Introduction

Infaunal invertebrate populations are greatly influenced by the properties of the environment in which they live (Rhoads 1974; Snelgrove and Butman 1994; Widdicombe et al. 2009; McArthur et al. 2010; Clements and Hunt 2018). Sediment properties that are commonly included in studies looking at their importance to infaunal invertebrates include grain size metrics (e.g. mean grain size, sorting, kurtosis), organic matter content, phosphorus content, nitrogen content, oxygen content, and sediment carbonate chemistry (Snelgrove and Butman 1994; McArther et al. 2010; Meseck et al. 2018; Drylie et al. 2019; Drylie et al. 2020; Bianichi et al. 2021). Grain size has long been considered one of the key sediment characteristics for infaunal invertebrates (Snelgrove and Butman 1994). Individual species exhibit preferences for sediment grain size (Snelgrove and Butman 1994). For example, some species, such as the amphipod Corophium volutator, prefer mud sediment (e.g. Barbeau et al. 2019), while soft-shell clam Mya arenaria recruits prefer mud and smaller rocky sediments (e.g. Morse and Hunt 2013). This is likely related to their ability to burrow, availability of food, and sediment stability (reviewed by Snelgrove and Butman 1994). Another sediment characteristic that has been widely considered is organic content (Snelgrove and Butman 1994; McArther et al. 2010; Bianichi et al. 2021). The presence of organic content is important for deposit feeders as it serves as a critical source of food (Snelgrove and Butman 1994). However, decomposition of a large amount of organic content can lead to reductions in oxygen availability for all marine life in that system, which can have devastating effects (Nixon 1995; Carstensen and Duarte 2019). Another abiotic variable that has been less studied compared to grain size and organic content is sediment carbonate chemistry. Studies have found that alkalinity and sediment pH are related to the abundance of bivalve recruits (Clements and Hunt 2018; Meseck et al. 2018). The impacts of sediment carbonate chemistry on additional taxa of infaunal invertebrates require further study.

For infaunal invertebrates, water column conditions, such as pH, temperature, oxygen availability, and salinity, are also important to consider in addition to sediment characteristics (Coyle et al. 2007; Widdicombe et al. 2009; Clements and Hunt 2018). Macrofaunal and nematode community composition was found to be significantly affected when exposed to differing water column pH values (pH treatments: 5.6, 6.5, 7.3, and 8.0) in a laboratory experiment, with species richness reaching a peak level between pH 7.0 and 7.5 after a 20 week exposure period (Widdicombe et al. 2009). In the Bering Sea, warmer temperatures were associated with decreases in the infaunal invertebrate community biomass, which may have been due to increased predation activity by fish (Coyle et al. 2007). Additionally, temperature changes can cause changes to invertebrate behaviour and dispersal (e.g. Clements et al. 2021), which could affect communities. Salinity gradients have been documented as resulting in spatial variation in infaunal invertebrate communities due to salinity thresholds of some species (e.g. Dethier et al. 2010). All three of these water column variables (pH, temperature, salinity) have an influence on infaunal invertebrates, the extent to which varies among species, as is the case for sediment characteristics as well. There is a great need for both water column and sediment characteristics to be considered simultaneously when determining variables that most influence infaunal invertebrate abundances and species composition.

While many studies have considered the influence of either water column or sediment characteristics on infaunal invertebrates, fewer studies have simultaneously examined the effect of both types of variables on infaunal invertebrates (but see Thrush et al. 2012; Meseck et al. 2018; Clements and Hunt 2018; Mevenkamp et al. 2018, Thrush et al. 2020). Studies of bivalves that incorporated both sediment and water column characteristics have found that sea surface temperature, air temperature, grain size, pH, and salinity all influenced juvenile bivalve recruitment (Meseck et al. 2018; Clements and Hunt 2018). When both sediment and water column properties are considered, variation in different species abundances may be better explained, as seen in Clements and Hunt (2018) and Meseck et al. (2018). Both of these studies focused on bivalve recruitment; therefore, it is critical for future research looking at both water column and sediment properties to study more taxa of infaunal invertebrates.

The main objective of our study was to determine which abiotic variables influence infaunal invertebrate communities and juvenile bivalve species abundances at intertidal sites in two regions in New Brunswick, Canada with contrasting oceanographic conditions, the Bay of Fundy and the Southern Gulf of St. Lawrence. We were interested in explaining how sediment grain size characteristics, sediment carbonate chemistry, water salinity, water temperature, and water column carbonate chemistry influenced infaunal community composition, biological metrics (species richness and total abundance), and recruitment of juvenile bivalves. Three species of bivalves are found in the two study regions, soft-shell clam (Mya arenaria), Baltic clam (Limecola balthica), and gem clam (Gemma gemma). Using data from 2020 and 2021, the abiotic variables influencing multivariate infaunal invertebrate community composition were determined. Additional analysis was completed to determine if there was spatial and temporal (seasonal and inter-annual) variation in community composition within the two regions. Generalized linear modeling was also completed to determine which abiotic variables were related to infaunal species richness and total abundance, total juvenile bivalve abundance, and soft-shell clam (Mya arenaria) abundance. Univariate and multivariate analyses were both completed to look at both the species- and community-level effects of the abiotic variables. By determining which abiotic characteristics best explained variation in each metric considered, better predictions of infaunal invertebrate responses to future oceanic conditions can be made.

