To study patterns of local adaptation to latitudinal temperature gradients in the ocean, we conducted two common garden experiments on the Chilean silverside (Odontesthes regia). Wild adult spawners from four locations (20 – 42°S) along the Chilean coast were used to produce offspring that was then reared in triplicates from fertilization to approximately 35 mm total lengths (TL) under a set of four common temperatures in the laboratory (11-23°C) and ad libitum feeding conditions. Larval and juvenile TL were measured via calibrated digital pictures (intermediate samples) or via calipers (final samples). Subsequently, the data will allow calculating population, temperature, and life-stage-specific growth capacities (i.e., TL growth rates at excess feeding conditions). This will reveal whether higher latitude populations have evolved genetically faster growth capacities than lower latitude populations – as has so far been shown only for northern hemisphere fishes. In addition, adults collected from each location were x-rayed to count the vertebrae of each specimen and subsequently test whether lower latitude populations have fewer vertebrae on average than their higher latitude conspecifics – a phenomenon known as Jordan’s Rule. This dataset includes measurements of total length (TL) and wet weight (wW), developmental stage (Stage), individual ID, and key dates including fertilization, main hatching, and sampling. It also contains population information (Species, Population, Fert_num), collection site metadata (Longitude, Latitude), and specific conditions of the common garden experiment, such as temperature treatment (Temp_treat), replicate number, and bucket ID. The supplemental tables provide additional data including adult total length and collection metadata (Table 1), vertebral counts from x-ray images and associated metadata (Table 2), and high-resolution temperature records from each experimental treatment tank throughout the common garden experiment (Table 3).

Table of Contents

• Coverage
• Dataset Description
  • Methods & Sampling
  • Data Processing Description
  • BCO-DMO Processing Description
  • Problem Description
• Data Files
• Supplemental Files
• Related Publications
• Parameters
• Instruments
• Project Information
• Funding

Field sampling of spawners, offspring production, transport 

At each of four locations along the Chilean coast (north to south: Iquique [IQ, 20°S], Caleta Sierra [CS, 31°S], Dichato [DI, 37°S], Puerto Montt [PM, 41°S]), flowing-ripe adults of the Chilean silverside (Odontesthes regia) (urn:lsid:marinespecies.org:taxname:281830) were obtained during the species’ spawning season in austral spring (September-October) in 2023 and 2025. With the help of local collaborators and/or local artisanal fishermen, adults were caught during night or early morning hours with gill nets of various lengths (2 cm stretched mesh). Fish were separated by sex and kept alive for up to one hour, before they were strip-spawned into plastic dishes filled with clean seawater. At least 20 spawners per sex and population were used if available. In 2023, the northernmost site of Iquique was sampled twice; we denoted this in the data as fertilization 1 and fertilization 2. Fertilized embryos were transported in cooler boxes or thermos containers to the rearing facility in Dichato via car (PM, DI, CS) or airplane (IQ). Upon arrival at the rearing facility no more than 24h post-fertilization, they were randomly distributed among temperature treatments and replicate rearing containers.

Common garden experiment

In both years, offspring were reared at the Dichato Marine Research Station; in 2023 the experimental setup was located at the INCAR facility (Centro Interdisciplinario para la Investigación Acuícola), whereas in 2025 the setup was housed at the Experimental Marine Biology Laboratory (aka ‘Acuario’) of the Universidad de Concepción. We used four 900-1200L recirculating tanks representing four temperature treatments (2023: 11, 14, 18, 23°C; 2025: 14, 17, 20, 23°C). Each tank housed up to 12 rearing containers that represented the 3 replicates for each of the 4 populations (4 temperatures × 4 populations × 3 replicates = 48 containers). The container design was identical to previous silverside experiments (e.g., Baumann & Conover 2011, Murray et al. 2014), i.e., we used round, white 20L containers equipped with individual airlines and mesh-screened (150 µm) holes, which guaranteed ~100% oxygenation and water exchange with the surrounding tank while retaining food. Tank water of ~ 30 psu was drawn from Coliumo Bay, filtered/UV-sterilized, and controlled for temperature via thermostats connected to commercial aquarium heaters or chillers. Each tank was equipped with a HOBO® Pendant MX2201 temperature logger (Onset) that recorded conditions every 30 minutes in 2023 and every 15 minutes in 2025. For biofiltration we used 4 large FX4-250 (Fluval) canister filters. The photoperiod was 15h light: 9h dark throughout the entire experiment.

Starting at hatch, silverside larvae were fed ad libitum (=excess) rations of newly hatched brine shrimp nauplii (Artemia salina, San Francisco strain, Brine Shrimp Direct) produced daily on site. Ad libitum rations were ensured by never allowing rearing containers to become nauplii depleted. 1-3 days post hatch, the number of fish in each rearing container was equalized to 130 to avoid density-dependent growth effects. Another standardization to n = 80 per container occurred when larvae reached approximately 15 mm total length (TL), and a final standardization to n = 40 occurred when larvae reach approximately 25 mm TL. When larvae reached approximately 35 mm TL, the rearing concluded and all survivors were sampled.

