An estimate of the sustainable rate of shell extraction from the Dutch Wadden Sea

Authors


J.J. Beukema (fax: 31 222319674; e-mail: jsr@nioz.nl ).

Abstract

1. Shell production by cockles Cerastoderma edule was studied to examine whether or not the present licensed rate of shell extraction in the Dutch Wadden Sea exceeds the current rate of shell addition to the exploitable stocks.

2. Long-term data on numbers of cockles and weights of their shells were used to estimate their annual production on Balgzand, a 50-km2 tidal flat area in the western-most part of the Wadden Sea. During the 1969–97 period, it amounted to an average of 125 g m–2, including 107 g m–2 of shells large enough to be exploitable for shell-lime fishery.

3. The very irregular annual recruitment of cockles was the main cause of the wide 95% confidence limits (74 and 140 g m–2 year–1) of this 28-year estimate. Moreover, high mortality rates in severe winters substantially reduced production per recruit in some year classes.

4. About one-third of the estimated production does not reach exploitable stocks, because it is fragmented by birds (particularly eider ducks), permanently buried in the sediment, or removed by the fishery for live cockles.

5. During the last few decades, the estimated mean amount added annually to the exploitable stocks was 88 million kg or 132 000 m3 of large cockle shells. This amount compares favourably with the current annual level of removal of 134 000 m3 of shells, three-quarters of which are cockles.

6. Even at temporarily lower production rates, the exploitation of shell stocks at its present rate is not expected to lead to a rapid exhaustion of the existing stocks in the tidal inlets of the Dutch Wadden Sea, as these stocks will be in the order of a few million m3.

Introduction

The fishery for shells removes about 150 000 m3 annually (equivalent to about 100 million kg) of coarse shell material from the Dutch Wadden Sea and its tidal inlets. These shells are sold for a variety of uses, such as path paving, drainage, insulation under floors and as a supply of grit for chicken farming. The shells are fished by suction dredges working in deeper parts of the tidal streams where shells tend to aggregate. The areas where shell dredging has been carried out during recent decades cover only about 1·5% of the total area of the Dutch Wadden Sea (Cramer 1998). Nearly all of the fished shells originate from bivalves that once lived in the shallow parts (mostly intertidal flats) of the Wadden Sea. As judged from the species composition of the accumulations (Essink, Eppinga & Tydeman 1996), only about 10% will have originated from offshore North Sea areas, whereas by far the majority (74%) are cockles Cerastoderma edule (L.), which occur in high densities only in the shallow parts of the Wadden Sea. Other species are either less common or produce shells that are either too small (e.g. Macoma balthica (L.)) or too brittle (Mytilus edulis L.). Thus the annual production of cockle shells determines almost completely the rate of renewal of the stocks that are suitable for shell fishery.

The exploitable stock of shells in the Dutch Wadden Sea was estimated around 1950 as 750 000 m3 (Anonymous 1952) but may be larger nowadays. Although the precision of this estimate cannot be assessed, it is clear that a stock of this size could be exhausted rapidly if production failed for prolonged periods. At the present rate of exploitation a non-renewed stock of the estimated size would be sufficient for only 5 years.

The shell fishery in the Netherlands is licensed. The current maximally allowable yield is fixed at 250 000 m3 year–1, 195 000 m3 of which can be fished from the Wadden Sea or its tidal inlets (Cramer 1998). This is close to an earlier (but rather rough) estimate of cockle shell production in the Dutch Wadden Sea (Beukema 1982a). The authorities who issue the fishing permits base their policy on the principle of sustainable yields. Therefore, they are interested in an updated estimate of the long-term mean of the annual production of shells that are exploitable for the shell fishery (such shells should be at least 1·5–2 cm long and transported whole by the tidal currents to the tidal-stream depots). To this end, we made an estimate of the long-term mean annual production of adult-sized cockle shells and the proportion of this production that is not fragmented by birds or otherwise, not removed by live-cockle fishery, and not permanently buried in the sediment. By order of Rijkswaterstaat (Department of Public Works) we prepared a report in Dutch with limited circulation (Beukema & Cadée 1997a). The present paper is based on this report. It poses the question of whether or not the shell fishery at its current level is exploiting a fully renewable resource or is using up the available stocks.