Methods

Study sites

Sampling was completed at intertidal sites in two regions in New Brunswick (NB), Canada with large differences in tidal ranges, oceanographic conditions, and sediment characteristics. Tidal ranges in the Bay of Fundy average 6.5 m, while in the Southern Gulf of St. Lawrence, tidal ranges average 0.8 m (Fisheries and Oceans Canada 2023). Oceanographically, the Southern Gulf of St. Lawrence has warmer summer water temperatures and colder winter temperatures compared to the Bay of Fundy. When it comes to intertidal sediment characteristics, the Bay of Fundy primarily has mudflats with some small rocks mixed in while the Southern Gulf of St. Lawrence is a mixture of rocky and sandy beaches. Samples were collected in the summers of 2020 and 2021 from sites in southern NB in the Bay of Fundy and from eastern NB in the Southern Gulf of St. Lawrence in 2021. The two regions were approximately 335 km apart. The main criterion for site selection was presence of infaunal bivalves. Juvenile bivalve recruitment was of particular interest due to their predicted sensitivity to climate change impacts (Clements and Hunt 2018; Leung et al. 2022). The sites chosen in the Bay of Fundy were selected based on past studies of bivalve recruitment (see Clements and Hunt 2018), while the sites in the Gulf of St. Lawrence were selected based on information from Clements et al. (2021) and K. Beardy (pers. comm.). In the Bay of Fundy, four sites were sampled: Red Head Road (RH; 45° 6′ 50.00″ N, 66° 36′ 37.63″ W), Pocologan (POCO; 45° 7′ 11.98″ N, 66° 35′ 27.81″ W), Little Lepreau Road (LL; 45° 7′ 55.12″ N, 66° 28′ 22.02″ W), and Cassidy Lane (CL; 45° 7′ 28.82″ N, 66° 28′ 17.97″ W) (Fig. 1). The sites were chosen to form two pairs of sites (POCO and RH; CL and LL) with sites in each pair 0.5–1 km apart while the pairs were approximately 10 km apart. In the Gulf of St. Lawrence, the sites sampled were: Lower Newcastle (LN; 47° 4′ 1.43″ N, 65° 25′ 37.86″ W), Anderson Bridge (BRIDG; 47° 2′ 9.02″ N, 65° 28′ 35.09″ W), Cote-a-Fabien (CAF; 46° 50′ 7.44″ N, 64° 55′ 54.89″ W), Callander Beach (CAL; 46° 48′ 34.74″ N, 64° 54′ 21.65″ W), Kelly Beach (KELLY; 46° 49′ 53.35″ N, 64° 54′ 53.44″ W), and Ryan Beach (RYAN; 46° 49′ 59.75″ N, 64° 55′ 17.46″ W) (Fig. 1). LN and BRIDG were approximately 6 km apart within the Miramichi River estuary, which was 60 km away from Kouchibouguac National Park where CAF, CAL, KELLY, and RYAN were all located a maximum of 5 km apart. Ryan’s Beach was only sampled once so it was not included in the statistical analysis, but it was included in the boxplots to examine abiotic characteristics in the Gulf of St. Lawrence region.

Fig. 1
figure 1

Location of study sites in the Bay of Fundy in southern New Brunswick and in the Southern Gulf of St. Lawrence in eastern New Brunswick, Canada

Field sampling design

From July to October of 2020, sampling was completed once a month at the four Bay of Fundy sites. In 2021, sampling in the Bay of Fundy occurred every 2 weeks from July to October. In the Southern Gulf of St. Lawrence in 2021, samples were collected biweekly from the five sites from late May to late August, except for RYAN that was only sampled once (August). The sampling periods between the two regions differed as we were interested in capturing bivalve settlement periods in both regions, which occurs in May–June in the Southern Gulf of St. Lawrence and September–October in the Bay of Fundy.

Both abiotic and biotic variables were measured. The water column characteristics included salinity, temperature, pH, and alkalinity. The sediment and porewater characteristics measured were sediment pH profiles, sediment alkalinity, organic matter content, carbonate content, and four sediment grain size metrics (mean grain size, sorting, kurtosis, and skewness). Sediment pH profiles and all water column characteristics were measured on each sampling trip, while analyses of sediment alkalinity, organic matter, carbonates, and sediment grain size were completed once a month. Invertebrate community samples were collected once a month and used to determine species richness, total abundance, bivalve abundance, and multivariate community composition of individuals > 180 μm in size.

A permanent 25-m transect was selected at each site that was parallel to the shore around the mid-intertidal zone and was chosen to represent as much of the variation as possible in sediment characteristics at each site. On each sampling date, six randomly selected points along the transect were sampled at each site. At each sampling point, one sediment core (diameter 6.4 × 5 cm deep) was collected to complete sediment pH profiles and sample the infaunal invertebrate community and placed in a plastic container of the same diameter. The core and sieve sizes were chosen as we were primarily interested in meiofauna and smaller macrofauna, such as juvenile bivalves, as these invertebrates are less studied and juvenile animals have been shown to be more sensitive to environmental conditions (Pandori and Sorte 2019). Three additional sediment cores of the same dimensions were collected, next to the first, middle, and last pH profile/invertebrate cores, to be used for analysis of sediment grain size, organic content, and carbonate content. Lastly, at each of the six sampling points, a test tube (15 mL; 1.6 cm diameter × 11.8 cm height) of sediment was filled with sediment scooped from the upper 2 cm to be used for porewater extraction and alkalinity titration. All the sediment samples were transported back to the lab on ice in a cooler to maintain the same temperature as on the mud/sandflat. At each site, salinity, temperature, and pH was measured using a YSI Pro1030 series model with the YSI 1001 pH sensor in a tide pool or the larger body of water (depending on site conditions) below the sampling transect as close to the receding tide as possible to characterize the water column conditions. One water sample per site was collected to be used for alkalinity measurement in the lab. Sampling was started while the tide was just beginning to recede (approximately 2.5 h beforehand) and was completed before reaching low tide.

Lab analyses

pH profiles were measured using a Unisense microelectrode pH-500 probe and Unisense pH/Redox Uniamp meter at 0.5 cm intervals from surface (0 cm) to 3 cm depth within each core. A Velmex Unislide system was used to position the probe to obtain precise depth measurements. Three profiles were measured in each sediment core. The cores were stored on ice in a cooler then allowed to warm slightly to complete pH profiles at the same temperature as the mudflat, between 10 and 18 °C for the Bay of Fundy and between 14 and 20 °C for the Gulf of St. Lawrence. All pH profiles were completed within a maximum of 6 h of collection. These measurements cannot be made in situ due to the fragility of the pH microelectrode tool used and the necessity of the positioning system to achieve measurements every 0.5 cm. There was the potential that transporting the cores back to the lab impacted the pH of the samples. While the extent to which this delay in measurement changed the pH is unknown, our measurements were made within the timeframe recommended for measurements of water pH (Pimenta and Grear 2018).

