Received: 10 August 2018 Accepted: 14 November 2018 DOI: 10.1111/jfb.13866 FISH REGULAR PAPER

Variation in thermal tolerances of native freshwater fishes in South 's Cape Fold Ecoregion: examining the east–west gradient in ' sensitivity to climate warming

Jody-Lee Reizenberg1 | Lesley E. Bloy2,3 | Olaf L. F. Weyl3,4 | Jeremy M. Shelton4,5 | Helen F. Dallas5,6

1Department of Biological Science, University of , Rondebosch, The Cape Fold Ecoregion (CFE) is a biodiversity hotspot with high levels of in its 2Department of Ichthyology and Fisheries freshwater fish fauna. This study examined inter and intra-specific variation in critical thermal

Science, Rhodes University, Grahamstown, maxima (TCmax) for eight native species of freshwater fish from the CFE. Cape Galaxias South Africa zebratus, Breede River redfin burchelli, redfin Pseudobarbus burgi, Clan- 3 Center for Invasion Biology, South African william redfin Pseudobarbus calidus and fiery redfin Pseudobarbus phlegethon were the most ther- Institute for Aquatic Biodiversity (SAIAB), –  Grahamstown, South Africa mally sensitive (TCmax = 29.8 32.8 C). Clanwilliam rock-catfish Austroglanis gilli, Eastern Cape 4DST/NRF Research Chair in Inland Fisheries redfin Pseudobarbus afer and Cape kurper Sandelia capensis were moderately sensitive  and Freshwater Ecology, South African (TCmax = 33.0–36.8 C). An increase in intra-specific thermal sensitivity of S. capensis was Institute for Aquatic Biodiversity (SAIAB), observed from east to west. The results were related to in situ water temperature, which influ- Grahamstown, South Africa enced TCmax for all species, suggesting that thermal history is a major driver of variation in ther- 5Freshwater Research Centre, Scarborough, mal tolerance amongst populations. These thermal tolerance data for freshwater fishes in the South Africa CFE demonstrate that resilience to climate warming follows a geographical cline and that the 6Nelson Mandela University, Port Elizabeth, South Africa more sensitive western species and regions are conservation priorities. Correspondence Jody-Lee Reizenberg, Department of KEYWORDS Biological Sciences, University of Cape Town, Anabantidae, Austroglanidae, Cape Fold Ecoregion, , , thermal limits Rondebosch 7701, South Africa. Email: [email protected] Funding information The authors gratefully acknowledge the Water Research Commission (WRC) of South Africa (Project K5/2337) for funding this research, which was awarded to the Freshwater Research Centre. The National Research Foundation (NRF) South Africa and the University of Cape Town are acknowledged for student funding.

1 | INTRODUCTION annual average precipitation and the most observable hydrological alterations (Filipe et al., 2013). Non-native species introductions and

Globally, the rise in atmospheric CO2 and temperature has been anthropogenic disturbance amplify the effects of these changes on recognized as one of the greatest threats to biodiversity Med-region freshwater ecosystems (Filipe et al., 2013). Consequently, (Intergovernmental Panel on Climate Change (IPCC), 2014), threaten- freshwater biodiversity has undergone acute bottlenecks resulting in ing the functioning of ecosystems at all levels of organization community structure changes and shifts in the distribution of many (Woodward et al., 2010). While climate warming is a universal phe- species (Carpenter et al., 1992; Dudgeon et al., 2006; Sunday et al., nomenon, the effects are disproportionately manifest across climatic 2012). Given their sensitivity to thermal alteration under projected regions (Woodward et al., 2010). Within the past decade, Mediterra- climate scenarios, the adaptability, vulnerability and distribution of nean climate regions (Med-regions) have experienced the most con- freshwater fauna has emerged as a primary research focus in Med- siderable increases in annual average air temperature, decreases in regions the world over.

J Fish Biol. 2019;94:103–112. wileyonlinelibrary.com/journal/jfb © 2018 The Fisheries Society of the British Isles 103 104 FISH REIZENBERG ET AL.

The Cape Fold Ecoregion (CFE) spans 87,900 km2 across the temperature at which physiological signs of heat stress are observed in south-eastern and south-western coasts of South Africa. It is a global ectothermic vertebrates and invertebrates (Becker & Genoway, 1979). biodiversity hotspot (Myers et al., 2000) and a priority biogeographic Physiologically, TCmax is the point at which behavioural thermoregula- unit for freshwater conservation (Abell et al., 2008; Nel et al., 2011). tion is impeded and the tissues of the organism begin to respond nega- The CFE is rich in headwater streams that provide the last remaining tively (Paladino et al., 1980). The behavioural manifestations of refuges for many small-bodied endemic fishes (Ellender & Weyl, physiological heat stress are generally consistent amongst fishes. Visible 2014). Since the formation of the Cape Fold Belt about 290 million symptoms (or biomarkers) of heat stress include the loss of righting years ago, freshwater fishes in the CFE have remained largely isolated, response (LRR) followed by the onset of muscular spasms (OS; Beitin- resulting in the high regional diversity and endemism of the freshwa- ger, 1990; Lutterschmidt & Hutchison, 1997). ter fish fauna. This fauna comprises 42 recognised taxa belonging to The widely-used protocol for determining TCmax is the critical the families Anabantidae, Austroglanidae, Cyprinidae and Galaxiidae thermal method (CTM). In the laboratory, the CTM is conducted over (Ellender et al., 2017). The majority of these endemic fishes (60%) are relatively short periods of time (1–2 h) and the thermal stress is short- on the IUCN Red List as either Endangered or Critically Endangered lived (given that the is removed from the experiment as soon (Ellender et al., 2017). With reference to climate change and freshwa- as visible symptoms of heat stress are observed); prolonged exposure ter ecosystem vulnerability, Dallas and Rivers-Moore (2014) recog- to TCmax in the wild however is lethal (Coutant & Brook, 1970; Eliason nised the CFE not only as a biodiversity hotspot, but a ‘hotspot for et al., 2011). Since TCmax is an indication of sensitivity to warming, it is concern’. Habitat degradation and invasion by alien fishes are the cur- useful for identifying thermally sensitive species under predicted cli- rent greatest threat to native fishes in the region (Ellender et al., mate change scenarios (Dallas & Ross-Gillespie, 2015; Somero, 2010). 2017), but the threat of climate warming may further affect these In the past, the CTM has been criticised for its ability to induce fishes adversely in the future (Shelton et al., 2018). stress in the test subjects, but the method produces data that are criti- The climate of the south-western CFE is classified as Mediterra- cal to understanding biological limits of organisms (Lutterschmidt & nean, with hot dry summers and cool wet winters (driven primarily by Hutchison, 1997). The CTM is appropriate for use with endangered winter frontal systems; Chase & Meadows, 2007; Gasith & Resh, species as alternative methods, such as the incipient lethal tempera- 1999; Reason & Rouault, 2005). Towards the south-east of the region, ture (ILT), requires death of the test individual (Becker & Genoway, rainfall is less predictable (following a year-round stochastic pattern; 1979; Beitinger et al., 2000; Dallas & Ketley, 2011) and are thus not Ellender & Weyl, 2015; Fauchereau et al., 2003). Global climate recommended for (Beitinger et al., 2000). Further- change models predict a summer maximum air temperature increase more, CTM allows for in situ experimental design, allowing for test of 2–4C and 4–6C in the eastern and western portions of the CFE organisms to be returned to the site of collection post-experiment, respectively, with an expected decrease and greater variability in rain- which is beneficial when working with endangered species. Alterna- fall respectively (Dallas & Rivers-Moore, 2014; Schulze, 2011). In tive methods, such as utilising stream temperature data to infer ther- mal limits of fish, have been used in the past, but field-based methods response to elevated air temperatures and decreased flows, stream do not account for individual responses and the ability of fish to miti- water temperatures are expected to increase throughout the CFE. In gate stressful conditions (Eaton et al., 1995). Furthermore, field data the eastern CFE where freshwater fishes, are and historically have are confounded by biotic interactions, whereas laboratory experi- been, exposed to greater variability in thermal habitat (Ellender & ments facilitate controlled exposure and the assessment of biological Weyl, 2015), this may translate into greater resilience with respect to responses that are more accurate (Todd et al., 2008). climate warming (when compared with the western conspecifics). At present, the lack of physiological information on native fishes in Temperature has been described as a master abiotic variable in the CFE is considered a major bottleneck in understanding their biol- aquatic ecosystems in that it has a particularly strong influence on fit- ogy, ecology, distributions and behaviour (Ellender et al., 2017), and ness, behaviour and life-histories of aquatic biota (Caissie, 2006; Dal- constrains the understanding of the potential consequences of climate las, 2008). Fitness, in the context of climate warming, refers to the change for their survival. For this reason, the current study examined physiological and behavioural performance of fishes at elevated tem- the T of eight CFE endemic freshwater fish species. Given the peratures and this differs among freshwater biota (Somero, 2010). Cmax potential for variation in thermal tolerance (as a result of differences in Species that are more sensitive to thermal alteration are more likely to the thermal regimes in which a species or population has evolved), and be adversely affected by warming. In the northern hemisphere, global because of finer-scale thermal variation among habitats, it was warming has resulted in a general decline in fitness across freshwater hypothesised that T would differ significantly between species and taxa (Heino et al., 2009). In the southern hemisphere, and in Cmax that intra-specific and inter-specific thermal tolerance would differ sig- South Africa particularly, data concerning the thermal sensitivity of nificantly between the eastern and western portions of the CFE. freshwater fishes remain largely unexplored (see Ellender et al., 2017). Since climate projections suggest differential effects from the east to west of the CFE, understanding the thermal sensitivity of native fish 2 | MATERIALS AND METHODS species throughout this region will assist in evaluating the potential ecological consequences of climate warming on freshwater fishes 2.1 | Study area across this environmental gradient.

