Animal ethics
All experiments were conducted under the UK Home Office licence P88687E07 and with approval from the University of Exeter Ethics Committee.
Fish husbandry
Juvenile rainbow trout (Oncorhynchus mykiss) (n = 42; body mass: 159.9 ± 5.2 g), were obtained from Houghton Spring Fish Farm (Dorset, UK) and housed in the Aquatic Research Centre at the University of Exeter (UK). Before transfer to individual experimental chambers, all fish were housed across two 400 L tanks (n = 21 per tank) supplied with recirculated fresh water for 14 days. During this 14 day acclimation period, fish were maintained at 15 °C and fed on a 1% ration of commercial trout feed (Aller platinum 4.5 mm (Aller AQUA ©) three times a week. Prior to experimentation, fish were fasted for seven days.
Acid buffering diets
Diets were prepared by adding one of three calcium-based salts, CaCO3, Ca3(PO4)2 or CaCl2 (as non-buffering control) with isomolar quantities of calcium to a commercial trout pelleted diet (Skretting 4.5 mm Horizon, Skretting, UK). The quantities of these salts used were designed to mimic the calcium content of the skeletal component of crustacean or bony fish prey.
Cameron (1985)50 estimated that the bone of teleost fish represents 16.3% of whole-body mass (and therefore soft tissue represents 83.7%). However, bone is not just calcium phosphate, but includes numerous organic components as well as water content. By comparing titrations of pure calcium phosphate salt and samples of ground-up teleost (rainbow trout) bone, we established that it required 10.25 times less calcium phosphate salt to achieve the same acid-buffering capacity as that of an equal mass of bone. We therefore created a diet that was supplemented with 1.9 g calcium phosphate for every 100 g of trout pellets (i.e. [16.3 g ÷ 10.25] x [100 ÷ 83.7 g] = 1.9 g), in order to match the bone content of calcium phosphate typically found in fish prey as a proportion of the soft tissue mass. This amounted to 18.4 mmoles of calcium phosphate salt (Ca3(PO4)2; M.W. = 310.2) per 100 g of trout pellets. For the two other diets we aimed to maintain the same molar amount of calcium cation added whilst varying the anionic component of the salt added. So, for the unbuffered version of the diet 2.7 g of calcium chloride (CaCl2.2H2O; M.W. = 147.0) was added, whilst for the calcium carbonate (CaCO3; M.W. = 100.0) buffered diet 1.84 g was added, per 100 g of trout pellets.
To form each diet, 100 g of Skretting 4.5 mm Horizon trout pellets were ground to a fine powder using a pestle and mortar. Following grinding, 1.9, 1.84 and 2.7 g of Ca3(PO4)2, CaCO3 and CaCl2 were added to the ground pellet and mixed. Then, 70 ml of ultrapure water was added to the dry material to form a paste. This paste was pressed into commercial 4 mm moulds, removed and dried at 70 °C for 24 h. An acid titration test was conducted to ensure that diets remained representative of the buffer capacity of prey and each calcium salt. For this test, 60 ml of ultrapure water were added to 1 g of each experimental diet and titrated down to pH 3.5 using 0.05 mol L−1 HCl. The CaCl2 diet treatment required 4.56 ml of the acid which was only slightly less than the 6.4 ml required to titrate the Ca3(PO4)2 diet. In contrast it took almost double the amount of acid (11 ml) to titrate the CaCO3 diet. In molar terms it took 228, 320 and 550 µmoles of HCl to titrate 1 g of the CaCl2, Ca3(PO4)2 and CaCO3 feeds to pH 3.5, respectively. To calculate the total acid-buffering consumed, the buffer capacity (per g of food) was multiplied by the actual ration ingested for each individual. Based on manufacturer details each diet had a gross energy of 23 kJ per gram of feed.
