Biological and life-history parameters for Labeo rosae Steindachner, 1894 and Oreochromis mossambicus (Peters, 1852) from Flag Boshielo Dam, Olifants River, South Africa

Inland fisheries contribute towards food security for rural communities living close to large aquatic systems. However, sustainable inland fisheries rely on accurate biological information for the target species at the proposed fishery location. In this study, the biological information for Labeo rosae Steindachner, 1894 and Oreochromis mossambicus (Peters, 1852) were determined at Flag Boshielo Dam, Olifants River system, Limpopo province, as part of a larger study to evaluate a small-scale gill net fishery at this impoundment. Nine fish surveys were conducted from February 2016 to April 2017 using mixed sampling gears. Labeo rosae exhibited positive allometric growth (b > 3) while O. mossambicus displayed negative growth (b < 3). For L. rosae, the asymptotic length was 391 mm L ∞, and the length-at-50% sexual maturity 165.8 mm. For O. mossambicus, the asymptotic length was 420 mm L ∞, and the length-at-50% sexual maturity 166.5 mm. Both species population stocks showed signs of exploitation. Total mortality for L. rosae was 0.57 yr−1 with a fishing exploitation of 0.21 yr−1, while total mortality of 0.78 yr−1 and fishing exploitation of 0.50 yr–1 were estimated for O. mossambicus. The biological data collected, apart from the length–weight parameters for O. mossambicus, are suitable for conducting a fisheries assessment for these species at Flag Boshielo Dam.

South Africa's National Development Plan 2030 identifies poverty alleviation and unemployment as major priorities (National Planning Commission 2013), and the South African government has identified inland fisheries as a potential priority area for social and economic development (Department of Agriculture Forestry and Fisheries 2020). Unlike the marine sector, inland fisheries are not well-established in South Africa, with initiatives for small-scale fisheries poorly developed (Hara and Backeberg 2014) and prone to failure (Barkhuizen et al. 2016). Being semi-arid, South Africa has approximately 4 700 reservoirs and a number of inter-basin transfer schemes to store and supply water to areas where domestic, agricultural and industrial demand is high (Hara and Backeberg 2014). However, the potential national fisheries yield from these impoundments was empirically estimated at 15 000 t yr -1 , a figure these authors cautioned should be verified through finer scale analyses.
The fish stocks in most South African impoundments are considered to be underutilised (Weyl et al. 2007;Ellender et al. 2010;McCafferty et al. 2012). However, their relatively low productivity precludes the development of large-scale commercial inland fisheries . Nevertheless, Britz et al. (2015) believe there is considerable scope for the development of small-scale fisheries in north-eastern Limpopo and northern KwaZulu-Natal provinces. A major limitation of empirical desktop assessments is that they fail to consider the species composition and size of fish stocks within the impoundments . The Rome Declaration: Ten Steps to Responsible Inland Fisheries strongly emphasises improving the assessment of biological production of fisheries to enable sciencebased management, appropriately valuing inland aquatic ecosystems, developing and improving science-based approaches to fishery management, improvin governance, especially for shared water bodies, and developing an Introduction action plan for global inland fisheries (FAO and Michigan State University 2016;Taylor et al. 2016). Consequently, the proper assessment and management of a fishery requires an understanding of the biology, life history, and distribution of the target species (King 2013). This information is vital for supporting the decision to develop a fishery (Weyl et al. 2007;Sara et al. 2017b); however, the lack of these data for many large impoundments in South Africa remains a major constraint to the sustainable development of inland fisheries. There is thus a need to collect the data required to support the development of sustainable fisheries in areas where the productivity of the impoundments is sufficient for such a fishery. Tapela et al. (2015) suggested that Flag Boshielo Dam, Limpopo province, South Africa (Figure 1), a mainstem impoundment on the Olifants River in the broader Limpopo River system, is a suitable candidate for the development of a sustainable small-scale fishery. The impoundment, however, falls on the boundary of a region with a high fishery potential recommended by Britz et al. (2015), and despite numerous studies having been conducted at the impoundment, knowledge of the fish stock composition and the