The critical thermal maximum (TCmax), which is used as a non-lethal The CFE lies in the extreme south-western and south-eastern regions measure of thermal sensitivity, is a parameter that indicates the upper of South Africa (33–35 S and 18–22 E; Cowling et al., 2003). Five REIZENBERG ET AL. FISH 105 rivers from the CFE were selected for this study: four mountain region and experience high flows during winter (Dallas & Rivers- streams in the and one in the Eastern Cape Province Moore, 2014). In contrast, rainfalls and, consequently flow in the (Figure 1). Fernkloof, on the eastern edge of the CFE, is episodic (Ellender & The Amandel River (33.5044 S, 19.4879 E and 33.5038 S, Weyl, 2015). These rivers therefore represent different environmental 19.4866 E) is a tributary within the Breede River catchment of the conditions and hydrological patterns across the geographical bounds Western Cape. The elevation of the collection sites fell between of the CFE, which translates into varying acclimatization conditions 444 and 448 m asl. In-stream habitat consisted predominantly of (including temperature) that are likely to be reflected in the tolerance alternating pool–riffle sequences generally < 1 m deep over small and limits of freshwater fauna (Caissie, 2006; Webb et al., 2008). large cobble with c. 20% canopy cover and dense, native riparian veg-  etation. The collection site on the Upper Berg River (33.9553 S, 2.2 | Indicators of thermal alteration 19.0733 E) was situated at an elevation of 268 m asl and in-stream habitat comprised predominantly of pools, runs and riffles generally Using 1 year of hourly water temperature data for each site, annual < 1 m deep over cobble substrates, with no canopy cover and sparse, descriptive statistics (temperature metrics) were calculated using the native, riparian vegetation at the site. Both the Driehoeks and Ronde- indicators of thermal alteration (ITA) approach (Rivers-Moore et al., gat Rivers are situated in the Olifants–Doring River catchment of the 2012, 2013). The data for computing these metrics were recorded Western Cape. The elevation of the Driehoeks River collection sites using Hobo UTB1-001 TidBit V2 loggers (Onset Computer Corpora- (32.4346 S, 19.1808 E; 32.4301 S, 19.1755  E; 32.4314 S, tion; www.onset.com) installed at the collection sites for at least 19.1484  E) fell between 903–909 m asl and the river had dense 1 year. Sub-daily water temperature data were converted to daily data riparian and in-stream native riparian vegetation. In-stream habitat (mean, minimum, maximum and range). From these daily data, thermal comprised long, shallow (< 1 m deep) pools alternating with stony rif- metrics were calculated to characterize and compare thermal signa- fles. The Rondegat River sites (32.3630 S, 19.0449 E and tures of the five rivers with respect to magnitude and duration of 32.3757 S, 19.0656 E) were situated at an elevation of 452–520 m thermal events. The derivation of thermal metrics also allowed for the asl and comprised cobble-bottomed pools and runs < 1 m deep with correlation of the fishes' thermal histories to thermal tolerance. open canopy. The Fernkloof River site (33.7180 S, 25.2895 E) was Annual descriptive statistics included Colwell's predictability index situated at 91 m asl. The Fernkloof Stream is an episodic headwater (ICP), which is used to detect periodicity in time series data such as stream characterised by isolated surface pools which are fed by water temperature (values closer one represent greater predictability) groundwater (Ellender & Weyl, 2015). The substrate is characterised (Colwell, 1974; Rivers-Moore et al., 2013). To calculate ICP, daily water   by bedrock, large boulders and some pebbles. The Fernkloof has a temperatures are assigned to 2 C classes between 2 and 36 C and predominantly closed canopy with an abundance of marginal the frequency of temperatures in the classes are then used as a basis vegetation. for calculating ICP (Rivers-Moore et al., 2013). The annual coefficient The rivers are known to differ in their thermal signatures based of variation metric was calculated for all the rivers to illustrate annual on water temperature logged over a 1 year period in each river (this temperature variability (Van Etten, 2009). Annual temperature metrics study). The broader prevailing hydrological conditions in each region also included mean annual temperature, SD of mean annual tempera- have been describe by Chase and Meadows (2007), Dallas and Rivers- ture, mean of daily range, mean of annual minima, mean of annual  Moore (2014) and Ellender and Weyl (2015). The Amandel, Berg, maxima and degree days [ d, calculated by subtracting the threshold Driehoeks and Rondegat Rivers are located within a winter rainfall temperature (10C in this case) from the daily mean temperature with

