The seasonal onset of reproduction is constrained in many systems by a need to first accumulate energetic reserves. Consequently, the observation that larger individuals reproduce earlier may be due to a negative relationship between size and mass-specific basal metabolic rate that is shared across diverse taxa. However, an untested prediction of this hypothesis is that individuals should be metabolically efficient enough to escape energetic constraints above a certain size threshold. Seasonally reproducing species, such as temperate fishes, that must recover winter energy losses before reproduction and exhibit indeterminate growth are ideal models to test this prediction. We harness decade-long behavioral data on parental male smallmouth bass, Micropterus dolomieu, to investigate contributions of energetic allometry to differences in reproductive timing. At the population level, peak seasonal reproductive timing (i.e., the median date on which eggs were found in nests each year) was negatively related to degree days–a measure of thermal energy experienced–before reproduction. At the individual level, degree days accumulated by males before reproduction was related to male size and condition in every year, but the impact of temperature on reproductive timing by the largest males was relaxed in most years. Additionally, we used our data to replicate the analyses of two previous studies of M. dolomieu populations and found virtually identical negative associations between male body size and degree days accumulated before reproduction. Our results suggest that in smallmouth bass the onset of seasonal reproduction is constrained by basal metabolic rate–as indicated by total length–and that large individuals can escape size-associated energetic constraints. We reveal a more complicated relationship between size and reproductive timing than earlier studies, which may be relevant for many species. Knowledge of this relationship is critical to understanding how a changing climate will influence population dynamics of economically, ecologically, and recreationally important species like M. dolomieu.

Study site

We conducted this study in 1999 and 2001–2009 on Pallette Lake, a 70-ha research lake in the Northern Highlands Fishery Research Area of north-central Wisconsin (46.067° N/89.604° W). The Wisconsin Department of Natural Resources manages this lake, which is closed to fish migration and has a maximum depth of approximately 20 m and a mean depth of approximately 10 m. Saunders et al. (2002) describe the benthic and limnological characteristics of the lake in detail. Anglers are subject to a mandatory creel census, where caught fish must be reported to ensure regulations are followed. From 1999–2006, regulations required anglers to return to the lake any male less than 41 cm total length and from 2007 onwards all males under 51 cm had to be returned. We had significant representation of males larger than 40 cm in our analyses (13.8%).  

Annual nest census

We initiated surveys each year when the water temperature approached 15 °C, typically in mid to late May, and continued until males ceased nest construction (late June to early July) and progeny in most nests had dispersed. Two to three crew members with snorkels searched for nests as they swam transects running from the shoreline out to a depth of about 4 m. This depth was well beyond both the average 1.7 m depth at which nests in the lake were found and the maximum depth of smallmouth bass nests found in studies of other Wisconsin lakes (Bozek et al. 2002). When each nest was discovered, snorkelers placed a numbered tag constructed from a strip of Rite-n-Rain paper tied to a sinker on the edge of the nest and recorded the date, stage of embryos (if present), depth, location and distinctive landmarks that might be useful to relocate nests.

Characteristics of parental males

We captured parental males from their nests with hand nets and recorded their total length (cm) and weight (g). Each male was marked with a uniquely numbered Floy FD-67C anchor tag (University of Wisconsin RARC protocol A-48-9700-L00173-2-04-99). These tags allowed us to track individuals across seasons and estimate the total number of males that spawned each year, although analyses on reproductive timing were limited to those males that were found with eggs in their nest to ensure estimates of degree days before reproduction were comparable across individuals (Table 1).

Temperature and degree days

We used temperature data to estimate the thermal energy experienced by males in each year. A thermograph positioned near the shoreline at an approximate depth of 1 m recorded the water temperature (°C) hourly when the lake was not covered by ice. Thermograph malfunctions or other disturbances (e.g., removal of the thermograph from the lake by a lake visitor) resulted in some missing temperature records over the 10 years. We estimated these records using regression equations that characterized the relationship between water temperature in our study site (Pallette Lake) and nearby Sparkling Lake (46.010° N/89.701° W), a North Temperate Lakes Long-Term Ecological Research lake (64 ha; maximum depth 20 m) that has a nearly identical temperature profile as Pallette Lake (range of adjusted R² over the 10-year study: 0.87–0.99; P < 0.0001; Supplementary Figure S1 and Table S1).