After the pH profiles were completed, these sediment cores were rinsed over a 180 μm sieve with freshwater to remove most of the sediment. Remaining sediment, algae, and invertebrates > 180 μm were transferred into a jar and preserved with 95% ethanol. These samples were then examined under a dissecting scope for invertebrates, which were identified to the lowest taxonomic level possible and counted. The species list and abundances were then used to determine species richness, total infaunal abundance, bivalve abundance, and multivariate community composition.

Porewater was extracted from the test tubes of sediment by spinning at 3000 rpm for 10 min in a centrifuge. Porewater was used to determine sediment alkalinity while the separate water sample from each site was used for the water alkalinity measurements. For both sediment and water alkalinity measurements in 2021, a spectrophotometric method for alkalinity determination was used based on Sarazin et al. (1999). In 2020, alkalinity of the porewater was determined by using a two-point titration procedure developed by Edmond (1970). Unfortunately, the total alkalinity values determined from the titration procedure could not be used due to chemical concentration problems and are not presented.

The additional sediment samples were processed to determine sediment grain size metrics, percent organic matter and carbonates, and porosity. Approximately 150–175 g of wet sediment was weighed then placed in a drying oven at 70 °C for at least 48 h after which it was weighed again to determine the porosity of each sample. These dried samples were then placed into a sediment shaker for the grain size analysis using sieve sizes 4, 2, 1 mm, 710, 500, 250, 125, and 63 μm. Mean (± standard deviation) grain size, sorting, skewness, and kurtosis were calculated from the sediment grain size analysis using the “granstat” function in the G2Sd package (Fournier et al. 2014) in R software. The percentage of organic matter was determined using the loss on ignition method described by Heiri et al. (2001). A 2–3 g subsample of each sediment sample was dried for 48 h at 70 °C and then weighed. Then, the subsample was placed into a muffle furnace for 4 h at 550 °C and reweighed once the sample cooled. The percent change in weight was the estimated percent organic content in the sample. The subsample was then placed into a muffle furnace for 2 h at 950 °C. It was then weighed, and the percent change in weight was used as an estimate of percent of carbonate matter within the sample.

Statistical analysis

Multivariate analysis

PRIMER (v7) with PERMANOVA was used for all multivariate analyses. Permutational ANOVAs, PERMANOVAs, and distance-based linear modeling were done for each region, Bay of Fundy and Southern Gulf of St. Lawrence, separate due to the lack of overlap in the sampling dates between the two regions.

PERMANOVAs were used to analyze the effects of date and site (random factors) on the combined regions’ multivariate infaunal community composition (n = 313 total samples; n = 128 for the Bay of Fundy, n = 185 for the Southern Gulf of St. Lawrence) and abiotic conditions (n = 118 total samples; n = 78 for the Bay of Fundy, n = 40 for the Southern Gulf of St. Lawrence) in 2021. Some sampling trips were split over 2 consecutive days for the Southern Gulf of St. Lawrence sites. These trips being split over 2 days were addressed by assigning each sampling trip to its start date. The interaction for site and date was included in these analyses despite all sites not being sampled on every sampling trip. When there was a significant interaction, pairwise PERMANOVA tests were carried out that excluded the missing site/date combination and, therefore, only looked at the overlaps in site and date. Non-metric multi-dimensional scaling (nMDS) plots were created to examine the similarity of the abiotic and invertebrate samples collected.

Prior to building any models, a correlation matrix for the abiotic variables was built. The only highly correlated variables were the sediment pH values measured at multiple depths (0–3, every 0.5 cm) within each core. The 1.5 cm depth was chosen for use in the analyses after comparing generalized linear mixed models and distance-based linear models with different single pH depths as the predictor variable. Models with pH at the 1.5 cm depth had the best Akaike information criterion (AICc) scores for each dependent variable in both univariate and multivariate analyses.

All models that included abiotic variables, including abiotic PERMANOVAs, had the following abiotic variables: water temperature, salinity, water column alkalinity, sediment pH at 1.5 cm, sediment alkalinity, organic matter content, carbonate content, and grain size metrics (mean grain size, sorting, skewness, and kurtosis). Sediment characteristic values were estimated from the closest core for invertebrate cores without matching sediment characteristic cores for dates when sediment characteristics were sampled. Samples missing any other single abiotic variable, such as missing alkalinity or sediment pH values for a single sediment pH depth, were excluded from the multivariate analysis.

The species abundance data was dispersion-weighted and square root transformed to adjust for the patchy distribution of some species of infaunal invertebrates, then a Bray–Curtis dissimilarity calculation was used to create resemblance matrices between samples. For the abiotic data, variables were standardized (ranging from 0 to 1 with all the means for all variables being the same) then Euclidean distances were calculated between samples. The PERMANOVA models contained ‘Site’ and ‘Date’ as random factors to examine spatial and temporal variation in abiotic and biotic variables.

To determine if there was a linear relationship between community composition and abiotic variables in 2021, distance-based linear modeling (DISTLM) was completed separately for the Bay of Fundy and Gulf of St. Lawrence regions due to the lack of overlap in the sampling dates between the two regions. DISTLM was completed to see how much of the variation observed in the multivariate community composition could be explained by sediment pH and sediment characteristics (mean grain size, sorting, percent organic matter, and percent carbonates). The same procedure for missing abiotic samples, both the exclusion and substitutions of sediment values, from the abiotic PERMANOVAs was followed for the DISTLM. Marginal tests were used to determine the proportion of variance in invertebrate community composition explained by each abiotic variable. Within the DISTLM analysis, AICc was then used to see which variables should be included in the best model. Lastly, BEST (Bio-Env) analysis was completed to determine which species drove the relationships between the abiotic variables and the community composition.

