Eelgrass surveys were conducted along the northern Pacific coast of Baja California in the three major wintering areas of black brant: Bahia San Quintin (BSQ), Laguna Ojo de Liebre (LOL), and Laguna San Ignacio (LSI; fig. 1). These three embayments combined contain the largest beds of eelgrass in the southern portion of its range in the eastern north Pacific (Wyllie-Echeverria and Ackerman, 2003) and support about 30 percent of the entire Pacific flyway brant population each winter (Olson, 2019; Palacios and Ávila, 2019). The climate of the region is arid with warm average annual air temperatures (15–27 degrees Celsius [^oC]) and low average annual precipitation (<15 centimeters [cm]; Phleger and Ewing, 1962; Ibarra-Obando and others, 2001; Cabello-Pasini and others, 2003).

BSQ is the smallest (4,900 hectares [ha]) and shallowest (maximum water depth of about 8 meters [m] at the mouth of the bay) of the three embayments (fig.1). The bay has undergone considerable alterations in support of agriculture, including a dike that extends almost completely across the bay and hundreds of racks and containers used for oyster and other bivalve farming (Ward and others, 2003). Additionally, much of the surrounding land has been plowed for farming, leading to an increase in sediment loading into the bay (Ward and others, 2003). The most common human activity in the bay is boating associated with waterfowl hunting and mariculture.

LOL and LSI are large (58 and 25 kilometers squared [km²], respectively), hypersaline, deeper water (maximum water depths of about 25 m deep) embayments (Ward and others, 2022 situated in the Vizcaino Biosphere Reserve, the largest federally designated protected area in Mexico (fig. 1). They are the principle calving and nursery areas for the northeastern Pacific population of gray whale (Eschrichtius robustus) and important nursery areas for harbor seals (Phoca vitulina) and California sea lions (Zalophus californianus; Rice and others, 1981). LOL and LSI are influenced by warm equatorial surface waters and experience higher sea and air temperatures, lower precipitation, and higher evaporation rates than in BSQ (Cabello-Pasini and others, 2003; Winant and de Velasco, 2003).

LOL has undergone considerable human alteration in support of one the largest salt producing operations in the world (Eberhardt, 1966). Large salt pannes and shrublands adjacent to the lagoon were converted to evaporation ponds or filled to construct buildings and roads. Smaller areas inside the lagoon (channels, salt marshes) were dredged to allow passage of transport barges or filled to accommodate roads and a shipping dock. Guerrero Negro, a town of about 14,000 residents, is located less than 2 km away from the lagoon. Although located in a biosphere reserve, LOL receives considerable human activity primarily from boats associated with transporting salt, fishing, clamming, and whale-watching.

LSI is the least altered of the three embayments and has had historically few permanent residents along its shores. A small community of about 100 people reside year-round at the lagoon. The most common activities in this lagoon are boating associated with fishing, whale-watching, and oyster-farming.

The abundance of eelgrass and widgeongrass (Ruppia maritima, a less dominant seagrass that also provides food for brant in Mexico; Ward and others, 2005; Ward and others, 2022) were conducted over a 9–19-day period in late autumn and early winter of 1998, 1999, 2005, and 2012 (table 1). In 1998 and 1999, a point sampling approach was used following a systematic random design, where points were distributed evenly across each of the embayments (BSQ: 500 m spacing between points; LOL and LSI: 1 km spacing between points; fig. 2). The survey area was restricted to points located within high intertidal zone (+1.5 m mean lower low water [MLLW]; upper limit of widgeongrass in the bay excluding salt marsh) to the shallow subtidal (–2.5 m MLLW; deepest depth reachable by snorkeling). This survey area encompassed the majority of the spatial extent of eelgrass (75–80 percent in LOL/LSI and >90 percent in BSQ; Ward and others, 2004; Ward and others, 2022 and all intertidal beds, which are the most critical to brant and at risk to impacts of climate warming (for example, desiccation and disease; Short and Neckles, 1999). In 2012 surveys were restricted to only BSQ and previously identified eelgrass areas within this bay because of time and logistical constraints. Thus, there was only one site estimate (BSQ) for eelgrass abundance in 2012.

Points were located by boat and sampled by snorkeling primarily during high tide. At each point, water depth was measured with a tape measure attached to a pole, and percent cover of eelgrass, widgeongrass and all macroalgae (seaweeds) species combined were estimated within four, 0.25-m² quadrats. We also estimated the distance a 20-cm diameter Secchi disk disappeared from the surface when lowered into the water. If eelgrass or widgeongrass was present at a point, five representative plants were collected from each quadrat and measured for shoot length. Water depth (m) measurements were standardized to MLLW by subtracting the measured water depth from the closest publicly available tidal gauges with consideration to observed offsets in the timing of tidal curves. Tidal gauge measurements at San Diego (Broadway), California were used for BSQ with no offset except in the back part of the bay behind the dike (1 hour delay). Tidal gauge measurements at San Carlos, Baja California were used for the other two lagoons with offsets of 0–3 hours in LOL and 0–2 hours in LSI, depending on the location in each lagoon (in other words, no delay at the mouth of lagoon, 1–2 hour-delay mid-lagoon, and 2–3 hour-delay at the back of the lagoon).