biology of potentially commercially exploitable species is limited. Research at the impoundment has primarily focused on the limnology (Dabrowski et al. 2014); fish parasites (Madanire-Moyo et al. 2012a;Madanire-Moyo et al. 2012b); water quality (Heath et al. 2010;Jooste et al. 2013); the metal content in the muscle tissue of selected fish species and the associated health risks to humans if consumed (Addo-Bediako et al. 2014a, b;Jooste et al. 2014Jooste et al. , 2015Marr et al. 2015;Lebepe et al. 2016;Marr et al. 2017b;Sara et al. 2017a;Sara et al. 2018;Sara et al. 2021b); and, the prevalence of pansteatitis in Oreochromis mossambicus (Peters, 1852) (Sara et al. 2020), a disease that results in the necrosis and hardening of adipose tissue (Huchzermeyer et al. 2017). Only Brits (2006) and Sara et al. (2017b) (Dabrowski et al. 2014;Tapela et al. 2015). In addition, Flag Boshielo Dam is an important recreational angling venue for indigenous species, such as O. mossambicus, and introduced species, such as Micropterus salmoides (Lacepède, 1802), C. carpio and H. molitrix, which contributes to the local economy. Sara et al. (2017b) found that L. rosae, rednose labeo, and O. mossambicus, Mozambique tilapia, were the dominant fish species in the impoundment, both in terms of abundance and biomass. Given that large cyprinids of the genus Labeo and cichlids of the genus Oreochromis are harvested in small-scale and commercial fisheries elsewhere (Booth and Weyl 2004;Weyl et al. 2010;Tweddle et al. 2015), the aim of this study was to collect the biological and life-history parameters for L. rosae and O. mossambicus necessary to evaluate the potential of a small-scale fishery targeting these species at Flag Boshielo Dam.

Permits and ethical approval
The project was executed under a research permit from the Limpopo Department of Economic Development, Environment and Tourism (ZA/LP/HO/3370) and Animal Ethics Approval from the University of Limpopo Animal Research Ethics Committee (AREC/09/2017:PG).

Study area
Flag Boshielo Dam (24°47′0″ S, 29°25′40″ E) is located at the confluence of the Olifants and Elands Rivers, approximately 30 km from the town of Marble Hall in Limpopo province, South Africa ( Figure 1). The dam was constructed in 1987 to secure water for agricultural, municipal and mining activities in neighbouring municipal districts (van Koppen 2019). In response to increasing water demand, the wall was raised by 5 m between 2007 and 2008, increasing the storage capacity from 100 to 188 million m 3 and the annual water supply from 56 to 72 million m 3 (Jooste et al. 2013). The lake's sub-catchment covers an area of 4 213 km 2 . Flag Boshielo Dam is surrounded by typical central sandy bushveld savanna vegetation (Mucina and Rutherford 2006) with many partly and fully submerged dead trees providing roosts for piscivorous birds like darters, herons, cormorants and fish eagles (Jooste et al. 2013). The Schuinsdraai Nature Reserve covers a large portion of the impoundment's western shore, restricting public access to recreational anglers and wildlife enthusiasts. In contrast, besides the state-run Tompi Seleka Agricultural College and the privately-owned Aloe Park Holiday and Fishing Resort, public access to the eastern shore of the impoundment is unrestricted. The lake and its tributaries fall within the southern temperate highveld freshwater eco-region (Abell et al. 2008). Categorised to be oligo-to meso-trophic, the impoundment is dominated by nitrogenfixing phytoplankton species, e.g. Cylindrospermopsis sp. (Dabrowski et al. 2014). Further, it hosts a considerable, albeit declining, population of Nile crocodile Crocodylus niloticus Laurenti, 1768 (Botha 2010;Botha et al. 2011).

Fish surveys
Between February 2016 and April 2017, nine fish surveys were undertaken at Flag Boshielo Dam. Fish surveys were conducted during April, May, June, August, September, October and November 2016, and February and April 2017. For each survey, nets were deployed for two nights, with the exceptions of June 2016, when the nets were only deployed for a single night, and April and October 2016, when the nets were deployed for three nights.
Sites were randomly selected within the riverine and lacustrine regions of the impoundment. At each site, in situ water temperature was recorded between 06:00 and 10:00 immediately below the surface and at 1 m and 2 m depths using a multi-parameter handheld meter (Model: YSI 554 Datalogger and multiprobe). In addition, the level of Flag Boshielo Dam (% capacity) was obtained from the Department of Water and Sanitation's (DWS) hydrology website for the impoundment (https://www.dwa. gov.za/hydrology/weekly/percentile.aspx?station=B5R002; accessed September 2016 and August 2017).
