A Fur Seal Simulation Model to Explore Alternative ...

A Fur Sea

ation Mode Management Strategies

Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by UNIV MANITOBA on 09/24/14 For personal use only.

Mdrine Biology Research Institute, University of Cape Town, Rondebosch, 7700, South Afrisa

Department of Fisheries and Oceans, Science Branch, P.O. Box 5667, St. john's, Newfoundland A 1 C 5x1, Canada

J. H. MeDavid Sea Fisheries Research Institute, Private Bag X2, Rogebaai, 80 12, South Africa

J. G. Field Marine BisEogy Research Institute, University of Cape Town, Rondebosch, 7700, South Africa

WeHeOosthuizen Sea Fisheries Research Institute, Private Bag X2, Roggebaai, 8012, South Afrisa

J-P. ROUX Sea Fisheries, P - 0 . Box 394, huderib, 9000, Namibia

and A. M. Starfield Department of Ecology and Behavioral Biology, 109 Zoology, 3 18 Church Street SE, Minneapolis, hfN 55455, USA

Wickens, P. A., P. A, Skelton, ). H. M. David, I. G. Field, W. H. Oosthuizen, I-P. Roux, and A. M. Starfield. 1992. A fur seal simulation model to explore alternative management strategies. Can. J. Fish. Aquat. Sci. 49: 1396-1 485. A simulation model is formulated for the South African fur seal (Arctocephalus pusil%uspusiIIus) to evaluate the appropriate management action when culling ts reduce population growth rate, culling to decrease fish consumption by seals, or harvesting to maximise numbers of seals removed. There is disturbance associated with bull sealing which increases pup mortality and reduces pregnancy rates, but this is not well quantified. Disturbance can be included or excluded from model runs. To reduce population growth, cow removal is most effective, but the population sex ratio becomes severely altered and this may be undesirable ecslogically. Rede~ctionof fish consumption is best achieved either by removing cows, with the same caveat regarding sex ratio, or by removing bulls and including disturbance effects. However, the acceptability of a reduction achieved by humans disrupting seals is questionable, and the continued removal of bulls may eventually lead to further decreases in pregnancy rate. To maximise a harvest, the relative commercial value of different seal products is csnsidered, and bull removal, excluding disturbance effects, followed by removal of pups achieves this aim most effectively. pus;I%us)afin Les auteurs ont elabort5 un modele de simulation concernant I'otarie geante (Asctocephaluspusil%us d'evaluer la mesure de gestion appropriee lors de I'abattage selectif en vue de reduire le taux de croissance dbmographique, I'abattage selectif en vue de reduire la conssmmation de poisson par les staries et la chasse en vue de maximiser le nombre d'otaries tuees. La perturbation qu'entraine la chasse des mzles mPne 21 une augmentation du taux de mortalit6 des jeunes otaries et 2 une baisse du taux de conception, mais cette perturbation n'a pas kt6 adequatement quantifiee. Elle peut etre incluse dans les passages de modPle, su en etre excluse. Afin de reduire le taux de croissance d6mographique, il est plus efficace d'abattre des femelles bien que la proportion relative des sexes en est fortement rmodifiee; ceci peut ne pas &re desirable au niveau kcologique. Afin de reduire la consommation de poissons par les otaries, i l est plus efficace d'abattre des femelles (et donc de modifier la proportion relative des sexes) ou des msles (y compris la perturbation correspondante). O n peut toutefois mettre en doute l'aspect acceptable d'une reduction des effectifs par I'entrernise d'une chasse, source de perturbations; l'klimination continue de m%lespeut t5ventuellement mener 3 de plus importantes baisses du taux de conception. Afin de maximiser le nornbre d'otaries tuees, on tient compte de la valeur commerciale relative de different5 produits de I'otarie; i% est donc plus efficace d'abattre en premier lieu des males (A l'exclusion de la perturbation correspondante), puss des jeunes otaries. Received February 7 5, 199 % Accepted lanuary 13, 1992

(JAS99)

Can. /. Fish. Aquat. Sci., Vol. 49, 1992


T

he recovery of seal populations in several marine systems following overexploitation and population declines in the eighteenth to mid-twentieth century has focussed attention on appropriate management measures, especially where seals are considered to affect commercial fisheries (e. g . Beverton 1985). Conservationists, responsible for acceleration of the recovery in recent years through lobbying to reduce commercial exploitation of seals, would like the population to remain unexploited whereas commercial sealers desire a maximum sustainable yield of seals. Some commercial fishermen, on the other hand, believe that seals eat too much fish and that these fish could otherwise be caught by the fishing industry. They would therefore like the seal population to be reduced to decrease the possible impact of seals on commercial fish yield, although to date, there is no evidence to suggest that reducing consumption of fish by seals will increase fishery yields (e.g. Butterworth et al. 1988; Anonymous 1991). Resolving conflicting objectives is a complex management issue. Three important questions arise: what management action should we adopt in order to achieve efficiently (1) culling to reduce seal population growth rate, (2) culling to reduce fish consumption, and (3) harvesting to maxirnise number of seals removed over a period of time. In this paper, we develop a simulation model to provide insight into the case of the South African (Cape) fur seal (Arctoceph&uspusillus pusillus) population. Because each of these three management objectives can be challenged by both scientific and ethical arguments, the aim here is not to justify the rationale behind each of them, but to provide an objective, even if theoretical, analysis of possible management action required to achieve differing future management scenarios with the use of simulation modelling. The benefit of using a modelling approach as opposed to empirical methods is that models create an ersatz world in which many trials, testing different management hypotheses, can be conducted in a short time without perturbing the system. We draw the distinction between "culling," when the objective is to reduce the seal population efficiently, and ""hesting"' in which seals are removed for the value of their products. We emphasise that action in which animals in one population are removed in order to provide benefit to another population is much less common in fisheries management than it is in the management of terrestrial systems.