Methods

A long-term monitoring programme of the macrobenthic fauna on Balgzand, a 50-km2 tidal-flat area in the western-most part of the Wadden Sea, provided the basic data. This comprised twice-annual estimates of numbers m–2 of cockles of the successive year classes and the mean weights of their dried shell doublets. Cockles can easily and accurately be allotted to age classes (Beukema 1989). From 1969 onwards, 15 permanent stations (scattered over Balgzand) have been sampled in late winter/early spring and in late summer. Details of sampling method, position and other characteristics of the sampling stations, including their faunal composition, can be found in Beukema (1988). All Balgzand data shown are averages of the 15 sampling places where every year a total of 13·5 (late winter) or 8·5 (late summer) m2 was sampled.

Methods of estimating shell carbonate production from data on numbers and weights of aged shells have been described in Beukema (1980). Production rates can be estimated both from growth increments (‘production’) and from losses through mortality (‘elimination’). The two methods yield nearly identical values (as shown for cockles by Beukema 1982a). In this paper we have calculated half-year rates of elimination (E) of shells of living animals because it is the most direct way to estimate the annual amounts of shells added to the stocks of dead shells. E is defined as the sum of the weights of the shells of all individuals that died during the period between two sampling occasions. In practice, we calculated the E-values using the equation E = z.C, where z is the instantaneous rate of mortality, and calcimass C is the mean of the total weight of shells of living cockles during the period of observation (in practice the average of the two weight estimates at the two successive sampling occasions). Because z differed between spat and adults, E was estimated separately for these two age groups. As only shells of adult size (> about 15 mm, individuals of at least approximately 1 year old) are fished, all subsequent data refer to such shells if not stated otherwise. Cockle shells are characterized by a low content of organic material. They are almost completely made up of carbonates, weight loss on ignition amounting to ≈ 2% or less (Evans 1977; Beukema 1982a; Glover & Kidwell 1993). Therefore, we did not distinguish between shell production and shell lime or carbonate production.

To judge whether the Balgzand stock of cockles was representative for the entire Dutch Wadden Sea (or the 1500 km2 of shallow areas where cockles can occur in high densities), our estimates of stock size obtained at Balgzand were compared with similar estimates from larger parts of the Wadden Sea, using data from surveys by Beukema (1976), De Vlas (1982) and the shellfish department of the Netherlands Institute for Fishery Research, RIVO-DLO (Van Stralen 1990; Kesteloo-Hendrikse & Van Stralen 1992, 1995, 1996; Van Stralen & Kesteloo-Hendrikse 1992, 1997; Kesteloo-Hendrikse 1994a,b; J.J. Kesteloo-Hendrikse, personal communication). Since 1990, RIVO has monitored the bivalve populations annually in May over about 1500 km2 of shallow (mostly intertidal) parts of the Dutch Wadden Sea by taking one sample per km2.

Results

Annual production in the balgzand area

Cockle densities found on Balgzand during the 28-year observation period were extremely variable. The mean numbers of adults ranged from 0 to 81 m–2. Values of calcimass C ranged from 0 to 254 g shells m–2 in late winter and from 2 to 339 g m–2 in late summer (Fig. 1a). Winters with an average air temperature of > 2 °C below the long-term average occurred in early 1979, 1985, 1986, 1987 and 1996. During the full year after these winters (including three sampling occasions: at the end of the specified winter, the subsequent late summer and the following late winter), observed C-values for adult cockles were extremely low (Fig. 1a). On the other hand, high C-values were observed whenever a strong year class reached adult size, e.g. the one of 1979 in the summer of 1980 (and a few subsequent years) and the one of 1987 in the summer of 1988 (and the subsequent year).

Figure 1.

Year-to-year variability in some characteristics of the stocks of adult (> 15 mm) cockles on Balgzand during the 1969–97 period of observation. (a) Calcimass C (amount of shell material present in live cockles) in g m–2 in late winter (○) and late summer (•); (b) annual mortality z (per year, instantaneous rate of mortality) calculated from late-winter to late-winter sampling, natural logarithm of the quotient n m–2 at the start/n m–2 at the end of the year; (c) annual shell production E in g m–2 for each year (late-winter to late-winter periods). The added open circles for the 1991–96 period are estimates for Balgzand with a cockle stock similar to the one of the entire Dutch Wadden Sea in this period.