Rondegat River Berg River 25° E C

D I South Africa Driehoeks River G 30° B Amandel River

A F Eestern N Cape H Western Cape

18° 24° 30°

Fernkloof River

K

Cape Fold Ecoregion J

FIGURE 1 Map showing location of sampling sites within the Cape Fold Ecoregion for: (a) Galaxias zebratus; (b) Pseudobarbus phlegethon; (c) Pseudobarbus burgi; (d) Sedercypris calidus; (e) Pseudobarbus phlegethon;(f)Pseudobarbus burchelli; (g) Austroglanis gilli; (h) Sandelia capensis; (i) Sandelia capensis; (j) Sandelia capensis; (k) Pseudobarbus afer (Fish images courtesy of SAIAB) 106 FISH REIZENBERG ET AL. the total degree days calculated by summing the differences for the variation in fitness, as recommended by Becker & Genoway (1979). The year; Dallas et al., 2017]. The number of degree days indicates the buckets containing river water were then transported to the field labo- probability of stress to aquatic organisms; a greater number of degree ratory (transportation time varied from 30 min to 1 h depending on the days reflects greater levels of stress (Rivers-Moore et al., 2012, 2013). distance from the field laboratory). All experiments were conducted at a From annual data, group one (monthly statistics) were calculated field laboratory near the collection site using water from the river sys- using the ITA method (Rivers-Moore et al., 2013). Group 1 statistics tem from which fish were sampled. include monthly magnitudes in temperature, namely mean, maximum Following collection, fish were retained in aerated buckets in a and minimum. Group 2 statistics (describing the magnitude and duration dark room at the field laboratory (set up within 17 km of the collec- of annual extreme water temperature conditions) include mean 7 day tion sites) for c. 24 h prior to commencement of the CTM experi- moving averages in temperature, maximum 7 day moving average of ment. This allowed fish to recover from any stress incurred during mean temperatures over a 1 year period, as well as maximum and mini- collection and handling. Waterproof Hobo UTB1-001 TidBit V2 log- mum moving averages allowing the determination of the most extreme gers were used to record hourly water temperature in the aerated periods (7, 30 or 90 days) over an annual cycle. These chronic stress bucket during this period. The temperatures recorded in the holding thresholds are based on an averaging statistic account for both magni- buckets were all between the rivers' daily mean and daily maximum tude and duration of exposure (Sullivan et al., 2000; Nelitz et al., 2007). temperatures recorded during the week when experiments were To derive metrics for the week prior to experiments, mean, maxi- conducted. Mean and maximum temperatures respectively for the  mum and minimum water temperature (C) were calculated from log- week preceding experiments were: 21.5 and 26.5 C(Amandel   ger data for the 7 days prior to the last catch. River), 22.1 and 28.3 C (Berg River), 18.3 and 21.9 C(Rondegat River), 17.3 and 23.8C (Driehoeks River) and 23.7 and 25.1C (Fernkloof River). 2.3 | Experimental protocol After the 24 h holding period, the CTMs commenced. Three to Eight species of fishes were selected for the study: Austroglanis gilli six fish were placed in a 5 mm mesh basket submerged in a water (Barnard 1943), Galaxias zebratus (Castelnau 1861), Pseudobarbus afer bath that was fitted with a circulating heater (Julabo; www.julabo. (Peters 1864), Pseudobarbus burchelli (Smith 1841), Pseudobarbus burgi com). For smaller (< 11 cm total length; LT) or less active species, a (Boulenger 1911), Pseudobarbus phlegethon (Barnard 1938), Sandelia maximum of six adult fish were selected for one trial, whereas for capensis (Cuvier 1829) and Sedercypris calidus (Barnard 1938). These larger (> 11 cm total length) and more active species, up to three species were chosen as representatives of the four families of native adult fish were used per trial. After a 30 min control period in which fishes known to be distributed within the CFE. temperature in the water bath equalised with the holding tempera- Fieldwork and experiments were undertaken during summer low- ture, water temperature was gradually increased at the standard lin- flow conditions November–December in 2015 (western CFE: Western ear heating rate for thermal tolerance studies of 0.3Cmin−1 using Cape) and in February 2017 (eastern CFE: Eastern Cape). Individuals the Julabo circulating heater (Perez et al., 2003). When an individual were collected using 4 mm mesh fyke nets (set overnight on the night showed signs of thermal stress (i.e., the biomarker; Supporting Infor- preceding the experiments and emptied between 0800 and 1000 h mation Table S1), it was removed from the water bath and placed in daily; Western Cape) or by seine netting (Eastern Cape). Both methods an aerated recovery chamber at the holding temperature. The end- are passive capture techniques for fishes (Hubert et al., 2012). Adult fish point of the experiment for an individual was the observed loss of of similar size and fitness (Table 1) were collected and transferred to righting response, where the individual lost complete ability to aerated buckets using dip nets to minimize handling. Fish selected for remain upright (Beitinger & Lutterschmidt, 2011; Lutterschmidt &  the thermal tolerance experiments were uniformly sized to minimize Hutchison, 1997). LRR constituted the TCmax ( C) which was

TABLE 1 Thermal tolerances (TCmax) of eight native freshwater fish species in the Cape Fold Ecoregion

 a  River Species TCmax ( C) Sample size (n) LT (mm) TH ( C) Rondegat Austroglanis gilli 33.0 16 (6) 12.0 20.7 Driehoeks Galaxias zebratus 29.8 32 (9) 4.9 19.3 Fernkloof Pseudobarbus afer 35.1 36 (6) 5.6b 19.9 Amandel Pseudobarbus burchelli 32.8 32 (6) 8.3 23.5 Berg Pseudobarbus burgi 32.1 41 (6) 8.0 24.3 Driehoeks Pseudobarbus phlegethon 30.3 27 (8) 6.9 18.0 Rondegat Pseudobarbus phlegethon 32.6 26 (8) 6.2 20.7 Amandel Sandelia capensis 34.8 30 (5) 8.2 24.4 Berg Sandelia capensis 35.3 39 (8) 8.5 24.5 Fernkloof Sandelia capensis 36.8 36 (6) 6.5b 19.9 Rondegat Sedercypris calidus 32.4 23 (8) 7.4 20.9

