Historical kelp forests in California over multiple centuries
Data files
Jun 27, 2024 version files 127.58 MB
-
coding_themes_categories_oral_histories.csv
2.08 KB
-
contemporary_data_sources.csv
5.85 KB
-
enso_recovery_lag.csv
52.54 KB
-
extent_historical_map.gpkg
59.47 MB
-
historical_collections_searched.csv
1.51 KB
-
kelp_data_sources.csv
3.40 KB
-
kelp_historical_expanded_20230612.csv
36.81 KB
-
kelp_shallow_final_canopy_only.gpkg
67.44 MB
-
otter_historical_location3.csv
53.94 KB
-
otter_historical_long_20230702.csv
70.27 KB
-
otter_kelp_oh_subset_long.csv
254.15 KB
-
pycno_contemporary_density_post-SSWD.csv
143 B
-
pycno_contemporary_density_pre-SSWD.csv
140 B
-
pycnopodia_historical_long_20240308.csv
51.11 KB
-
README_Selgrath_et_al_Historical_Kelp_Forests.txt
1.94 KB
-
README.md
6.55 KB
-
respondentsYear.csv
829 B
-
urchins_data_micheli_2002.csv
2 KB
-
urchins_data_published.csv
72.50 KB
-
urchins_historical_20240308b.csv
56.45 KB
Abstract
Kelp forests have deteriorated globally due to anthropogenic stressors. There is an urgent need to extend baselines, to understand the processes that underlie the persistence and recovery of kelp forests, and to distinguish the normal range of ecosystem variability from more extreme changes. Using a mixed-method, historical ecology approach we integrate archival data, oral histories, and contemporary ecological data to examine the dynamics of kelp forests over a multi-decadal to multi-century time period in central California. We focus on sea otters, sunflower seastars, sea urchins, kelp cover, kelp species dynamics, and climate. From 1826 to 2020 kelp was highly variable. There were seven periods of low kelp cover and two periods of exceptionally low kelp cover (1896-1899; 2014-2016) following El Niño-Southern Oscillations (ENSOs). Exceptionally low kelp cover did not occur when two predators – seastars and sea otters – were present. In all cases, kelp recovered following times of extremely low cover, with a lag, which was extended by the duration of warm water anomalies. We present the concept of an ENSO Recovery Lag - a metric indicating the time it takes for kelp to recover following ENSO events. Kelp remained low for approximately two years following 80% of ENSOs. The greatest kelp decline (12-fold) was in Santa Cruz (northern Monterey Bay). Herbivore populations (sea urchins) were highly variable over the past century and exhibited short and long-term changes in abundance. Sunflower seastars were present in low, stable abundances prior to seastar wasting disease (1938-2013 mean density: 0.02/m2) when they declined by 97.5%. Insights from this reconstruction indicate that kelp recovery following extended warm water anomalies exhibits a lag, and occurs over multiple years.
https://doi.org/10.5061/dryad.xpnvx0khq
For this research, we integrated hundreds of datasets to look at the historical changes in kelp forests through time. The data sets we considered ranged from 1602 to 2020.
