VU Research Portal

Predation on intertidal mussels

Waser, A.M.

2018

document version

Publisher's PDF, also known as Version of record

Link to publication in VU Research Portal

citation for published version (APA)

Waser, A. M. (2018). Predation on intertidal mussels: Influence of biotic factors on the survival of epibenthic

bivalve beds.

General rights

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain

• You may freely distribute the URL identifying the publication in the public portal ?

Take down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

E-mail address:

intertidal soft-sediment bivalve

beds: No evidence for ecosystem

collapse in the Dutch Wadden Sea

Jaap van der Meer, Norbert Dankers, Bruno J. Ens, Marnix van Stralen,

Karin Troost and Andreas M. Waser

Abstract

Recruitment and fate of all 1436 mussel and oyster beds of the Dutch Wadden Sea were studied for the period 1999-2013. Cox’s proportional hazard rate model with fixed-time covariates such as orbital speed, inundation time and bed size and type, showed that large, shallow lying beds that experience a low orbital speed, live longer. Yet the most striking result was that mixed beds do have a much lower hazard rate than pure oyster or mussel beds. Simulation studies, using the observed recruitment series, which was very variable, and the estimated baseline survival curve, showed large variability and strong serial correlation in total bed area, implying that the present area, though it is lower than before, does not point to a systematic deviation from the pre-1990 situation, i.e. the situation before intensive fisheries and the disappearance of most beds around 1990. Claims that we have witnessed ecosystem collapse as a result of the fisheries and that bivalve bed recovery is impossible without restoration efforts are premature and not supported by our analysis. On the contrary, the observed high survival rate of mixed beds and the expectation that mixed beds will predominate in the near future, can easily result in much larger future bed coverage than what has ever been measured before.

Introduction

Birth and death processes are at the core of population ecology and demography. Obviously, most attention has been paid to the birth and death rates of individual organisms. Sometimes, however, it might be more convenient to work at a higher level of biological organization (Boughton & Malvadkar 2002). Students of colonial insects such as ants and termites, often focus on the birth and death of colonies. The life time of an ant colony may coincide with that of the queen, in which case one might argue that work is still done at the level of the individual organism. But, as queen supersedure within a colony occurs in many species, such is not necessarily true. In plant ecology, Watt’s seminal paper on the dynamics of plant communities, resulted in the idea of gap-phase dynamics, which still plays a central role in forest ecology (Watt 1947, Gratzer et al. 2004, Hunter et al. 2015). In forests, gaps appear as a result of, for example, storms or of the fall of a single canopy individual. The regeneration rate of the gap is then related to the size of the gap and characteristics of the surroundings (van der Maarel 1996). Similar thoughts developed in coastal ecology, where hard substrata are discontinuous pieces of habitat surrounded by sand and mud. These hard substrata can almost entirely be covered by epifaunal invertebrates such as sponges, tunicates and mussels. Clearings are regularly made either by physical forces or by predators. The recolonization of such clearing will then, just as in the case of forest gaps, depend upon the size of the clearing and the characteristics of the surroundings (Connell & Keough 1985). Eventually these ideas evolved in a more general theory of patch dynamics, where the birth, growth or shrinkage, and death of patches rather than of individual organisms is the central issue (Levin & Paine 1974, Levin et al. 1993, Pickett & White 1985). Metapopulation ecology, introduced by Hanski, makes a further step by using entire populations, connected into a meta-population, as the main unit of interest (Hanski & Gilpin 1991). That is, the birth and death of connected populations are studied.