TCmax: critical thermal maxima; LT: total length; TH: holding temperature. a Values in parentheses denote the number of trials for the species. b Fork length. REIZENBERG ET AL. FISH 107 recorded for each fish. Each trial lasted approximately 90 min during days, which reflect the cumulative temperature experienced by an which the water was continuously aerated by the Julabo circulating organism above a threshold temperature was highest for Fernkloof pump and an air stone to ensure that the dissolved oxygen levels (2,914 d) and Rondegat Rivers (2,900 d) and lowest for the Drie- remained at c. 70% throughout the experiment (Dallas et al., 2015). hoeks River (1863 d). The Amandel, Berg and Rondegat Rivers had Following recovery, each fish was measured, either as total length similar highest recorded 7 day moving averages of daily mean (24.6,  (Western Cape: TL)orforklength(LF; Eastern Cape). After each 24.2 and 24.0 C, respectively) compared with the lower Driehoeks experiment, individuals were kept for a 24 h recovery period and (23.7C) and higher Fernkloof Rivers (26.2C). The 7 day moving aver- then returned to their collection site. ages of daily maximum was highest in the Amandel and Berg Rivers  Trials were repeated until TCmax had been determined for all the (31.6 and 29.6 C, respectively) and lowest in the Driehoeks individuals collected (Table 1). The number of trials conducted for River (24.8C). each species were A. gilli (6), G. zebratus (9), P. afer (6), P. burchelli (6), Water temperature data for the month preceding experiments P. burgi (6), P. phlegethon Driehoeks (8), P. phlegethon Rondegat (8), (Table 3) and the week of thermal experiments (Table 4) showed that S. capensis Amandel (5), S. capensis Berg (8), S. capensis Fernkloof the Fernkloof River had the highest weekly and monthly mean tempera-  (6) and S. calidus (8). The data were tested for normality using Shapiro- tures of 23.7 and 23.9 C, respectively, while the Amandel and Berg Riv-  Wilk tests and were found to be non-parametric. A Kruskal-Wallis ers had the highest weekly maximum temperatures of 26.5 and 28.3 C, one-way ANOVA was used to compare median TCmax between spe- respectively. The Driehoeks and Rondegat Rivers had the lowest weekly  cies and within species. Pairwise comparisons were then carried out and monthly mean temperatures of 17.3 to 18.3 C, respectively. Seven- using the Nemenyi-test with χ2 (ties-corrected) approximation for day moving averages of daily mean temperatures, calculated for the independent samples. month preceding the experiments, were highest in the Fernkloof (24.9C), while 7 day moving averages of daily maximum temperatures were highest in the Fernkloof (26.8C) and Berg Rivers (26.3C). 3 | RESULTS 3.2 | Inter and intra-specific variation 3.1 | Water temperature and indicators of thermal There was a significant difference in T amongstthespeciesatthe alteration Cmax 95% CI (Kruskal-Wallis χ2 = 233.25, df =10;P < 0.001). Of the eight Thermal metrics for 1 year (Table 2) showed that mean annual tem- species examined, G. zebratus was the most thermally sensitive with a  Æ  perature (Tann SD) was lowest in the Driehoeks River TCmax of 29.8 C(DriehoekRiver),whileS. capensis was most thermally    (15.1 Æ 5.1 C) and highest in the Fernkloof River (18.0 Æ 4.4 C). Var- tolerant species with a TCmax of 34.8, 35.3 and 36.8 C in the Amandel, iability was highest in the Driehoeks River and lowest in the Rondegat Berg and Fernkloof Rivers, respectively (Table 1); post hoc tests between

River. Colwell's predictability index (ICP) was similar for four of the five species are shown in Supporting Information Table S2. TCmax differed sig- rivers (range 0.56–0.67) but substantially lower for the Driehoeks nificantly among populations of S. capensis (Kruskal-Wallis χ2 = 54.6; df =

River (ICP = 0.37) suggesting this river is the least predictable. Degree 2; P < 0.001) and post hoc tests showed the Fernkloof population to be

TABLE 2 Thermal metrics for 1 year calculated from hourly temperature data for five Cape Fold Ecoregion rivers

River Amandel Berg Driehoeks Fernkloof Rondegat   Mean annual temperature (Tann C) 17.4 16.6 15.1 18.0 17.9   Standard deviation of Tann ( C) 4.8 4.4 5.1 4.4 3.6 Annual coefficient of variability 27.6 26.4 33.6 24.3 20.1

Colwell predictability index (ICP) 0.67 0.56 0.37 0.61 0.64 Mean daily temperature range (C) 4.9 5.6 2.2 1.7 4.2 Mean annual temperature minima (C) 15.6 14.3 14 17.2 16.3 Mean annual temperature maxima (C) 20.5 19.9 16.2 18.9 20.5 Degree days 2712 2412 1863 2914 2900 Magnitude and duration of annual extreme Mean_7 24.6 24.2 23.7 26.2 24.0 water temperature conditions (Sustained Min_7 8.2 8.7 7.1 9.9 10.6 extreme conditions, either high or low) (C) Min_30+ 8.7 9.3 7.8 9.9 12.3 Min_90+ 9.9 10.1 8.0 9.9 12.8 Max_7 31.6 29.6 24.8 28.2 28.3 Max_30+ 29.7 27.8 24.0 25.8 27.5 Max_90+ 28.5 26.7 23.1 25.1 26.7

Mean_7: 7 day moving average of daily temperatures; Min_7: 7 day moving average of daily minimum; Max_7: 7 day moving average of daily maximum; Min_30+: 30 day moving average of daily minimum; Max_30+: 30 day moving average of daily maximum; Min_90+: 90 day moving average of daily mini- mum: Max_90+: 90 day moving average of daily maximum. 108 FISH REIZENBERG ET AL.

TABLE 3 Daily water temperatures for the month preceding experiments calculated from hourly temperature data for five Cape Fold Ecoregion rivers

Mean (C) Minimum (C) Maximum (C) Range (C) Min_7 (C) Max_7 (C) Amandel River 18.3 15.2 25.1 8.1 19.9 24.4 Berg River 19.1 12.9 27.8 10.5 21.5 26.3 Driehoeks River 17.4 12.5 22.4 4.3 20.1 21.9 Fernkloof River 23.9 22.9 25.3 2.4 24.9 26.8 Rondegat River 17.3 13.3 23.3 5.8 18.6 21.4

Min_7: 7 day moving average of daily maximum; Max_7: 7 day moving average of daily maximum.

TABLE 4 Daily water temperatures for the week of thermal TABLE 5 Inter and intra-specific variation in thermal tolerances of experiments calculated from hourly temperature data for five Cape eight species of native fishes in the Cape Fold Ecoregion Fold Ecoregion rivers Species Rivers P df KW     Mean ( C) Minimum ( C) Maximum ( C) Range ( C) Pseudobarbus spp. Amandel, Berg, Driehoeks, < 0.001 5 92.6 Amandel River 21.5 18.7 26.5 7.8 Fernkloof, Rondegat Berg River 22.1 16.0 28.3 10.5 Pseudobarbus phlegethon Driehoeks, Rondegat < 0.001 1 34.1 Driehoeks River 17.3 12.6 23.8 10.6 Sandelia capensis Amandel, Berg, Fernkloof < 0.001 2 54.6

Fernkloof River 23.7 22.7 25.1 2.4 KW: Kruskal-Wallis test statistic. Rondegat River 18.3 15.1 21.9 6.8 Examining first the effect of present habitat temperature on