Acid secretion in the stomach and the blood alkaline tide
To investigate the effect of dietary buffer capacity on the blood acid–base chemistry (alkaline tide) and gut secretions, blood and gut samples were taken from fish to determine blood gas and acid–base balance and haematology variables of fish fed each experimental diet. Fish were fasted for 7 days and then fed a 2.5% ration of one of three experimental feeds. Diet was randomly allocated to each individual (n = 6 per diet). At 24 and 48 h following meal ingestion fish were anesthetised using benzocaine (100 mg l−1). Once fish had lost equilibrium and were un-responsive to a tail pinch, fish were transferred to a gill irrigation system dosed with a lower concentration of benzocaine (75 mg l−1). Fish were placed upside down within the irrigation chamber so that the head was fully submerged, and the entire gill basket covered. A micro pump was used to artificially ventilate the gills via a tube placed into the fish mouth. This allowed for the continuous ventilation of fish gills and ensured there was no build-up of CO2 or lactic acid during blood sampling that could unintentionally affect blood acid–base status. Blood was then drawn into a sodium-heparinised syringe via caudal puncture. Fish were then euthanased via pithing and dissected to collect stomach and intestinal contents. Gut samples were centrifuged to isolate gastric and intestinal juices.
Blood and gastric pH were measured using an Accumet CP-620-96 MicroProbe (Accumet Engineering Corporation, USA) connected to a Hanna HI 8424 m (Hanna Instruments, Woonsocket, Rhode Island, USA). Whole blood PO2 was measured using a Strathkelvin 1302 electrode, housed within a thermostatted glass chamber (Strathkelvin), and connected to Strathkelvin 781 m (Strathkelvin Instruments Ltd., Scotland)51. Blood was drawn into three micro-haematocrit tubes (Hawksley) via capillary action and anaerobically sealed using Hawksley Critaseal Wax Sealant, then centrifuged (Hawksley microhaematocrit centrifuge, 10,000 rpm for 2 min) and then used to record haematocrit and held on ice before using the plasma. Plasma and intestinal total CO2 was then measured using a Mettler Toledo 965 carbon dioxide analyser and together with blood and intestinal pH measurements was used to calculate plasma and intestinal HCO3− and PCO2 by rearranging the Henderson–Hasselbalch equation and using values for solubility and pKapp from Boutilier et al. (1985)52.
Net acid–base fluxes to the external water
The effect of diet on the net flux of acid–base relevant ions to the external water was measured in a separate subset of juvenile rainbow trout (n = 10, 161.8 ± 6.9 g). Prior to measurements fish were weighed and transferred to individual 25 L chambers supplied with recirculated freshwater maintained at 15 °C. Following a 3-week acclimation period, fish were fed weekly on a 2.5% ration of one of three experimental feeds, with diet order randomised to each individual (See Supplementary Table 4). Initial and final water samples were taken from each chamber over six flux periods each week for three weeks (−23 to 1 (fasted), 0–6, 7–23, 24–47, 48–71 and 72–96 h post feed). Water inflow to each chamber was turned off during each flux period whilst aeration was maintained. Following the final measurement from each flux period, tanks were flushed with dechlorinated freshwater for 60 min so to ensure solid faeces and dissolved waste products (e.g., ammonia) were removed.
Total ammonia was measured in triplicate on 200 µL water samples using the colourimetric salicylate-based method adapted from Cooper and Wilson (2008)19 and Verdouw et al. (1978)53 and the Infinite 200 PRO microplate reader (Tecan Trading AG Switzerland ©). Titratable alkalinity was measured in 20 ml water samples using an auto-titrator with autosampler (Metrohm 907 Titrando with 815 Robotic USB Autosampler XL) running double titrations with 0.02 mol l−1 of HCl and 0.005 mol l−1 NaOH. The double titration method calculates titratable alkalinity based on the difference in HCl required to titrate each water sample down to pH 3.9 and the amount of NaOH required to bring the sample back to the starting pH. During the titration, the sample is continuously bubbled or ‘purged’ with the inert gas N2 to remove any CO2. The net fluxes of titratable alkalinity (JTalk) and total ammonia (JTamm) were calculated using the following equation from Cooper and Wilson 2008:
$${J}_{mathrm{net}}mathrm{X}=frac{[left(left[{mathrm{X}]}_{i}-{left[mathrm{X}right]}_{mathrm{f}}right) times Vright]}{(M times t)}$$
(1)
where Xi and Xf are the initial and final ion concentration in each tank (μmol l−1) from each flux period, V is the tank volume (L), M is the animal mass (kg) and t is the flux duration (h).
