Data from: From kelp forests to turf reefs: patterns, drivers, and impacts on functional diversity
Data files
Mar 12, 2026 version files 18.99 MB
-
Biomass_21_22_23.csv
1.54 MB
-
Coordinates.csv
1.34 KB
-
cover_env_reg.csv
172.58 KB
-
fd.csv
2.66 KB
-
Heatwave_Timeseries_Temp.zip
1.49 MB
-
Percent_Cover_21_22_23.csv
1.21 MB
-
README.md
10.16 KB
-
Script1_Turf_Spread_MS_Temp_v1.ipynb
1.40 MB
-
Script2_Turf_Spread_MS_Percent_Cover_v4.ipynb
1.04 MB
-
Script3_Turf_Spread_MS_Diversity_v3.ipynb
8.65 MB
-
Script4_Turf_Spread_MS_SEM_v1.ipynb
3.47 MB
Jul 06, 2026 version files 16.33 MB
-
Biomass_21_22_23.csv
1.23 MB
-
Coordinates.csv
1.34 KB
-
cover_env_reg.csv
172.59 KB
-
fd.csv
2.66 KB
-
Heatwave_Timeseries_Temp.zip
1.49 MB
-
Percent_Cover_21_22_23.csv
1.21 MB
-
README.md
10.45 KB
-
Script1_Turf_Spread_MS_Temp_v2.ipynb
1.27 MB
-
Script2_Turf_Spread_MS_Percent_Cover_v5.ipynb
1.48 MB
-
Script3_Turf_Spread_MS_Diversity_v4.ipynb
5.90 MB
-
Script4_Turf_Spread_MS_SEM_v2.ipynb
3.55 MB
Abstract
Kelp forests are declining in many regions due to ocean warming, predator loss, and other anthropogenic stressors. In areas of rapid ocean warming, including the southern Gulf of Maine, these ecosystems have transitioned to a novel state dominated by low-lying mats of turf algae. However, the pace, drivers, and ecological consequences of this transition remain unclear. Here, we used field surveys from 32 sites over five years (2018–2023) to reveal a continuation of kelp forest collapse and northward expansion of turf algae across Maine’s coast. Next, we united data on benthic cover with key environmental variables in a structural equation model to show that turf algae were directly enhanced by higher ocean temperatures and decreased wave disturbance and indirectly enhanced by a warming-induced loss of kelp cover. Lastly, the shift from kelp to turf yielded a seaweed assemblage dominated by traits associated with rapid growth, high surface-area-to-volume ratios, and markedly reduced canopy height, indicating declines in habitat provisioning and carbon storage with kelp forest loss. Our findings highlight the accelerating impacts of climate change on temperate reef ecosystems and the vital services they provide. Further, we provide insights into a new state shift that is now occurring globally and underscore the need for urgent actions to mitigate further loss of foundational kelp forests.
Dataset DOI: 10.5061/dryad.rr4xgxdk2
Description of the data and file structure
Coordinates.csv
Description: Coordinates of sites and associated metadata.
Variables
- Site: name of each site.
- Subregion: name of each subregion (Casco Bay, Midcoast, Penobscot Bay, Mount Desert Island, Downeast).
- Lat: Latitude (decimal degrees).
- Long: Longitude (decimal degrees).
Biomass_21_22_23.csv
Description: This dataset contains measurements of seaweed biomass collected from 1m² quadrats across study sites over three years. (See “Macroalgal percent cover and biomass” in the methods section for detailed methods). Taxonomic designation for the species, as inferred from an unpublished version of the NEAS key (draft 2021viii20) by Gary Saunders and the associated website "http://seaweedcanada.wordpress.com".
* To note for this paper "Turf" communities (i.e., those comprised of red, green, and/or brown filamentous and uniseriate algae, ~1 cm to ~15 cm in canopy height, that form upright filaments, mats, or tufts) can include any of the following species: Acrosiphonia_spp, Antithamnion_spp, Antithamnionella_spp, Audouinella_spp, Bonnemaisonia_hamifera, Carradoriella_elongata, Ceramium_spp, Cladophora_spp, Dasysiphonia_japonica, Ectocarpus_spp, Kaprunia_schneideri, Leptosiphonia_spp, Melanothamnus_harveyi, Other, Polysiphonia_stricta, Pterothamnion_plumula, Red_tufts, Rhodomela_spp, Scagelia_pylaisaei, Spermothamnion_repens, Sphacelaria_spp, Stylonema_alsidii, Vertebrata_fucoides, Vertebrata_nigra, Vertebrata_spp
* Kelps include Agarum_clathratum, Alaria_esculenta, Laminaria_digitata, Saccharina_latissima, and Laminarian_juvenile. All other species are non-kelp canopy-forming species (Desmarestia_viridis and Desmarestia_aculeata) and understory species (all others).
