately and then averaged over all the stomachs from each species. The VM was calculated in the same way as the numeric method.

An index of relative importance (IRI) was calculated for each food item by the formula IRI = (%NM + %VM) × (%FO) (Pinkas et al. 1971). IRI was expressed as a proportion of the sum of the IRI values of all prey items (%IRI). The most important food item has the highest IRI. The %IRI ranges from 0 to 100% where zero indicates no overlap and 100% complete overlap.

The proportion of the detritus and sand in the stomachs of O. niloticus and O. mortimeri were tested for similarity using ANOVA in SPSS version 18 statistical package. The proportion was arcsine transformed because percentage or proportions form a binomial rather than a normal distribution.

Diet overlap

Diet overlap between O. niloticus and O. mortimeri was estimated using a modified similarity index D (Schoener 1970) and the overlap coefficient Q (Pianka 1973):

� � � � ��∑ /

�

��

𝑷𝑷

��

� 𝑷𝑷

��

/� � �

∑�𝑷𝑷�� 𝑷𝑷�� ∑ 𝑷𝑷��∑ 𝑷𝑷��

where n = number of food categories, Pij = proportion (% by volume) of food category

(8)

63

4

the diet of species ‘k’. Values of 0 or 1 for D or Q indicate zero, or full overlap

respectively. The degree of overlap in diet between two species is considered significant when the index is greater than 0.60 (Wallace 1981).

The similarity of the diet was further checked by plotting the proportions of one fish against another. The goodness of fit of the trendline (R2 value) was used to test difference in the diets. The Reduced Major Axis Regression (RMA) was used because both variables (the diets) were measures with error (Smith 2009). Contrary to the univariate regression where the errors are minimized along the vertical or y-axis, in RMA errors in both the y- and x-axis are minimized using the product of the y and x distance of the observation from the trend line. The trend line was fitted using SPSS version 18 statistical package.

Relative digestion efficiency

The stomach and hindgut contents were used to determine the relative protein digestive efficiency. Bowen (1981) and De Silva (1985) reported that in O. mossambicus assimilation does not occur in the stomach, so the stomach contents were used to represent the food ingested. The hindgut content was taken to represent faecal material. The stomach was cut open to remove contents and those of the hindgut were withdrawn from the fish by stripping with a needle and then dried at 70 °C to constant mass. Crude protein was determined using the Kjeldahl method (Bremner & Malvaney 1982). The digestive coefficient (DC), modified from Blackburn and Southgate (1981), for each fish was calculated as:

DC = (S – F) / S × 100

where DC is the digestive coefficient, S the amount (mg) of protein per mg of stomach contents and F is the amount (mg) of protein per mg of faeces.

Results

A total of 51 O. niloticus and 48 O. mortimeri containing food items were used in the diet comparison. Of the stomachs collected using gill-nets, 17 out of 34 and 12 out of 28 O. niloticus and O. mortimeri stomachs, respectively, were empty, as compared to only two (n = 35) and zero (n = 32) from electrofishing, respectively. Because of the small sample sizes, the fish from both gears were pooled for the diet analysis. The mean stomach fullness was 70.9 ± 9.4 and 73.4 ± 9.8 for O. niloticus and O. mortimeri (mean ± SD), respectively, and the difference between species was not significant. Length of O. niloticus caught by electrofishing varied between 3.0 and 10.5 cm (mean = 6.4cm, SD = 0.6) and those by gillnet between 9.0 and 32.7 cm (mean = 21.6 cm, SD = 3.4). The range of O. mortimeri caught by electrofishing was 2.5 to 18.5 cm (mean = 6.5 cm, SD = 1.2) and those by gillnets 7.5 to 33.7 cm

(9)

64

(mean = 24.7, SD = 7.1). Four fish caught by electrofishing could not be identified as either O. niloticus or O mortimeri and were excluded from the analysis.