The net acid–base flux was calculated as the difference between the flux of titratable alkalinity (JTalk) and the flux of total ammonia (JTamm).
Measuring the SDA
Intermittent flow-through respirometry was used to determine the rate of oxygen consumption (MO2) by juvenile rainbow trout fed voluntarily on a 2.5% ration of three experimental feeds. Prior to measurements, juvenile rainbow trout (n = 8, 162.2 ± 7.5 g) were weighed and transferred to individual 25 L chambers supplied with recirculated freshwater at 15 °C for 3 weeks. During this acclimation period, fish were fed weekly on a 2.5% ration of Skretting 4.5 mm Horizon trout pellets (Skretting UK). Following this acclimation period, measurements were conducted after 7 days of fasting. Each fish was fed once per week on all three diets over a 3-week period, with diet order randomised for each individual.
During experimentation, fresh water was supplied continuously to two aerated 160 L sumps each fitted with a ballcock valve and overflow. Aerated freshwater was then pumped from the sump to the eight respirometry chambers in a loop for the duration of the testing period. Water within each fish chamber was continuously mixed using a submerged mini-pump (WP300; Tetra Werke, Melle, Germany). During measurements, water inflow to each chamber was shut off and the decline in O2 was recorded by PO2 OxyGuard Mini Probe (OxyGuard ® International, Denmark) connected directly to the mini-pump. Oxygen partial pressure values were logged continuously by Pyro Oxygen Logger software (Pyroscience GmBH, Germany) which interfaced with a respirometry software package (AquaResp3: aquaresp.com, see Svendsen et al. 2016 54) to instantaneously convert PO2 into O2 content and calculate the rate of oxygen consumption (MO2, mg O2 kg−1 body mass h−1) based on the fish body mass in kg (m), chamber water volume in L after discounting the fish body volume (Vresp), and the slope (s) of the decline in oxygen concentration (kPa O2 h−1) versus time using the following equation from Svendsen et al. (2016)54:
$${MO}_{2}= {sV}_{Resp}{alpha m}^{-1}$$
where:
$$s= frac{{O}_{2}, initial- {O}_{2}, final}{time, initial-time, final}$$
Following each closed measurement period, the chamber was automatically flushed with freshwater from the aerated sumps by two AquaMedic Ocean Runner pumps (Aqua Medic, Ocean Runner 6500). The length of the flush and measurement periods was controlled by two USB- 4 Cleware switches (Cleware GmbH, Germany) which were also interfaced with the AquaResp software to ensure that the partial pressure of oxygen (PO2) within the respirometry chambers never fell below 90% of the starting value. This meant that the measurement period of 15 min was followed by a flushing period of 2 min and a wait time of 60 s.
Prior to feeding a baseline 24 h period of standard metabolic rate (SMR) was recorded. The mean SMR of each individual was calculated using the R package ‘fishMO2’ and the ‘calcSMR’ function. Following Chabot et al. (2016)55, the coefficient of variation (CVmlnd) was used to determine whether the mean of the lowest normal distribution (MLND) or the quantile method (P = 0.2) was used to estimate SMR for each individual. Following the SMR measurement, fish voluntarily fed on a 2.5% ration of experimental feed and MO2 recorded continuously for six days. This procedure was repeated for two more consecutive weeks to measure MO2 in fish fed all three experimental diets. Background oxygen consumption was recorded overnight (18 h) in blank (no fish) chambers. Oxygen consumption was not corrected for background respiration as it was considered negligible (< 1% of resting fish MO2).