It is interesting to note that the choice of Watt was not so much based on theoretical, but merely on practical considerations. He acknowledges that ‘the ultimate parts of the community are the individual plants’, but adds ’but a description of it in terms of the characters of these units . . . is impracticable . . . ’ (Watt 1947). In fact, the same argument is used by students of colonial insects, subtidal epifauna and island populations. It is much easier and less data demanding to describe the fate of patches, communities or populations than that of of individual organisms. Our interest is in the dynamics of intertidal bed-forming bivalve populations living on the soft-sediments of the Dutch Wadden Sea, such as the blue mussel (Mytilus edulis) and the recently introduced Pacific oyster (Crassostrea gigas). These intertidal bivalve beds are the habitat of many benthic and epibenhtic species and attract numerous birds and fishes, for which these beds provide a rich food source (Goss-Custard et al. 1982). As such they have a high conservation value, and conservationists were worried to see that in the early 1990s, in a period with low spatfall and ongoing fisheries, hardly any intertidal bed was left (Herlyn & Millat 2000, Ens 2006). Since then these beds have partly recovered (Dankers et al. 2001), but doubts remain to what level compared to the pristine situation (Dankers et al. 2001). It has even been claimed that due to sediment disturbance by mechanical fishing activities, the Wadden Sea has undergone drastic changes and collapsed into an alternative stable state, where recovery of bivalve beds is basically impossible without large-scale restoration projects (Eriksson et al. 2010). Recently, quite some effort has been put in initiating artificial restoration projects, but without much success (van der Heide et al. 2014, de Paoli et al. 2015).

state, due to the invasion of the Pacific oyster (Troost 2010). Although the Pacific oyster was introduced in the Dutch Wadden Sea already around 1978, it did not really spread until the late 1990s, i.e. well after the disappearance of the intertidal mussel beds (Reise 1998, Diederich 2005, Nehls et al. 2006, Troost 2010). Whereas in the past bivalve beds consisted entirely of mussels, nowadays bivalve beds may consist entirely of mussels or oysters, or a mixture of these two species. Apart from studying present-day recruitment and survival of bivalve beds in general, we will also estimate these parameters for different types of bivalve bed.

Finally, we will show that there is no evidence at all for an ecosystem collapse and that present recruitment and survival rates of mussel-oyster beds are sufficient to achieve and maintain historic population levels. Due to the presence of the invading Pacific oyster, we even predict an increase in bed area in the long run.

Materials and methods

Definition of a mussel bed

In 2002, Dutch, German and Danish scientists agreed upon a common definition of littoral mussel beds to facilitate trilateral comparisons (CWSS 2002, Herlyn 2005, Nehls et al. 2009b). A mussel bed was defined as a benthic community structured by blue mussels that may consist of an irregular collection of more or less protruding smaller patches, separated by open spaces. Patches should be larger than 1 m in diameter and within 25 m distance from each other. A collection of humps smaller than 1 m in diameter should have an areal coverage of more than 5% in order to be grouped within one patch (Figure 3.1). Following the expansion of the Pacific oyster into many mussel beds of the Wadden Sea the original definition gained a more general character and is now used to define different types of bivalve beds with varying amounts of mussels and oysters. In the Netherlands, beds are defined as pure mussel beds if the cover of Pacific oysters is less than 5% and as pure oyster beds if the cover of mussels is below 5%. In all other cases, beds are denoted as mixed beds.

A

d

e

B

B

d

e

B

Figure 3.1: Visual representation of the demarcation of mussel beds (adapted from Nehls et al. 2009b).

they are further away from each other than 25 meter (Figure 3.2). Within this definition, beds can neither split nor merge. They can only be born, survive or die. The definition requires an iterative procedure to uniquely classify all patches in beds.

1999.048

Figure 3.2: Changes over time of mussel bed 1999.048. Each colour represents a different year. Note that some green patches are only grouped within the same bed, as they overlap with the same blue bed, that in itself consists of patches that only form one bed because they overlap with the same red bed. The thick red line indicates a distance of 50 m.

Field data

Each spring Wageningen Marine Research (formerly IMARES) and the private company MarinX map all the littoral mussel and oyster beds in the Dutch Wadden Sea. During low tide, researchers walk around the beds to demarcate the contours and their tracks are recorded by GPS and imported into GIS. Characteristics of the beds, such as percentage cover of mussels and oysters, are recorded. Bed locations are determined by a previous inspection of the area from an aero-plane flying at an altitude of 500 m. Due to time constraints not all locations can be visited each year. For unvisited locations, use may be made of the data set from the previous year or the year after. For example, if in year 2008 a 2-year old bed is found at a particular location which was not recorded in 2007, the 2007 data set is updated to include that bed. Bed distribution data for the period 1999–2013 are used. Details of the sampling procedure are in Steenbergen et al. (2006).

Survival analysis

only classified as mussel beds when they were categorized as such for all years in which they existed. The same holds for oyster beds. All other cases are classified as mixed beds. So, for example, a bed that experiences a single transition from a mussel bed to an oyster bed only, is classified as a mixed bed in the survival analysis. The quantitative covariates (longitude, log bed size, depth, orbital speed, inundation time) are also considered as fixed-time covariates, which means that they are assumed not to change over the study period.