Variables:
- Year: The year in which the survey took place (2021, 2022, or 2023).
- Month: The month (1–12) when the site was surveyed.
- Day: The day of the month the site was surveyed.
- Site: The site at which the survey took place.
- Meter_Mark – The specific meter mark along the transect where the survey took place (5, 10, 15, 20, 25, 30, 35, and 40). Surveys happened at 4-6 meter marks per site.
- name: The name of the organism species or complex identified (e.g., red_tufts, red_tubes). Other = indiscernible mixture of taxa that could not be teased apart.
- wet_weight_g: The weight of the seaweed in grams after being spun to remove excess water. 0 = true absence, 0.001 = present but below scale detection limit (.01g).
- phylum: The phylum the seaweed or complex belongs to. Other = indiscernible mixture of taxa that could not be teased apart.
- class: The class the seaweed or complex belongs to. Other = indiscernible mixture of taxa that could not be teased apart.
- order: The order the seaweed or complex belongs to. Other = indiscernible mixture of taxa that could not be teased apart.
- genus: The genus the seaweed or complex belongs to if identifiable. Other = indiscernible mixture of taxa that could not be teased apart.
- species: The species the seaweed or complex belongs to, if identifiable.
- survey_area_m2: The area the seaweed was surveyed in (1m² for large brown seaweed, 0.25m² for all other seaweeds).
- m2_weight_g: Standardized biomass per 1m². Since some seaweeds were surveyed in 0.25m², those values were multiplied by 4 to be comparable to 1m².
Percent_Cover_21_22_23.csv
Description: This dataset contains measurements of seaweed percent cover collected from quadrats across study sites over three years. (See “Macroalgal percent cover and biomass” in the methods section for detailed methods). Taxonomic designation for the species was inferred from an unpublished version of the NEAS key (draft 2021viii20) by Gary Saunders and the associated website "http://seaweedcanada.wordpress.com".
- Year: The year in which the survey took place (2021, 2022, or 2023).
- Month: The month (1–12) when the site was surveyed.
- Day: The day of the month the site was surveyed.
- Site: The site at which the survey took place.
- Meter_Mark: The specific meter mark along a 50 meter transect tape where the quadrat was placed. N=8 per sampling event.
- name: The name of the algae, invertebrate, substrate, or complex identified (e.g., red_tufts, red_tubes).
- PCT_Cov: The percent cover of the specific algae, invertebrate, substrate, or complex. Note percent cover can be over 100% due to the multilayered nature of the seaweed community.
- Subregion: name of each subregion (Casco Bay, Midcoast, Penobscot Bay, Mount Desert Island, Downeast).
Heatwave_Timeseries_Temp.zip
Description: This dataset contains satellite-derived sea surface temperature data (NASA's MUR SST) from all of our study sites from 2002-2024 (the full extent of the datasets records. (See “In situ and satellite environmental data” in the methods section for detailed methods). Each .csv file within the zip corresponds to a site and has the same layout.
Variables
- Time (UTC format): Year-Month-Day hours:Minutes:Seconds
- Latitude (Degrees_north): Latitude of the point
- Longitude (Degrees_east): Longitude of the point
- Analysed_SST (Degree_C): Sea Surface Temperature in Celcius.
cover_env_reg.csv
Description: This dataset contains biotic (kelp and turf percent cover; recorded at the quadrat-level) and environmental data (temperature and significant wave height; recorded at the site level). This dataset is used to model the direct and indirect drivers of turf abundance through space and time using a Structural Equation Model. Included in this dataset are group-mean centered metrics and sub-regional means for these predictors. (See “Statistical analysis: Identifying drivers of turf spread using SEM” in the methods section for detailed methods).
Variables
- year: Calendar year for each measurement.
- date: Measurement date at each site (MM/DD/YYYY).
- subRegion: Subregional grouping of sites used in the analysis.
- site – Individual sampling site within each subregion.
- meter_mark: The specific meter mark along a 50 meter transect tape where the quadrat was placed. N=8 per sampling event.