Fish surveys were conducted using a fleet of four composite multi-filament gill nets. Each gill net comprised five randomly positioned panels of 44, 60, 70, 100 and 144 mm stretch mesh attached to form a single 25 m net, each panel 5 m wide and 2 m deep. To avoid and limit crocodile interactions (Sara et al. 2017b), all four gill nets were deployed approximately two hours before sunset and left to soak for four to six hours. Whilst deployed, the nets were monitored for crocodile activity. Fyke nets were deployed overnight in habitat types different to those surveyed by the gill nets. Due to the presence of crocodiles, seine nets were not used. On removal from the net, the specimens were identified according to Skelton (2001), Maake et al. (2014) and Skelton (2016), the total length (TL: mm) and live mass (g) recorded by mesh size. Following Sara et al. (2017a), target species were identified as the species that comprised more than 20% of the total biomass caught. Selected specimens of the target species were retained after each survey for the biological assessments. All unused fish and bycatch were returned live to the water.  where k is the number of categories and p i the proportion of observations found in category i. and species evenness (J') using (Zar 1999):

Length-weight relationship
The length-weight relationship for the target species was described using the standard power curve (Anderson and Neumann 1996): where W is the total body mass (g), L a is the length-at-age (mm) and a and b are parameters determined by a linear regression of the log-log transformed length and weight using the lm function in R.

Age and growth estimations
The fish were sacrificed by severing the spinal cord directly behind the head region. Sagittal otoliths were removed, cleaned and stored dry in labelled envelopes. In the laboratory, the otoliths were mounted medial side down in clear casting resin that was subsequently sectioned transversely through the nucleus using a double-bladed diamond-edged saw and mounted on a microscope slide using DPX mountant. Growth was determined based on the number of opaque and translucent margins following Weyl and Hecht (1998), Winker et al. (2010), and Winker et al. (2012). Counts of the age were determined by two researchers, independently, without reference to the length of the specimen and verified following Campana (2001).
Growth parameters for each species were determined using length-at-age by fitting the three parameter von Bertalanffy growth function (VBGF) of the form following Beverton and Holt (1956).
where L a the length-at-age a, L ∞ is the predicted asymptotic length, t o is the age-at-"zero" length; and K the Brody growth co-efficient (Ricker 1975) that determines the rate at which L ∞ is attained. Following Winker et al. (2012), the model was fitted by minimising the negated normal log-likelihood function using the nls function in R to determine the nonlinear least-squares estimates of the parameters of the VBGF.

Sexual maturity
Specimens retained for examination were sacrificed, dissected, sexed, and the developmental stage of gonads was macroscopically determined and categorised following Weyl and Booth (1999): Stage 1 Juvenile; Stage 2 Resting; Stage 3 Developing, Stage 4 Ripe; and Stage 5 Spent (see Table S1, supplementary material). Length-at-maturity was expressed as the proportion of reproductively active females (developing, ripe and spent reproductive stages) in each 10 mm size class. The proportion of sexually mature individuals (ψ) by length (L) and age was fitted to the logistic curve (Booth and Hecht 1997) based on the assumption that all mature individuals are in reproductively active condition at the same time (i.e. asymptotic maturity (m ∞ ) is unity): where Ψ is the percentage of mature fish at length (L), L m50 is the mean length-at-50%-sexual maturity and δ is the width of the logistic ogive. This was implemented in R by using the glm function (family binomial) to fit an ogive to the proportion of mature females as a function of length. The parameters L m50 and δ were determined from the coefficients generated by the glm. A 95% confidence interval was estimated as the logit of the predicted ogive ±1.96 the standard error of the predicted ogive.
Total mortality (Z; yr -1 ) was calculated by determining the fraction or percentage of the population surviving at age t described by the negative exponential function used by Ricker (1975) of the form: where N t is the number of fish at time t, N 0 is the number of fish at time t 0 , and Z is a first approximation of the instantaneous rate of the total mortality rate estimated by catch curve analysis (Ricker 1975). To determine total mortality, catch curve analysis was applied to length frequency distributions, which were converted to age frequency distributions by means of a normalised age-length key (Butterworth et al. 1989). The natural log of specimens in each age class was then plotted across the age-structure of the population and a linear regression fitted to the descending section of the catch curve as described in Ricker (1975).