Univariate analysis

All univariate data analyses were completed in R Studio v. 1.3.1.093 (R Core Team 2013). All univariate modeling was completed for the two regions separately.

Generalized linear models (stats package, glm function, R Core Team 2013; GLMs) were completed to look at the temporal and spatial variation in species richness, total infaunal abundance, and bivalve abundance in both regions studied, with site and date as fixed factors. For each dependent variable a series of models with several potential distributions (discrete data: Gaussian, Poisson, negative binomial, zero-inflated Poisson, and zero-inflated negative binomial; continuous data: Gaussian, gamma, and beta) were constructed to determine which distribution was best to model each dependent variable. Testing multiple distributions allowed us to determine what the best fit for a specific variable was, such as zero-inflated vs. regular Poisson. For the four Bay of Fundy sites, there were five sampling dates in 2020 and six sampling dates in 2021 with 215 samples total for both years. For the Gulf of St. Lawrence sites, there were eight sampling dates in 2021 with a total of 221 samples. The interaction between site and date was included in this analysis as well. As with the multivariate analysis, some sampling trips were split over 2 consecutive days for the Southern Gulf of St. Lawrence sites which was addressed by assigned each sampling trip to its start date. For the GLMs, Tukey’s HSD comparisons (multcomp package, glht function; Hothorn et al. 2008) were used to assess significant main effects. As not all sites were sampled on each sampling trip, Tukey’s comparisons could not be used to assess significant interactions.

Generalized linear mixed modeling (GLMM) was also completed to investigate which abiotic variables were related to species richness, total abundance, Mya arenaria abundance, and total bivalve abundance in 2021 for the Bay of Fundy (n = 122 samples) and Gulf of St Lawrence (n = 81 samples) regions separately (“glmmTMB” function, “glmmTMB” package, Brooks et al. 2017; “glm” function, “stats” package, R Core Team 2013). It was not possible to examine the abundance of each bivalve species individually due to low abundances of Limecola balthica and Gemma gemma. Limecola balthica, Mya arenaria, and Gemma gemma were combined for the total bivalve abundance variable; G. gemma were not found at the Bay of Fundy sites. The abiotic variables used in the GLMMs were the same as those used in the multivariate analyses. There were two random effects (random intercepts) included in the GLMM models, site and date, to statistically account for temporal and spatial variation in the biotic and abiotic data without making specific comparisons between the sites and dates sampled.

In the GLMM models for invertebrate cores without matching sediment characteristic cores, the same procedure was followed as done in the multivariate analysis (See Multivariate Analysis for more information). Samples were excluded if any other abiotic variable was missing for a particular sampling trip.

For each dependent variable (species richness, total abundance, bivalve abundance, and Mya arenaria abundance), a series of GLMM models with all fixed factors and the potential distributions were constructed to determine which distribution was best to model each particular dependent variable, as was done for the GLM models. Models were then compared based on AICc scores to identify best fit and a distribution was selected for use in subsequent models. To determine the most important abiotic variables, a global model was constructed, using the best distribution identified previously, then each abiotic variable was removed from the global model and the AICc scores of the global and reduced (by one independent variable) models were compared. This was repeated for each abiotic variable. If the AICc score improved with the removal of the abiotic variable, then that variable was removed from further consideration. The variables that decreased the AICc score were then combined into new models with all possible combinations of these variables. The AICc score determined the best fit model while considering the number of variables included in the model to prevent overfitting. Once the top three models for a particular dependent variable were identified, key model parameters and pseudo-R2 values were determined (“summary$AICtab” function, “AICcmodavg” package, Mazerolle 2020; “pR2” function, “pscl” package, Jackman 2020). Chi-square and p-values for the fixed factors in the models were obtained through likelihood ratio tests due to the unbalanced nature of the data (car package, Anova function; Fox and Weisberg 2019). The assumption of homogenous variances and of the degree of correlation between independent variables were tested for all GLMs and GLMMs by examining residual plots and correlation matrices.

Results

Infaunal invertebrate communities

A total of 20 species of infaunal invertebrates were identified from the study sites in the Bay of Fundy and Gulf of St. Lawrence regions. The Bay of Fundy had 17 species and the Southern Gulf of St. Lawrence had 14 species, with 11 species found in both regions (Supplemental Information: Table S4). These species were a combination of polychaetes, oligochaetes, bivalves, amphipods, copepods, nemertean worms, and gastropods. The number of individual invertebrates within each core ranged from 0 to 481. The extremely high abundances, upwards of 400–600 individuals per sample, at some sites were composed primarily of polychaete and oligochaete worms. In general, there were fewer individuals collected per core in the Southern Gulf of St. Lawrence sites compared to the Bay of Fundy sites (Fig. 2).

Fig. 2
figure 2

Boxplots of species richness per sample core (upper panels) at sites in the A Bay of Fundy and B Southern Gulf of St. Lawrence, and total abundance per sample core (lower panels) of infaunal invertebrates at sites in the C Bay of Fundy and D Southern Gulf of St. Lawrence (n = 215 for Bay of Fundy, 221 for Southern Gulf of St. Lawrence; volume = 160.85 cm3) in 2020 (Bay of Fundy only) and 2021. Data are shown at the month level for visual clarity, with there being 1–2 sampling dates per month

Species richness was lower in the Southern Gulf of St. Lawrence region at all sites compared to the Bay of Fundy sites (Fig. 2). In the Southern Gulf of St. Lawrence sites, CAF had significantly higher species richness compared to the remaining Kouchibouguac National Park sites, CAL and KELLY (Fig. 2, Table 1, Supplemental Information: Table S1). Temporally, early June and mid-June had significantly higher species richness compared to July and August in the Southern Gulf of St. Lawrence (Fig. 2, Supplemental Information: Table S1). For species richness in the Bay of Fundy, there was a significant interaction between site and date. While species richness was generally lowest at CL throughout both years, the site with the greatest species richness was not consistent over time (Table 1, Fig. 2). There were minimal differences in the species richness at the sites in the Bay of Fundy between 2020 and 2021. In 2020, species richness was relatively similar between LL, POCO, and RH in the Bay of Fundy throughout the sampling period (Fig. 2). Meanwhile in 2021, there were clear seasonal increases in the species richness at LL, POCO, and RH (Fig. 2).