Diet composition

Most O. niloticus (76.5%) and O. mortimeri (83.3%) consumed detritus. Almost as many O. niloticus (72.5%) as O. mortimeri (70.5%) had sand in their stomachs. Macrophytes were found in 37.3 and 31.3% of O. niloticus and O. mortimeri stomachs, respectively. The proportion of detritus was 37.1 ± 8.5 and 33.2 ± 8.5, while the proportion sand was 19.0 ± 6.9 and 16.6 ± 5.6 in O. niloticus and O. mortimeri stomachs, respectively (mean ± SD). There was no significant difference in the average proportion of detritus (F1,79= 2.21, p = 0.14, ANOVA) and sand (F1,79= 0.05, p = 0.82,

ANOVA) in the stomach content of both fish. One hundred and eight genera were identified among the food items, many of which were diatoms (Table 4.1).

Table 4.1 Diet class and genera of O. mortimeri and O. niloticus as percentage frequency of occur‐

rence (FO), numerical relative abundance (NM), volumetric relative abundance methods (VM) and index of relative importance (IRI). Class values are highlighted. (x = observed only; O. mortimeri stomachs: n = 48; O. niloticus stomachs: n = 51).

Class/Genera FO NM VM IRI FO NM VM IRI

O. mortimeri O. niloticus Arcellinida 2.1 Difflugia x Asplanchnidae 2.1 Asplanchna 2.1 0 0 0 0 0 0 0 Bacillariophyceae 100 65 58 73 100 68 67 77 Achnanthes 10 0 0 0 12 0 0 0 Achnanthidium 10 0 0 0 6 0 0 0 Amphora 29 1 1 1 24 1 2 1 Asteroniella 0 0 0 0 10 0 0 0 Brachysira 4 0 0 0 6 0 0 0 Caloneis 0 0 0 0 8 1 1 0 Cocconeis 17 1 0 0 8 0 0 0 Cymatopleura 31 1 4 1 14 1 2 0 Cymbella 75 10 8 12 73 9 7 11 Diatoma 88 18 2 15 71 15 2 12 Encyonema 54 4 4 4 41 4 4 3 Entomoneis 0 0 0 0 4 0 0 0 Epithemia 50 2 1 1 26 1 0 0 Eunotia 21 1 0 0 10 0 0 0 Fallacia 2 0 0 0 2 0 0 0 Fragilaria 71 7 1 5 69 8 1 6 Gomphonema 92 7 2 7 86 6 2 7 Gyrosigma 13 0 1 0 20 1 3 1

(10)

65

4

O. mortimeri O. niloticus Navicula 60 4 2 3 65 6 3 6 Neidium 2 0 0 0 0 0 0 0 Nitzschia 13 0 0 0 16 1 0 0 Pinnularia 67 2 15 10 73 4 14 13 Rhopalodia 35 1 8 3 45 1 13 6 Sellaphora 2 0 0 0 0 0 0 0 Stauroneis 2 0 0 0 2 0 0 0 Surirella 69 6 9 9 43 7 9 7 Tryblionella 33 1 0 0 35 2 4 2 Brachionida 4 Keratella x Branchiopoda 4 Cladocera x Bosmina x Charophyceae 2 Chara x Chlorophyceae 56 4 6 3 49 3 5 2 Bulbochaete 0 0 0 0 8 0 0 0 Characium 0 0 0 0 4 0 0 0 Chlamydomonas 0 0 0 0 2 0 0 0 Chlorococcum 25 2 1 1 6 0 0 0 Coelastrum 4 0 0 0 0 0 0 0 Botrycoccus x Euastrum 0 0 0 0 2 0 0 0 Eudorina 0 0 0 0 4 0 0 0 Microspora 25 0 0 0 24 1 0 0 Oedogonium 23 0 5 1 16 1 4 1 Pediastrum 8 0 0 0 4 0 0 0 Scenedesmus 25 1 0 0 20 1 0 0 Ulothrix 6 0 0 0 6 0 0 0 Chrysophyceae 6 0 2 0 2 0 0 0 Mallomonas 6 0 2 0 2 0 0 0 Conjugatophyceae 48 2 2 1 39 3 2 1 Closterium 13 0 0 0 8 0 1 0 Cosmarium 27 0 0 0 24 0 0 0 Micrasterias 2 0 0 0 0 0 0 0 Mougeotia 8 0 0 0 20 1 1 0 Staurastrum 35 1 2 1 24 0 1 0 Zygnema 4 0 0 0 0 0 0 0 Coscinodiscophyceae 8 0 0 0 10 0 0 0 Melosira 6 0 0 0 8 0 0 0 Stephanodiscus 0 0 0 0 2 0 0 0 Cryptophyceae 0 0 0 0 4 0 0 0 Cryptomonas 0 0 0 0 4 0 0 0