Table 3.1: Censoring. For those beds ‘born’ in 1999 or still alive in 2013, the exact lifetime is not known. One can only say that they ‘lived’ for at least a specified number of years, i.e. 4+ means a life time of at least 4 years. 1999 2000 2001 2002 2003 . . . 2011 2012 2013 1999 1+ 2+ 3+ 4+ 5+ 13+ 14+ 15+ 2000 1 2 3 4 . . . 12 13 14+ 2001 1 2 3 . . . 11 12 13+ 2002 1 2 . . . 10 11 12+ 2003 1 . . . 9 10 11+ . . . 2011 1 2 3+ 2012 1 2+ 2013 1+

Simulation model

We developed a very simple simulation model of bed dynamics, where in each year of the simulation one number is randomly selected from the observed 14-year series (2000-2013) of recruitment area data. These data are visualized in Figure 3.5. The sampled number is the initial size of a simulated cohort. For each cohort and each subsequent year, the fraction of the initial recruitment area that had survived was simply taken from the overall survival curve, which is visualized in Figure 3.4 by the black line. So the stochasticity in the predicted series is entirely determined by the recruitment, and the predicted variability should therefore be considered as a minimum. In practice variability in survival will also contribute. The population was simulated for a period of 1000 years.

Software

All analyses were performed using the R platform (R Development Core Team 2015). We used the packages sp, maptools, rgeos, rgdal, raster, and spdep for spatial analysis and the package survival for survival analysis. Scripts are available from the corresponding author.

Results

Table 3.2: The original data can be summarized by a 1436 by 15 matrix where each element contains the type of bed; M is mussel, C is oyster, B is mixed, and 0 indicates that the bed has not yet been established or has disappeared. 1999 2000 2001 2002 2003 2004 . . . 2010 2011 2012 2013 1999.001 M 0 0 0 0 0 . . . 0 0 0 0 1999.002 M M M B B C . . . B C B B 1999.004 M M M M 0 0 . . . 0 0 0 0 1999.007 M 0 0 0 0 0 . . . 0 0 0 0 1999.008 M 0 0 0 0 0 . . . 0 0 0 0 1999.009 M 0 0 0 0 0 . . . 0 0 0 0 . . . . ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● 1 2 3 4 5 6 7 1 2 3 4 5 6 7 log10 Area inm 2 in year i log 10 Area in m 2 in y ear i + 1 ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● _● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ● ●●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● _● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ●_● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●_● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● _● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● _●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ●● ● ● ● ● ●● ●● ●● ● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●_● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ●_● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●_● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ●_● ● ● ● ● ● ● ● ● ● ● ● _● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●_●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ●● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ●● ● ● ● ● ● ● ● _● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ●●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●● ● ● ●● ● ● ● ● ● ● ● ● ● ●● ●● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

Figure 3.3: The area of individual beds in two succeeding years. Points above the diagonal line point to growth and those below the line to shrinkage of beds.

Table 3.3: Year of ‘birth’ (rows) versus year of ‘death’ (columns). 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 1999 55 34 9 12 5 2 0 0 2 0 1 0 1 1 23 2000 0 60 9 10 3 1 1 0 1 1 2 0 2 2 9 2001 0 0 15 12 2 1 0 1 1 1 0 0 0 1 2 2002 0 0 0 138 35 16 12 15 3 4 3 0 2 0 35 2003 0 0 0 0 42 3 3 0 2 0 0 0 0 0 8 2004 0 0 0 0 0 99 23 24 2 7 5 0 1 5 37 2005 0 0 0 0 0 0 62 13 1 1 1 0 2 1 15 2006 0 0 0 0 0 0 0 74 21 8 9 1 3 3 8 2007 0 0 0 0 0 0 0 0 43 13 8 1 2 3 2 2008 0 0 0 0 0 0 0 0 0 9 8 2 6 2 15 2009 0 0 0 0 0 0 0 0 0 0 29 8 6 8 13 2010 0 0 0 0 0 0 0 0 0 0 0 44 22 14 19 2011 0 0 0 0 0 0 0 0 0 0 0 0 22 1 18 2012 0 0 0 0 0 0 0 0 0 0 0 0 0 8 30 2013 0 0 0 0 0 0 0 0 0 0 0 0 0 0 51

0

5

10

15

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Year

Sur

viv

al

All

Mussel

Mussel/Oyster

Oyster

Figure 3.4: Survival curves of all beds together, and of the different bed types separately; mussel beds, mixed beds and oyster beds. Dashed lines show average ± SEs.