Fishing mortality (F; yr -1 ) was calculated by using the estimates of Z and M (Ricker 1975, Pauly andMorgan 1987) such that: and fishing exploitation (u) calculated as (Ricker 1975):

Catch-per-unit-effort
Mean hourly CPUE (kg m -1 hr -1 ) was calculated for each species per gill net deployment following Pollock et al. (1994): where C i is the catch mass (kg) per species i, E i is the effort (soak time in hours), and n is length of the gill net (25 m). To estimate the catch rates for commercial gill nets, which are generally 100 m in length, the hourly CPUE was multiplied by 100 and expressed as kg 100-m net -1 hr -1 for a 100 m gill net comprised of equally sized panels of 44-, 60-, 70-, 100-and 144-mm stretch mesh.
To determine whether CPUE varied between surveys, the data were first tested for normality and homogeneity of variance using the Shapiro-Wilk and Kolmogorov-Smirnoff tests using the shapiro.test and ks.test functions in R, respectively. One-way analysis of variance (ANOVA) was used for normally distributed data and the Kruskal-Wallis test was used for non-parametric data using the aov and kruskal.test functions in R, respectively. Where the ANOVA or Kruskal-Wallis tests returned a significant result, post-hoc tests were conducted to determine which of the pair-wise contrasts between survey pairs contributed to the significant result. The Tukey HSD test was used for the ANOVA and the Dunn test for the Kruskal-Wallis test using the Tukey HSD function in R and the DunnTest function from the DescTools package for R (Signorell et al. 2019), respectively.

Results
This research commenced when the water level in Flag Boshielo Dam was less than 50% of capacity; see Sara et al. (2021a) and Marr et al. (2022). A year prior to the study, January 2015, the impoundment was at 100% capacity (Sara et al. 2021a;Marr et al. 2022). Through 2015, the impoundment dropped steadily to about 40% of capacity to when the study commenced in February 2016. The first four surveys were conducted at about 40% of capacity. On only three occasions between 1998 and 2018 had the impoundment fallen below 40% of capacity (see Marr et al. 2022). Between the fourth and the fifth surveys, July 2016, the capacity dropped to below 30% of capacity and, over the next four surveys, decreased to 17% of capacity in December 2016. Between the eighth and ninth surveys, January 2017, the level increased to about 50% of capacity. It took a further year before the level increased to 100% capacity and spilled in April 2018.
A total of 1 376 specimens, from 11 species in six families, were captured using gill nets (Table 1). Species composition was dominated by four cyprinids; L. rosae, southern papermouth Enteromius rappax (Steindachner, 1894), C. carpio and the lowveld largescale yellowfish Labeobarbus marequensis (A. Smith, 1841), followed by two cichlids; O. mossambicus and Coptodon rendalli (Boulenger, 1896). The catch included two alien species, C. carpio and M. salmoides. The total biomass captured using gill nets was 247.6 kg with L. rosae and O. mossambicus contributing 58.7% and 22.0% towards the total catch, respectively, while C. gariepinus contributed 7.2% and Synodontis zambezensis Peters, 1852 6.0% to the total catch (Table 1). About 3% of the total catch was damaged by crocodile activity. Species diversity of H′ = 0.54 and an evenness of J′ = 0.5 were considerably lower than those reported by Brits (2006) and may, however, be attributed to the assortment of gears used by Brits (2006).
