MATERIALS AND METHODS
Study site: We selected sites adjacent to two southern California long-term monitoring sites; Navy South (32.68306°N, -117.24963 °W; hereafter NASO) and Navy North (32.692312 °N, -117.25297°oW; hereafter NANO). These Multi-Agency Rocky Intertidal Network sites (MARINe) contain dense patches ofSilvetia , perhaps because of the rarity of some stressors such as trampling (Denis 2003, Tydlaska & Edwards 2022) and runoff (Whitaker et al. 2010). We surveyed the Silvetia assemblages, collected the algae and grazers used in the mesocosm experiment, and conducted the field experiment at NASO. We then added NANO as a secondary collection site. Silvetia at both sites grows on emergent substrata at intertidal elevations between 0-1 m above Mean Lower Low Water (hereafter MLLW). Average water temperatures at these sites are ~18 °C and maximum summer water temperatures reach ~24 °C (SeaTemperatures 2023).
Mesocosm experiment: To examine the impacts of projected changes in ocean temperature and pH on Silvetia assemblages, we conducted a mesocosm experiment at San Diego State University’s Coastal Marine Institute and Laboratory (CMIL) that exposed the assemblages to three ocean climate conditions (Ambient, RCP 2.6, RCP 4.5). Ambient conditions represent current levels of temperature and pH. RCP 2.6 is a global emissions pathway representing low levels of climate change that will be experienced in the year 2100 (in line with the theoretical stabilization of global emissions by ~2020 leading to an average change of +1 °C/-0.1 pH units on global oceans). RCP 4.5 represents moderate levels of climate change (+2°C/-0.2 pH units). Importantly, our experiments used flow-through seawater, which allowed for natural variation in ambient conditions. Thus, our future scenarios that manipulated pH and temperature relative to ambient conditions also experienced such variation.
Each mesocosm consisted of a clear plastic box (15 x 15 x 7.6 cm; l*w*h) that had three 5-cm diameter holes in each of two opposite sides. Window screen mesh covered these holes and the box tops to retain box contents and allow water exchange. We crossed climate scenario (Ambient, RCP 2.6, RCP 4.5) with Silvetia canopy (Present, Absent) treatments. Replicate mesocosms (n=10) were randomly assigned to three outdoor water tables (1.8 x 0.9 x 0.3 m; l*w*h) that received flow-through seawater from San Diego Bay. Each water table was then randomly assigned to one of the three climate scenarios. Because water temperatures in San Diego Bay are warmer than the average water temperatures at rocky shores whereSilvetia occurs, we chilled the incoming seawater using a flow-through seawater chiller (Aqualogic, Inc.) but still allowed the temperature to vary with natural ambient fluctuations (Fig. 1). Seawater delivered to the future scenario mesocosms (i.e., RCP 2.6, RCP 4.5) was then altered in a header tank using aquarium heaters and CO2 injections before entering experimental mesocosms. Our goals were for 1) seawater in the RCP 2.6 treatments to be heated 1°C and acidified 0.1 pH units relative to Ambient conditions, and 2) seawater in the RCP 4.5 treatments to be heated 2°C and acidified 0.2 pH units relative to Ambient conditions. Seawater was delivered to each header tank at 2270 L/h, which then flowed via gravity to the experimental mesocosms. To create realistic tidal conditions, ball valves connected to drains were opened and closed using a digital watering timer (DIG Model C002, DIG Corporation), which resulted in the mesocosms being submerged at tide heights 0.5 m above MLLW, and emerged at tide heights below this. This tide height is representative of intertidal elevations whereSilvetia occurs in southern California (Littler 1980).
