Green-lipped mussel associations with macroalgae on a mussel reef in New Zealand
Description
The goal of this study was to develop and test a tool that has the potential to aid restoration practitioners in the selection of suitable natural substrates for overcoming substrate limitation at mussel reef restoration sites in New Zealand and potentially elsewhere across the globe. This study aimed to; 1) create an index that scores the morphological features of macroalgae for their potential to support juvenile mussel recruitment based on prior published data, 2) compare the relative performance of various macroalgae species with differing morphology as attachment substrates for three size classes of juvenile green-lipped mussels or Perna canaliculus on two remnant mussel reefs, and 3) use these findings to evaluate the accuracy with which the index can predict the presence of attached juvenile mussels among macroalgal substrates based on the morphology of the macroalgae. This data details the presence of three different size classes of juvenile green-lipped mussels (i.e., <10 mm, 10 - <20 mm, and 20 - <30 mm in shell length) attached to different species of macroalgae sampled on two remnant intertidal mussel reefs in northeastern New Zealand (Waipu Cove and Pakiri Beach). The presence of juvenile mussels in each size class on macroalgal substrates is reported as total counts and as the proportion (%) of each species of macroalga observed during the reef sampling that exhibited juvenile mussel attachments. The macroalgae species that supported the highest juvenile mussel presence per mussel size class have the most promise for use in overcoming substrate limitation at mussel reef restoration sites. Scores (1-3) for each of the eight morphological features of macroalgae (i.e., canopy cover, canopy height, holdfast complexity, clustering frequency, planes of branching, branch spacing, degree of branching, and branch width) were generated for each species of macroalga by the Macroalgal Morphology Index (MMI). These scores were tested against the presence of juvenile mussels attached to each species of macroalga using a linear discriminant analysis (LDA) to determine the accuracy of the MMI's scoring criteria. The LDA revealed that collectively, the index scores for each morphological feature were able to predict the likelihood of mussel attachments to macroalgae for juveniles <10 mm in shell length with 75% accuracy but with only 40-60% accuracy for juveniles 10 – <30 mm. Additionally, a 10-fold cross-validation was employed to assess the relative importance of each morphological feature of macroalgae as a predictor of juvenile mussel presence. This revealed that holdfast complexity and canopy cover were the strongest predictors of attached juvenile presence for all three mussel size classes, and planes of branching, branch width, degree of branching, and branch spacing were generally strong predictors of the presence of attached juvenile mussels <10 mm.
Files
Steps to reproduce
Between May 2023 and June 2024, sampling was conducted in the intertidal during low tide on two remnant intertidal mussel reefs in northeastern New Zealand (Waipu Cove and Pakiri Beach), in which a series of 13 m alongshore transects were laid across the width of the mussel reef to ensure that a range of available macroalgae substrates were sampled for attached juvenile green-lipped mussels or Perna canaliculus (<30 mm in shell length). At 0.5 m intervals along each transect, 10 × 10 cm quadrats were placed over the substrate to standardise the area of macroalgae searched for cryptic juvenile mussels, resulting in 26 quadrats per transect. Each macroalga species with a holdfast located within the quadrat was searched along the entire length of its thallus for any P. canaliculus juveniles attached to it by byssal threads. Once a juvenile mussel was located, its shell length was measured using callipers. For each macroalga species within a quadrat, the presence or absence of juvenile mussels within three size classes was recorded, i.e., <10 mm, 10 – <20 mm, and 20 – <30 mm in shell length. For each macroalga species in a quadrat, once one mussel in a size class was observed, the presence of mussels in that size class was recorded, and any subsequently observed mussels in the same size class were not recorded. Only macroalgae species that were observed in at least five quadrats per site were included in the analysis. For statistical analysis, the presence of juvenile mussels attached to different macroalgae species were tested against the scores generated for each macroalga species by the Macroalgal Morphology Index (MMI) based on their morphological features using a linear discriminant analysis (LDA) to determine the accuracy of the MMI's scoring criteria. The predictor variables assessed in the LDA were the eight morphological features of macroalgae listed in the MMI (i.e., canopy cover, canopy height, holdfast complexity, clustering frequency, planes of branching, branch spacing, degree of branching, and branch width). The outcome variable used for this analysis was juvenile mussel presence converted from proportional data to binary categorical data (i.e., “high presence” or “low presence”) based on the median for juvenile presence in each size class at each site, allowing relative differences in presence to be assessed without the interference of broader trends in juvenile presence. The LDA was implemented using the lda function in the MASS package in R version 3.5.2. To assess the performance of the LDA and avoid overfitting, a 10-fold cross-validation was employed. The relative importance of each morphological feature of macroalgae as a predictor of juvenile mussel presence was quantified using the varImp function in the caret package in R, which evaluates the relative contribution of each predictor based on its influence on the outcome variable.