FILE | TYPE | DESCRIPTION | YEARS |
---|---|---|---|
coding_themes_categories_oral_histories.csv | analysis | themes used to code oral history data | NA |
contemporary_data_sources.csv | sources | contemporary ecological and ENSO data sources | 1770-2020 |
enso_recovery_lag.csv | dataset | data about the ENSO recovery lag. This metric integrates published sources about the intensity of El Nino Southern Oscillations and Marine Heat Waves in central California | 1800-2020 |
extent_historical_map.gpkg | gis | extent of historical maps in specific years | 1852-2016 |
historical_collections_searched.csv | sources | collections searched for data | NA |
historical_organism_abundances.csv | dataset | values from archival sources documenting the presence and abundance of kelp, sea otters, sunflower seastars, and urchins in central California | 1602-2017 |
kelp_data_sources.pdf | sources | historical and contemporary maps used to determine the spatial extent of kelp | 1856-2016 |
kelp_historical_expanded_20230612.csv | dataset | historical and descriptive data about the abundance of canopy forming kelp | 1774-1967 |
kelp_shallow_final_canopy_only.gpkg | gis | kelp canopyfrom all sources in years under study | 1852-2016 |
otter_historical_location3.csv | dataset | historical and descriptive data about the abundance of sea otters | 1815-1914 |
otter_historical_long_20230702.csv | dataset | historical and descriptive data about the abundance of sea otters | 1602-1929 |
otter_kelp_oh_subset_long.csv | dataset | oral history observations about kelp forests in central California | 1939-2020 |
pycno_contemporary_density_post-SSWD.csv | summary statistics | density estimates for sunflower seastars after sea star wasting disease (2013) | 2013-2020 |
pycno_contemporary_density_pre-SSWD.csv | summary statistics | density estimates for sunflower seastars before sea star wasting disease (2013) | pre-2013 |
pycnopodia_historical_long_20240308.csv | dataset | historical data about the abundance of sunflower seastars (Pycnopodia) | 1835-1978 |
qualitative_abundance_species.csv | analysis | examples of qualitative descriptions of species abundances and their assigned relative abundance values | NA |
respondentsYear.csv | dataset | number of oral history respondents in each year | 1939-2020 |
urchins_data_micheli_2002.csv | dataset | survey data of sea urchin abundances | 2002-2003 |
urchins_data_published.csv | dataset | published sea urchin densities | 1964-1989 |
urchins_historical_20240308b.csv | dataset | historical and descriptive data about the abundance of purple and red sea urchins | 1854-1961 |
Sharing/Access information
Links and/or references to original datasets and sources can be found in: \
+ contemporary_data_sources.csv\
+ kelp_data_sources.csv\
+ historical_organism_abundances.csv\
+ are referenced in other .csv files
Code/Software
Code is available on GitHub and Zenodo (links in the related works section).
The driver.R file provides the overview of the inputs, outputs, and goals of each code chunk.
Study Area
Our research focused on kelp forest ecosystems in central California, USA (here, Año Nuevo (37°07'35.4"N, 122°19'35.4"W) in the north to Big Sur (36°06'16.2"N, 121°37'19.8"W) in the south. This cold water, upwelling-driven ecosystem is characterized by high biodiversity. The social-ecological history of central California has been documented for over two hundred years by historical and contemporary sources due to the region’s history of Tribal nations, exploration, European colonization, maritime trade, and Western science. We divided central California into three regions, north to south: Santa Cruz, Monterey Peninsula, and Big Sur.
Data Sources
To identify historical baselines and longitudinal dynamics of kelp forest ecosystems, we assembled a multi-decadal to multi-century dataset that integrated historical and contemporary information focused on central California largely from the early 1800s onward. We evaluate trends in the relative abundance of sea otters (Enhydra lutris nereis; hereafter ‘otters’), predatory sunflower seastars (Pycnopodia helianthoides), herbivorous purple urchins (Strongylocentrotus purpuratus, hereafter ‘urchins’), and two species of canopy forming kelp (giant kelp Macrocystis pyrifera and bull kelp Nereocystis luetkeana; hereafter ‘kelp’). We did not include red urchins (Mesocentrotus franciscanus) in formal analyses over historical time periods because there was insufficient information about their abundances over a multi-decadal period. Instead, we qualitatively discuss changes in their abundances.
Qualitative and quantitative data sources include narrative accounts (1602 onward), archival maps (1852-1934), 48 oral histories (1939-2020), ENSO reconstructions (1820-2020), kelp harvest records (1931-1999), and ecological surveys (1985-2020). We use archeological records and Indigenous knowledge as reference where available, but focus on integrating sources onward from the exploration and colonization of California by Europeans. Sources for these datasets are avaliable in the article's supplementary materials.
Archival Documents
Diverse historical accounts provide early narrative descriptions of the social-ecological system of central California. We searched collections that focused on west coast history for accounts of early explorers, traders, and religions figures who were active in central California. Next, we searched early natural history and scientific books, field notes, newspaper articles, historical art, and reports that provided accounts of natural history and environmental conditions. Since historical documents are not indexed for ecological terms, historical sources were identified using the expert knowledge of librarians at Stanford University’s Hopkins Marine Station (Pacific Grove, California) and authors (JP, JTC, TT), and were supplemented with searches for historical literature referencing the three regions and four focal species. We reviewed identified documents to assess if they contained descriptions of natural history.