Table 1 ANOVA table of generalized linear model results showing the spatial and temporal variation in species richness, total abundance of infauna, and bivalve abundance at sites in both the Bay of Fundy and Southern Gulf of St. Lawrence regions (n = 215 for Bay of Fundy, 221 for Southern Gulf of St. Lawrence; volume = 160.85 cm3)

Juvenile bivalve abundance in the Southern Gulf of St. Lawrence had significant temporal and spatial variation, while in the Bay of Fundy, there was significant temporal variation. CAL and RYAN had significantly lower bivalve abundance compared to the remaining Southern Gulf of St. Lawrence sites (Fig. 2, Supplemental Information: Table S1). The significant temporal variation in bivalve abundance was limited to a single comparison, where late June had higher abundance compared to late August (Table 1, Supplemental Information: Table S1). Bivalve abundance was low at all Bay of Fundy sites in 2020 (Southern Gulf of St. Lawrence was not sampled) and 2021, except for a spike in bivalve abundance in September 2020 at LL (Supplemental Information: Fig. 1, Supplemental Information: Fig. 2). There was significant temporal variation in bivalve abundance in the Bay of Fundy sites, with early September having significantly higher abundance compared to all other months sampled in 2020 and 2021 (Table 1, Supplemental Information: Table 1). There were no significant differences among sites in bivalve abundance in 2021 in the Bay of Fundy region (Supplemental Information: Table 1).

Total infaunal abundance in the Southern Gulf of St. Lawrence sites was similar to that in the Bay of Fundy and averaged 50 individuals per core or less (Fig. 2). The Southern Gulf of St. Lawrence had significant spatial variation in total abundance but did not have a significant effect of date or an interaction of date and site (Table 1), with almost all sites being significantly different from one another in the Southern Gulf of St. Lawrence (Supplemental Information: Table S1). There was a significant interaction between date and site for total abundance in the Bay of Fundy, which was likely driven by the spike in total abundance in RH in July and August of 2021 (Table 1, Fig. 2).

Of the ten most abundant individual infaunal species, there were some species that exhibited temporal variation, spatial variation, or an interaction of the two in the two study regions. At the Bay of Fundy sites, the most abundant infaunal species were less abundant in 2021 compared to 2020 (Supplemental Information: Fig. 1). Abundances varied among the sites, with POCO having higher amounts of Hediste diversicolor and RH having more Corophium volutator later in the sampling season (Supplemental Information: Fig. 1). There was some temporal variation present with Streblospio benedicti abundances dropping by October (Supplemental Information: Fig. 1). There was greater temporal variation present in the Southern Gulf of St. Lawrence sites, with some species like juvenile Mya arenaria, Nematoda spp., and Spio benedicti declining in abundance by July and August 2021 (Supplemental Information: Fig. S2). Cerebratulus sp. and S. filicolnis were rarely observed in the samples from the Southern Gulf of St. Lawrence except for in June and August 2021, respectively (Supplemental Information: Fig. S2).

nMDS plots were used to display the multivariate similarities of infaunal invertebrate communities between samples for sites in both the Bay of Fundy and Southern Gulf of St. Lawrence in 2021 (Fig. 3). The invertebrate community showed a slightly different pattern in the similarities between sites in the two regions. While the Southern Gulf of St. Lawrence sites grouped together and three of the four Bay of Fundy sites were grouped together (Fig. 3), the invertebrate assemblage at the Bay of Fundy site CL was more similar to the Southern Gulf of St Lawrence sites than it was to the other Bay of Fundy sites (Fig. 3).

Fig. 3
figure 3

Non-metric MDS plot of the infaunal invertebrate community for both the Bay of Fundy and the Southern Gulf of St. Lawrence sites (n = 128 cores for Bay of Fundy, 185 cores for Southern Gulf of St. Lawrence; volume = 160.85 cm3) in 2021 with data shown at the month level for visual clarity, with there being between 1 and 2 sampling dates per month. Sites LN, BRIDG, CAF, CAL, and KELLY are in the Southern Gulf of St. Lawrence while POCO, RH, CL, and LL are in the Bay of Fundy

PERMANOVAs indicated there was a significant interaction between sampling date and site for both the Bay of Fundy and Southern Gulf of St. Lawrence regions (Table 2). In the Bay of Fundy, there were significant differences between almost all sampling dates at one site and some significant differences between dates at another site (Fig. 3). In the Southern Gulf of St. Lawrence, there were significant differences between most sites at each sampling date (Fig. 3).

Table 2 PERMANOVA results showing the sources of variation in multivariate community composition of infauna at sites in both the Bay of Fundy and Southern Gulf of St. Lawrence regions in 2021 (n = 128 for Bay of Fundy, 185 for Southern Gulf of St. Lawrence; volume = 160.85 cm3)

In the Bay of Fundy, the amphipod Corophium volutator, the oligochaete worms Tubificoides benedii and Clitellio arenarius, the polychaete Strebliospio benedicti, and insect larvae were the five taxa identified as key for the differences in the multivariate community composition between the samples (see Supplemental Information: Table S2). In the Southern Gulf of St. Lawrence sites, the polychaete Hediste diversicolor was identified as the key species differentiating samples (see Supplemental Information: Table S2).