We added realistic assemblages of understory algae to each mesocosm. To determine the species that comprised these representative assemblages, we surveyed natural understory algal communities in the field at NASO. We collected, identified, and weighed all the understory algae found within six 0.15 x 0.15 m quadrats that were placed beneath haphazardly selected Silvetia individuals. This identified five genera that made up 83% of the total understory algal biomass; namelyChondracanthus , Centroceras , Corallina ,Gelidium , and Laurencia . Because we were unable to find enough Gelidium during future collections for our experiment, we removed it from the study. The remaining four genera made up 76% of total understory biomass. To create realistic understory assemblages, we calculated the biomass density of each genera in the field (grams per m2) and scaled these calculations to match the surface area of the mesocosm floors. Using this approach, each mesocosm received 4 g of Centroceras , 10 g of Chondracanthus , 9 g ofCorallina , and 2.5 g of Laurencia . Additionally, because invertebrate grazers can alter algal-algal interactions (Rogers & Breen 1983, Hoffmann et al. 2020), they were included in all mesocosms. To add ecologically realistic densities of these grazers relative toSilvetia biomass, we scaled field densities (# of grazer individuals per gram of Silvetia ) reported in a previous study (Jones 2016) to our mesocosms. As a result, we added six Tegulafunebralis , six Lottia strigatella , six Lottia scabra , ten Littorina scutulata , and one Cyanoplax hartwegii to each mesocosm.
Grazers and understory algae were collected during the establishment of the field experiment (see below) and held at CMIL for a 10-day acclimation period. Each of the four understory seaweeds were weighed to the predetermined biomass (± 5%), attached to a rock with superglue, and placed into one of the four corners of the mesocosms. Additional rocks covered the bottom of the mesocosms to provide a refuge for grazers. We alternated the position of each algal type between replicates in a Latin Square design. For the Silvetia Present treatments, pre-weighed Silvetia (72.5 ± 1.3 g, mean ± SE) were laid across the assemblage inside mesocosm containers. During our 42-day experiment (August 5th-September 17th, 2021), we measured pH and temperature of the seawater as it flowed from each header tank into the experimental mesocosms every morning using a probe (Oakton 300 Series pH/DO meter), except on days 32, 38, and 39, which were not measured due to logistical constraints.
After 42 days, we ended this experiment as most of the understory algae in the Silvetia Absent treatments had bleached or disintegrated. We categorized the algae as being either bleached (dead) and unbleached (living) and measured the biomass of each group in each replicate after blotting them dry. The remaining biomass of each understory genus was then calculated as the percentage of final unbleached tissue weight relative to its initial weight. To assess Silvetia health, we measured quantum yield [a ratio of variable fluorescence (Fv) to maximal fluorescence (Fm)], which estimates the light harvesting efficiency of photosystem II (PS II), using a pulse amplitude modulated (PAM) fluorometer (sensu Edwards and Kim 2010, Bews et al. 2020). Because we observed within-individual variation in tissue health, we measured the quantum yield of each individual at five randomly selected sections of each thallus and averaged these measurements for eachSilvetia replicate.
To understand seasonal differences in how the Silvetia assemblage responded to climate change, we repeated this experiment in the winter (November 9th-December 20th, 2021). We followed the same protocols described above but made three changes: 1) We shortened the acclimation period from ten to five days, 2) we collected algae and grazers from a nearby site (NANO instead of NASO), and 3) the water tables were randomly reassigned different climate treatments. Pre-weighedSilvetia for this experiment averaged 71.0 ± 1.6 g. Although we did not see as much understory degradation in the Silvetia Absent treatments during this experiment, we maintained the 42-day experimental duration to facilitate comparisons between the two trials (hereafter summer and winter).