(11)

66

O. mortimeri O. niloticus Cyanophyceae 50 6 7 4 59 7 6 5 Anabaena 27 2 1 1 33 2 1 1 Anabaena mat 2 0 0 0 0 0 0 0 Aphanizomenon 0 0 0 0 14 0 0 0 Aphanocapsa 8 0 1 0 0 0 0 0 Arthrospira 0 0 0 0 10 1 0 0 Calothrix 4 0 0 0 14 0 2 0 Chroococcus 6 0 0 0 6 0 0 0 Cyanodictyon 6 0 0 0 2 0 0 0 Cylindrospermum 4 0 0 0 4 0 0 0 Dactylococcopsis 2 1 0 0 0 0 0 0 Dichothrix 4 0 1 0 2 0 0 0 Gloeotrichia 6 0 0 0 8 0 0 0 Hydrocoleus 0 0 0 0 2 0 0 0 Hassalia 2 0 0 0 0 0 0 0 Johannesbaptistia 0 0 0 0 2 0 0 0 Leptopogon 0 0 0 0 2 0 0 0 Lyngbya 4 0 0 0 12 0 0 0 Merismopedia 10 0 0 0 10 0 0 0 Microcystis 23 1 1 0 8 1 0 0 Nodularia 6 0 0 0 8 0 0 0 Nostoc 4 0 0 0 4 0 0 0 Oscillatoria 10 0 3 0 10 0 2 0 Phormidium 4 0 0 0 8 1 0 0 Planktolyngbya 0 0 0 0 4 1 1 0 Planktothrix 8 0 0 0 20 0 0 0 Pseudanabaena 8 0 0 0 10 0 0 0 Rivularia 10 0 0 0 4 0 0 0 Snowella 2 0 0 0 4 0 0 0 Trichodesmium 0 0 0 0 2 0 0 0 Desmidiaceae 6 2 1 0 4 0 0 0 Microcoleus 6 2 1 0 0 0 0 0 Spaerozosma 0 0 0 0 4 0 0 0 Diatomaphyceae 60 4 3 2 47 2 2 1 Aulcoseira 44 1 1 1 28 1 1 1 Cyclotella 29 2 2 1 24 1 1 0 Dinophyceae 17 1 2 0 10 0 2 0 Ceratium 2 0 0 0 0 0 0 0 Gymnodinium 4 0 0 0 2 0 0 0 Peridinium 13 1 2 0 8 0 1 0 Peridinopsis 4 0 0 0 4 0 0 0 Euglenoidea 4 0 0 0 10 0 0 0 Euglena 4 0 0 0 8 0 0 0 Euglenophyceae 6 0 0 0 12 0 0 0

(12)

67

4

O. mortimeri O. niloticus Phacus 4 0 0 0 6 0 0 0 Strombomonas 2 0 0 0 0 0 0 0 Trachelomonas 4 0 0 0 4 0 0 0 Fragilariophyceae 94 13 17 16 94 13 13 14 Frustulia 81 5 6 8 59 6 4 6 Meridion 21 1 0 0 2 0 0 0 Synedra 85 7 11 13 69 6 9 11 Tabellaria 6 0 0 0 14 1 0 0 Maxillopoda 2 0 Copepoda x Nauplii 8 1 0 0 0 0 0 0 Synurophyceae 0 4 Synura 0 0 0 0 2 0 0 0 Oocystis 0 0 0 0 2 0 0 0 unknown Platyhelminthes 0 0 0 0 2 0 0 0 unknown 75 3 0 2 49 2 0 1 Xanthophyceae 0 0 0 0 4 1 0 0 Goniochloris 0 0 0 0 4 1 0 0 Zygnemophyceae 2 0 0 0 10 0 1 0 Spirogyra 4 0 0 0 10 0 1 0

