Documentation for datasets that accompany the EVOLUTION publication: Phosphorus limitation does not drive loss of bony lateral plates in freshwater stickleback (Gasterosteus aculeatus) Sophie L. Archambeault, Daniel J. Durston, Alex Wan, Rana W. El-Sabaawi, Blake Matthews and Catherine L. Peichel Data are deposited on Dryad. There are two data files associated with this paper: 1. "Archambeault-et-al_2020_Evolution_Phosphorus_experiment_Growth-data_Dryad.csv" 2. "Archambeault-et-al_2020_Evolution_Phosphorus_experiment-Excretion-data_Dryad.csv" The first file (growth data) includes the data required to do all the analyses except the phosphorus excretion analysis. The second file (excretion data) has the data for the phosphorus excretion analysis. The two files are not combined because the excretion data is in a long format because each fish had multiple samples taken. These data have been trimmed of weight and SL measurements from individual fish that were bent or "stumpy". These fish were removed because they fell outside the normal relationship between length and weight. However, these fish were kept in the survival analyses. These data have also been trimmed of families that had such high mortality that they no longer had a complete cross design, i.e., if all fish in one of the conditions died, the entire family was removed. Excerpts from Methods: "Growth, intestine length, and bone phosphorus analyses" Growth was measured in two ways for fish from all three timepoints: standard length (SL measured in mm) and body condition. Intestine length and bone phosphorus content (%P) were measured in fish taken at Timepoint 3 only. Each of these measures were then analyzed for differences between treatments and genotypes using a linear mixed-effects model with family and tank as random effects, and water, diet, sex, genotype, and the interactions between water x genotype, water x diet, and diet x genotype as fixed effects using th lmerTest package in R. Analyses of fish from Timepoint 3 also included age as a fixed effect, since the fish ranged in age between 155 - 175 dpf. Body condition and intestine length were analyzed by running the model for the natural log of weight (or intestine length) and including the natural log of standard length as an additional fixed effect. Group means and variation for genotype, diet and water conditions were estimated using the R package emmeans. "Excretion data analyses" Excretion rates were calculated for the 0 - 60 minute sampling interval as well as the 60 - 120 minute sampling interval using the equation: Excretion rate = (P_f*V_i - P_i*V_i)/t where P_i and P_f are the μg/L phosphorus in the water at the start and end of the sampling interval, respectively, V_i is the volume of water (in liters) at the start of the sampling interval (before the end sample was removed), and t is the length of the sampling interval in minutes. The phosphorus excretion rate is therefore reported as μg/ min. During data exploration, we found that the excretion measurements varied from batch to batch on the Skalar San++ Continuous Flow P/N analyzer. We also found that excretion rates in the second time interval (60 - 120 minutes) were lower than in the first time interval (0 – 60 minutes) likely because the fish ceased excretion after 60 minutes (Authors, pers. observation). Therefore, the data were trimmed to only include samples from the first time interval, and in which the 60 minute sample was analyzed in the same batch as the 0 minute sample for each individual fish. Therefore, effects of time interval and analysis batch were removed. Excretion rates for the 0 – 60 minute time interval were analyzed for differences between treatments and genotypes using a linear mixed-effects model. Family and collection day were set as random effects; water, diet, sex, weight, and genotype were fixed effects; and the interactions between water x genotype, water x diet, and diet x genotype were also fixed effects. Column names for Growth data: date Date that the fish was sampled dpf Days post fertilization of the sampling date Timepoint Three experiments, or timepoints were run: TP1=pre-plate development, TP2=during plate development, TP3=during plate development+excretion, gut length and bone measurements fishID Unique ID for each fish SL.mm Standard length in mm ln.SL Natural log of standard length weight.mg Weight in mg ln.weight Natural log of weight family Family was used as a random effect in the model water Saltwater or freshwater condition (see ms for details) tank Tank was included as a random effect in the model diet Diet was either HIGH or LOW (see ms for details) dead Some fish were found dead in tank and collected for DNA, however these were removed from the analysis Stn382 Genotype at Stn382: CC=homozygous for the marine allele, LL=homozygous for the freshwater allele, CL=heterozygous. See Colosimo et al. (2005) Science SOM for details IDH Genotype at IDH was used as a molecular marker of sex: M(ale) or F(emale) LRR Genotype at LRR was also used as a molecular marker of sex: M(ale) or F(emale) sex Concatenated information from IDH and LRR columns gut.length Gut length in mm gut.weight Gut weight in mg L.plates Count of bony plates on left side of body P.PO4_preAsh Percent phosphorus of bone dry weight notes Any notes about the sample or the collection that may be of interest Column names for Excretion data: date Date that the fish was sampled dpf Days post fertilization of the sampling date Timepoint Three experiments, or timepoints were run: TP1=pre-plate development, TP2=during plate development, TP3=during plate development+excretion, gut length and bone measurements fishID Unique ID for each fish SL.mm Standard length in mm weight.mg Weight in mg gut.length Gut length in mm gut.weight Gut weight in mg family Family was used as a random effect in the model, plus family of the fish was recorded for the blank samples that were collected with the fish (see also "Collection" column water Saltwater or freshwater condition (see ms for details) tank Tank was included as a random effect in the model diet Diet was either HIGH or LOW (see ms for details) Stn382 Genotype at Stn382: CC=homozygous for the marine allele, LL=homozygous for the freshwater allele, CL=heterozygous. See Colosimo et al. (2005) Science SOM for details sex Concatenated information from IDH and LRR columns: molecular marker of sex: M(ale) or F(emale) L.plates Count of bony plates on left side of body P.PO4_preAsh Percent phosphorus of bone dry weight notes Any notes about the sample or the collection that may be of interest time.interval Either AB or BC for first 60-minute interval (AB) or second 60-minute interval (BC) minutes.food.1st.sample Time from feeding to first sample collected in the batch control Either NA for data collected from a fish, or "control" for the blank controls start.mean.ug.P.L Mean ug of phosphorus per liter of water in the start sample (time 0 for AB time interval, or time 60 for BC interval), mean of two technical replicates of P measurement start.stdev.ug.P.L Standard deviation ug of phosphorus per liter in the start sample (time 0 for AB time interval, or time 60 for BC interval), stdev of two technical replicates of P measurement end.mean.ug.P.L Mean ug of phosphorus per liter of water in the end sample (time 60 for AB time interval, or time 120 for BC interval), mean of two technical replicates of P measurement end.stdev.ug.P.L Standard deviation ug of phosphorus per liter in the end sample (time 60 for AB time interval, or time 120 for BC interval), stdev of two technical replicates of P measurement ug.P.excreted Calculated ug of phosphorus excreted during the time interval. For AB, calculated as (end.mean.ug.P.L*0.27L)-(start.mean.ug.P.L*0.27L). For BC interval, calculated as: (end.mean.ug.P.L*0.24L)-(start.mean.ug.P.L*0.24L). This is because 30mL were removed for each sample P.excreted.stdev Standard deviation of excreted phosphorus, calculated from technical replicates as: (volume in liters)*(sqrt((end.stdev.ug.P.L)^2 + (start.stdev.ug.P.L)^2)) P.excr.rate Calculated ug of phosphorus excreted per minute P.excr.rate.stdev Calculated standard deviation of excretion rate, calculated as: P.excreted.stdev/minutes Collection Samples and controls were collected in batches, and this column designates each batch Analysis.plate Technical replicates of samples were measured on plates, 100 at a time. There was significant plate to plate variation, so if this column is NA, then the start and end samples from that time interval (AB or BC) were not run on the same plate and were not used in the analysis. If this column has a Plate listed, then both the start and end samples were run on the plate listed