TCmax, inter-specific variation may reflect the influence of different responsible for this result (Supporting Information Table S3). T also Cmax thermal signatures of, and thus acclimatization of fish (the adjustment differed significantly among species within the genus Pseudobarbus Smith of organisms to their current natural conditions) to their natural habi- 1841(Kruskal-Wallis χ2 = 92.6; df =5;P < 0.001; Table 5). Post hoc tests tat in the five rivers. Beitinger et al. (2000) reviewed thermal tolerance showed that P. afer was responsible for the significant difference data for 116 northern-hemisphere fish species and found acclimation (Supporting Information Table S3). Between two populations of temperature to have influenced TCmax for the large majority of species P. phlegethon, T also differed significantly (Kruskal-Wallis χ2 = 34.1, Cmax studied. In those studies, fish were exposed to dynamic or constant df =1;P < 0.001; Table 5). Overall, G. zebratus, P. burchelli, P. burgi, P. cali- temperatures in the laboratory over a set period of time (days, weeks, dus and P. phlegethon were most thermally sensitive (TCmax = 29.8–32.8 months). Thus, TCmax represented the adjustment of organisms to pre- C), while A. gilli, P. afer and S. capensis were only moderately thermally defined conditions over the predetermined time period. In the present –  sensitive (33.0 36.8 C; Figure 2). study fish were not acclimated, but ITAs showed distinct differences in thermal habitat, indicating an acclimatization effect. Mean and max- 4 | DISCUSSION imum summer temperatures, for the months preceding experiments, increase from west to east (Table 3). This is consistent with the

This paper presents the first research on the thermal sensitivity of observed increase in thermal tolerance of fishes from the western to CFE fishes and explores variation in thermal tolerance among and the eastern CFE. A second source of variation among species may be evolutionary within species across an east-west gradient of the CFE. The results history and thermal environment. The climate and hydrology of the indicate that fishes from the perennial streams in the western CFE are CFE differs markedly from east to west and it is within these two cli- more thermally sensitive than those from the eastern region (which matic regions of the CFE that genetic divergence of the native fresh- experiences episodic flows) and that the source of variation in TCmax water fish species have been occurring for the last 18,000 years may be a result of both the thermal environments in which these spe- (Chakona et al., 2013; Swartz et al., 2007, 2009). The Mediterranean cies evolved and their short-term thermal history (or acclimatization). climate in the western CFE was however only established 5,000 years ago (Meadows & Baxter, 1999), with the climate grading into a sum- 4.1 | Inter-specific variation in T cmax mer rainfall pattern towards the east. It is possible that this cline in Variation in thermal maxima may arise from several sources. These habitat also resulted in a thermal tolerance cline, with eastern species include adaptation of physiological mechanisms across geo-climatic having inherited a greater resilience to thermal alteration. Habitats clines, disparity in how local abiotic variables interact, acclimation occupied by P. afer typically experience large fluctuations in flow and effects, or even variation in sampling protocol (Lutterschmidt & temperature (Ellender & Weyl, 2015), which may have facilitated the Hutchison, 1997; Pörtner, 2002). In this study, the potential sources species' adaptation to greater temperature tolerance than the western of inter-specific variation in TCmax were identified as habitat (acclimati- Pseudobarbus species. Further research on population-level differ- zation effect), evolutionary history, and physiology. While variation in ences between tributaries with different flow and thermal regimes for thermal tolerance is not limited to these factors, our data are suitable this species could provide important additional insights into their to address the three sources mentioned. potential to adapt to warming. REIZENBERG ET AL. FISH 109

populations to major flooding events was related to their evolution in 38 river systems characterised by environmental stochasticity. Environ- ment rather than genetic similarity or relatedness is therefore likely to 36 be the primary driver of the observed differences in thermal tolerance amongst the Pseudobarbus species. 34 The third potential source of variation in TCmax may be attributed to differences in the physiology of species. Sandelia capensis is well-

(°C) 32 adapted to environments that are harsh and highly variable, using an cmax T accessory breathing organ to survive in rivers that are more oxygen- poor (Skelton, 2001). Since oxygen demands increase with an increase 30 in temperature (Chatterjee et al., 2004), this physiological adaptation is advantageous in terms of the fish's ability to avoid stress. On the 28 other hand, G. zebratus displayed wide habitat tolerances in the field (Cowling et al., 2003; Skelton, 2001), but exhibit the lowest thermal 26 tolerance. From an evolutionary perspective, G. zebratus is one of the

ABCDEFGHI J K oldest fish species to have evolved in Africa (before the splitting of Gondwana; Linder et al., 2010) and such species are known to have FIGURE 2 Boxplot showing median, 25th and 75 quartiles, range retained their stenothermic adaptations (Ross-Gillespie, 2014). As ( ) and outliers (O) of thermal tolerance (TCmax) for eight species of opposed to S. capensis, it is likely that the extreme stenothermy of native fishes in order of decreasing sensitivity to warming: (a) Galaxias zebratus; (b) Pseudobarbus phlegethon; (c) Pseudobarbus burgi; G. zebratus is related to its evolutionary adaptation and not its (d) Sedercypris calidus; (e) Pseudobarbus phlegethon;(f)Pseudobarbus physiology. burchelli; (g) Austroglanis gilli; (h) Sandelia capensis; (i) Sandelia capensis; (j) Sandelia capensis; (k) Pseudobarbus afer 4.2 | Intra-specific variation in TCMAX

Previous studies have shown climate warming to disfavour spe- Intra-specific variation in thermal tolerance was observed in both cies that are evolutionarily warm-adapted (and closer to their thermal S. capensis and P. phlegethon. The former being widespread across the limits; Somero, 2010). Sandelia capensis (Fernkloof) and P. afer showed CFE, while the latter has two genetically distinct populations occurring the greatest thermal tolerances within the group, however their in the Olifants and Doring catchments of the Olifants River system observed thermal maxima are well above summer habitat tempera- (Swartz et al. 2007). For the present study, both P. phlegethon popula-  tures (even with a 2 C conservative estimated increase in summer air tions were assessed to enhance our understanding of the intra- temperatures as predicted by Dallas & Rivers-Moore, 2014). Given specific similarities and differences in thermal maxima. that the level of threat by climate warming decreases with an in For S. capensis, the Fernkloof population was found to be the increase in thermal tolerance (Somero, 2010), these two eastern CFE main source of variation in TCmax. Thermal metrics (showing recent populations may experience a lower level of threat relative to their thermal history) confirm that these systems differ in that mean tem-  western counterparts. In contrast, the conservative 4 C temperature perature for the week and month preceding experiments was signifi- increase scenario for the western CFE (Dallas & Rivers-Moore, 2014), cantly higher in the Fernkloof River. Differences in TCmax may thus  identifies P. burgi as the most sensitive species as it is within 0.3 Cof reflect differences in the thermal regimes of the systems where the its thermal limit in summer if warming continues. This is consistent species occur. Fangue et al. (2006) found evidence for intra-specific with the hypothesis of Somero (2010) concerning eurythermal species variation in thermal tolerance between northern and southern popula- living at their thermal extremes. tions of the killifish Fundulus heteroclitus (L. 1766). The tolerance range While stenothermy and eurythermy are rooted deep in the evolu- of the southern population encompassed the entire natural range of tionary history of species (viz. the Gondwanan fauna), thermal toler- daily and seasonal temperatures (Fangue et al., 2006) and thus thermal ance itself may be more strongly influenced by historical thermal history significantly influenced the thermal limit for the two popula- habitat. Amongst the species studied, P. afer (sensu stricto) is most tions. This may also be true for S. capensis exposed to greater fluctua- closely related to P. phlegethon (Swartz et al., 2004), but the species is tion in thermal regime and fluvial stochasticity in the eastern CFE. most similar to P. burchelli in terms of TCmax. Thermal habitat charac- At a catchment scale, the significant difference in thermal tolerance teristics may explain this result given that the Amandel and Fernkloof between the Olifants and Doring lineages of P. phlegethon, split by allo- Rivers were most similar in their summer mean temperatures when patric speciation in more recent geological time (Swartz et al., 2004), experiments were conducted; a trend that is largely consistent across indicate that thermal tolerance was not retained as an evolutionary relic, marine and freshwater taxa (Somero, 2010). This could also indicate but rather evolved as an adaptive one. The Driehoeks River (Doring lin- that the more recent evolutionary thermal environment for these spe- eage of P. phlegethon) is the cooler river over the summer low-flow cies were most similar (in terms of habitat temperature after the period, with the lowest minimum temperature over the same period. establishment of the western Mediterranean climate 5,000 years ago), The Rondegat River (Olifants lineage of P. phlegethon)isthewarmer thereby selecting for greater temperature tolerance independently. river over the summer period, however Colwell's predictability index Ellender and Weyl (2015) hypothesised that the resilience of P. afer showed the Driehoeks River to be the least predictable. This is reflected 110 FISH REIZENBERG ET AL. in the thermal tolerance and the thermal tolerance breadth observed in distinct differences in responses to warming. These differences may both P. phlegethon populations. Higher temperatures in the Rondegat be attributable to regional patterns in climate and differences in ther-