Diatoms in the classes Bacillariophyceae, Diatomaphyceae and Fragilariophyceae occurred in all stomachs of both fish species and contributed most (83%) to the diet of O. niloticus by number and volume (Figure 4.2; Table 4.1) when excluding detritus, sand and macrophytes. In O. mortimeri, the contribution of diatoms was 82% by number and 78% by volume. For both O. niloticus and O. mortimeri the IRI of diatoms was the highest of all the other groups in the diet. The diatoms belonging to the genera Gomphonema, Cymbella, Pinnularia, Diatoma, Fragilaria, Synedra, Navicula and Frustulia occurred in more than half the stomachs of O. niloticus (Table 4.1). Surirella, Encyonema and Epithermia were also common diatom genera in the diet of O. mortimeri. Diatoma was the most numerous genus whilst Pinnularia contributed the highest to the proportion by volume in both species (Figures 4.3 and 4.4). Many food categories were rare and contributed little to the numbers and volume of the stomach contents.

The most important taxa in the diet of O. niloticus according to the IRI were benthic epilithic genera Pinnularia (13%), Diatoma (12 %), Cymbella (11%), Synedra (11%) and Sirirella (7%). These taxa were also the top most important in the diet of O. mortimeri, though they differ in their order of importance. Diatoma (15%) was most important, followed by Synedra (13%), Cymbella (12%), Pinnularia (10%),

(13)

68

Figure 4.2 The a) percentage frequency of occurrence (FO) b) % numerical relative abundance c)

volumetric relative abundance and the d) % index of relative importance of the food classes consumed by O. niloticus and O. mortimeri.

Table 4.2 Percentage protein in the extract from the stomach and hindgut and the Apparent

Digestion Coefficient (ADC) of O. mortimeri and O. niloticus (CI = 95% confidence interval). Data type

% protein % protein

stomach hindgut ADC stomach hindgut ADC

O. mortimeri O. niloticus Individual 30.3 28.4 6.2 27.7 23.4 15.6 Individual 21.6 15.8 26.7 22.1 8.4 62.0 Individual 22.1 3.5 84.2 22.6 16.8 26.0 Individual 43.1 21.2 50.8 38.7 19.0 50.9 Individual 36.6 25.9 29.2 21.9 11.9 45.7 Individual 39.1 16.4 57.9 37.5 15.3 59.3 Individual 35.8 21.9 38.6 48.9 29.8 39.2 Individual 17.4 14.1 19.0 23.8 11.3 52.6 Individual 19.7 7.5 61.9 10.0 7.1 28.8 Individual 12.4 3.5 71.9 Individual 16.0 10.8 32.4 Mean and CI 29.5 ± 7.3 17.2 ± 6.3 41.6 ± 18.7 25.6 ± 8.0 14.3 ± 5.1 44.0 ± 11.6 a) Frequency of occurance % F re qu en cy o f o cc ur an ce 0 20 40 60 80 100 c) % Volume % vo lu m e 0 20 40 60 80 100

d) % Index of Relative importnce

Bacil lario phyc eae Chlor ophy ceae Chrys ophy ceae Conju gato phyc eae Cosc inodi scop hyce ae Cryp toph ycea e Cyan ophy ceae Desm idia ceae Diat omap hyce ae Dino phyc eae Eugle noid ea Eugle noph ycea e Frag ilario phyc eae Xanth ophy ceae Zygn emop hyce ae % In de x of R el ati ve Im po rta nc e 0 20 40 60 80 100 O. niloticus O. mortimeri b) % Number Bacil lario phyc eae Chlor ophy ceae Chrys ophy ceae Conju gato phyc eae Cosc inodi scop hyce ae Cryp toph ycea e Cyan ophy ceae Desm idia ceae Diat omap hyce ae Dino phyc eae Eugle noid ea Eugle noph ycea e Frag ilario phyc eae Xanth ophy ceae Zygn emop hyce ae % N um be r 0 20 40 60 80 100 O. niloticus O. mortimeri