Samples

All collected adults – both spawners and non-spawners - were measured for TL and then preserved frozen (-20°C) for later vertebral number analysis. Experimental samples were first taken on the temperature- and population-specific day of peak hatch, when up to 20 hatchlings were preserved in 95% ethanol (0-hatch). The first and second samples were taken at an average TL of approximately 15 mm TL (1-15mm, larvae) and 25 mm TL (2-25mm, early juveniles) and preserved in 95% ethanol as well. All individuals from these intermediate samples were measured for TL using the open-source software ImageJ and digital pictures taken with an iPhone 11 in front of a gridded (7mm grid size) white background. The 95% ethanol was replaced 24h after initial fixation. The final sample was taken when individuals on average had reached approximately 35 mm TL (3-35mm). In 2023, survivors were measured for TL using calipers (nearest 0.1 mm) and blotted wet weight (nearest mg) and then preserved frozen (-20°C) in individual plastic bags. In 2025, final samples were only measured for TL via digital pictures in ImageJ 1.53a and preserved in 95% ethanol (same as intermediate samples).

Vertebral number

The number of vertebrae was determined for all adult fish, regardless of whether they were used to produce offspring or not. Fish were X-rayed at the Oceanside Animal Hospital in Sandwich, MA, using a Vet Ray veterinarian X-ray system (SEDECAL, model A6504-25) and settings that were slightly modified from those recommended for small exotic pets (lizards). Specifically, we used a voltage of 60 kVp, power of 320 mA, and a shutter speed of 16 milliseconds. Digital x-ray pictures were analyzed using the multipoint tool in ImageJ to mark each vertebra between but excluding the basioccipital and the urostyle. In 2025, additional populations from Chimbote and Ancón in Peru were x-rayed by collaborator Victor Aramaya from the Universidad Nacional Mayor de San Marcos, Facultad de Ciencias Biológicas, Lima, Peru.

Outlier removal. After data compilation, we first removed stunted or malformed fish via outlier analysis. For that, we computed the mean (TLmean) and standard deviation (TLSD) of TL for each population (IQ, CS, DI, PM), temperature (11, 14, 17, 18, 20, 23°C), replicate (1, 2, 3), and sample type (hatch, 1-15mm, 2-25mm, 3-35mm), then removed all specimens with a TL < TLmean – 2 × TLSD from all further analyses.

Growth capacities. We computed 3 different growth capacities, i.e., TL growth rates (mm per day) at excess feeding conditions; GR1 refers to the growth rate of each sampled individual since hatch, while GR2 describes growth since the 15mm subsample, and GR3 describes growth since the 25 mm subsample. Therefore:

GR1ijkl = (TLijkl – TL_Hkl) / dphkl

GR2ijkl = (TLijkl – TL_15jkl) / dp15kl

GR3ijkl = (TLijkl – TL_25jkl) / dp25kl

For the ith individual of the jth replicate of the kth temperature of the lth population, where TL_H is the mean TL at hatch, TL_15 is the mean TL at the 15mm subsample, and TL_25 is the mean TL at the 25mm subsample. The length differentials were then divided by the number of days post hatch (dph) or the number of days since the 15mm subsample (dp15) or the number of days since the 25mm subsample (dp25mm).

Actually experienced temperature conditions. Logged temperatures were first averaged per day, month, year and then used to calculate actually experienced temperature means during each stage (i.e., hatch-15mm, 15-25mm, 25-35mm) for each population and temperature-treatment.

**Version 1**

- Imported NSF_OCE 2313288 - BCO-DMO source data.xlsx, sheet indexes 5, 4, 3, 2, and 1, into the BCO-DMO system
- Converted all dates to YYYY-MM-DD ISO format
- Converted lat and lon to decimal degrees format
- Rounded lat and lon values to 4 decimal places
- Renamed fields to remove spaces and units from parameters to comply with BCO-DMO system and style requirements
- Removed Month, Day, and Year from sheet 4, to avoid duplication
- Exported files as 956677_v1_common_garden_exp.csv (main file), adult_total_length.csv (supplemental file), adult_vertebral_number.csv (supplemental file), and experimental_temperatures.csv (supplemental file)

Accepted species identifier confirmed on 2025-04-05.