Mortality rates varied strongly too, with high values occurring particularly in years including a severe winter (Fig. 1b). The values of z ranged from 0·2 to 4·9 year–1, with a weighted average of 1·1 year–1. There was no clear relationship between the annual values of C and z.

Annual values of E (= z.C) for cockle shells of > 15 mm ranged from 7 to 355 g m–2 (Fig. 1c), with an average of 107 g m–2. The standard error of this 28-year average amounted to 16, giving 95% confidence limits of the estimate for mean annual shell production by adult cockles of about 74 and 140 g m–2. In addition, there was a mean annual production of smaller cockle shells of 18 g m–2. On Balgzand, cockle production has not shown a clear trend during the last few decades (Fig. 1c).

The wide confidence limits of E were a consequence of the high variability of both annual recruitment (Fig. 2) and annual mortality (Fig. 1b). Figure 2 shows the dependence of life-time production of shell carbonate of each of the 27 studied cockle year classes on their initial strength (mean number of individuals m–2 in their first summer). As one would expect, strong year classes were more productive than weak ones. The variability in productivity per year class (ranging from 4 to 628 g m–2) was governed largely by the variability in recruitment (ranging from 7 to 507 young-of-the-year m–2). The positive relationship shown in Fig. 2 was statistically highly significant (r = +0·79, n = 27, P < 0·001, Spearman rank test).

Figure 2.

Relationship between year class strength (n m–2 of young-of-the-year in late summer) and subsequent life-time production of shell material in g m–2. One point for each of the 27 years classes studied, with open circles referring to cohorts struck by a severe winter (causing high mortality) early in their life.

Some year classes produced small amounts of shells relative to their initial strength as a result of suffering a severe winter early in their life (the open circles in Fig. 2). If no cold winters occurred early in life, the production per recruit amounted to about 1 g of shell material (Fig. 2).

The strength of a year class, estimated as the mean number m–2 of recruits in August, varied strongly from year to year (Fig. 2). On Balgzand, a negative relationship was observed between the size of the adult stock (expressed either in biomass or calcimass units) and recruit density in the same summer (Fig. 3: r = –0·47, n = 28, P < 0·02, Spearman rank test).

Figure 3.

The relationship between the size of the adult stocks (expressed in g m–2 of shells of living cockles of > 15 mm length) and the number of recruits m–2 in August of 28 years (1969–96, incl.) on Balgzand.

Are the balgzand data representative for the entire dutch wadden sea?

Several data sets are available for comparison with the Balgzand data on cockle numbers and mass of cockle shells. In particular, the annual large-scale samplings of the Wadden Sea cockle stocks by RIVO-DLO that started in 1990 provide a wealth of data on cockle abundance at no less than about 1500 stations, 70 of which are located at Balgzand. A comparison of the latter with our 15-station data shows a close similarity of numerical densities (Fig. 4a), denoting that the two data sets will both give a fair representation of the actual sizes of the cockle stocks on Balgzand. Amounts of shells were slightly higher in the RIVO samples (Fig. 4b), but this difference can easily be explained by the 2–3-month difference in sampling time. The cockles had already started their seasonal growth in May (RIVO), but not yet in February/March (Netherlands Institute for Sea Research; NIOZ).

Figure 4.

A comparison between results of two independent sets of stock assessment of live cockles on Balgzand during the years 1990–96, inclusive. Horizontal axis: 15-station means by the first author (NIOZ); vertical axis: 70-station means by RIVO (J.J. Kesteloo-Hendrikse, personal communication). (a) in n m–2; (b) in g of shells m–2. The sampling occasion is indicated by the year. Sampling by NIOZ in February/March; by RIVO in May. The lines indicate exactly equal estimates.

The comparison between the Balgzand data and those for the entire Dutch Wadden Sea (Fig. 5) shows a fair correspondence for eight out of the 11 years that could be compared (1971, 1972, 1980, 1990, 1991, 1992, 1996 and 1997). However, in the 3 successive years 1993, 1994 and 1995, all Balgzand values (both from NIOZ and RIVO samplings) were much (up to 10 times) higher than those for the entire Dutch Wadden Sea. For these 3 years, Balgzand data of the cockle stock were clearly not representative for the Dutch Wadden Sea as a whole.