Table 3.4: Results of Cox proportional hazards model. β exp(β) se(β) z p Longitude 0.43 1.04 0.035 1.25 0.21 Bed size -0.46 0.63 0.033 -14.01 < 0.001 Depth -0.05 0.95 0.042 -1.24 0.21 Orbital speed 0.09 1.09 0.037 2.39 0.017 Inundation 0.14 1.15 0.047 2.98 0.003 Mixed -2.17 0.12 0.153 -14.17 < 0.001 Oyster -0.70 0.50 0.165 -4.21 _{< 0.001}

Table 3.5: Transition matrix showing annual transitions from (rows) to (columns) a specific bed type, all 1436 beds by 14 transitions per bed.

None Mussel Mixed Oyster

None 14855 1055 44 52

Mussel 1122 1718 31 4

Mixed 64 111 494 51

Oyster 105 10 18 370

Table 3.6: As Table 3.5, but None now stands for no bed or for a bed younger than 4 years old. Mussel, Mixed and Oyster refer to beds older than 3 years.

None Mussel Mixed Oyster

None 18289 115 19 12

Mussel 242 591 22 3

Mixed 32 80 406 33

Oyster 58 1 11 190

of 720 cases, which is 9% (Table 3.5). Oyster beds in 105 out of 503 (21%) and mussel beds even in 39%, i.e. in 1122 out of 2875 cases (Table 3.5). Restricting the analysis to beds older than 3 years yields similar disappearance rates of 6%, 22% and 28%, respectively (Table 3.6). So, the differences are large, but not as large as the proportional hazards model indicates.

1999

2002

2005

2008

2011

Year

Area ne

w beds

km

0

5

10

15

Figure 3.5: Surface area of new beds in spring versus year. Grey indicates the area of pure mussel beds.

Discussion

400

450

500

550

600

0

10

20

30

40

Year

Area m

usselbed

km

800

850

900

950

1000

0

10

20

30

40

Area m

usselbed

km

Figure 3.6: Simulated total surface area of all mussel beds for two arbitrary periods. Grey line shows the true total bed area over the study period 1999–2013.

We set out to study survival and recruitment of bivalve beds to investigate the claim that the ecosystem had collapsed after overfishing in the late 1980s. We assume that ecosystem collapse implies that nowadays bed areas are much lower and bed dynamics very different from the past. Thus, we must compare our results to historical information.

The simulation study showed that the large observed variability in annual recruitment of new beds produces strong serial correlation, where periods of several decades of bed area far above or far below average bed area are no exception. Even in the well-studied Dutch Wadden Sea, data before 1990 of total bed area are uncertain. Historic data spanning several decades are entirely lacking. In an extensive overview of all data reported in governmental reports, Dankers et al. (Dankers et al. 2003) conclude that in the 1970s and 1980s between 10 and 56 km2of bed area was present. A slightly different approach by the same authors arrived at a somewhat smaller, but still large interval of 17–48 km2(Dankers et al. 2003). The indicated uncertainty is not so much due to interannual variability, but merely to the use of different methods, such as aerial or ground sampling, and different ways of demarcating beds. Our approach yielded a range of 10–30 km2, which is entirely due to interannual variability. Other long-term studies also point to a huge variability and a very right-skewed distribution of mussel recruitment among years. For example Beukema’s 1970–2008 series at Balgzand, a tidal flat in the westernmost Dutch Wadden Sea, only contained four years with good mussel recruitment (Beukema et al. 2010). A recent study by Beukema confirms these results (Beukema et al. 2015). This is a general phenomenon for most bivalve species in soft-sediment habitats. The same Beukema series revealed six outstanding recruitment years for the Baltic tellin Limecola balthica, and also only four years for the cockle Cerastoderma edule. In the Wash, spatfall of cockles and mussels was poor in most years between 1990 and 1999. Significant recruitment that increased the shellfish stock occurred in only 14% and 19% of the study years for mussels and cockles, respectively (Dare et al. 2004). Thus, as already indicated by the simulation study, full recovery of bivalve bed area after the disappearance in the early 1990s to pre-1990 levels, might take some time. And if one also considers the uncertainty about the pre-1990 area, it seems premature to assume ecosystem collapse and invest effort in costly large-scale restoration programs.