River may underly the high TCmax observed in that population, while the mal history among the different habitats sampled. The expected less predictable Driehoeks population has the wider thermal tolerance climate-induced increase in water temperature for the region is likely breadth. This suggests that the population most vulnerable to climate- to affect more sensitive species in the western CFE, where water tem- induced stress is the Driehoeks population. peratures are cooler and flow is more predictable, but where pre- dicted increases in water temperature and reduced runoff may 4.3 | Examining the east-west gradient in sensitivity exacerbate thermal stress of native fishes over the warm, dry summer to warming period. In the eastern CFE, P. afer and S. capensis are expected to show greater resilience to temperature fluctuation. This finding is con- Sensitivity in the ecological context refers to the amount of change an sistent with Ellender and Weyl (2015) concerning the response of organism or ecosystem can tolerate before biological and ecological P. afer to disturbance events. The focus of more in-depth research in processes are disrupted (Meyer et al., 1999). Mediterranean ecosys- the future should be to undertake more comprehensive studies to tems and the species they support, have been described as particularly determine the thermal sensitivities within species and species com- sensitive to climate warming (Thuiller, 2007). The findings of this plexes, such as the Cape Peninsula Galaxiidae (Wishart et al., 2006), as study support this as the western CFE (a Med-region) fishes showed this study showed that spatial variation is likely to differ amongst greater vulnerability to thermal warming. populations. Investigating the mechanisms behind these inter and Chevin et al. (2010) note that the current rate of human-induced intra-specific variations in thermal tolerance offers an opportunity to environmental change may exceed the adaptive, genetic and develop- identify sensitive fish populations and thereby direct conservation mental capacity of organisms (i.e., their plasticity). Furthermore, popula- initiatives. tions have evolved demographic mechanisms to cope with environmental change, but these mechanisms are being pushed to their limits (Chevin et al., 2010). Under these selection pressures, the resilience ACKNOWLEDGEMENTS or sensitivity of each species will determine their adaptive capacity, and The Freshwater Research Centre, CapeNature and the South African although aquatic taxa have approximately twice the plasticity of terres- Institute for Aquatic Biodiversity, are acknowledged for institutional trial biota (Gunderson and Stillman, 2015), plasticity is still evolutionarily support. O.L.F.W. and L.E.B. acknowledge use of infrastructure and constrained. More stenothermic species (G. zebratus and certain Pseudo- equipment provided by the SAIAB Research Platform and the funding barbus spp.) are likely to fare poorly under the current climate change scenarios, while the more thermally plastic species may already survive channelled through the NRF-SAIAB Institutional Support system as – close to their thermal maxima (Somero, 2010). Pseudobarbus burgi is an well as funding by the National Research Foundation South African example of such a species, having the largest thermal tolerance breadth Research Chairs Initiative of the Department of Science and Technol- of all species examined, but most sensitive to a probable increase in ogy (Inland Fisheries and Freshwater Ecology, Grant No. 110507) and temperature. the DST/NRF Centre of Excellence in Invasion Biology. CapeNature Prior to this study, east-west patterns in thermal sensitivity had and the Eastern Cape Parks and Tourism are acknowledged for per- not been determined for freshwater fishes and while latitudinal varia- mits to collect and experiment on the fish used in this study under tion in thermal tolerance is well documented, east-west variation is permits 0056-AAA041-00116, NOS. CRO 37/17CR, CRO 38/117CR, poorly understood. Filipe et al. (2013) suggest that stream biota in CRO 27/16CR and CRO 28/16CR, respectively. The authors also Med-regions may have the potential to alter several life-history traits acknowledge the two anonymous reviewers whose comments greatly to improve resilience and resistance to new habitat conditions. These improved the manuscript. include being short-lived, small and resistant to low streamflow and desiccation (Filipe et al., 2013). Indeed, in the west, where flow is ORCID expected to reduce drastically, these adaptations or traits may already Jody-Lee Reizenberg https://orcid.org/0000-0001-6931-1710 be evident but further research is warranted to confirm which physio- Olaf L. F. Weyl https://orcid.org/0000-0002-8935-3296 logical traits are subject to the strongest selection pressure. In the Jeremy M. Shelton https://orcid.org/0000-0001-7174-5446 eastern CFE it is possible that certain traits are rooted deep in the Helen F. Dallas https://orcid.org/0000-0001-8133-3365 species' evolutionary history and have been retained, resulting in their increased thermal tolerance. Although intuitively these beneficial REFERENCES traits improve competitive advantage, the rate of adaptation for both eastern and western CFE species may be insufficient to cope with the Abell, R., Thieme, M. L., Revenga, C., Bryer, M., Kottelat, M., Bogutskaya, N., … Stiassny, M. L. (2008). Freshwater ecoregions of the rate of environmental changes in the future (Filipe et al., 2013). world: A new map of biogeographic units for freshwater biodiversity In conclusion, this study has produced some of the first data on conservation. Bioscience, 58(5), 403–414. thermal tolerance for native fishes in the CFE. The inter and intra- Becker, C. D., & Genoway, R. G. (1979). Evaluation of the critical thermal maximum for determining thermal tolerance of freshwater fish. Envi- specific variation of TCmax in the CFE fishes indicate that sensitivity to ronmental Biology of Fish, 4(3), 245–256. thermal alteration can differ between species across geographical Beitinger, T. J., & Lutterschmidt, W. I. (2011). In A. P. Farrell, E. D. Stevens, clines. Within species and across regions, fishes have displayed J. J. Cech, & J. G. Richard (Eds.), Measures of thermal tolerance. REIZENBERG ET AL. FISH 111