(14)

69

4

Figure 4.3 Relationship between numerical relative abundance (NM) and volumetric relative

abundance methods (VM) of food categories occurring in more than 35 % of stomachs, consumed by a) O. niloticus and b) O. mortimeri.

(15)

70

Figure 4.4 Relationship between the percentage frequency of occurrence (FO) and volumetric

relative abundance methods of food categories occurring in more than 35 % of stomachs, consumed by a) O. niloticus and b) O. mortimeri.

(16)

71

4

Figure 4.5 Relationships between the importance of food items found in the stomachs of O.

niloticus and O. mortimeri, as demonstrated by: a) the frequency of occurrence (R2_{= 0.89, t =}

28,1) b) the abundance (numerical method) (R2_{= 0.92, t = 35.3) c) the volume (volumetric}

method) (R2_{= 0.89, t = 29.2) and d) the index of relative importance (IRI) (R}2_{= 0.91, t = 32.5) (p <}

0.0001 for all plots).

and Sirirella (9%). Zooplankton taxa such as copepods were rare and were only found in two stomachs, while the cladoceran Bosmina sp., the rotifer Keratella sp., an un-identified cladoceran and a flatworm (Platyhelminthes) were found only once.

The importance of different food items based on relative frequency (FO), abun-dance (NM), volume (VM) and relative importance (IRI) all correlated strongly between O. niloticus and O. mortimeri (FO: R2_{= 0.89, t = 28.1, p = 0.001; NM: R}2

= 0.92, t = 35.3, p = 0.001; VM: R2_{= 0.89, t = 29.2, p = 0.001; IRI: R}2_{= 0.91, t =}

32.5, p = 0.001; Figure 4.5). Comparing the slope of the reduced major axis regression line of O. niloticus and O. mortimeri for the FO, NM, VM and IRI, shows that the differences in the diet were small (1.18, 1.10, 1.00 and 1.03 for FO, NM, VM and IRI, respectively).

a) Frequency of occurance O. niloticus 0 20 40 60 80 100 O. m or tim er i 0 20 40 60 80 100 b) numerical method O. niloticus 0 5 10 15 20 25 O. m or tim er i 0 5 10 15 20 25 O. niloticus vs O. mortimeri isometric line linear regression c) Volume method O. niloticus 0 5 10 15 20 25 O . m or tim er i 0 5 10 15 20 25 d) IRI O. niloticus 0 5 10 15 20 25 O. mo rtime ri 0 5 10 15 20 25

(17)

72

Diet overlap and relative digestion efficiency

The diets of Oreochromis niloticus and O. mortimeri overlapped significantly, with

D and Q similarity indices of 0.75 and 0.95, respectively. These values are well

above the biologically significant limit of 0.60 suggested by Wallace (1981). There was no significant difference in the amount of protein in the stomach and the intestine as well as in the digestion coefficient of the two species (stomach: t = 0.80, df = 18, p = 0.44; intestine: t = 0.83, df = 18, p = 0.42; digestion efficiency: t = 0.26, df = 18, p = 0.80).

The mean quantities (± SD) of protein in extracts from the stomach and hindgut of O. niloticus were 25.6% ± 8.0 and 14.3% ± 10.8, respectively, whilst the mean DC was 44.0% ± 11.6 (Table 3.2). Comparable amounts were found in O. mortimeri (stomach: 29.5% ± 7.3; hindgut: 17.2% ± 13.7; DC: 41.6% ± 18.7).