The respirometry chambers used in this study were open to the atmosphere (water exposed to air) meaning O2 exchange could have occurred at the surface. Therefore, prior to placing fish into respirometry chambers, experiments were conducted to determine the maximum rate of exchange of O2 at the water surface in the current study and its influence on observed rates of fish oxygen consumption. Oxygen was purged from each respirometry chamber down to 50% air saturation by bubbling water with N2 and left to re-equilibrate back up to 95–100% air saturation. Re-equilibration was recorded using the respirometry software described above. This revealed that at water O2 levels typically observed during respirometry measurements with fish (decline in air saturation from 100 to a minimum of 90%), the rate of O2 diffusion from air into the water would have been equivalent to at most 1.8% of the O2 removal by fish respiration. Also, this worst-case-scenario rate of O2 diffusion would only have occurred when the diffusion gradient was largest between the water and the air, i.e. at the end of the 15-min measurement period. Therefore, the rate of O2 diffusion prior to this point, would have been slower and somewhere between zero and 1.8% of the fish respiration rate (MO2). Given the negligible impact on rates of oxygen consumption, fish oxygen consumption was therefore not corrected for O2 diffusion. This is similar to the conclusions of a previous study by McKenzie et al. (2007)56 that used open top respirometry to measure oxygen consumption in rainbow trout maintained at 10 °C. Following a similar test on rates of O2 diffusion, their study determined that surface exchange would have modified the decline in O2 concentration caused by fish respiration by less than 2%, and they also considered this negligible and not requiring any correction.
If the slope of change in O2 over time used to calculate MO2 had an R2 < 96% it was removed from the data set for that fish. The total SDA was measured as the area under the curve of oxygen consumption rate versus time from the time of feeding until MO2 values returned to SMR. The energetic cost of digestion for each individual meal ingested was standardised to kilojoules (kJ) using the total magnitude of the SDA and the conversion factor of 1 mg O2 = 14 J3,57,58,59. The SDA scope was calculated by dividing the post-prandial peak in MO2 by SMR, and the SDA coefficient (energy cost of digestion relative to the energy content of the meal) was calculated by dividing the total SDA (in kJ) by meal energy.
Growth
To determine whether an acid buffering diet influenced growth efficiency, 18 fish (n = 6 per diet) were isolated into individual 25 L tanks supplied with dechlorinated freshwater and fed daily on a 1% ration of one of three experimental feeds for 21 days. Initial and final body mass were recorded at the beginning and end of the experimental period and total feed consumed was calculated each day. The feed conversion ratio (FCR) and specific growth rate (SGR; % growth per day) of each individual was then calculated as follows:
$$FCR=frac{total, mass, of, feed, consumed}{(final, fish, mass-initial, fish, mass)}$$
(2)
$$SGR=frac{In, final, weight (g) – In, initial, weight (g)}{experimental ,days } times 100$$
(3)
Statistical analyses
All statistical analyses were performed in R version 4.0.3 (R Development Core Team, 2020) in the RStudio environment Version 1.3.1093–1 ‘Apricot Nasturtium’ (RStudio, Inc 2020). All graphics were produced in Prism Version 9.00 for Mac (GraphPad Software, La Jolla California USA). Analyses were conducted following a D’Agostino & Pearson normality test. The R package ‘fishMO2′52 (Version 0.43) was used to determine standard metabolic rate (SMR) and the magnitude, duration and peak MO2 of the SDA from each individual (where τ = 0.2, λ = 1). MO2 values with an R2 < 0.96 were removed from further SDA analyses55. A repeated-measures one-way analysis of variance (RM- ANOVA), and Tukey’s multiple comparisons test was used to determine differences in the SDA (total magnitude, duration, peak, time to peak, SDA coefficient and SDA scope) and cumulative acid–base fluxes (ammonia (JTamm), titratable alkalinity (JTalk) and net acid or base flux) between diets. A standard one-way ANOVA and Tukey multiple comparisons test was performed to assess differences between diets for changes in mean blood pH, HCO3−, the partial pressure of CO2 (pCO2), hourly fluxes of JTalk and JTamm, gut pH and HCO3− concentration, the feed conversion ratio (FCR) and specific growth rate (SGR). Where applicable comparisons to fasted animals were conducted using a two-sample t-test. As an additional measure, a linear mixed effects model was used to examine the relationship between total acid-buffering consumed (μmol HCl required to titrate the food ingested (per 100 g of fish) to pH 3.5) and the SDA. Similarly, a simple linear regression was performed to assess the relationship between total acid-buffering consumed, gut pH, HCO3−concentration, cumulative fluxes, FCR and SGR. Model selection was determined using the AIC function, and where suitable diet order, fish mass, tank and/or individual were included in the model as a fixed or random factor. Data are expressed as means ± SE where n = number of fish or samples. Significance was accepted at P < 0.05.
Source: Ecology - nature.com