**Version 2**

- Loaded sheet 5 from "NSF_OCE 2313288 - BCO-DMO source data 2023-25.xlsx"; missing values flagged as "" and "nd"
- Converted "Fertilization date", "Hatch date", and "Sample date" fields from "%d-%b-%y" format to ISO "%Y-%m-%d" date format
- Converted Longitude (forced West) and Latitude (forced South) from degrees-minutes-seconds to decimal degrees; rounded both to 4 decimal places
- Rounded "Longitude" and "Latitude" to 4 decimal places
- Renamed fields to remove spaces and special characters
- Loaded sheet 2 from same Excel file as resource "adult_total_length"; blank headers ignored; missing values flagged as "" and "nd"
- Converted "Sampling date" from "%m/%d/%y" format to ISO "%Y-%m-%d" date format
- Converted Longitude (forced West) and Latitude (forced South) from degrees-minutes-seconds to decimal degrees; rounded both to 4 decimal places
- Loaded sheet 3 from same Excel file as resource "adult_vertebral_number"; missing values flagged as "" and "nd"
- Converted Longitude (forced West) and Latitude (forced South) from degrees-minutes-seconds to decimal degrees; rounded both to 4 decimal places
- Converted "Sampling date" from "%m/%d/%y" format to ISO "%Y-%m-%d" date format (UTC)
- Loaded sheet 4 from same Excel file as resource "experimental_temperatures"; blank headers ignored; missing values flagged as "" and "nd"
- Converted "datetime" field from "%m/%d/%y %H:%M" format to ISO "%Y-%m-%dT%H:%M" datetime format in Chile/Continental timezone, output to new field "ISO_DateTime_Local"
- Deleted fields "Month", "Day", "Year", and "datetime" from "experimental_temperatures"
- Output resources saved as: 956677_v2_common_garden_exp.csv, adult_total_length.csv, adult_vertebral_number.csv, and experimental_temperatures.csv

Accepted species identifier confirmed on 2026-03-17.

**V2 Dataset changes**

-New 2025 values added to dataset
-The Methods section was updated to include the latest protocols
-Information about data gaps was added
-The Data Processing section was updated with information about outlier removal

Primary Datafile:
-The parameter PopLabel was previously used to denote multiple sampling events at the same site. This is now represented with the column "Fert_num" (Fert#)
-The parameter Temp is now "Temp_treat"; mainhatchday is now "Hatch_date"; sampleday is now "Sample_date"; ID is now "Ind"; Sample is now "Stage"
-The parameter BucketID is now Bucket_ID and is formatted with a B in front of the number
-Additional columns include Year, Tank, GR1_from_hatch, GR2_from_15MM, GR3_from_25MM, T1_fromH, T2_from15MM, T3_from25MM

adult_total_length.csv:
Total length is now represented in Millimeters

experimental_temperatures.csv:
Log ID was added

Poor fertilization success for PM and insufficient spawners for CS required reduced replication

The malfunction of chillers on 8 November 2023 and 12 December 2023 required the premature termination of the 11 and 14C treatments, respectively. Hence, final samples are not available for all populations for these temperatures.

Data Gaps and Issues:

In 2023, the offspring for ‘Caleta Sierra (31°S)’ was produced from a single female strip-spawned with 5 males, while in 2025, no spawning-ripe fish were caught at all at the site. In both years, the majority of ‘Puerto Montt (41°S)’ adults were not fully spawning ripe and fertilization success was very low, resulting in lower replicate and stocking numbers. In both years, the experiments had to be terminated before Christmas (12/20/2023; 12/22/2025), when fish in the coldest treatments had not yet reached 35 mm TL. In 2023, ambient temperatures at the INCAR facility exceeded the capacity of the chillers to maintain 11°C on November 8th and 14°C on December 13th, resulting in the termination of these temperature treatments.

Website: https://befel.marinesciences.uconn.edu/expanding-the-silverside-system

Coverage: laboratory-based

NSF Award Abstract:
Many species have evolved adaptations to latitudinal climate gradients and studying these sheds light on how species will evolve in response to global climate change. To investigate adaptation in Chilean silversides, offspring of wild fish from four locations along the Chilean coast are being reared at four common temperatures. Differences in growth rates, vertebral number, and mercury uptake among populations from different latitudes indicate genetic differences due to local adaptation. The research team is integrating these Chilean silverside data with prior data from northern hemisphere silverside species to better understand the relationship between climate gradient strength and adaptation strength. This project provides training for two graduate students through a US-Chile graduate cross-cultural exchange. The research is being integrated into graduate and undergraduate courses taught in Chile and the research team is sharing the results with the public through a website and magazine articles.

This project advances the understanding of two forms of local adaptation, co-gradient variation (CoGV) and counter-gradient variation (CnGV), which underlie adaptation to large-scale, latitudinal climate gradients. Using a common garden experiment, the team is examining the relationship between temperature and genetic variation in Chilean silversides. Newly-fertilized offspring obtained from wild founders at four locations along the Chilean coast are being reared at four common temperatures to a common juvenile size. Differences among populations in trait measurements (growth capacity, vertebral number, and total mercury concentration) to test the hypotheses that 1) growth capacity increases with temperature and latitude (CnGV) and 2) vertebral number increases with latitude (CoGV). These data are being integrated with existing evidence from northern hemisphere silverside species to determine if there is a relationship between CnGV/CoGV strength and latitudinal gradient strength.