Figure 5.

A comparison between several independent assessments of cockle stocks (expressed in g m–2 of shells of live cockles) present on Balgzand (about 50 km2) and in the shallow parts of the entire Dutch Wadden Sea (about 1500 km2). Sampling occasion indicated by year. Balgzand data from Fig. 1(a) (samplings by NIOZ; •) or Fig. 4(b) (samplings by RIVO; ○). Wadden Sea data recalculated from various authors: 71 and 72 from Beukema (1976), 80 from De Vlas (1982), 90 from Van Stralen (1990), 91 from Van Stralen & Kesteloo-Hendrikse (1992), 92 from Kesteloo-Hendrikse & Van Stralen (1992), 93 from Kesteloo-Hendrikse (1994a), 94 from Kesteloo-Hendrikse (1994b), 95 from Kesteloo-Hendrikse & Van Stralen (1995), and 96 from Kesteloo-Hendrikse & Van Stralen (1996). Various conversion factors have been used to convert data on biomass (ash-free dry weight of soft parts) or total weight (shells plus soft parts and water) to shell weight (Beukema & Cadée 1997a).

The temporarily higher cockle stock on Balgzand was caused by the unusual distribution of cockle recruitment in the entire Dutch Wadden Sea in 1991 and 1992, as can be deduced from the maps resulting from the RIVO surveys (Fig. 10 in Kesteloo-Hendrikse 1994a; Fig. 11 in Kesteloo-Hendrikse 1994b). In these 2 years, high recruit densities were restricted everywhere to relatively sheltered coastal tidal flats with a rather silty sediment and an intertidal height close to mid-tide level (J.J. Beukema, personal observations on Balgzand, deductions from maps of the RIVO surveys). In all other years, cockle recruitment was maximal (both on Balgzand and in the remainder of the Wadden Sea) on tidal flats that were more sandy, lower in the intertidal, and situated in a more offshore and more exposed zone (compare, for example, Fig. 10 in Van Stralen 1990 and Figs 12 and 13 in Van Stralen & Kesteloo-Hendrikse 1997). As the proportion of sheltered coastal flats is much higher on Balgzand than in the remainder of the Wadden Sea (as evidenced by the distribution of sediment types – compare with the map in De Glopper 1967 – showing an over-representation of sediment with a silt content of > 2% on Balgzand), the cockle cohorts of 1991 and 1992 were strongly overrepresented on Balgzand, causing relatively high amounts of large cockles in the 1993–95 period. In these years, the presence of such dense populations of adult cockles was very restricted in the remainder of the Wadden Sea (Kesteloo-Hendrikse 1994a,b; Kesteloo-Hendrikse & Van Stralen 1995). Therefore, we present in Fig. 1(c) not only the actually observed values for shell production on Balgzand (closed circles), but also reduced figures (open circles) that would more realistically reflect production in the entire Dutch Wadden Sea. These reduced figures were obtained by fitting the 1992–96 points for NIOZ estimates of Balgzand cockle stocks to the line of equality with Wadden Sea stocks in Fig. 5.

Annual production in the dutch wadden sea

If the Balgzand estimate of observed shell production in the form of adult-sized cockle shells of 107 g m–2 year−1 is simply extrapolated to all shallow parts of the Dutch Wadden Sea (1500 km2), the 28-year mean annual production of large cockle shells would have amounted to 160 million kg. If the ‘abnormal’ years of the 1992–96 period are excluded, the 23-year mean annual amount was 95 instead of 107 g m–2. If the reduced values are used for the 1992–96 period, it becomes 84 g m–2 (as an average of 28 years). We consider the estimate of 84 g m–2 to be the most representative for the 1500 km2 of shallow parts of the entire Dutch Wadden Sea. With this estimate for annual production per m2, the total annual production in the Dutch Wadden Sea would have amounted to an average of 126 million kg during the 1969–96 period.

Losses between production and fishery of shells

Not all of the shells produced end up in the fished accumulations. Various processes intervene, such as (i) fishery for live cockles; (ii) fragmentation of shells by birds; (iii) fragmentation or dissolution by physical or chemical processes; and (iv) more or less permanent burial of shells in stable sediments. Below, we try to quantify these losses mostly on the basis of published information.

