Impacts on fisheries yield

14.3.2 Impacts on fisheries yield

The evidence for greater yield in fisheries adjacent to MPAs is based on few studies, in contrast to the evidence for greater abundance and body size of target species in MPAs (Table 14.3). The empirical work has been on reef fisheries in Kenya and the Philippines (Alcala and Russ 1990; McClanahan and Kaunda-Arara 1996), and demersal shelf spe- cies such as plaice in the North Sea (Borley et al. 1923; Margetts and Holt 1948), and shrimp in Florida (Klima et al. 1986). McClanahan and Kaunda-Arara (1996) showed that although catch per unit effort increased within 2 km of the bound- ary of the Mombasa Marine Park, total catch declined because the MPA consituted a large part of the ground and fishing effort per unit area in- creased; seven years after closure, the catch was still reduced (T.R. McClanahan and S. Mangi, un- published data). The yield of pink shrimp (Penaeus duorarum ) following establishment of a fishery MPA in the Tortugas of Florida was greatly affected by recruitment, but compliance by fishermen with the closure was also poor (Klima et al. 1986). One of the most striking examples of the effects of closure

301 are those of the ‘great fishing experiments’ based (1999) included both Schaefer and Beverton–Holt

Marine Protected Areas

on fisheries yield before and after both world wars stock–recruitment relationships in a model as well in the North Sea (Smith 1994 and Smith, Chapter as transfer across a hypothetical MPA boundary,

4, this volume). Although there are problems with they showed that at the fishing mortality corre- interpretation of the effort data, the evidence for sponding to maximum sustainable yield (MSY), these demersal fisheries is that with large-scale the yield is likely to be less when part of the ground closure of entire grounds tens to hundreds of kilo- is protected from fishing. However, when fishing metres across, substantial increases in catch rate mortality was double that producing the MSY, and total catch occur over time-spans of much less the yield was maintained in the presence of the than 10 years (Borley et al. 1923; Margetts and Holt MPA and exceeded that without a protected area 1948). Other more recent examples of large-scale (Guénette and Pitcher 1999). The suggestion is closures in the North Atlantic include those of that the recruitment effects of MPAs will be more herring (Clupea harengus) and capelin (Mallotus important for enhancement of yield than the villosus ) (ICES 1990; Bailey 1991), but the North spillover effects, but as yet there are no data about Sea plaice (Pleuronectes platessa) box had failed to recruitment dynamics at the scale typical of MPAs. produce an increase in yield or stock biomass four

Intense fishing should often increase year-class years after permanent closure following earlier variability by making stock biomass more prone to seasonal closure (ICES 1999). Another example of a annual recruitment (but see, e.g. Rijnsdorp et al. large-scale temperate demersal fishery closure is 1992). As a result MPAs may under certain circum- that of the extension of the seasonal ban on trawl- stances reduce this variability by building spawn- ing off the coast of Cyprus, where the MPA sub- ing stock biomass. Only modelling results are stantially increased yield within two years of the currently able to provide specific predictions of the extension because the fishing mortality on fish in extent to which this will happen. Thus in a simu- the first few months post-recruitment had prob- lated cod (Gadus morhua) stock, the number of ably been very high (Garcia and Demetropoulos years with poor recruitment is likely to be reduced 1986).

when some 10% or more of the ground is closed to In the absence of extensive data both on spill- fishing (Fig. 14.2). Using a simple logistic model, over and recruitment through increased egg and Lauck et al. (1998) have shown that the likelihood larval output, modelling studies have provided of maintaining a stock above, say, 60% of carrying predictions of yield under different MPA and fish- capacity for at least 20 years, is likely to fall as the ing conditions, taking account of varying biologi- temporal variability in the catch increases. The cal features of the animals involved (Table 14.3). proportion of a ground that must be protected to With increase in MPA size, the spawning stock bio- maintain the stock in this way is likely to be more mass per recruit can be expected to increase, but than 50%. the yield per recruit outside the MPA will not in- crease, unless there is a decline in fishing effort; in

14.3.3 Under what conditions will

fact, if the fishing effort remains the same, the yield will decline, because the effort per unit of fished

fish and fisheries benefits be detected?