We considered four types of abundance values: (1) presence only; (2) relative abundance; (3) quantitative abundance; and for kelp, (4) percent cover (methods described in the Kelp Map sections below) (Table S7). First, when qualitative descriptions of species’ existence could not be translated to relative abundances, we considered these records to represent presence-only data. We used presence-only data to triangulate against other data sources and used the information in qualitative assessments of change. Second, we developed a standardized rubric for translating qualitative descriptions of abundance to relative abundance values (scale: 0-5). Third, for quantitative abundance records (e.g., ecological transects, published data) we assigned densities to relative abundances (range: 0-5) based on the mean and standard error values for all records of that species (Table S4). When data were only available in graphs, we estimated the density values from the figures. We considered relative abundances derived from both qualitative and quantitative sources to be consistent within species, but not comparable across species.
El Niño-Southern Oscillations and Marine Heat Waves
To assess long term ENSO and MHW patterns, we used historical ENSO time series (1770 – 1983) and continued the historical series with contemporary MEI (multivariate ENSO index) estimates (1984 – 2020) and MHW records (Quinn et al. 1987; Gergis and Fowler 2009; NOAA 2020). We considered that a year was influenced by an ENSO when the year had a MEI 2.0 value >1.5, or when the ENSO was classified as Medium + or greater by historical analyses (Quinn et al. 1987; Gergis and Fowler 2009; NOAA 2020).
Sea Surface Temperature Records
To assess changes in SST we used daily sea surface temperature measured at Hopkins Marine Station since 1919 (Breaker and Miller 2023).
Kelp Maps
Kelp Maps: Archival Sources.
We used a total of seven archival maps (five nautical charts and two kelp surveys), which we considered to represent presence-only data (Costa et al. 2020). Nautical charts which documented kelp were surveyed from 1856 to 1934 (Fig. 1; Table S6). The US Department of Agriculture conducted two surveys of kelp (1911, 1912), which included maps with species-level resolution (Table S6). For all archival maps, we documented the map name, number, scale, year of the last survey, survey date (where available), and first year of publishing (Table S6). For charts published in multiple years, we only considered the first edition.
To extract kelp data from archival maps, we first photographed paper maps and downloaded maps that had been previously scanned by libraries and NOAA. Second, we georectified digital maps using satellite images and control points from stable features (e.g., rock outcrops) to reduce root-mean-square error (ArcGIS 10.7.1) (Costa et al. 2020). Third, we demarcated kelp areas using on-screen digitization. We traced the edges of the kelp symbology to create polygons depicting the area of kelp beds (Fig S2). Where species information was available, we assigned kelp bed polygons to one of three groups: giant kelp (Macrocystis); bull kelp (Nereocystis); or mixed beds (similar amounts of Macrocystis and Nereocystis).
Kelp Maps: Contemporary Sources.
The contemporary spatial extent of canopy forming kelp was documented by the California Department of Fish and Game (now California Department of Fish and Wildlife, CDFW) during a series of aerial kelp surveys (1989–2016), available as shapefiles (Table S6; Movie S1). When both canopy and subsurface canopy data were available in later years (2008-2016) we restricted maps to canopy to improve data consistency through time (R. F. Miller, CDFW, pers. comm., 2020). We attempted to find earlier aerial photos taken by CDFW (e.g., 1972-1977), but the agency no longer maintained these records (R. F. Miller, CDFW, pers. comm., 2020). We documented methods and survey months when such data were available in shapefile metadata.
Kelp Maps: Estimates of Error.
To assess the locational accuracy of the archival maps we set a reliability criterion that considered the local depth in which kelp was located. We used a depth threshold of 40 m to account for known depth ranges, kelp movement with currents and tides, residual processing errors, and cartographic variability (e.g., line thickness) (Costa et al. 2020). We overlaid the polygons on a 10m resolution bathymetry map of California. We classified map reliability based on the percentage of kelp that was mapped in shallow (< 40m) vs deep (> 40m) areas (four classes: Very High (>99% of kelp mapped in shallow areas), High (>80%), Medium (>60% mapped in shallow areas), and Low (>40% mapped in shallow areas)).
