The Interspecific Abundance–Occupancy Relationship in Invertebrate Metacommunities Associated with Intertidal Mussel Patches

Abstract

To explain the distribution and abundance of species, ecology searches for general models. A pattern often encountered in nature is the interspecific abundance–occupancy relationship (AOR), which describes how the mean local abundance of species relates to the proportion of local sites that each species occupies. Both are central variables in ecology and are often positively correlated, although exceptions have been found. As most AOR research has been conducted with terrestrial systems, recent studies are testing for its occurrence in marine systems. This contribution tests the AOR for invertebrate metacommunities associated with intertidal mussel patches. Using data from six coastal locations in Nova Scotia (Canada), this study shows that the negative binomial model properly describes the relationship between abundance and occupancy for these systems. The degree of wave exposure (wave-sheltered versus wave-exposed habitats) had some influence on the shape of the AOR. Overall, these findings extend the applicability of the AOR to intertidal invertebrate metacommunities. The raw data are included as part of this article to help future syntheses on the AOR, which will need data for a variety of terrestrial and aquatic environments.

1. Introduction

In its quest to explain the distribution and abundance of species, ecology searches for general models that can predict outcomes for unstudied systems. A commonly observed pattern is the interspecific abundance–occupancy relationship (AOR), which relates the mean local abundance of species to the proportion of local sites that each species occupies. AOR theory is thus relevant because it aims to make inferences or predictions between species abundance and occupancy, which are central variables in ecology. Normally, these two variables are positively related, but the absence of a relationship has also been found [1,2,3]. At present, the factors driving the shape of the AOR are incompletely understood and include, for example, niche breadth, vital rates, body size, migration, dispersal, successional stage, and urbanization [4,5,6,7,8,9,10,11,12,13,14,15].

Historically, surveys have documented the AOR predominantly in terrestrial systems, as exemplified by the studies cited above. Thus, in recent years, AOR surveys have been conducted in marine systems, such as muddy [16,17] and rocky [18] shores. Studying a wide diversity of systems is necessary to understand how ubiquitous the AOR is in nature. An equally important step is the publication of AOR datasets. Currently, there is a scarcity of such datasets available from public sources, probably because data publishing is a relatively recent phenomenon in science. These are important gaps to address because future syntheses of the AOR will require information from different environments. With these notions in mind, this study investigates the AOR for invertebrate metacommunities associated with intertidal mussel patches, making the underlying data freely available.

Intertidal systems are those between the highest and lowest tide marks on marine shores [19,20]. On many temperate rocky shores, mussels form intertidal patches of densely packed individuals that host many small species of invertebrates thanks to the food and shelter that such patches provide [21]. As mussel patches are discrete habitat units for the associated invertebrates, these invertebrate assemblages are spatially structured as metacommunities. A metacommunity is a group of discrete communities that are connected by some degree of migration and dispersal [22]. The spatial structure of these systems makes them ideal for evaluating the AOR and possible sources of variation on marine shores. The present study demonstrates the occurrence of the AOR in these metacommunities and examines if the shape of the AOR depends on wave exposure, an important factor in intertidal ecology [23,24,25]. Overall, this study expands the domain of application of the AOR and provides data that will be useful for future syntheses as data for a wider diversity of systems accumulate.

2. Materials and Methods

To address this study’s objectives, I used data that were originally collected to describe the invertebrate species occurring in rocky-intertidal mussel stands from Nova Scotia, Canada [26]. The following paragraph summarizes aspects of the sampling design of that study that are relevant to understand the present study (see full methodological details in [26]).

Between early September and early October 2012, six rocky-intertidal locations spanning 317 km of the Atlantic coast of Nova Scotia were sampled during low tides. These locations belong to the Northwest Atlantic cold-temperate biogeographic region [27,28]. Three locations are wave-exposed because they face the open ocean directly: Tor Bay Provincial Park (45.182894° N, 61.353258° W; TB hereafter), Crystal Crescent Beach (44.447372° N, 63.622214° W; CC), and Kejimkujik National Park (43.818614° N, 64.834747° W; KE). The other three locations are semienclosed without open waters visible from the shore, so they are wave-sheltered: Webber Cove (45.188333° N, 61.354722° W; WC), Casino Nova Scotia (44.651944° N, 63.573611° W; CN), and Halifax Harbourfront (44.648056° N, 63.570278° W; HH). A map indicating these locations is shown in Figure 1. On this coast, two intertidal mussel species (Mytilus edulis and M. trossulus) can form dense patches on the rocky substrate, with body sizes being consistently larger at wave-sheltered habitats than at wave-exposed habitats [29] (see this size difference in Figure 2 published in [26]). Both mussel species are almost visually indistinguishable [30] and occur in mixed stands in these habitats [29,31,32]. At the mid-intertidal zone of each studied location, the mussels and all associated invertebrates were collected from 15 patches that were fully covered by mussels. Each collected patch had an area of 100 cm² and was tens of cm or several m apart from other patches. The associated invertebrates larger than 0.5 mm were identified, and the number of individuals of each such species was measured for each mussel patch. These data are available from a recent data paper [33].