The length-frequency distribution for L. rosae was bimodal and dominated by adults (Figure 2a) with little evidence of recent recruitment; no individuals < 170 mm TL were captured. In contrast, the O. mossambicus distribution was unimodal, peaking at 170-190 mm TL (Figure 2b). The length-weight relationship for L. rosae exhibited positive allometric growth (b > 3, R 2 = 0.93) (Figure 2c), while that for O. mossambicus exhibited negative allometric growth (b < 3; R 2 = 0.96) (Figure 2d). The b value derived for O. mossambicus was low at 2.59. The L. rosae catch was skewed towards males with females comprising 38.6% of the catch. There was a significant difference between gender size (Kruskal-Wallis p < 0.001) with females being significantly larger than male specimens (Figure 3a). The O. mossambicus catch was skewed slightly towards males with females comprising 47.9% of the catch. There was no significant difference in size between the genders (Kruskal-Wallis p = 0.451) (Figure 3b).  (2016) and ** indicates alien species. The maximum age for L. rosae was 10 years. An analysis of the age structure of L. rosae found that no 1-year age class specimens were recorded in the study. Initial estimates for the VBGF parameters resulted in a very poor fit. To compensate for the missing age classes, the age-at-zero-length was set to 0 as recommended by Ogle (2018). This revision resulted in a vastly improved fit for the VBGF. The asymptotic length for the revised VBGF (L ∞ ) was estimated at 391 mm TL and the Brody coefficient (K) at 0.314 (Figure 4a; Table 2). Marginal zone analysis of the otoliths revealed that opaque growth dominated the otolith margins in August and September with a single opaque zone deposited yearly (Supplementary material: Figure S1a). The maximum age for O. mossambicus was 10 years and the L ∞ estimated at 420 mm TL, Brody coefficient (K) at 0.227, and the age-at-zero-length at 0.073 years ( Figure 4b; Table 2). Marginal zone analysis of the otoliths revealed the deposition of single opaque zone between April and June, annually (Supplementary material: Figure S1b). The L. rosae age assessment was skewed towards males with females comprising 35.3% of the sample. There was a significant difference between gender age with the females of the species being significantly older than those recorded for the males (Kruskal-Wallis p < 0.001; Figure 4c). The O. mossambicus catch was skewed towards males with females comprising 42.5% of the sample. There was a significant difference in age between the genders (Kruskal-Wallis p = 0.007) with males being older than females (Figure 4d).
The mean length-at-50%-sexual maturity for L. rosae was estimated to be 165.8 mm TL at an age of 1.8 years ( Figure  5a and Table 2). There was evidence of spawning occurring between September and November 2016. Natural mortality for L. rosae was estimated at 0.36 yr -1 , total annual mortality at 0.57 yr -1 for individuals aged three to eight years ( Figure  6a), and fishing mortality at 0.21 yr -1 . The exploitation for L. rosae was calculated to be 0.37.
The length-at-50%-sexual maturity for O. mossambicus, was estimated to be 166.5 mm TL at an age of 2.3 years (Figure 5b and Table 2). There was evidence of spawning  occurring between September and November and between January and March. Natural mortality was estimated at 0.28 yr -1 , total mortality at 0.78 yr -1 for ages three to nine years (Figure 6b), and fishing mortality at 0.50 yr -1 . The exploitation for O. mossambicus was calculated to be 0.64.

Catch-per-unit-effort
The mean CPUE over the study for all species was 9.0 kg 100-m net hr -1 . For L. rosae, the mean CPUE over the study was 5.5 kg 100-m net -1 hr -1 and was significantly different between surveys (Kruskal-Wallis p = 0.002); Figure 7a

Discussion
The water levels observed for Flag Boshielo Dam were below average for most of the study. The impoundment level dropped from 100% to 17% of capacity over two years due to an extended period of low and, subsequently, no inflow from both the Olifants and Elands rivers; see Marr et al. (2022). This severed connectivity between the impoundment and its rivers and resulted in the gauging weirs on the Olifants and Elands rivers upstream of the dam becoming impassable for fish migrations to spawning sites. Migrations between reaches are critical to the characin Micralestes acutidens (Fowler, 1934) and native cyprinids present in Flag Boshielo Dam (Bok et al. 2007); e.g. Engraulicypris brevianalis (Boulenger, 1908), Labeo spp., Labeobarbus spp., and most Enteromius spp. Migration for cichlids, e.g. O. mossambicus, is not considered critical within their life cycles (Bok et al. 2007); however, recruitment is flood dependent (Weyl and Hecht 1998). The receding water level resulted in the constant shifting of the littoral zone into areas that had not been exposed for long periods. At a capacity of 40%, the water level is about 6 m below the spillway level, but at 17% capacity, the water level is 11 m below the spillway level.   The constant receding littoral zone is likely to reduce the aquatic macroinvertebrates to more r-selected taxa, such as chironomids, and moderately fast moving taxa such as aquatic hemiptera and swiming coleoptera (Carmignani and Roy 2017). Slower moving species such as crawling Coleoptera, e.g. Elmidae, and burrowing species such as clams, e.g. Corbiculidae and Sphaeriidae, and burrowing mayflies, e.g. Caenidae, are likely to be reduced in abundance (Carmignani and Roy 2017 The receding water level left a bare zone between the high mark and current water levels because these rooted plants did not colonise the exposed substrate. Furey et al. (2004) found fundamental changes in the biochemical characteristics of exposed littoral sediments, including loss of fine sediments, nutrients and organic matter, with the organic matter becoming more allochthonous in origin. The impact of receding water levels on benthic algae has not been sufficiently studied, but the few studies published show mixed responses (Carmignani and Roy 2017). Receding water levels are advantageous to fish with pelagic and opportunistic diets and habitat generalists, but detrimental to invertivores (Carmignani and Roy 2017). It is unclear how receding water levels would impact detritivorous fish such as the Labeo spp. It is clear that there is considerable scope for research on the impact of receding water levels on lakes and impoundments in South Africa. Further, the declining water volume of the impoundment increased densities of fish and crocodiles within the impoundment, increasing fish susceptibility to predators (e.g. crocodiles and birds) and human exploitation. Job losses in the agricultural sector during periods of drought leads to an increase in subsistence fishing . Through the study, an increase in the number of gill nets set by informal fishers in the impoundment was noted (JRS pers. obs.) indicating that groups, other than local subsistence fishers, may be utilising the resource. The impacts of the aforementioned factors on the estimation of biological parameters for L. rosae and O. mossambicus are, however, difficult to quantify and beyond the scope of this work.