In the 1984–95 period, the cockles fished for human consumption in the Dutch Wadden Sea yielded a mean annual amount of 4 million kg of soft parts (Anonymous 1996). The weight of their shells will have amounted to 10 million kg. Slightly more than half of these cockles (J.D. Holstein, personal communication) were transported to locations outside the Wadden Sea before the soft parts were separated from the shells. In this way, an annual average of 6 million kg of large cockle shells was removed from the Dutch Wadden Sea. This amount is equivalent to 5% of the mean annual production of such shells.

Small amounts of live cockles are transported from the Wadden Sea to the land by birds such as oystercatchers and gulls (Cadée 1989). The role of birds is much more important in fragmentation of shells. Eider ducks in particular crack huge amounts of large cockles in their well-muscled stomachs (Swennen 1976), producing shell fragments of mostly 2–8 mm (Cadée 1994). The estimate by Cadée (1994) of this loss amounts to 25 million kg year–1WP extended char 4,78. A recalculation with more realistic conversion factors (Beukema & Cadée 1997a) reduces this amount to 20 million kg. Another estimated 2 million kg of large cockle shells is fragmented annually by gulls (Cadée 1995). Thus the proportion of the cockle shell production that is fragmented by birds amounts to an estimated 22 × 100/126 = 17%.

Bivalve shells frequently show signs of partial dissolution by the action of algae, fungi or bacteria (Cutler & Flessa 1995; Knauth-Köhler, Albers & Krumbein 1996). However, due to their low contents of organic material, cockle shells are relatively resistant to bio-erosion (Glover & Kidwell 1993). Indeed, Van Straaten (1954) did not find indications of decalcification of sediments on tidal flats of the Dutch Wadden Sea. Therefore, we assume that weight loss by dissolution is negligible in Wadden Sea cockle shells. Eroded shells might become more liable to destruction by physical forces, such as strong currents and waves. In the relatively sheltered Wadden Sea, the proportion of large cockle shells seriously damaged in this way will be low. Exact proportions are not known. We assume that a small percentage of the amounts of large cockle shells produced will get lost by chemical or physical destruction. Thus the total proportion of large cockle shells that is fragmented to parts that are too small to be useful for shell fishery will amount to nearly 20% of the production, mostly as a consequence of fragmentation by birds.

Most of the cockle shells produced in shallow areas will be transported sooner or later by tidal currents and wind to tidal streams, where they aggregate to dense accumulations (depots) in the deeper parts, located in the tidal inlets between the Frisian Islands. Transport routes as depicted by Krause (1950) for a part of the German Wadden Sea will in principle apply also to the Dutch Wadden Sea. A minor fraction will get buried for long periods at depths of more than 10–20 cm below the sediment surface, among others by the activities of lugworms (Van Straaten 1952). In the Dutch Wadden Sea lugworms Arenicola marina (L.) bioturbate a sediment layer of on average 6–7 cm annually (Cadée 1976), but this may be an underestimate for more recent years because lugworms became more numerous during the 1980s (Beukema 1991). During periods of net sedimentation (as a consequence of sea level rise or bottom subsidence) substantial amounts of shells may get buried to depths beyond the reach of waves and currents. Only changes in the course of major tidal streams would mobilize such deeply buried shells.

It is difficult to estimate the proportion of shells produced that gets more or less permanently buried. If all shells produced had been buried locally during the time of existence of the Wadden Sea tidal flats (about 1000 years), the average stock within the tidal-flat sediments would amount to some 100 000 g m–2. In fact, the amounts of dead cockle shells actually observed in tidal-flat sediments were generally smaller by one or two orders of magnitude. In productive areas, around 10 or a few 10s of kg m–2 were frequently present, whereas far less than 1 kg m–2 was observed in areas where living cockles were always rare or absent. Therefore, we guess that the fraction of large cockle shells that gets more or less permanently buried amounts to only about 5% of the production. Essink (1996) shares this opinion.

Once whole shells are deposited in the depots in the deeper parts of the tidal streams, cockle shells are unlikely to become lost by processes other than shell fishery. Shell half-lives of several hundreds of years (Meldahl, Flessa & Cultler 1997) appear not to be uncommon in marine assemblages and may be expected in the environments of the Wadden Sea depots, where most shells remain buried below layers of other shells.