These temperature data were used to compute degree days, a measure of the thermal energy experienced by individuals (Chezik et al. 2014a, b).  In particular, we used the sum of positive differences between the average daily water temperature at 1 m and the threshold T₀ = 10 °C for each day until the date that the average water temperature reached 15 ºC, the temperature at which seasonal reproductive activity is typically initiated, to evaluate the population-level reproduction response to temperature each year (Shuter et al. 1980; Ridgway et al. 1991; Lukas and Orth 1995). The sum of degree days each year from the first date on which the water temperature was favorable for energy recuperation (i.e., average exceeded 10 °C) until the date at which eggs were found in the nest of a male was also computed to produce a measure of the seasonal thermal energy experienced by each parental male prior to reproduction and evaluate individual responses to temperature (Shuter et al. 1980; Ridgway et al. 1991). In addition, we calculated the average and maximum temperature of the growth season for every year, defined as the period between the first day the average temperature climbed above 10 ºC and the last day before the average temperature dropped below 10 ºC, and the duration of winter, defined as the number of days between growth seasons (Shuter et al., 1980; Ridgway et al., 1991; Table 1). These latter measures of thermal energy were used to explore how temperature may control a size threshold at which males are released from energetic constraints on seasonal reproductive phenology.

Statistical analyses

Estimating body condition

Individual condition, often measured as some function of body mass, may provide an estimate of energetic reserves for individuals of a specified size and, hence, influence reproductive phenology (Kaufman et al. 2007, Schloesser and Fabrizio 2017). Individual condition was calculated as the residual of the relationship between weight (natural log [ln] transformed) and length (ln transformed) for each year, which has been shown to successfully predict energy content in a broad set of fish species (Kaufman et al. 2007, Schloesser and Fabrizio 2017).

Population-level responses to thermal energy

We used linear regression to predict annual population response times, the time interval between the date on which the mean water temperature reached 15 ºC and the median date on which eggs were found in nests, from the number of degree days accumulated before the first 15 ºC day each year.  If thermal energy controls individual reproductive readiness, then we expect the slope of this relationship to be negative (Shuter et al. 1980).

Individual behavioral response to thermal energy

We first used a linear mixed model structured similarly to those used in prior studies on smallmouth bass reproductive timing, where size was regressed on reproductive timing and year, to generate a comparable measure of the relationship between male total length (ln transformed) and the number of degree days (ln transformed, threshold T₀ = 10 °C) that had accumulated up to the date on which a male spawned (i.e., the date a nest was found to contain eggs) (Ridgway et al. 1991; Lukas and Orth, 1995). Individual identifiers were included as a random effect because 316 individuals bred in more than one season. If a temperature-dependent metabolic process is the main driver of reproductive timing, species-specific constants in allometric scaling laws predict the slopes of these relationships to be similar across studies (West et al. 1997). 

Next, to examine predictions related to the presence of an energetic threshold and to explore whether incorporating condition improves predictions of reproductive timing, we fit and competed a set of models predicting the degree days a male accumulated prior to reproduction. Alternative models included as predictors all combinations of condition, length (ln transformed), and length² (ln transformed), which allowed the effect of body size on reproductive timing to vary. Indicators for years and a random effect for individuals were included in all models. Models were tested with and without year interactions with condition or the two body length terms. Akaike information criterion (AIC) was used to choose between models (Akaike 1973). Some males spawned more than once within a season (N=127). In these cases, we restricted the analysis to the first attempt within a year. We excluded data from approximately one-third of males (34.6%) whose nests were discovered after eggs had hatched because timing of reproduction could not be definitively determined for these males. The mean size and range of sizes of these males were qualitatively very similar to those included in the study so we did not expect this to bias our results. If larger males experience less severe winter energy losses, then the slope of the relationship between male length and degree days accumulated before reproduction is expected to be negative in each year (Ridgway et al. 1991). If there is a size threshold above which males can escape energetic constraints on reproduction, we expected the quadratic length term to be included in the chosen model. Similar models including condition, length (ln transformed), and length² (ln transformed) as predictors of reproductive timing were conducted for each year independently to validate results of our comprehensive model (Supplement; Supplementary Table S3).

Post hoc linear models were conducted to test the hypothesis that in years where environmental conditions favored the accumulation of greater energetic reserves, energetic constraints on reproduction should be restricted to smaller individuals (Supplementary Table S4). Specifically, we investigated the relationship between the size threshold at which there was no relationship between size and degree days before reproduction in a given year–given by the untransformed body length of the parabolic vertex in the best regression models–and environmental variables that we hypothesized might influence the size threshold at which individuals are released from metabolic constraints (Supplementary Table S4). These environmental variables included the degree days accumulated before 15 ºC, the duration (days) of the winter, the duration (days) of the previous season, the number of degree days accumulated during the previous growth season, and the average temperature of the previous growth season.

All data files are in .csv format and can be loaded and viewed in RStudio or any common spreadsheet application such as Microsoft Excel. R file is best viewed in RStudio.