Since the studied locations are separated by the coastal seascape, each location can be viewed as hosting a metacommunity of invertebrates associated with mussel patches. Thus, I calculated the AOR separately for each of the six surveyed locations. Although different models have been explored to describe the AOR, the negative binomial model has an advantage because it often yields the greatest fit [34] and is based on the degree of spatial aggregation, which is a common feature of biological populations [35]. In fact, this property makes the negative binomial model a more realistic alternative than models that are based on random distribution patterns or on restricted conditions of aggregation that exclude natural variation [34]. The negative binomial model is P = 1 − (1 + µ/k)^−k, where P is the occupancy of a species (proportion of mussel patches occupied by a species), µ is the average abundance of a species (mean number of individuals per patch, averaged over all surveyed patches), and k is a spatial aggregation parameter [34]. I parameterized the AOR for each metacommunity (location) through nonlinear least-squares regression using PRISM 6 software. The data on species abundance and occupancy for the six studied metacommunities are freely available from the figshare online repository [36].

3. Results

The 2012 survey on which the present study is based encountered a total of 50 invertebrate species in intertidal mussel beds from Nova Scotia. The results of that survey (including lists of species and their abundance in each mussel patch) have been published elsewhere [26,33] and are therefore not repeated here. The results that address the specific objectives of the present study are described below.

The AOR was significant (p < 0.001) for the six surveyed metacommunities (

Table 1. Summary results of the AOR obtained for each surveyed location (invertebrate metacommunity), including the model parameter k and its 95% confidence interval, the adjusted R² with its p-value, and the sample size N (number of species encountered at each location).

Location	k (95% C.I.)	Adjusted R2 (p)	N
Webber Cove (wave-sheltered)	0.32 (0.13–0.51)	0.51 (0.0004)	20
Halifax Harbourfront (wave-sheltered)	0.65 (0.29–1.02)	0.94 (<0.0001)	12
Casino Nova Scotia (wave-sheltered)	0.43 (0.11–0.76)	0.67 (0.0006)	13
Tor Bay Provincial Park (wave-exposed)	1.02 (0.59–1.45)	0.92 (<0.0001)	28
Crystal Crescent Beach (wave-exposed)	1.23 (0.66–1.79)	0.94 (<0.0001)	24
Kejimkujik National Park (wave-exposed)	1.00 (0.64–1.36)	0.95 (<0.0001)	22

, Figure 2). The degree of fit (adjusted R²) varied among locations, from a lowest value of 51% at WC to a highest value of 95% at KE. The spatial aggregation parameter (k) was lower for wave-sheltered locations than for wave-exposed locations, but significant differences (indicated by the lack of overlap of 95% confidence intervals of k; Table 1) only occurred between WC (a wave-sheltered location) and TB, CC, and KE (the three wave-exposed locations).

4. Discussion

The first conclusion of this study is that the negative binomial model [34] properly describes the relationship between abundance and occupancy for invertebrate metacommunities associated with intertidal mussel patches. Therefore, this adds to recent studies that have demonstrated the occurrence of the AOR in other coastal marine systems [16,17,18], enriching the knowledge base that used to be based until recently almost exclusively on terrestrial systems [4,5,6,7,8,9,10,11,12,13,14,15]. The data attached to this article [36] should help future syntheses of the AOR once enough datasets are produced for different ecological systems.

The second conclusion of this study is that wave exposure was related to some extent to variation in the single parameter (k) of the negative binomial model that describes the AOR. While two wave-sheltered locations (CN and HH) did not differ significantly from the three wave-exposed locations (TB, CC, and KE), the other wave-sheltered location (WC) differed statistically from the three wave-exposed locations. A previous study using rocky-intertidal communities of primary space holders and their mobile consumers (no species living in mussel patches were considered) found no significant effects of wave exposure on k [18]. Thus, it remains to be investigated under what specific conditions wave exposure can influence the aggregation parameter. Interestingly, as found for some of the locations analyzed for this study, high levels of model fit were also found for marine communities from sedimentary shores [17].

Understanding the precise causes of the differences in AOR observed among the surveyed metacommunities remains extremely challenging. On the one hand, the drivers of the shape of the AOR are incompletely understood in general, as noted above. On the other hand, the drivers that structure invertebrate assemblages in intertidal mussel stands are poorly known, since intertidal research has typically aimed at understanding patterns for primary space holders and their consumers [37,38,39,40]. In general terms, metacommunity structure depends on spatiotemporal changes in environmental filtering, biotic interactions, food supply, and dispersal and migration among patches [22,41,42,43]. For the small invertebrates living in intertidal mussel stands, however, knowledge gaps exist at almost all levels, from abiotic tolerances in such environments, to biotic interactions, to recruitment within such stands. It is also worth noting that snapshot abundance data (such as the data commonly used for AOR research) often cannot easily reveal the complex interaction webs that underlie community structure [44,45,46], for which controlled experiments are required [47]. Therefore, these areas need dedicated research efforts to yield further advances.