Discussion

Diet compositions of the introduced Oreochromis niloticus and the native O. mortimeri in Lake Kariba were very similar and dominated by benthic diatoms. Most O. niloticus (76.5%) and O. mortimeri (83.3%) consumed detritus. Both Schoener’s

D (1970) and Pianka’s Q (1973) similarity indices, indicated an almost complete

overlap in diet and the digestion efficiencies of the species were also highly comparable. This suggests that the species have very similar diet niches and that competition for food is likely if food resources are limited.

Diet composition

Tilapiines are known to consume a wide range of food that include algae, detritus, macrophytes, zooplankton and aquatic insects; a typical generalist diet of Oreochromis niloticus (Lowe-McConnell 2000; Moriarty 1973; Trewavas 1983). Therefore, the diet of O. niloticus and O. mortimeri found in this study is typical. Confirming earlier findings in Lake Kariba, diatoms are common in the diet of both O. niloticus and O. mortimeri even though other algae species, macrophytes, zooplankton and detritus are consumed (Chifamba 1998). Matthes (unpublished) found that adults O. mortimeri in Lake Kariba mainly feed on algae gathered from substrate and that higher plants were incidentally ingested whereas juveniles (< 7 cm) were omnivo-rous and had a diet that included small invertebrates such as insect larvae and algae. There is not much published data on the diet of O. mortimeri, but from this study and the available literature its diet appears to be similar to that of O. niloticus, and mostly derived from scraping and suction at the bottom.

Detritus nutrition is derived from the detritus itself, associated organisms which include epiphytic and benthic algae, and non-protein amino acids (Bowen 1982). Oreochromis mortimeri of different sizes feed by dredging the bottom rather than browsing (van der Lingen 1973). This feeding strategy may explain the high

(18)

73

4

proportion of stomachs that contained detritus, sand and benthic diatoms found in

this and other studies. O. niloticus also ingests detritus and other benthic organisms (Peterson et al. 2006; Tudorancea et al. 1988). The most frequent items in the stomachs of O. niloticus caught in the Mississippi watershed were amorphous debris, detritus, sand grains and mud clumps (Peterson et al. 2006). Benthic diatoms such as Synedra, Cymbella, Neidium and Gomphonema blue greens were the main diet of juvenile O. niloticus (10 to 50 mm) in Lake Awassa, Ethiopia (Tudorancea et al. 1988). Adults (18 - 32 cm) in the same lake fed mostly on Chroococcus, Oscillatoria and Botryococcus (Getachew & Fernando 1989).

Phytoplankton and periphyton are also important sources of energy and the diet of Oreochromis niloticus depends on what dietary items are available in a water body (Huchette et al. 2000; Huchette & Beveridge 2003; Lowe-McConnell 1958; Trewavas 1983). In Lakes George and Turkana where phytoplankton is abundant, it is the dominant food compared to Lake Albert where the main food is epiphytic diatoms (Trewavas 1983). The diet of O. niloticus in Lake Kariba differs from that in Lake Chivero, Zimbabwe where the fish fed mainly on blue-greens (Ndebele 2003; Zengeya & Marshall 2007). Microcystis sp. and Melosira sp. were dominant, whilst Cyclotella and Pediastrum sp. were common (Ndebele 2003). Numerical dominance of the plankton community in Lake Chivero by Microcystis aeruginosa (64%) and Melosira sp. (19.3%) may have influenced the diet. Because of low nutrients concen-trations in Lake Kariba, blue-green algal blooms were present only in rich rivers and estuaries (Cronberg 1997). A total of 40 genera of diatoms were found on the submerged macrophyte Valisneria aethiopica. The most abundant genera were Achnanthidium and Gomphonema, which made up 23.4 and 42.9% of the diatom count, respectively (Phiri et al. 2007). Diatoms are therefore an available food resource in Lake Kariba explaining their dominance in the diet of O. niloticus and O. mortimeri, as found in this and previous studies (Chifamba 1998).