Annual production of exploitable shells

The added estimates of losses amount to 5 + 20 + 5 = 30% of the production of adult-sized cockle shells, leaving 70% to be exploited by the fishery for shells. An annual amount of 0·70 × 84 = 59 g m–2 would then become available for this fishery. For the 1500 km2 of shallow area of the Dutch Wadden Sea this is equivalent to 88 million kg or 132 000 m3.

A balance of production and extraction

For the last few decades, data are available on both production (see above) and extraction of large cockle shells. During the 26-year period 1970–95 inclusive, a mean of 134 000 m3 of shells was extracted annually from the Dutch Wadden Sea and its tidal inlets (Rijkswaterstaat (Department of Public Works), personal communication). These amounts will have included at least 74% of cockle shells (Essink, Eppinga & Tydeman 1996). The cumulative amount of cockle shells fished from the Dutch Wadden Sea during the last 28 years can be estimated at 0·74 × 28 × 134 000 = 2·8 million m3. During this period, the production of exploitable cockle shells can be estimated at 28 × 132 000 = 3·7 million m3.

According to these estimations, the reserve of exploitable cockle shells may have increased by about 0·9 million m3 since 1970. In the 1950s the magnitude of the stocks of such cockles was estimated at minimally 0·75 million m3 (Anonymous 1952). The exploitable reserve may thus have doubled or even tripled (if production exceeded extraction to the same extent in the 1950s and 1960s). At the present level of extraction, a stock of about 2 million m3 of exploitable shells could serve as a reserve of about 15 years for a failing production. In addition, there is an inaccessible stock of similar magnitude of shells buried in the sediments. This stock becomes exploitable at a very low rate, usually by deep erosion as a consequence of channel dredging or natural changes in the courses of tidal streams.

Discussion

The cockle productivity on balgzand

Cockle abundance (Fig. 1a) and shell production (Fig. 1c) fluctuated strongly from year to year. Such high variability appears to be characteristic for cockles, as it has been observed in several other European populations (Hancock 1973; Dörjes, Michaelis & Rhode 1986; Ducrotoy et al. 1991; Coosen et al. 1994). The high year-to-year variability in cockle abundance originated in particular from strong fluctuations in annual mortality (Fig. 1b) and annual recruitment (Fig. 2). Year-to-year differences in growth rate were minor and appear to be less important in explaining the high interannual variability in productivity.

Cockles are sensitive to low winter temperatures in the Wadden Sea (Beukema 1979, 1985, 1992; Dörjes 1992). This explains the high mortality values observed in 1979, 1985, 1986, 1987 and 1996 (Fig. 1b). The high value of z in 1990–91 (with a close-to-average winter) was caused by intensive fishing for live cockles in mid-1990 followed by high predation pressure by birds on the few remaining cockles (Beukema 1993; Beukema & Cadée 1996). Annual mortality appears to be almost unrelated to cockle density, as also observed by Hancock (1973).

If a cockle cohort was struck by a severe winter early in its life, its production was dramatically reduced (Fig. 2, open circles). If not, life-time production per recruit amounted to about 1 g of shell material. This amount did not seriously decline at high initial cohort densities, as would have been the case if cockles competed intensively for food at the higher densities. On Balgzand, densities of adult cockles rarely exceeded a few hundred m–2 and at such densities negative effects on growth rate are slight or absent (Kamermans 1993). Cockle growth is significantly reduced only at very high densities (thousands per m2) (Hancock 1973; Jensen 1992, 1993; De Montaudouin 1996). Thinning or reseeding results in significantly higher growth rates only at such exceptionally high densities (Dijkema, Bol & Vroonland 1987).