The number of species reported in this study were lower than those reported by Brits (2006) and Sara et al. (2017b). However, the catch composition was still dominated by L. rosae and O. mossambicus, confirming the findings of Sara et al. (2017b) that these two species held the greatest fishery potential at Flag Boshielo Dam. The results are similar to those of Sara et al. (2017b); however, L. rosae had increased from 37% to 59% of the catch and O. mossambicus from 16% to 21% of the catch. It should be noted that Sara et al. (2017b) conducted a biodiversity assessment, whereas, in this study, the two species were specifically targeted in order to collect biological data.

Oreochromis mossambicus
Oreochromis mossambicus is a maternal mouth brooding cichlid, endemic to eastern southern Africa (Skelton 2001). This species has been widely used in aquaculture and is targeted in commercial and subsistence fisheries (Russell et al. 2012). Key biological characteristics include a broad ecological tolerance, generalist diet, rapid reproduction with maternal parental care, and an aggressive behaviour, allowing it to successfully compete with other fish species (Russell et al. 2012). The biology of O. mossambicus has been extensively studied because of its importance as an aquaculture species (see Froese and Pauly 2022b). Oreochromis mossambicus was the original tilapia aquaculture species, but fell out of favour and was replaced by Oreochromis niloticus (Linnaeus, 1758) and Oreochromis sp. hybrids (Fitzsimmons et al. 2011;Russell et al. 2012). The flesh of Oreochromis spp. remains highly desirable to consumers because the flesh is white, firm, mild tasting and has no intermuscular bones  (Perschbacher 2014). Oreochromis mossambicus thrives in standing waters and breeds in areas with sandy substrate (Skelton 2001), and lake level does not appear to affect spawning periodicity (Weyl and Hecht 1998). The species grows rapidly, attaining 400 mm in length, and can breed within their first year (Skelton 2001). It is listed as "Vulnerable" on the IUCN red list (Bills 2020) mostly due to the loss of population integrity through hybridisation with the introduced O. niloticus (Zengeya et al. 2013;Marr et al. 2017a;Bills 2020).
Oreochromis mossambicus displayed negative allometric growth (b = 2.59), which was at the lower limit of the 2.5 to 3.5 range expected for this parameter (Ogle 2018). The value is also considerably lower than the 3.13 determined in 2009 at Flag Boshielo Dam (Addo-Bediako et al. 2014a). At the commencement of the study, the b value was 2.69 and decreased over the study. Although values of 2.4 to 2.6 have been reported for this species from reservoirs in Sri Lanka (Froese and Pauly 2022b), values closer to 3.0 have been reported from southern Africa (de Moor et al. 1986;Harrison 2001;Booth and Khumalo 2010). Growth in O. mossambicus is highly plastic (Lorenzen 2000), responding quickly to changing conditions throughout life depending on the quality and quantity of food (Marshall 2011), the size and stability of the environment (James and Bruton 1992;Brummett 1995), and the density of fish (James and Bruton 1992;Lorenzen and Enberg 2002). The low growth values may indicate that the O. mossambicus population was stressed by the receding water levels, the high fish densities, or that the quality and quantity of the food was sub-optimal for the species. However, these hypotheses were not evaluated in this study.