A rapid, visual estimation technique was used to assess abundance of seagrasses and seaweeds. This method is quick yet robust and highly repeatable, thereby minimizing among observer differences in estimates of percent cover of seagrasses and seaweeds (Kenworthy and others, 1993). The cover of seagrasses and seaweeds present in a quadrat was assigned a Braun-Blanquet (BB; Braun-Blanquet, 1972) visual cover score (BB score of 0 percent=0; solitary shoot=0.1; 2–5 shoots=0.5; 1–5 percent=1; 6–25 percent=2; 26–50 percent=3; 51–75 percent=4; 76–100 percent=5). From these cover estimates three statistics were computed for eelgrass, widgeongrass and total seaweeds: density, abundance, and frequency of occurrence according to algorithms presented by Fourqurean and others (2001). An abundance index (mean BB score times mean shoot length) was also calculated for each quad with seagrass to estimate aboveground biomass at each point. For quadrats where widgeongrass was present but not measured for shoot length, the mean length of widgeongrass for all quadrats in the year of the survey at a site was used instead.

To establish the relationship between the abundance index and aboveground biomass for each seagrass species, eelgrass and widgeongrass were harvested annually within 12–37 calibration quadrats (0.25-m²) in each embayment. Percent cover (BB score) was assessed and seagrass shoots (aboveground parts only) within each quadrat were removed, cleaned, and 10 representative shoots were measured for shoot length. Samples were then dried to constant mass and weighed to determine biomass per quadrat, which was scaled to grams per square meter (g/m²). Finally, linear regression was used to determine the annual slope of the relationship between the abundance index and aboveground biomass across calibration quadrats for each embayment (eelgrass) or all embayments combined (widgeongrass). An estimate of aboveground biomass was then back-calculated for each seagrass species in a quadrat.

Summary statistics of environmental and seagrass parameters were calculated for each embayment and year after first averaging values of the four quadrats at each point. Means and standard errors are reported unless otherwise noted.

Eelgrass was the dominant seagrass both intertidally (0.0–0.8 MLLW) and subtidally (< 0.0 MLLW) across all survey years in BSQ. Mean tidal depth was lower at survey points with eelgrass (–0.18+0.03 m [range = –1.8 m to 0.8 m; n=392]) than at survey points with widgeongrass, 0.5 m+0.05 m (range= –0.7 m to 1.3 m; n=49; fig. 3) in BSQ, reflecting the generally discrete spatial variation in distribution of these seagrasses in embayments of northern Baja California (fig. 3; Ward and others, 2004; Ward and others, 2022).

During the first 2 years of the survey period (1998, 1999), eelgrass was the most common macrophyte in the bay, occurring on a majority of points (mean=0.74) and with a high density (mean=3.22; table 2). Widgeongrass and seaweeds were less common on points (mean=0.15 and 0.34, respectively) and occurred at a lower density (mean=0.34 and 1.01, respectively) during the first 2 years of our surveys (1998, 1999; table 2). However, this pattern of eelgrass dominance changed during the last 2 years (2005, 2012) of the study, in which we observed fewer points with eelgrass (mean=0.47) with a lower density (mean=1.51) and an increase in the number points with seaweeds (mean=0.58) with a greater density (mean=2.23; fig.4A–B). Moreover, mean aboveground biomass of eelgrass declined between the early (1998, 1999) and later years (2005, 2012) of the survey years (table 2; fig.4C). Widgeongrass aboveground biomass also declined over the 3 years of surveys (table 2; fig. 4D). There was no assessment of widgeongrass biomass in the last year of the study.

The increase in seaweeds abundance in 2005 and 2012 was largely driven by an expansion of Ulva spp., which was the most frequently encountered seaweed in the BSQ across all years. The temporal increase in seaweed abundance coincided with an overall decline in biomass of eelgrass (Spearman’s r= –0.184; P=0.001) and widgeongrass (Spearman’s r= –0.214; P=0.001) on survey points. Declines in eelgrass biomass appeared to be similar between intertidal and subtidal areas of the bay.

Seagrass declines that were first detected in the 1990s (Ward and others, 2003) continued through the early 2000s in BSQ (Zertuche-González and others, 2009). Much of the eelgrass losses can be directly attributed to the proliferation of Ulva, particularly in the southern and western parts of the bay (Ward and others, 2003). Large accumulations of drift macroalgae, such as Ulva, can negatively impact eelgrass growth and survival in embayments that become nutrient over-enriched. These fast-growing macroalgae can out compete seagrasses for nutrients and light and eventually displace them as the dominant macrophyte in nutrient enriched embayments (Short and Burdick, 1996; Hauxwell and others, 2001; McGlathery and others, 2001).