Thus cohort density appears to be the key factor governing cockle production. Magnitude and variability in cockle production were largely governed by year-to-year variability in year class strength (Fig. 2). In cockles, recruit densities tend to fluctuate heavily from year to year (Hancock 1973; Beukema 1982b; Ducrotoy et al. 1991). Recruitment success is related to unpredictably changing weather conditions such as temperature in winter (Beukema 1992) and wind in spring (Young, Bigg & Grant 1996). There is, however, also a relationship with adult densities, with high recruitment occurring at low adult stocks and failing recruitment at high adult abundance (Fig. 3). Similar negative relationships between stock and recruitment have been observed in other cockle populations (Hancock 1973; André & Rosenberg 1991; Bachelet, Guillou & Labourg 1992). Such relationships contribute to a regulation of the cockle stocks, preventing extremely high densities (that might otherwise arise from a close succession of strong cohorts) or long periods with very low densities (that might occur if successful recruitment was rare at low adult stocks). Nevertheless, the abundance of cockles was highly variable from year to year (Fig. 1a). In addition to the strong fluctuations in recruitment success, the short life span of cockles (Beukema 1989; Dörjes 1992) and the high year-to-year variability in mortality (Fig. 1b) were major causes of the strong and unpredictable changes in the abundance and production of cockles.

The updated estimate of a mean annual production of 107 (adults) + 18 (first year) = 125 g m–2 is close to the earlier estimate of 114 g m–2 for the same area (Beukema 1982a) and partly the same (14 out of 28) years. It is, however, much higher than an early and rough estimate for the western part of the Dutch Wadden Sea by Verwey (1952).

An extrapolation to the entire dutch wadden sea

Although the data set collected on Balgzand during nearly three decades represented the size of the local cockle population fairly (Fig. 4), it is questionable whether it was also sufficiently representative of the entire Dutch Wadden Sea. The correspondence between the 50- and 1500-km2 areas was fair in 8 out of the 11 years with data available for both areas (Fig. 5). Moreover, the deviations observed in the remaining 3 years could be explained: they were due to an unusual geographical pattern of recruitment that was limited to the cohorts of 1991 and 1992 and that occurred all over the Dutch Wadden Sea. Therefore, we think that it is warranted to reduce the Balgzand values of the aberrant years to obtain figures that can be used as representative for the entire Dutch Wadden Sea (the open circles in Fig. 1c).

If any trend in annual cockle shell production can be deduced from Fig. 1(c), it would be a downward trend for the 1974–97 period (taking into account the reduced values for the 1990s). However, the existence of strong year-to-year fluctuations precludes any firm statement on long-term trends in cockle productivity. During the last few decades, several other zoobenthic species have shown upward long-term trends as a consequence of eutrophication in the western part of the Dutch Wadden Sea (Beukema 1991).

The production of exploitable shells

Accepting the reduced Balgzand production estimate of 84 g m–2 year−1 of adult-sized cockle shells as representative for all shallow parts of the Dutch Wadden Sea, the total annual production would amount to an average of 126 million kg. This amount, however, will not turn up in the depots that are accessible to the shell fishery. Substantial losses take place between the production and the fishery. Some of these losses could be estimated quite precisely, such as the 5% loss to the fishery for live cockles and the 17% fragmented by birds. Other losses could not be quantified, such as the (undoubtedly low) proportion fragmented by bio-erosion or physical forces and the proportion permanently buried (probably also low, as otherwise the tidal-flat sediments would be more shelly). The total of the losses will not be lower than 20% of the production, but the stated total of 30% is not more than an informed guess.

Another uncertainty is the rate of natural disappearance from the accumulations (e.g. by dissolution). We recommend a study not only of the sizes of the present stocks of exploitable shells but also of the age distribution of these shells and thus of the half-lives of these shell assemblages, along the lines depicted by Meldahl, Flessa & Cutler (1997). For the time being, we assume that large cockle shells stay sufficiently long within the accumulations in an unaltered condition to be fully exploitable for the shell fishery. For periods of years or decades this appears to be true, but not for centuries. Simple calculations show that the amounts accumulated over the last 1000 years must have been in the order of 100 million m3 if the present rate of production had prevailed during that period and extraction had been negligible (the present intensive fishery started less than a century ago). Although the only available estimate of 0·75 million m3 was almost certainly too low, a hundred-fold underestimate appears improbable. Thus either the half-lives in the depots were relatively short (in the order of decades rather than centuries) or production rates in former centuries were significantly lower than at present. Half-lives may have been affected by slow chemical dissolution (not measurable at short time scales) or by abrasion during transport when the position of the tidal streams shifted. Present production rates probably overestimate former ones as a consequence of the present organic enrichment of the Wadden Sea, causing elevated stocks and productivity in macrobenthic animals (Beukema 1991; Beukema & Cadée 1997b).