The asymptotic length estimated for O. mossambicus is consistent with the 400 mm SL recorded in Skelton (2001) indicating that the VBGF parameters may be a good representation of the growth of the species. Estimates of the asymptotic length for South African populations of this species listed in FishBase (Froese and Pauly 2022b) varied from 214 mm in Swaziland (now Eswatini; Booth and Khumalo 2010) to 410 mm listed for South Africa and Lake Kariba (Lévêque 1997). Stunting of populations is known for this species (Marshall 2011) and may account for the numerous studies listed in FishBase that report asymptotic lengths considerably lower than the maximum length reported by Skelton (2001). Oreochromis mossambicus populations have demonstrated alternative life-history patterns, with some populations in an area being altricial and long-lived with large asymptotic lengths while other populations are precocial and short-lived with smaller asymptotic lengths (James and Bruton 1992). The population in Flag Boshielo Dam appears to be large, altricial and long-lived, while the populations on the coastal plains appear to be smaller, precocial with shorted life spans, e.g. Lake Sibaya in northern KwaZulu-Natal (Bruton and Allanson 1974) and Lake Chicamba in Mozambique (Weyl and Hecht 1998). Hecht and Zway (1984) found a maximum size of 100 mm for O. mossambicus from an isolated hot spring in the Kruger National Park, where females as small as 35 mm had developing ovaries. The population in this hot spring are an example of the precocial extreme of O. mossambicus.
The length-at-50%-sexual maturity was found to be 166.5 mm TL at an age of 2.3 years. The smallest female O. mossambicus from Flag Boshielo Dam found with developing ovaries was 70 mm TL. In Lake Chicamba, O. mossambicus had a length-at-50%-sexual maturity of 223 mm TL (Weyl and Hecht 1998), an indication that the Flag Boshielo Dam population reaches maturity at a smaller size. However, the warmer temperatures in Mozambique would result in the fish growing faster with a lower asymptotic length than those in Flag Boshielo Dam, possibly indicating a more precocial life history. The spawning season for O. mossambicus in Flag Boshielo Dam was from August to March, with spawning peaks in November and February. This is consistent with observations from Lake Chicamba, where the reproductive activity was confined to the summer period from September to May (Weyl and Hecht 1998).
Marginal analysis of otoliths showed predominantly opaque margins, indicative of periods of suppressed growth during winter and spawning (Campana and Thorrold 2001), between April and August when water temperatures were cooler, with the latter period coinciding with the onset of rainfall, warmer temperatures, and the commencement of spawning. Oreochromis mossambicus from Nwanedi-Luphephe dams in South Africa deposit two opaque zones annually, one formed during winter and the other during the peak reproductive period (Hecht 1980). However, Lake Chicamba fish deposit a single opaque band because the low winter temperatures and the onset of the spawning coincide (Weyl and Hecht 1998). Mouth-brooding cichlids become aggressive and do not feed while brooding (Oliveira and Almada 1998), which may correspond to the deposition of an opaque ring during parental care (Hecht 1980). Mortality rates are determined by a combination of environmental factors (e.g. temperature, food availability, presence of predators), and factors rooted in the behaviour, physiology and genetics of the individual organism (e.g. choice of habitat, predator avoidance behaviour) (Lorenzen 2000). The estimated natural mortality (0.28 yr -1 ) was considerably lower than the total mortality of 0.78 yr -1 , resulting in an estimated fishing mortality of 0.50 yr -1 with an estimated utilisation of 0.64. The natural and total mortality are considerably lower than those found by De Silva (1991) for an impoundment in Sri Lanka (natural mortality 1.01 yr -1 and total mortality 2.84 yr -1 ) and Wakwabi and Marine (1988) for the Lower Tana River in Kenya (natural mortality 1.53 yr -1 and total mortality 2.98 yr -1 ), while Bhanbhro et al. (2017) found lower values for total mortality in the Indus River in Pakistan (natural mortality 1.07 yr -1 and total mortality 1.35 yr -1 ), all calculated using the same techniques used in the current study. However, Khumalo (2006) reported a total mortality of 0.49 yr -1 for an unexploited population of O. mossambicus in Swaziland (now Eswatini). The presence of juveniles and adults in the catches, and the lack of significant differences in catch efforts over the study, suggest that the O. mossambicus population of Flag Boshielo Dam may be self-sustaining.