Encyclopedia of fish physiology-from genome to environment Ellender, B. R., & Weyl, O. L. (2014). A review of current knowledge, risk (pp. 1695–1702). San Diego, CA: Academic Press. and ecological impacts associated with non-native freshwater fish Beitinger, T. L. (1990). Behavioral reactions for the assessment of stress in introductions in South Africa. Aquatic Invasions, 9(2), 117–132. fish. Journal of Great Lakes Research, 16(4), 495–528. Ellender, B. R., & Weyl, O. L. (2015). Resilience of imperilled headwater Beitinger, T. L., Bennett, W. A., & McCauley, R. W. (2000). Temperature stream fish to an unpredictable high-magnitude flood. Koedoe, tolerances of north American freshwater fish exposed to dynamic 57(1), 1–8. changes in temperature. Environmental biology of fish, 58(3), 237–275. Fangue, N. A., Hofmeister, M., & Schulte, P. M. (2006). Intraspecific varia- Caissie, D. (2006). The thermal regime of rivers: A review. Freshwater Biol- tion in thermal tolerance and heat shock protein gene expression in ogy, 51(8), 1389–1406. common killifish, Fundulus heteroclitus. Journal of Experimental Biology, Carpenter, S. R., Fisher, S. G., Grimm, N. B., & Kitchell, J. F. (1992). Global 209(15), 2859–2872. change and freshwater ecosystems. Annual Review of Ecology and Sys- Fauchereau, N., Trzaska, S., Rouault, M., & Richard, Y. (2003). Rainfall vari- tematics, 23(1), 119–139. ability and changes in southern Africa during the 20th century in the Chakona, A., Swartz, E. R., & Gouws, G. (2013). Evolutionary drivers of global warming context. Natural Hazards, 29(2), 139–154. diversification and distribution of a southern temperate stream fish Filipe, A. F., Lawrence, J. E., & Bonada, N. (2013). Vulnerability of stream assemblage: Testing the role of historical isolation and spatial range biota to climate change in mediterranean climate regions: A synthesis expansion. PLoS One, 8(8), e70953. of ecological responses and conservation challenges. Hydrobiologia, Chase, B. M., & Meadows, M. E. (2007). Late quaternary dynamics of 719(1), 331–351. southern Africa's winter rainfall zone. Earth-Science Reviews, 84(3), Gasith, A., & Resh, V. H. (1999). Streams in Mediterranean climate regions: – 103 138. Abiotic influences and biotic responses to predictable seasonal events. Chatterjee, N., Pal, A. K., Manush, S. M., Das, T., & Mukherjee, S. C. (2004). Annual Review of Ecology and Systematics, 30(1), 51–81. Thermal tolerance and oxygen consumption of Labeo rohita and Cypri- Gunderson, A. R., & Stillman, J. H. (2015). Plasticity in thermal tolerance nus carpio early fingerlings acclimated to three different temperatures. has limited potential to buffer ectotherms from global warming. In Pro- – Journal of Thermal Biology, 29(6), 265 270. ceedings of the Royal Society B., 282(1808), 20150401. https://doi. Chevin, L. M., Lande, R., & Mace, G. M. (2010). Adaptation, plasticity and org/10.1098/rspb.2015.0401 in a changing environment: Towards a predictive theory. Heino, J., Virkkala, R., & Toivonen, H. (2009). Climate change and freshwa- PLoS Biology, 8(4), e1000357. ter biodiversity: Detected patterns, future trends and adaptations in Colwell, R. K. (1974). Predictability, constancy and contingency of periodic northern regions. Biological Reviews, 84(1), 39–54. phenomena. Ecology, 55, 1148–1153. Hubert, W. A., Pope, K. L., & Dettmers, J. M. (2012). Passive capture tech- Coutant, C. C., & Brook, A. J. (1970). Biological aspects of thermal pollution niques. In Zale, A. W., Parrish, D. L., & Sutton, T. M. (Eds.). Fisheries I. Entrainment and discharge canal effects. Critical Reviews in Environ- techniques (3rd ed.), pp. 223–265. Bethesda, MD: American Fisheries mental Science and Technology, 1(1–4), 341–381. Society. Cowling, R. M., Pressey, R. L., Rouget, M., & Lombard, A. T. (2003). A con- Intergovernmental Panel on Climate Change. (2014). Climate change servation plan for a global biodiversity hotspot—the Cape floristic 2014–impacts, adaptation and vulnerability: Regional aspects. Cam- region, South Africa. Biological Conservation, 112(1), 191–216. bridge, UK and New York, NY: Cambridge University Press. Dallas, H., Shelton, J., Paxton, B., Weyl, O., Reizenberg, J., Bloy, L., & Linder, H. P., Johnson, S. D., Kuhlmann, M., Matthee, C. A., Nyffeler, R., & Rivers-Moore, N. (2017). Assessing the effect of climate change on native Swartz, E. R. (2010). Biotic diversity in the southern African winter- and non-native freshwater fishes of the Cape Fold Ecoregion, South Africa rainfall region. Current Opinion in Environmental Sustainability, 2(1), (WRC Report No. K5/2337). Pretoria, South Africa: Water Research 109–116. Commission. Lutterschmidt, W. I., & Hutchison, V. H. (1997). The critical thermal maxi- Dallas, H. F. (2008). Water temperature and riverine ecosystems: An over- mum: Data to support the onset of spasms as the definitive end point. view of knowledge and approaches for assessing biotic responses, with Canadian Journal of Zoology, 75(10), 1553–1560. special reference to South Africa. Water SA, 34(3), 393–404. Meadows, M. E., & Baxter, A. J. (1999). Late quaternary palaeoenviron- Dallas, H. F., & Ketley, Z. A. (2011). Upper thermal limits of aquatic macro- ments of the southwestern Cape, South Africa: A regional synthesis. invertebrates: Comparing critical thermal maxima with 96-LT50 values. – Journal of Thermal Biology, 36, 322–327. Quaternary International, 57, 193 206. Dallas, H. F., & Rivers-Moore, N. (2014). Ecological consequences of global Meyer, J. L., Sale, M. J., Mulholland, P. J., & Poff, N. L. (1999). Impacts of climate change for freshwater ecosystems in South Africa. climate change on aquatic ecosystem functioning and health. JAWRA – South African Journal of Science, 110(5–6), 01–11. Journal of the American Water Resources Association, 35(6), 1373 1386. Dallas, H.F., Rivers-Moore, N.A., Ross-Gillespie, V., Ramulifho, P., & Myers, N., Mittermeier, R. A., Mittermeier, C. G., Da Fonseca, G. A., & Reizenberg, J. (2015). Adaptability and vulnerability of riverine biota to Kent, J. (2000). Biodiversity hotspots for conservation priorities. – climate change- developing tools for assessing biological effects (WRC Nature, 403(6772), 853 858. Report No. 2182/1/15). Pretoria, South Africa: Water Research Nel, J.L., Strydom, W.F., Maherry, A.M., Petersen, C., Hill, L., Roux, D.J., ... Commission. Smith-Adao, L.B. (2011). Atlas of freshwater ecosystem priority areas in Dallas, H. F., & Ross-Gillespie, V. (2015). Sublethal effects of temperature South African maps to support sustainable development of water on freshwater organisms, with special reference to aquatic insects. resources (Water Research Commission Report No. TT500/11). Preto- Water SA, 41(5), 712–726. ria, South Africa: Water Research Commission. Dudgeon, D., Arthington, A. H., Gessner, M. O., Kawabata, Z. I., Nelitz, M. A., MacIsaac, E. A., & Peterman, R. M. (2007). A science-based Knowler, D. J., Lévêque, C., … Sullivan, C. A. (2006). Freshwater biodi- approach for identifying temperature-sensitive streams for rainbow versity: Importance, threats, status and conservation challenges. Biolog- trout. North American Journal of Fisheries Management, 27, 405–424. ical Reviews, 81(2), 163–182. Paladino, F. V., Spotila, J. R., Schubauer, J. P., & Kowalski, K. T. (1980). The Eaton, J. G., McCormick, J. H., Goodno, B. E., O'brien, D. G., Stefany, H. G., critical thermal maximum: A technique used to elucidate physiological Hondzo, M., & Scheller, R. M. (1995). A field information-based system stress and adaptation in fish. Revue Canadienne de Biologie, 39(2), for estimating fish temperature tolerances. Fisheries, 20(4), 10–18. 115–122. Eliason, E. J., Clark, T. D., Hague, M. J., Hanson, L. M., Gallagher, Z. S., Perez, E., Diaz, F., & Espina, S. (2003). Thermoregulatory behavior and crit- Jeffries, K. M., … Farrell, A. P. (2011). Differences in thermal tolerance ical thermal limits of the angelfish Pterophyllum scalare (Lichtenstein) among sockeye salmon populations. Science, 332(6025), 109–112. (Pisces: Cichlidae). Journal of Thermal Biology, 28(8), 531–537. Ellender, B. R., Wasserman, R. J., Chakona, A., Skelton, P. H., & Weyl, O. L. Pörtner, H. O. (2002). Climate variations and the physiological basis of (2017). A review of the biology and status of Cape Fold Ecoregion temperature dependent biogeography: Systemic to molecular hierar- freshwater fish. Aquatic Conservation: Marine and Freshwater Ecosys- chy of thermal tolerance in . Comparative Biochemistry and tems, 27(4), 867–879. https://doi.org/10.1002/aqc.2730 Physiology A, 132(4), 739–761. 112 FISH REIZENBERG ET AL.