There is seasonality in algal composition with blue-greens dominating the algal community during the warm-rainy season and diatoms during the cold-dry season (Cronberg 1997). Significant seasonal difference in the abundance of Chroococcus and Oscillatoria was found in the stomach contents of O. niloticus caught in Lake Awasa Ethiopia (Getachew & Fernado 1989). Therefore, a seasonal analysis of diet might have revealed seasonality in the algal species in the diet.

A diet shift attributed to environmental change was observed in O. niloticus in lakes Nabugabo and Victoria (Njiru et al. 2004; Bwanika et al. 2006). In these studies, omnivory was observed in lakes Nabugabo and Victoria where Nile perch (Lates niloticus) was introduced and had reduced the populations of haplochromine cichlids, and phytoplanktivory where L. niloticus was absent. Before L. niloticus was introduced in Lake Victoria, O. niloticus was herbivorous feeding mostly on algae but following the introduction of L. niloticus the diet diversified to include insects,

(19)

74

fish, algae and plant material. Hence, the diet of O. niloticus displays both spatial and temporal variation, a valuable trait for an invasive fish species.

Different diet items eaten by O. mortimeri and O. niloticus require different feeding techniques for their acquisition (Trewavas 1983; Yamaoka 1991). Peri-phyton is obtained by scraping or rasping it from the substrate, and in the process large quantities of the substrate are also consumed (Yamaoka 1991). Oreochromis niloticus consumes phytoplankton by gulping water and collecting plankton by means of a mucus- trap mechanism (Trewavas 1983). Copepods as well as terrestrial and aquatic insects are caught by actively pursuing the prey whilst detritus is consumed in a pecking motion. Ability of Oreochromis spp to utilize a variety of feeding methods is related to the functional morphology of their mouths, that are intermediate in size compared to predominantly visually feeding cichlids and suction feeders (Beveridge & Baird 2000). The use of diverse feeding methods by Oreo-chromis niloticus has enabled diet plasticity, which is an important attribute in an invasive species.

Diet overlap

Above a true overlap of 0.76 the Schoener index tends to underestimate the overlap, whilst the value from the Pianka equation is more accurate (Linton et al. 1981). This means that the higher overlap value from the Pianka equation may be closer to the true diet overlap of the two species. In Lake Chivero, both Oreochromis niloticus and O. macrochir fed mostly on blue-green algae (> 50%) and their diets, in all size classes, overlapped almost completely (Zengeya & Marshall 2007). Dietary overlap (Pianka 1973) coefficient for Lake Chivero was 0.98 (close to the value of 0.95 for O. niloticus and O. mortimeri in Lake Kariba), indicating an almost complete overlap in diet between the species.

Having similar diet does not always prevent coexistence because factors such as feeding behaviour, feeding site and habitat may reduce competition (Yamaoka 1991). In its native habitat in a River Nile canal in the Egyptian delta, the diet of O. niloticus overlapped significantly with that of Tilapia zillii (Gervias), particularly in the age groups 10 – 19.9 cm (Khallaf & Alne-na-ei 1987). The main food for O. niloticus was macrophytes (97%) compared to T. zillii (92%). Competition between the species may be reduced by specialization of T. zillii on aquatic insects. The diet of three cichilids, Orochromis andersonii, O. macrochir and O. sparmanii, was dominated by vegetative detritus and yet these fish species cohabited the Upper Zambezi River and floodplain (Winemiller & Kelso-Winemiller 2003). Yamaoka (1991) reviewed studies where commensal and mutualistic relationships may have allowed coexistence of cichlid species in African Great Lakes. Mechanisms for sharing food in species recently been placed in the same environment may be lacking, leading to displacement of one of the species with competitive disadvantage.