Abiotic characteristics

As seen in Fig. 4, at the Bay of Fundy sites, there were differences between the two years in some abiotic characteristics. Sediment pH remained relatively stable over time at all four sites in 2020; however, in 2021, there was much greater variability in the pH among months at RH, POCO, and LL. CL remained stable in sediment pH between months in both 2020 and 2021 but the pH was almost a full pH unit higher in 2021 (Fig. 4). There was also greater variation in pH between the sites in 2021 than in 2020 (Fig. 4). Sediment alkalinity was only measured in 2021 due to equipment problems in 2020. Alkalinity remained stable across all sites, with minimal variation within and between sites, during the 2021 sampling period (Fig. 4). Salinity remained relatively stable at all sites in both years, except RH which had a great deal of variation when compared to the other sites in 2021 (Fig. 4). Sediment grain size sorting and kurtosis were also consistent with a slight increase across the sampling period in both years, except for CL which had a decrease in sorting in 2021 compared to 2020 (Fig. 4). Organic content, carbonate content, mean grain size (mm), and sorting all showed differences among sites that were consistent between years (Supplemental Information: Fig. S1). Water alkalinity varied greatly at each site in August 2021 but was relatively stable in the other 3 months within each site (Supplemental Information: Fig. S1). Water temperatures were warmer across all sites in 2021 compared to 2020, especially in September (Supplemental Information: Fig. S1).

Fig. 4
figure 4

Boxplot of abiotic characteristics (n = 215 for pH and alkalinity and 84 for sediment grain size metrics; volume = 160.85 cm3) for the Bay of Fundy sites for 2020 and 2021 (date shown at the month level for visual clarity, with 1–2 sampling dates per month)

Within the Southern Gulf of St. Lawrence sites, the LN and BRIDG sites were located within the Miramichi River estuary, so they had much lower salinity than the sites in Kouchibouguac National Park due to the higher amount of freshwater input from the larger Miramichi River. The sediment chemistry characteristics were consistent across the sites with there being minimal differences in sediment pH and sediment alkalinity (Fig. 5). There were differences in the sediment characteristics, primarily sorting and kurtosis, between the sites in the Miramichi estuary and the Kouchibouguac National Park estuary (Fig. 5). The Miramichi estuary sites were much rockier at LN and full of peat at BRIDG, unlike the sand-dominated sites sampled in Kouchibouguac National Park. Boxplots of the remaining abiotic characteristics of the Southern Gulf of St. Lawrence sites, including water temperature, water alkalinity, percent organic matter, percent carbonates, mean grain size, and skewness, can be found in Supplemental Information Fig. S2. There was some variation between sites in these characteristics.

Fig. 5
figure 5

Boxplot of abiotic characteristics (n = 221 for pH and alkalinity and 63 for sediment grain size metrics; volume = 160.85 cm3) Southern Gulf of St. Lawrence sites for 2021 (date shown at the month level for visual clarity, with 1–2 sampling dates per month)

nMDS plots of the multivariate distances of the core samples based on abiotic conditions showed that the Southern Gulf of St. Lawrence and Bay of Fundy regions were generally differentiated from one another, although LL grouped with the Southern Gulf of St. Lawrence sites in August (Fig. 6). PERMANOVA results indicated there was a significant interaction between month and site for both regions (Table 3).

Fig. 6
figure 6

non-metric MDS plot of the abiotic conditions for both the Bay of Fundy and the Southern Gulf of St. Lawrence sites (n = 78 in Bay of Fundy; n = 40 in Southern Gulf of St. Lawrence; volume = 160.85 cm3) in 2021 with date shown at the month level for visual clarity. Sites LN, BRIDG, CAF, CAL, and KELLY are in the Southern Gulf of St. Lawrence while POCO, RH, CL, and LL are in the Bay of Fundy

Table 3 PERMANOVA results showing the sources of variation in the abiotic conditions of both the Bay of Fundy and Southern Gulf of St. Lawrence sites (n = 78 in Bay of Fundy; n = 40 in Southern Gulf of St. Lawrence; volume = 160.85 cm3)

Relationships between abiotic variables and infaunal invertebrate data

Generalized linear mixed models

For the Bay of Fundy sites, water alkalinity was included in the best models for all four dependent variables (Fig. 7, Supplemental Information Table S3). For the Southern Gulf of St. Lawrence sites, sediment pH, water alkalinity, and sediment sorting were the three most frequently appearing abiotic terms in the top models, with one or more of them included in each model (Fig. 7, Supplemental Information Table S3). Alkalinity, either water column or sediment, was included in the top model for all four dependent variables in both regions. For the Bay of Fundy region, juvenile bivalve abundance was poorly explained (total bivalve pseudo-R2 value = 0.06 and Mya arenaria pseudo-R2 value = 0.04) and total abundance well explained with pseudo-R2 value = 0.835 (Supplemental Information Table S3). For the Southern Gulf of St. Lawrence region, all models had pseudo-R2 values around 0.25, except species richness which had a pseudo-R2 value = 0.10 (Supplemental Information Table S3). Overall, most of the univariate models explained less than 25% of the variation in each dependent variable.

Fig. 7
figure 7

Coefficient plots of terms included in three best generalized linear mixed models for total bivalve abundance for A Southern Gulf of St. Lawrence and B Bay of Fundy, Mya arenaria abundance for C Southern Gulf of St. Lawrence and D Bay of Fundy, species richness for E Southern Gulf of St. Lawrence and F Bay of Fundy, and total abundance for G Southern Gulf of St. Lawrence and H Bay of Fundy (n = 122 for Bay of Fundy, 81 for Southern Gulf of St. Lawrence; volume = 160.85 cm3). These models show the relationships between each dependent variable and the abiotic characteristics. Terms with p < 0.05 are shown in blue

Distance-based linear models

The overall best model explained 47% of the variation in the similarity of the infaunal community among samples for the Bay of Fundy (Table 4). In the Bay of Fundy region, all variables were included in the best multivariate model with sediment pH (17%), carbonates content (12%), alkalinity of sediment (8%), and alkalinity of water (7%) explaining the most variability in the multivariate similarity in infaunal invertebrate communities among samples in the Bay of Fundy (Table 4). The overall best model for the Southern Gulf of St. Lawrence region explained 51% of the variation among samples (Table 4). In the Southern Gulf of St. Lawrence region, the best model also contained all the variables with salinity (10%), mean grain size (8%), skewness (5.5%) explaining the most variation in infaunal invertebrate community composition (Table 4).