2011

2012

2013

Spr

ing

A

utumn

Length (mm)

Density

(

m

)

The much higher survival probability of oyster beds and particularly mixed beds than that of pure mussel beds, even suggests that total bed area may increase to much higher levels compared to the pre-1990 situation. Further simulation studies (not presented here) arrived at an area close to 100 km2when all beds become mixed ones, but density-dependent processes will probably prevent that such large values will be reached (Folmer et al. 2014). Nevertheless, our analysis points to larger bed areas in the near future than presently observed.

Using beds as the unit of observation, and estimating their recruitment and survival, provided some insight in the underlying processes and relevant environmental factors that govern total bed area. Though of interest to managers and conservationists, it does not tell much about the fate, in terms of recruitment and survival, of individual mussels and oysters. Beds are continuously replenished with new cohorts (Figure 3.7), and we have not investigated the link between bed survival and recruitment and survival of individual animals. So we do not know, for example, whether most recruitment in terms of individuals, occurs in existing beds or in newly formed beds. It also cannot be ruled out that stable beds with low hazard rates are linked with relatively low survival of individuals, but high recruitment of new cohorts. Perhaps beds that are constantly renewed are more stable than beds that lack regular replenishment. The link between these two levels of biological organization, the individual bivalve and the bed, is an interesting topic for future studies.

Acknowledgments

Box 3.1

Recruitment, growth and survival of mussels

When studying the progression of specific bivalve beds it is also of interest to consider the fate of individual mussels. Information on recruitment, survival and growth of the mussels allows to get more insight into the link between bed survival and the fate of individual mussels as well as on the predator-prey relationships, and secondary production of the mussels.

Mussel bed (W001_A1) Oyster bed (W001_A0)

Spr ing 2010 A utumn 2010 Winter ₂₀₁₁ Spr ing 2011 A utumn 2011 Winter ₂₀₁₂ Spr ing 2012 A utumn 2012 Spr ing 2013 A utumn 2013 0 10 20 30 40 50 60 0 10 20 30 40 50 60 200 400 600 200 400 600 200 400 600 200 400 600 200 400 600 200 400 600 200 400 600 200 400 600 200 400 600 200 400 600

Length (mm)

Density

(

m

)

As part of long-term investigations of several epibenthic bivalve beds (see Chapter 4: Waser et al. 2016a, for an overview map and sampling details), mussels were repeatedly sampled in order to investigate several parameters (i.e., birth, growth and death) of the different cohorts within the mussel population on a given bivalve bed. In bivalves, concentric growth rings on the shell surface, which are usually produced annually during the winter period of ceased shell growth, are commonly used to age and assign the individuals to a specific cohort. In mussels, however, age determination via growth lines is complicated, since growth rings are either absent or difficult to discern from irregular occurring disturbance rings (storms, dredging, predation) (reviewed in Richardson 2001).

Length frequency distributions of mussels

Alternatively, the separation of distinct cohorts can be achieved with the help of length frequency distributions, in which mussels can be statistically assigned to a specific cohort on the basis of their length (Bhattacharya 1967, Wanink & Zwarts 1993). Figure B3.1 shows the length frequency distributions of mussels sampled at two different bivalve beds sampled in 2010–2013. Both beds are located at the northern tip of Texel and differ in their composition of bivalve species. While bed W001_A0 is characterized by a high biomass of Pacific oysters, bed W001_A1 consists almost entirely of mussel recruitment from the year 2009.

This analysis requires that the number of individuals sampled is big enough and reflects the size structure of the actual population. Where recruitment is seasonal and variability in individual growth rates is low, individual year classes can be identified as distinct modes and can be followed over time, as in the case for bed W001_A1 (Figure B3.2). In contrast, the bed rich in oysters (W001_A0) showed a variability in mussel distribution between the different sample dates (Figure B3.1). This variability was observed on all investigated oyster dominated beds, presumably caused by high fluctuations in individual growth rates or size specific predation (see e.g., Chapter 6: Waser et al. 2015), and prevents detailed analysis of mussel cohorts.