Labeo rosae Labeo rosae is endemic to the Limpopo, Crocodile and Phongolo River systems of eastern southern Africa. This species is similarly sized to O. mossambicus, attaining 410 mm in length (Skelton 2001). Labeo rosae prefer sandy substrates of large perennial and seasonal rivers and undertake upstream spawning migrations in summer when the rivers are swollen (Pienaar 1978;Skelton 2001;Marshall 2011). Sexual maturity begins at a length of about 150 mm and both mature and immature fish take part in the spawning migration (Pienaar 1978;Skelton 2001). The flesh of L. rosae quickly becomes rancid if the fish is not gutted soon after landing (Sara et al. 2021a) and is thus not preferred and targeted less intensely by subsistence fishers (Jooste et al. 2013). No data on the growth rates, age and length at maturity, and mortality, important in determining the sustainable harvest rates of fisheries, are available for L. rosae (see Froese and Pauly 2022a). Labeo rosae is listed as "Least Concern" on the IUCN red list (Bills et al. 2017).
Labeo rosae in Flag Boshielo Dam had a bimodal population dominated by adult specimens, which may be an artefact of the low impoundment levels and/or the gears used. Labeo rosae catches peaked at the dam inflow during late August 2016 indicating a spawning aggregation, which was confirmed by the majority of females collected being 'ripe' at the time and having opaque otolith margins. However, the lack of connectivity between the impoundment and its rivers through the study eliminated opportunities to spawn in the river, potentially resulting in putative recruitment failure for the 2015-2016 and 2016-2017 spawning season; which may explain the absence of specimens smaller than 170 mm in the catches.
The positive allometric growth for L. rosae in this study was lower than the 3.10 and the 3.19 determined for Flag Boshielo Dam in 2009 and 2013, respectively (Jooste et al. 2013;Lebepe et al. 2016). These values are similar to those reported for the closely related, and similarly sized, Labeo altivelis Peters, 1852 in Lake Kariba, Zimbabwe (Marshall 2011). At the commencement of the study, the b value for L. rosae was 2.86 in February 2016, but, contrary to that of O. mossambicus, increased through the study. This low value at the commencement of the study may indicate that the receding water levels in the months prior to the study resulted in sub-optimal conditions for L. rosae.
The asymptotic length estimated for L. rosae was considerably consistent with the 410 mm SL recorded in Skelton (2001) when the age-at-zero-length was set to 0 in the VBGF. The lack of specimens of 1 year age in the sample resulted in an overestimation of the asymptotic length when the age-at-zero-length was included in the VBGF. Although the length-at-50%-sexual maturity, and associated age, both appear to be high, the length-at-50%sexual maturity is similar to that recorded for L. altivelis in Lake Kariba (Kolding et al. 1992).
The estimated natural mortality (0.36 yr -1 ) was higher than that estimated for O. mossambicus. The estimated total mortality (0.57 yr -1 ) is similar to the 0.6 yr -1 found for L. altivelis in Lake Kariba (Kolding et al. 1992). Due to the putative low estimate of the natural mortality, fishing mortality is believed to be overestimated. The utilisation was estimated to be 0.37, which is lower than that estimated for O. mossambicus. This is consistent with the preference of the fishers for O. mossambicus, which is expected to be targeted more intensely.

Conclusion
The study confirmed that L. rosae and O. mossambicus were the most abundant species in Flag Boshielo Dam, both exceeded 20% of the catch, and were thus potential fishery species (Sara et al. 2017b). The study also showed that both species are currently being exploited. Even though the study was conducted during a period of exceptionally low impoundment levels, the biological data collected will allow for the estimation of the fishery parameters for Flag Boshielo Dam.
The biology and life history of most Labeo species in southern Africa is poorly studied. Indeed, studies on the biology and life history of most of southern Africa's freshwater fishes are lacking, with the exceptions of C. gariepinus, O. mossambicus, Labeobarbus spp., Orange River Labeo spp., and Hydrocynus vittans Castelnau, 1861. This study is the first on the life history of L. rosae, for which only two length-weight relationships (Jooste et al. 2014;Lebepe et al. 2016) and anecdotal data were available (Pienaar 1978;Skelton 2001;Marshall 2011). If the establishment of inland fisheries is to be achieved as major priority for South Africa, the biological and life history parameters, including behavioural and migration data, for species with fishery potential, and the species likely to be by catch, should be compiled regionally to ensure the sustainable management of the country's inland fishery stocks.