The drop in eelgrass abundance also occurred in the eastern part of the bay, but a large portion of this reduction can be linked to ENSO flood events in the winters of 1997–98 and 2009–10 that led to significant sediment loading of eelgrass beds located at or near the mouth of the Simon River (Ward and others, 2003). Reasons for the proliferation of Ulva in BSQ are unknown but may be due to increased nutrient-enrichment from the expansion of oyster farming in the bay (Ward and others, 2003; Zertuche-González and others, 2008), but see Jorgensen and others (2010) for other possible reasons for the Ulva proliferation.

Eelgrass was the dominant seagrass species in LOL and largely occurred in the low intertidal to subtidal areas of the lagoon. Mean tidal height of eelgrass across survey points was –0.25+0.03 m (range= –1.4 m to 0.6 m; n=273), slightly lower, but statistically similar to the mean tidal height of eelgrass in BSQ across similar survey years (fig. 3). Widgeongrass occurred primarily in the mid to low intertidal areas in LOL. Mean tidal height of widgeongrass was 0.19+0.03 m (range= –1.0 m to 1.1 m; n=273), which was lower than the mean tidal height of widgeongrass in BSQ (fig. 3). Seaweeds were mainly distributed where eelgrass occurred. Mean tidal height of seaweeds was –0.15+0.04 m (range= –1.4 m to 1.1 m; n=157).

Mean aboveground biomass and shoot length of eelgrass was consistently lower in LOL than in BSQ across common survey years (1998, 1999 and 2005; table 2 and 3). On average over these 3 years, aboveground biomass was three times lower in LOL (24.9 grams [g] per square meter, m²) than in BSQ (75.8 g/m²). The lower eelgrass biomass in LOL was likely driven by the lower density and shorter shoot length of eelgrass in this lagoon (density score: mean=2.0+0.10; length: mean=14.7+0.52 cm) compared to BSQ (density score: mean=2.5+0.11; length: mean=47.0+1.38 cm).

The frequency of occurrence and density of eelgrass decreased in LOL across survey years (fig. 5A–B). This reduction of eelgrass appeared linked to a general increase in frequency of occurrence and density of seaweeds (fig. 5A–B). Nevertheless, a negative trend in aboveground biomass of eelgrass was not detected across survey years (table 3; fig. 5C). It is unclear why a decrease in aboveground biomass was not detected, but it could be related to the stability of shoot lengths in eelgrass, which were stable across survey years (table 3). Aboveground biomass of widgeongrass was highly variable and no trend was detected in this seagrass across the three survey years (table 3; fig. 5D).

Eelgrass was also the dominant seagrass in LSI, but unlike the other two sites, monotypic beds of eelgrass occurred primarily in subtidal areas. Mean tidal height of eelgrass across survey points was –0.60+0.03 m (range= –1.8 m to 0.4 m; n=136), lower than the mean tidal height of eelgrass at the other two sites (fig. 3). Widgeongrass was the dominant seagrass of intertidal areas in LSI (Ward and others, 2022). Mean tidal height of widgeongrass was –0.16+0.04 m (range= –2.0 m to 0.7 m; n=122), also lower than the mean tidal height of widgeongrass at the other sites (fig. 3). Mean tidal height of seaweeds was –0.39+0.03 m (range= –1.4 m to 0.5 m; n=165).

The deeper depth distribution of eelgrass in LSI was consistent with findings of Cabello and others (2003), who reported that eelgrass shoots grew about 1 m lower within the water column in LSI than in BSQ. They attributed the greater depth distribution of eelgrass to higher water temperatures, irradiance levels and clarity of the more tropical waters in these Vizcaino embayments compared to the more temperate waters and environmental conditions in BSQ.

Mean aboveground biomass of eelgrass was considerably lower per annum in LSI than in BSQ, and slightly higher than aboveground biomass levels in LOL across survey years (tables 2, 3). In contrast, mean aboveground biomass of widgeongrass was two times higher in LSI (6.3+0.5 g/m²) than in BSQ (2.7+0.5 g/m²) and LOL (3.0+0.4 g/m²) across survey years, as reflected by the higher densities and frequency of occurrence of this seagrass in LSI when compared to the other two sites (tables 2, 4).

While frequency of occurrence of eelgrass was highly variable (fig. 6A), mean eelgrass abundance score and aboveground biomass declined across survey years in LSI (table 4; fig. 6B). This reduction in abundance appeared to be driven largely by a declining eelgrass density in all three years and to a lesser extent shoot length, which dropped in 2005 (table 4; fig. 6C).

The decline in eelgrass abundance in LSI, however, doesn’t appear to be linked to an increase in seaweed abundance as was found in BSQ because seaweeds also declined over the survey years (table 4; fig. 6C). Moreover, the decline in eelgrass abundance was not similarly detected in widgeongrass; the aboveground biomass of this seagrass increased over the same survey years (table 4; fig.6D).