Von Bertalanffy growth

The classification of mussels into specific age classes can be used to assess the cohort specific survival rates (Figure B3.3) and to estimate the mussel growth over time (Figure B3.4). In bivalves, the Von Bertalanffy growth equation is commonly used to describe individual growth. However, the original equation does not take seasonal variations in growth rate into account. A popular modification considering fluctuations in seasonal growth is Somers’ Von Bertalanffy growth model (Somers 1988, García-Berthou et al. 2012). This growth model is described as

L(t ) = L∞(1 − exp(−K (t − t0) − S(t) + S(t0))),

with S(t ) = (C K /2π)sin(2π(t − tS)),

so S(t0) = (C K /2π)sin(2π(t0− tS)),

where L(t ) is the expected length at time t ; L∞is the asymptotic length; K is the exponential

rate to approach the asymptotic length; t0is the theoretical time at which the average length

would be zero; C modulates the amplitude of the growth oscillations (i.e., C = 0: no seasonal oscillation; C = 1: stopped growth e.g. in winter); tsis the time between time 0 and the start

0.00 0.06 0.12 0.00 0.04 0.08 0.00 0.04 0.08 0.00 0.04 0.08 0.00 0.03 0.06 0.00 0.02 0.04 0.00 0.02 0.04 0.00 0.03 0.00 0.03 0.06 0 10 20 30 40 50 60 0.00 0.02 0.04 Length (mm) 2010 – Apr 2010 – Aug 2011 – Jan 2011 – Mar 2011 – Nov 2012 – Jan 2012 – Apr 2012 – Sep 2013 – May 2013 – Oct Probability Length (mm)

100 500 1000 5000 0 1 2 3 4

Age (years)

Density

(

m

)

Figure B3.3: Catch curves of different mussel cohorts on bivalve bed W001_A1 sampled in 2010–2013. It is assumed that new recruitment stages are born each year in May.

0 1 2 3 4 10 20 30 40 2009 2010 2011 2012 2013 Cohort =−0.086 = 0.075 K = 0.39 = 55 C = 1 L_∞ t0 ts

Age (years)

Shell length (mm)

Secondary production

Finally, cohort specific information resulting from the length frequency analysis in combination with the relationships between size and flesh weight can be used to estimate secondary production of the different mussel cohorts. Secondary production can be estimated by adding either the growth increments (increment-summation) or the weight losses (removal-summation) caused by size-dependent mortality (Crisp 1984). Both methods provide identical results (see van der Meer et al. 2013). In Table B3.1, production of the 2009 cohort in the period 2010–2013 is estimated following the removal-summation method. This method considers both the matter that leaves the cohort by mortality, and the difference between the total biomass of the cohort at the end of the observation period and at the start of the period.

Table B3.1:Production calculations for the 2009 cohort of Mytilus edulis on the bivalve bed W001_A1. Samples were taken solely from mussel covered patches. For the estimation of secondary production on the entire bed area (including bare patches) the production on mussel covered patches was multiplied by the fraction of mussel cover of the entire bed area. All mass indications refer to ash free dry mass (AFDM) measurements.

Date Density mussel patch (m-2) Mass (g) Mass gain (g) Mean density (m-2) Production mussel patch (g m-2) Mussel cover (%) Production total bed (kg ha-1) 2010–Apr 5269 0.027 2010–Aug 3745 0.117 0.091 4507 409.3 8.25 337.5 2011–Jan 2867 0.079 -0.038 3306 -126.5 8.49 -107.4 2011–Mar 2272 0.088 0.009 2570 23.5 10.26 24.1 2011–Nov 1602 0.181 0.092 1937 178.7 12.40 221.6 2012–Jan 1208 0.210 0.030 1405 41.9 12.21 51.1 2012–Apr 1237 0.198 -0.012 1222 -14.8 16.78 -24.8 2012–Sep 776 0.323 0.125 1007 125.5 17.76 222.9 2013–May 904 0.295 -0.028 840 -23.5 15.00 -35.3 2013–Oct 830 0.506 0.211 867 182.5 16.01 292.1 Sum 796.6 982.0 Concluding remarks