Kelp Maps: Estimates of Kelp Cover.
To account for the fact that different extents were mapped across years, we restricted our analysis of kelp canopy cover to areas where the kelp canopy was documented by archival maps for a minimum of three years (Fig. 1).
We calculated kelp area as the total area where kelp-forming canopy was documented in one year (t) in the spatial extent covered by the mapped area (m) in each Region (r),
(i) Kelp Areatmr = Total area of canopy-forming kelp in time (t) in the mapped area (m) in each region (r)
To standardize estimates of kelp area, we next identified the maximum area of kelp canopy that was ever mapped in any year (t1-ti) using any method for each region (r),
(ii) Maximum Area t1-ti r = maximum area of kelp mapped across all years (t1-ti) in each region (r)
Finally, for each map, we calculated the proportion of maximum area in time (t) as the Kelp Area in time (t) divided by the maximum area of the kelp canopy that was ever mapped in a region (r),
(iii) Proportion of Maximum Areat = Kelp Areatmr / Maximum Area t1-ti
We used the proportion of maximum area as the standardized measure of kelp cover in all quantitative analyses. Additionally, we calculated a second standardized kelp cover metric as the maximum extent to which kelp occurred in any year (Maximum Extent) within the mapped area (m). This yielded similar trends, and we thus proceeded with the method described above.
Oral Histories
To document changes in coastal marine species and conditions of the coastal ecosystem we conducted semi-structured oral history interviews (n = 48 respondents). We used snowball sampling to select respondents who had first-hand experience with the marine biota and ocean conditions of the central Californian coast. Initial respondents were identified through personal networks of the authors (JP, JC, JTC, TT and FM) who had worked in central California for decades. We included individuals who fell into two categories of ocean-knowledge: people who had worked in the region for (a) a long-term period (e.g., scientists, historians based in the region) or (b) a finite period (e.g., former ocean-focused graduate students). We obtained informed consent from all participants. Although several respondents had not lived in central California for many years, the ocean in the region made a strong impression on their memories. Most respondents (54%) resided along the central California coast for less than ten years (duration: range=3-54 years; median=8 years), placing their memories within discrete date ranges.
We conducted interviews both in person and virtually through Zoom (stanford.zoom.us). We also received written survey responses from six individuals. Interview and survey data (hereafter ‘oral history data’) were transcribed and respondents were provided the opportunity to make corrections. Final data were thematically coded using grounded theory. We used NVivo and documented the presence, abundance, and location of species, including observations of changes over time (Table S8). Oral history data also included distribution information regarding both species of canopy-forming kelp, Macrocystis and Nereocystis. Species density estimates from oral histories were typically qualitative (e.g., ‘a lot’; ‘hardly any’). Following the methods for historical data (see above), we standardized responses by assigning specific descriptions to different abundance levels (range: 0-5) (Table S3). For kelp records, we weighted observations from three kelp experts higher than other respondents, which allowed us to capture their specific experiences with ENSO dynamics that were not mentioned by other respondents. Abundance levels were relative to each species. All interviews were reviewed for accuracy by JP and JCS.
Contemporary Ecological Data
We compiled contemporary ecological data (sea otters, purple urchins, sunflower seastars, kelp cover, kelp species) from several sources: USGS otter surveys (1985-2019), unpublished field data (2002, F.M.), published field data (various), PISCO database (1999-2020), ReefCheck database (2006-2019), CDFW kelp harvest data (1916-2001), and CDFW aerial kelp surveys (2003-2016) (sample sizes in supplement). To facilitate data comparison, we converted density data (e.g., purple urchins per m2) to relative abundance values (range: 0-5). For kelp maps we used equal intervals to set relative abundances. For contemporary datasets, we set relative abundances based on the mean and standard deviation of survey values for each species considered. We then integrated data from archival sources, oral histories, and contemporary ecological surveys.