- kelp_cover: Percent cover of kelp at each site (%).
- turf_cover: Percent cover of turf algae at each site (%).
- mean_sst_1yr: Mean site sea surface temperature anomaly over the previous year (°C).
- dd_gt10: Degree-days above 10°C (°C·days) at a site over the previous year leading up to the survey.
- dd_gt15: Degree-days above 15°C (°C·days) at a site over the previous year leading up to the survey.
- dd_lt5: Degree-days below 5°C (°C·days) at a site over the previous year leading up to the survey.
- mean_wave_1yr: Site-level mean significant wave height anomaly over the past year (meters).
- mean_wave_30d: Site-level mean significant wave height anomaly over the past 30 days (meters).
- subregion_mean_sst: Subregional mean sea surface temperature (°C) the previous year leading up to the survey date.
- subregion_mean_wave: Subregional mean wave height (meters) the previous year leading up to the survey date.
- subregion_mean_kelp: Subregional mean kelp percent cover (%).
- subregion_mean_turf: Subregional mean turf percent cover (%).
- mean_sst_centered_reg: Site-level SST centered by the Subregional mean (°C).
- wave_1yr_centered_reg: Site-level annual wave height centered by Subregional mean (meters).
- wave_30d_centered_reg: Site-level 30-day wave height centered by Subregional mean (meters).
- kelp_centered_reg: Site-level kelp cover centered by the Subregional mean (%).
- turf_centered_reg: Site-level turf cover centered by the Subregional mean (%).
fd.csv
Description: This dataset contains species or complex-specific trait measurements for 7 different traits. (See “Seaweed functional trait measurements” in the methods section for detailed methods).
Variables
- Species: The name of the individual species/genus or complex
- TDMC: Avergae thallus Dry Matter Content (no units): obtained by dividing dry mass (g) by fresh mass (g)
- Thickness: Average thickness (mm) taken haphazardly along the blades of a sample and avoiding, when applicable, the midrib. Thickness measured via calipers when the alga was big enough or cross-section was done on small filamentour forms.
- Branching_order: Average number of divisions of the main axes of a macroalga from its holdfast to the tip of the blades taken haphazardly within the sample; no units.
- SA:V: Average surface Area to Volume ratio (mm2/ mL -1): obtained by dividing the area (mm2) of a sample by its volume (mL).
- STA: Average specific Thallus Area (mm2/g-1): obtained by dividing the area (mm2) of a sample by its dry mass (g).
- C:N: Average Carbon to Nitrogen ratio (no units): obtained by dividing Carbon content (%) by Nitrogen content (%).
- Maximum_length: Average maximum length (cm) of a sample, from the base of the holdfast (or any other anchoring system) to the tip of the longest blade.
Code/Software
Script1_Turf_Spread_MS_Temp_v2.ipynb
Description:
Code (Jupyter notebook file) which uses the long-term (2002-2024) temperature data. Using these data, we calculate the rise in average temperature and marine heatwave metrics over space and time.
Script2_Turf_Spread_MS_Percent_Cover_v5.ipynb
Description:
Code (Jupyter notebook file) which uses the Percent cover and biomass data. In this script we analyze turf/kelp cover trends over space and time (GLMMs) and use the biomass data to analyze turf taxonomic richness (GLMM) over space and time.
Script3_Turf_Spread_MS_Diversity_v4.ipynb
Description:
Code (Jupyter notebook file) which uses the biomass data and functional trait measurements. In this script, we calculate our Simpson's diversity index, functional diversity metrics, and community weighted means.
Script4_Turf_Spread_MS_SEM_v2.ipynb
Description:
Code (Jupyter notebook file) which uses the Percent cover and environmental (waves and temperature) data to construct a Piecewise SEM to identify the direct and indirect drivers of turf spread.
Changes after Mar 12, 2026: Edited README.md.
Edited README.md and updated CSV headers from region to subregion.
This edit was made after the manuscript was published. The only change was to replace the term “Region” with “Subregion” to reduce confusion for readers. In the manuscript, the Gulf of Maine is treated as the broader region, while Casco Bay, Midcoast, Penobscot Bay, Mount Desert Island, and Downeast are referred to as subregions. However, the underlying CSV files previously used the header region for these subregional groupings.
This update changes only the wording of the metadata and CSV headers. The underlying data, statistical analyses, results, and figures remain unchanged.