Reason, C. J. C., & Rouault, M. (2005). Links between the Antarctic oscilla- across multiple river systems in South Africa. Journal of Biogeography, tion and winter rainfall over western South Africa. Geophysical 34(12), 2086–2099. Research Letters, 32(7). https://doi.org/10.1029/2005GL022419 Swartz, E. R., Skelton, P. H., & Bloomer, P. (2009). Phylogeny and biogeog- Rivers-Moore, N. A., Dallas, H. F., & Morris, C. (2013). Towards setting raphy of the genus Pseudobarbus (Cyprinidae): Shedding light on the environmental water temperature guidelines: A South African example. drainage history of rivers associated with the Cape floristic region. Journal of Environmental Management, 128, 380–392. Molecular Phylogenetics and Evolution, 51(1), 75–84. Rivers-Moore, N. A., Mantel, S., & Dallas, H. F. (2012). Prediction of water Thuiller, W. (2007). Biodiversity: Climate change and the ecologist. Nature temperature metrics using spatial modeling in the Eastern and Western Reports Climate Change, 448, 550–552. Cape, South Africa. Water SA, 38(2), 167–176. Todd,A.S.,Coleman,M.A.,Konowal,A.M.,May,M.K.,Johnson,S., Ross-Gillespie, V. (2014). Effects of water temperature on life history traits of Vieira, N. K., & Saunders, J. F. (2008). Development of new water temper- selected South African Aquatic insects: Implications for the ecological ature criteria to protect Colorado's fisheries. Fisheries, 33(9), 433–443. reserve (Doctoral dissertation, University of Cape Town). Van Etten, E. J. (2009). Inter-annual rainfall variability of arid : Schulze, R.E. (2011). Climate change and the South African water sector: Set- Greater than elsewhere? Australian Geographer, 40(1), 109–120. ting the scene on a 2011 perspective. A 2011 Perspective on Climate Webb, B. W., Hannah, D. M., Moore, R. D., Brown, L. E., & Nobilis, F. Change and the South African Water Sector. WRC Report No. TT (2008). Recent advances in stream and river temperature research. 518 (12), 1–399. Hydrological Processes: An International Journal, 22(7), 902–918. Shelton, J. M., Weyl, O. L., Chakona, A., Ellender, B. R., Esler, K. J., Wishart, M., Hughes, J., Stewart, B., & Impson, D. (2006). Extreme levels Impson, N. D., … Van Der Walt, J. A. (2018). Vulnerability of Cape Fold of intra-specific divergence among Cape Peninsula populations of the Ecoregion freshwater fishes to climate change and other human Cape galaxias, Galaxias zebratus Castelnau 1861, reveals a possible impacts. Aquatic Conservation: Marine and Freshwater Ecosystems, species complex. African Journal of Aquatic Science, 31(1), 99–106. 28(1), 68–77. Woodward, G., Perkins, D. M., & Brown, L. E. (2010). Climate change and Skelton, P. H. (2001). A complete guide to the freshwater fishes of Southern freshwater ecosystems: Impacts across multiple levels of organization. Africa. Cape Town, South Africa: Struik. Philosophical Transactions of the Royal Society of London B: Biological Somero, G. N. (2010). The physiology of climate change: How potentials Sciences, 365(1549), 2093–2106. for acclimatization and genetic adaptation will determine ‘winners’ and ‘losers’. Journal of Experimental Biology, 213(6), 912–920. SUPPORTING INFORMATION Sullivan, K., Martin, D. J., Cardwell, R. D., Toll, J. E., & Duke, S. (2000). An analysis of the effects of temperature on salmonids of the Pacific North- Additional supporting information may be found online in the Sup- west with implications for selecting temperature criteria. Portland, OR: porting Information section at the end of the article. Sustainable Ecosystems Institute. Sunday, J. M., Bates, A. E., & Dulvy, N. K. (2012). Thermal tolerance and the global redistribution of animals. Nature Climate Change, 2(9), How to cite this article: Reizenberg J-L, Bloy LE, Weyl OLF, 686–690. Shelton JM, Dallas HF. Variation in thermal tolerances of Swartz, E. R., Flemming, A. F., & Mouton, P. L. F. N. (2004). Contrasting genetic patterns and population histories in three threatened redfin native freshwater fishes in South Africa's Cape Fold Ecoregion: species (Cyprinidae) from the Olifants river system, western examining the east–west gradient in species' sensitivity to cli- South Africa. Journal of Fish Biology, 64(5), 1153–1167. mate warming. J Fish Biol. 2019;94:103–112. https://doi.org/ Swartz, E. R., Skelton, P. H., & Bloomer, P. (2007). Sea-level changes, river capture and the evolution of populations of the eastern Cape and fiery 10.1111/jfb.13866 redfins (Pseudobarbus afer and Pseudobarbus phlegethon, Cyprinidae)