(20)

75

4

There are conditions under which competition is likely, such as the distribution of

the food resource in space and time (Milinski & Parker 1991). Even though it has been suggested that a diet of detritus may not be limited in the environment, its quality has been shown to vary in space (Bowen 1979, 1980, 1981). Bowen (1979) found that juveniles of Sarotherodon mossambicus that feed in the shallow water in Lake Sibaya, South Africa, have access to detritus with higher protein level than adult fish that inhabit and feed in deep water. This leads to malnourishment of adults whilst the juveniles had good growth rates. Such spatial differences in resource introduce the possibility for contest for optimal feeding sites. Though the spatial distribution of food in Lake Kariba was not mapped, a spatial difference in the distribution of O. niloticus and O. mortimeri was observed indicating displacement of the native species (Chifamba 2006).

Digestion efficiency

Protein content and digestion efficiency could be a factor that confers competitive advantage to O. niloticus over O. mortimeri. It has been shown that protein content is the factor limiting the growth of herbivorous and detritivorous fish. Whilst the prey of carnivorous fish contain > 80% protein, the protein content in the food of tilapias ranges from < 1% to 50% only (Bowen 1980, 1982; Bowen et al. 1995). Protein content in the stomachs of O. niloticus (mean = 25.6% ± 8.0) and O. mortimeri (mean = 29.5% ± 7.3) measured in this study fell within this range. The mean digestive coefficient of proteins for O. niloticus in Kariba (44.0% ± 11.6) was lower than that found in Lake Chivero (62% ± 9.1) but that of O. macrochir in Lake Chivero was even lower (39.4% ± 9.1; Marufu & Chifamba 2013). Differences between the lakes may have arisen from the differences in food sources; whereas the difference between O. macrochir and O. niloticus may be attributed to the superior digestion capability of blue-green algae by O. niloticus (Moriarty 1973). This study found no significant difference in the DC of O. niloticus (44.0% ± 11.6) and O. mortimeri (41.6% ± 18.7). Therefore, differences in the species competitiveness is not likely to be due to protein content of the diet or the digestion efficiency. Furthermore, the stomach fullness was high for both O. niloticus (70.9% ± 9.4) and O. mortimeri (73.4% ± 9), which suggests that neither species has an inadequate diet as a result of feeding on similar food items.

Oreochromis niloticus seems to have competitive superiority to congeneric species wherever it was introduced in Africa, and similarity in diet seems to be a common factor (Balirwa 1992; Zengeya & Marshall 2008). In Lake Victoria, it displaced Oreochromis esculentus (Graham) and Oreochromis variabilis (Boulenger), and other native cyprinids in Lake Luhondo (Balirwa 1992,). Wherever it was introduced in Zimbabwe, O. niloticus became the dominant cichlid species by replacing O. macrochir, O. mossambicus and O. mortimeri (Marshall 1999; Chifamba 2006). Even though similarity in diet is a common factor among these species it may

(21)

76

not be the only factor determining the outcome of the competition. Other traits such as fast growth rate and aggression levels might be of importance (Chapter 3 – Chifamba & Videler 2014; Chapter 5 – Chifamba & Mauru 2017).

Implication to fish introductions

Diet similarity between O. mortimeri and O. niloticus demonstrates that the intro-duced O. niloticus does not occupy a new niche in the Lake Kariba ecosystem. Consequently, fish production was not enhanced by introducing this alien species. Therefore, we recommend that introduction of other fish species into Lake Kariba or O. niloticus into other waterbodies should be preceded by examination of niches that are not occupied by the native species to avoid their displacement. Since it is not possible to reverse the consequence of O. niloticus introduction, management must focus on the sustainable exploitation of this fish. In addition, refugia populations of O. mortimeri need to be identified and protected to prevent loss of biodiversity. Acknowledgements

We are grateful to the crew of MV Erika, the staff of University Lake Kariba Research Station and of Lake Kariba Fisheries Research Institute, especially the late Mr Mushaike and Mr Chisaka. We thank Mr Ronald Bijkerk for providing tables of plankton biovolumes used in estimating biovolume. Mr Edwin Tambara assisted in the processing of the fish samples. Collection of gillnet samples was funded by the International Foundation for Science (IFS) Grant A/3159-1. Electrofishing samples were obtained during a VLIR scientific expedition.