Table 4 Marginal test results of DISTLM to determine the proportion of variation in the multivariate infaunal invertebrate community composition explained by the abiotic variables

Discussion

Both sediment and water column abiotic variables were important in determining the infaunal community composition at our study sites in both the Bay of Fundy and Southern Gulf of St. Lawrence regions. At sites in both regions, carbonate chemistry was associated with variation in intertidal infaunal invertebrate communities, as seen for both multivariate community composition and univariate biological metrics. For multivariate community composition, sediment pH was the best explanatory abiotic variable for the Bay of Fundy sites (17% of the variation). In the Southern Gulf of St. Lawrence, sediment pH explained only 0.5% of variation, with alkalinity of water (5.5%) and sediment (1.1%) explaining slightly more variation among samples. Meanwhile for univariate models of species richness, total abundance, and juvenile bivalve abundance, the abiotic variables explained considerable variation (for example, total abundance in the Bay of Fundy pseudo R2 0.835) for some of the univariate metrics but almost none for others (for example, juvenile Mya arenaria in the Bay of Fundy pseudo R2 0.039). Water alkalinity was included in all top models for the Southern Gulf of St. Lawrence region and most top models for the Bay of Fundy.

The two carbonate chemistry variables measured, pH and alkalinity, are part of a charge balance equation (Jones et al. 2016). Total alkalinity in coastal zones, both water column and sediment, is driven by production and dissolution of CaCO3, organic alkalinity, nutrient cycling, primary production, and respiration (Jones et al. 2016). Sediment pH on the other hand is driven by nutrient cycling, respiration, photosynthesis, and the dissolution of CaCO3 (Silburn et al. 2017). There is a great deal of overlap between the processes influencing total alkalinity in sediment and sediment pH in coastal oceans, further emphasizing the relationship between these two carbonate variables. However, alkalinity is usually not completely correlated with pH due to slight differences in driving forces (Jones et al. 2016). Carbonate chemistry is known to affect marine invertebrates in differing ways, therefore there is potential for carbonate chemistry to impact infaunal communities directly and indirectly through the differing effects among species. For example, the vast majority of calcifiers, including bivalves, gastropods, and crustaceans, have shown neutral responses to acidified conditions in several physiological processes including calcification, growth in size, and larval growth, while others had positive or negative effects (reviewed in Leung et al. 2022). To the best of our knowledge, there are no studies that have examined if carbonate chemistry has effects on entire infaunal communities in mudflat ecosystems.

In addition to carbonate chemistry, other abiotic factors influence the abundance and composition of infaunal invertebrate communities. For our Bay of Fundy sites, all abiotic variables were included in the best model of infaunal community and explained 47% of the multivariate variation with sediment pH, sediment carbonate content, and alkalinity of sediment being the three variables that explained the most variation. Community composition at our Southern Gulf of St. Lawrence study sites was similarly well explained (51%) by the abiotic variables. All the abiotic terms were included in the best model of infaunal community composition for the Southern Gulf of St. Lawrence, with water salinity, mean grain size, and sediment grain skewness being the three variables that explained the most variability. Past studies also found abiotic variables to be associated with infaunal community composition. For example, research conducted in New Zealand found that increasing organic matter enrichment, brought about by nutrient addition, resulted in lower sediment pH and reduced total abundance of macrofauna (Drylie et al. 2019). Spatial and temporal variation in abiotic conditions is also associated with differences in infaunal community composition. A study conducted by Gerwing et al. (2015) in the upper Bay of Fundy found that infaunal communities at their study sites were driven primarily by structural factors (~ 79%), which included spatial and temporal factors such as date and site. While our analysis did not include both abiotic variables and site and date in the same analysis to partition the variance, we did find significant spatial and temporal variation both for the multivariate infaunal community data and for univariate measures. In the Gerwing et al. (2015) study, abiotic variables explained 11% of the multivariate community structure while top-down factors, like predator populations, accounted for an even smaller amount of the variation (~ 6%). This suggests that in the upper Bay of Fundy community composition is more influenced by abiotic conditions than biotic conditions; however, the effect of these factors were minute in comparison to the spatial and temporal factors (Gerwing et al. 2015). In the present study, abiotic characteristics explained a large proportion of the variation (47%) in the multivariate model for our Bay of Fundy sites, which were located in the lower portion of the Bay. The upper Bay of Fundy, where Gerwing et al.’s (2015) study was conducted, has different biotic and abiotic factors, such as more bird predation and larger tidal ranges compared to our lower Bay of Fundy sites, which can all contribute greatly to the variability exhibited by the community composition. The connection between carbonate chemistry and infaunal invertebrate communities needs to be investigated further in other regions.

Univariate models found abiotic variables explained 2.2 to 83.5% of the variation in species richness and abundances. Water alkalinity was the only variable included in almost all top models for species richness and total abundance for both the Bay of Fundy and the Southern Gulf of St. Lawrence sites, having a slightly positive relationship between the variables, but it did not greatly explain the variation. For the Southern Gulf of St. Lawrence sites, sediment pH and sorting were both included in most of the top models. Grain size metrics were also important for the Bay of Fundy biological metrics and juvenile bivalve abundances. Previous work looking at the abiotic drivers of benthic species richness found that abiotic characteristics, such as percent gravel and depth, can play key roles in accounting for variability in biodiversity (reviewed by McArthur et al. 2010). The varied relationship of biodiversity metrics to abiotic characteristics in the present study may be due in part to responses of different species to their environment. For example, the best model of total abundance at sites in the Bay of Fundy region included five variables and explained ~ 84% of the variation in abundance. There was relatively little variation in the abundance at the sites in both years except for a spike at RH in the first 2 months of the 2021 sampling period. This coincided with a green macroalgal bloom observed at RH and a spike in the abundance of anoxia-tolerant species of polychaeta, such as Tubificoides benedii. This population spike at RH occurred during the warmest water temperatures observed in both years and a decline of sediment pH, which has been connected to anoxia in previous studies (Middelburg and Levin 2009; Widdicombe et al. 2009). Given the likelihood that different species are sensitive to certain abiotic factors (see review by MacArthur et al. 2010), examining individual species’ responses to abiotic conditions can provide valuable information.