Field experiment: Experimental field plots were established at NASO to simulate the effect of climate change-mediated loss ofSilvetia on its understory assemblage. Because the effect of canopy loss on the assemblage could depend upon the successional stage of the assemblage, we also manipulated the assemblage biomass of the understory by clearing half of the plots at the start of the experiment. We crossed Silvetia Canopy (High, Partial, None) with the initial state of the Understory (Full, Cleared); n=10. We established these plots in the summer (July 2021) because we hypothesized that the effects of Silvetia loss should be most pronounced during the less favorable summer conditions. Plots containing Silvetia (0.15 x 0.15 m) were marked at their corners with Z-spar Splash Zone epoxy and were randomly assigned to the different treatments. Plots were positioned just below the existing Silvetia holdfasts to study the understory species beneath where the Silvetia canopy drapes over the substrate during low tide. Prior to manipulations, we recorded the percent cover of each genus within the plots using 25-point intercepts within 0.15 x 0.15 m quadrats.
Plots assigned to the No Silvetia Canopy treatments simulated the effects of climate change-mediated loss of Silvetia by trimmingSilvetia to its holdfast using shears. This allowed the thallus to eventually regrow, while still subjecting the assemblage to any effects associated with an absent canopy for the duration of the experiment. In previous mesocosm experiments, future climate conditions caused Silvetia to discolor, shrivel, and lose biomass across its entire thalli (J.D. Long 2015 [unpublished data]). To examine the consequences of partial Silvetia loss, we trimmed Silvetiain Partial Canopy treatments from multiple layers originating from a single holdfast to a single thallus layer. The remaining plots containing Silvetia were left unmanipulated and represented our High Canopy treatments. However, because 1) we observed large within treatment variation and 2) the Full and Partial Silvetia Canopy treatments provided similar canopies, we pooled Full and PartialSilvetia treatments into a single “Silvetia Present” treatment and compared this pooled treatment to the “SilvetiaAbsent” treatment. To manipulate the understory assemblages, the existing assemblages in half of the plots of each Silvetiatreatment were removed using scrapers and chisels (Understory Cleared treatments) while the assemblages in the other half were left unmanipulated (Understory Full treatments). We measured the percent cover of the understory assemblages in October (hereafter fall) and December (hereafter winter) 2021.
Statistical analyses: All data were analyzed using R-Studio and Primer + PERMANOVA 7. Prior to analyses, data were checked for normality and heteroscedasticity using Shapiro-Wilk’s and Levene’s tests, respectively. For the mesocosm experiment, measurements of quantum yield required square-root transformation to meet assumptions of normality.Silvetia biomass and measurements of quantum yield within the mesocosms were compared among the three climate treatments using separate one-way ANOVAs (for each season). This was done as separate analyses rather than a two-way ANOVA that included season as a factor because the experimental mesocosms were broken down, cleaned, randomized, and reassigned with new assemblages prior to the winter trial. Tukey’s HSD post-hoc tests between pairs of climate treatments were then used when the ANOVAs returned significant differences. To visualize shifts in the understory algal assemblages between the Climate and Silvetia canopy treatments within each trial, Principal Coordinates Analysis (PCoA) was used to map similarities in the algae comprising each assemblage. Two-way PERMANOVAs were then used to determine if the assemblage shifts differed between the Climate andSilvetia canopy treatments. Due to a high number of zeroes for certain taxa in the Silvetia Absent treatments, the data were square-root transformed and the PERMANOVAs were run with a zero-inflated Bray-Curtis similarity indices using a dummy variable of 1. A priori post-hoc permutation tests were then used to examine pairwise differences in the assemblages between Climate and Silvetiacanopy treatments. SIMPER analyses were used to identify the relative contribution of each understory taxon to assemblage dissimilarity between treatments. As discussed above, these analyses were run separately for the summer and winter trials. For the field experiment, a three-way PERMANOVA was used to assess differences in the understory communities (based on percent cover) between Silvetia canopy treatments, Understory treatments, and Seasons. Unlike the mesocosm experiments, season was included as a factor because the field experiment was run continuously. Following the PERMANOVA, a priori permutation post-hoc tests were used to determine differences in understory assemblages between the Silvetia canopy treatments within each Understory treatment and season. SIMPER analyses were used to determine the percent contribution of each general to the observed differences. All analyses were evaluated at an ɑ-level of 0.05.