habitat will have increased. MPAs can scarcely The science of MPAs will no doubt be long in the

be expected greatly to increase yield (DeMartini development. Except in broad outline, extrapola- 1993), except where they are large and fishing tion of data from one site to another is unwise, but mortality is very high (e.g. Russ et al. 1992). The it is reasonable to ask what general factors may simplest yield-per-recruit models do not take ac- affect the likelihood of detecting fish and fisheries count of recruitment variations in relation to stock effects. (Shepherd and Pope, Chapter 8, this volume). This

Proper demonstration of fish abundance was true of the models used by Russ et al. (1992) changes or fishery effects of MPAs requires com- and DeMartini (1993). When Guénette and Pitcher parison of the state before and after introduction of

Chapter 14

(a) Beverton–Holt Ricker 60

Years of poor recruitment 10

Spawner biomass 100

Reserve size

No reserve 150

0 0.1 0 0.2 0.3 0.4 0.5 0.6 0.7 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Exploitation rate

Exploitation rate Fig. 14.2 Changes in (a) the number of years with poor recruitment, (b) spawner biomass, and (c) yield, of an Atlantic

Cod (Gadus morhua) fishery as a function of exploitation rate and the proportion of the fishing ground (‘reserve size’) set aside as an MPA. Exploitation rate u is given by the equation u = 1 - e -F , where F = the fishing mortality. (Source: from Guénette and Pitcher 1999, p. 299.)

the management regime and with unmanaged con- taining to yield itself fewer still. Warmer waters trols for comparison. This has seldom been possi- provide the opportunity of direct visual underwa- ble. In both tropical and temperate waters, the data ter surveys (e.g. Polunin and Roberts 1993; Russ on changes in abundance are few, and those per- and Alcala 1996 a point also mode by Jones et al.,

Chapter 16, Volume 1), but, at least in clear waters elsewhere, video is being recognized as a means of discriminating spatial differences in abundance (Willis et al. 2000) and size (Harvey and Shortis 1996) data. In many cases, including many fishery species, abundance and size estimation has to rely on sampling techniques such as fishing gears, including traps and nets (Jones et al., Chapter 16, Volume 1). Catch-per-unit-of effort data offer a relative measure of abundance, but the precision of this will typically be low (Appeldoorn 1996; Schnute and Richards, Chapter 6, this volume; Sparre and Hart, Chapter 13, this volume).

Given spatial and measurement sources of vari- ability, statistical power for detecting most differ- ences when they exist is small, where statistical power is the probability of correctly rejecting a null hypothesis of no difference between MPAs and unprotected areas. Put another way, only when large differences exist will they be detected (Fairweather 1991). It follows that in many cases, lesser but significant differences will not be detect-

ed. High spatial variability, small sample areas and numbers, differences in life history and behaviour, and high measurement error, perhaps caused by differences in procedure or acuity between divers, help to explain apparent differences among species in response to the creation of MPAs (e.g. Polunin and Roberts 1993). Yet, as noted by the meta- analysis of Mosquera et al. (2000), in MPA assess- ments to date, even basic statistics such as vari- ances in means are sometimes not reported. There has also been very little attention to sources of measurement error, and full details of sample de- sign are rarely given, yet the latter is crucial to defining spatial and temporal variability as a basis for assessing management effects. Differences in abundance between areas subject to different lev- els of management have been much more readily detected when species abundance data have been aggregated into trophic or taxonomic groups (e.g. Russ and Alcala 1996; Polunin and Jennings 1998).

Assessments of yield from fished areas are pos- sible through several means. Surplus production models are simple, but preferably require long time series including good contrasts between catches and effort (Hilborn and Walters 1992;

Appeldoorn 1996, Chapter 2.6; Schnute and Richards, Chapter 6, this volume). Yield- per-recruit models rely on catch, mortality and age data (Sparre and Hart, Chapter 13, this volume; Shepherd and Pope, Chapter 8, this volume). However, confident estimation of the underlying growth and mortality data is difficult in many areas (Appeldoorn 1996). Such data tend to be even more highly aggregated biologically and spatially than the best results from underwater visual work, and there is therefore a difference in spatial size of the sampling unit involved between using yields and using visual methods.