The balance of production and exploitation

For an evaluation of the present balance, we accept a value of 70% as the exploitable proportion of the adult-sized cockle shells produced and an annual production level of 126 million kg. The resulting annual amount added to the exploitable stock will then be 88 million kg, equivalent to 132 000 m3. The amounts fished annually during the last few decades amounted to 0·74 × 134 000 = 99 000 m3 cockle shells (the other 26% belonging to other species such as Spisula subtruncata). Thus present rates of additions to the reserve will have surpassed extraction rates by more than 30% (Fig. 6). Even if only slightly more than half (53% instead of 70%) of the produced cockle shells actually reached the exploitable depots (66 million kg or 99 million m3), the size of the reserves would have remained at the usual level. Any higher (> 53%) proportion of shells produced reaching the depots would result in an extension of the exploitable stocks.

Figure 6.

A comparison of the cumulative amounts of cockle shells (•) added to (by production) and (○) extracted from (by shell fishery) the exploitable stocks of large (> 15 mm) cockle shells. Added = 70% of the total production of large cockle shells in the Dutch Wadden Sea; extracted = 74% of the fished shells (see text for explanation).

There is indeed some evidence that the total reserves in the Dutch Wadden Sea have increased during the last decades. Shell fishery was mostly restricted to depots located in the major tidal streams of only one tidal basin (the Vlie), yielding more than 60% of the total (Cramer 1998). Although these depots will have received less than half of the cockle shells produced in the entire Dutch Wadden Sea, the fishery could continue for decades at an almost constant yield. Therefore, we think that the present fishery intensity is lower rather than higher than the actual rate of addition to the total of Dutch Wadden Sea depots. Because the shell fishers do not complain of exhausted depots in the intensively fished Vlie area, we cannot believe that the above estimates of addition rates for the total Dutch Wadden Sea are too high.

Conclusions

The licensed fishery for shells in the Dutch Wadden Sea provides a good example of a sustainable fishery on a renewable resource. The rates of renewal and extraction have been in balance during at least the last three decades. Rijkswaterstaat (Department of Public Works) annually grants a small number of shell fishery companies with a licence for a limited amount of shells. The former licences were also based on data published in Beukema (1982a). Fortunately, the production rates estimated at that time proved to be realistic for the next 15 years.

The maximal amounts to be fished have been (and will be) based on advice from independent experts who regularly estimate the annual rates of shell production. This is a flexible system, allowing rapid restrictions of the fishery intensity whenever cockle production might fail for prolonged periods. Considering that (even at zero production) the present size of the depots is sufficient for a continuation of extraction at the present rate for probably some 10 or 20 years, there will be ample time to adapt the fishing intensity to reduced production rates, should they occur.

Whether or not the fishery will maintain the present balance and the ample reserves will depend primarily on the future rates of production of cockle shells. The main single factor securing a high level of cockle production is the regular appearance of a new strong cohort of cockles. After the summer of 1987, however, there has been no cockle recruitment that was really successful all over the Dutch Wadden Sea. As a consequence, shell production has shown a declining trend since 1988 (Fig. 1c, open points for the 1991–96 period). At present, the unusually long period of failing cockle recruitment is a matter of concern both to fishermen and nature conservationists. Any expectation of future production rates should include a probable reduction of the present organic enrichment of the Wadden Sea (nutrient levels have declined for the past decade) and a possible consistent warming of the local climate (the mildest winters of the century occurred as recently as around 1990, resulting in failing recruitment in several bivalve species, including the cockle; Beukema 1992). Therefore, it is questionable whether the shell fishery is sustainable at its present level on time scales longer than one or two decades.

Acknowledgements

We are grateful to J.J. Kesteloo-Hendrikse for making available the detailed RIVO data on Balgzand cockle densities. We thank K. Essink for several useful suggestions. Y.J. Zijlstra encouraged the preparation of the original report in Dutch and put several Rijkswaterstaat (Department of Public Works) reports at our disposal. R. Dekker kindly supplied some recent data of the NIOZ sampling on Balgzand. W. de Bruin skilfully prepared the figures. Part of the data used were collected within a monitoring programme executed by NIOZ by order of Rijkswaterstaat-RIKZ.

Received 8 November 1997; revision received 26 November 1998

Ancillary