The abundance of juveniles of all three bivalve species and of Mya arenaria individually were negatively related to water alkalinity at our sites in the Bay of Fundy and Southern Gulf of St. Lawrence regions. The models explained 19–26% of the variation in juvenile bivalve abundance at sites in the Southern Gulf of St. Lawrence, but only 2–6% of the abundance in the Bay of Fundy. The amount of variation explained in bivalve abundances in the two regions suggests there are additional factors that need to be considered and incorporated into future analysis, especially at the Bay of Fundy sites. Meseck et al. (2018) found that juvenile M. arenaria abundance in Long Island Sound was best explained by salinity, grain size, and total alkalinity; their best fit model explained 14%. The inclusion of alkalinity in the best model and the low amount of variance (0.10) explained by alkalinity in Meseck et al.’s (2018) model is consistent with the best model for M. arenaria abundance in the present study. Meseck et al. (2018) also looked at juvenile Nucula spp. abundance and found that grain size, pH, and sediment phosphate levels were key determinants. Two additional studies have looked at the importance of carbonate chemistry for bivalve species abundances. Clements and Hunt (2018) found that sediment pH was a predictor of juvenile M. arenaria abundance at study sites in the Bay of Fundy, including some of the same sites as the present study. The difference in key abiotic characteristics between the present study and Clements and Hunt (2018) may be due to differences in the variables measured and to differences in environmental conditions as the two studies were conducted 4 years apart. In Green et al. (2013), larval Mercenaria mercenaria settlement was greatly improved as aragonite saturation state increased in the field. Given the coupling of pH and alkalinity, the relationship of sediment pH and carbonate saturation state to bivalve species abundance in Green et al. (2013) and Clements and Hunt (2018) could have been an indicator of the importance of carbonate variables in general. The results of these past studies match the results we observed in the Gulf of St. Lawrence for M. arenaria but not what we found in the Bay of Fundy where sediment and water column carbonate chemistry were not influential for total bivalve or M. arenaria abundance. More work is clearly needed to determine what abiotic characteristics are drivers of individual bivalve species abundances.

While the infaunal community composition of the two regions overlapped, the sampling sites in the Bay of Fundy and Gulf of St. Lawrence regions had fairly distinct abiotic characteristics. The Bay of Fundy site CL had a community composition and species richness more like the Gulf of St. Lawrence sites than it did the remaining Bay sites, yet the abiotic conditions at CL were more similar to the other Bay of Fundy sites according to the multivariate analysis. While CL was the sandiest site in the Bay region, suggesting it would be more like the Gulf region sites abiotically in terms of sediment characteristics, other environmental variables drove abiotic multivariate differences between the two regions. Potentially there are biotic drivers that have more influence on the community composition at CL, and other sites, than the abiotic drivers. Alternatively, there are abiotic characteristics not measured that were more influential, as almost all the abiotic characteristics measured helped separate CL from the remaining Bay of Fundy region sites when looking exclusively at the Bay of Fundy sites.

The differing results in the univariate modeling for all the dependent variables between the Southern Gulf of St. Lawrence and Bay of Fundy study sites were not surprising due to the differences in the abiotic conditions between the sites. The Gulf of St. Lawrence sites had higher water temperatures and greater organic matter content in addition to a slightly smaller range of mean grain size, which may have also contributed to the importance of alkalinity. Muddy sediments can have positive alkalinity fluxes, meaning more alkalinity is released than absorbed, compared to sandy sites which generally have negative alkalinity fluxes (Van Dam et al. 2022). These alkalinity fluxes could be part of the reason why water alkalinity is a main explanatory variable in all the models completed for the Gulf of St. Lawrence. The differences in sediment characteristics between the two study regions, mudflat vs sandflat, likely play a role as sandier sediments such as those observed at the Gulf of St. Lawrence in the Kouchibouguac National Park sites generally have smaller abundances of infaunal species due to the influence of sand on sediment chemistry, organic matter, and grain size metrics (Braekman et al. 2014). Another potential cause of the differences between the two regions was the difference in sampling periods. As we were trying to capture the bivalve settlement periods in each region, we conducted the sampling at different times in the summer. This difference in sampling periods likely also contributed to the differences detected in the infaunal community between the two regions. It is important in future research to collect matching biotic and abiotic data for each sampling point to better determine smaller-scale influence of abiotic variables on infaunal invertebrate communities.

In the present study, there was temporal and spatial variation in the infaunal invertebrate communities and abiotic conditions, especially in the Southern Gulf of St. Lawrence sites. Additionally, sediment and water carbonate chemistry were associated with variation in infaunal invertebrate communities in the two regions. There were similarities in the important abiotic variables between the regions and between the types of models, specifically the inclusion of carbonate chemistry variables. Water column alkalinity was the carbonate chemistry variable included in all top models, both univariate and multivariate. Sediment pH was the variable that best explained the variance in the multivariate community composition at study sites in the Bay of Fundy region while in the Gulf of St. Lawrence percent carbonates was the best explanatory variable. The differences in the important sediment and water column abiotic variables between study sites in the two regions was not surprising given the different environmental conditions and oceanography. These results suggest that the response of invertebrate communities to current day drivers and future oceanic conditions may differ spatially. For example, infaunal communities at our sites in the Bay of Fundy region may be more impacted by changes to sediment carbonate chemistry while those at our sites in the Gulf of St. Lawrence region may be more impacted by both water column and sediment carbonate chemistry changes. Our study is the first to focus on the impacts of water column and sediment variables, especially carbonate chemistry, on the infaunal invertebrate community, specifically larger meiofauna and small macrofauna, in the Bay of Fundy and Southern Gulf of St. Lawrence. Understanding which abiotic characteristics most affect infaunal invertebrates is critical to predicting future responses to climate change and human disturbance, which is needed to manage important infaunal invertebrate species and communities on a more localized scale.