Population Geography Only one seal species, the South African fur seal, breeds off the west and south coasts of southern Africa, and the range of this population (Fig. 1) extends from northern Namibia (B8"S) southwards as far as 34% and along the south coast of South Africa as far as 26%. There are 25 breeding colonies, 19 of which are on islands and the remaining six are mainland colonies. Ten known nonbreeding colonies were described by Oosthuizen and David (1988) and these are distinguished from breeding colonies by the absence of pups or minimal md erratic pup production. One of these colonies, Paternoster Rocks, has subsequently been classed as a breeding colony, bringing the total number of breeding colonies from 24 (David 1987) to 25. None of the colonies is located where human disturbance occurs. The island colonies are protected by their inaccessibility. Of the six mainland areas, five are in diamond-mining areas to which access is restricted and the remaining one, Cape Cross, is in a nature reserve. David (1 989) suggested that Kleinsee and Atlas Bay are likely to be the largest mainland seal colonies in the world.

Following the near extirpation of the seals prior to 1900, numbers of pups have grown exponentially since 1972 at an average rate of 2.7% per annurn (Butterworth and Wickens 1998). Breeding space does not appear to be a limiting factor at present, mainly because of access to extensive reasonably protected mainland breeding sites, particularly in Namibia. Food is more likely to be a factor that causes limitation to the population, but to date, this does not appear to have been a significant constraint to overall population growth, although pup numbers at some colonies have shown slowed growth in specific yeas (Wickens et al. 1991) and this can be attributed to food shortages. Bearing in mind that fluctuations in pup numbers do not necessarily mirror those of the population as a whole, population growth therefore appears to be density independent at present.

Life History Bulls are polygynous and arrive at the colonies to set up territories from mid-October (Shaughnessy 1985), with the females arriving slightly later. The bulls are only present for the duration of the breeding season. Each territory consists of a small fixed area defended by a single bull, which may be visited by varying numbers of females. The females come and go at any time and are not tied to any particular territory. The pupping season lasts approximately 1 mo (Shaughnessy 1979; Bavid 1987), with the median pupping date in early December (Bavid 1987). Mating takes place, on average, 6 d after pupping (Bavid and Rand 1984) a d the gestation period is almost 1 yr, with delayed implantation of about 4 mo (Shaughnessy 1979). Each cow produces one pup which is fully weaned between the ages of 8 and 11 mo (David and Rand 1986; David 1987). The cows make regular feeding trips to sea alternating with visits to the colony to feed their pups, until the pups are weaned. Pups usually start foraging in the waters close to the colony at an age of 5-4 mo (David and Rand 1986). Outside the breeding season the animals that can be seen at a colony comprise the unweaned pups and their mothers, some juveniles of both sexes, and some other adults.

During the present century, up to and including 1990, a total of 2 748 574 seals comprising 2 534 009 pups (92.2%), 205 009 bulls (7.5%), and 9556 cows (0.3%) have been killed (Wickens et al. 1991). Figure 2 shows the number of pups and bulls harvested since 1980. The number of pups killed increased dramatically from the 1938's and levelled off in the early 1970's. Since 11 983, sealing has taken place mainly at the major mainland colonies, Cape Cross, Kleinsee, Wolf Bay, and Atlas Bay, and changes in the numbers of both pups and bulls h a vested from these colonies are largely a consequence of changes in the market place as well as in management policy (Wickens et al. 1991). Prior to 1983, the management policy for the South African fur seal was one of achieving a maximum sustainable yield of pups, the pelts of which were used in the fur trade. In 1983, a worldwide collapse in the market for furs caused a large drop in the trade of South African seal pup pelts, so sealing of pups was no longer as financially viable. Between 1984 and 1987, various actions were taken without the support of scientific advice to reduce the seal population with the objective of increasing the commercial fish yield, e.g. harvesting cows and increasing the h m e s t of bulls. Since 1988, quotas have been


Albatross Rock

Hollarns Bird Island

SOUTH AFRICA

Paternoster Rock

Rock

'

EAST

FIG. 1 . Breeding colonies of the South African (Cape) fur seal (circles) tlwoughout the extent s f their range. The inset shows the colonies in the Luderitz area.

awarded according to scientific advice. The harvesting of bulls has increased in importance in recent years because the geni~ taken for their talia can be sold freely in the Far East. h p are pelts which are marketed as leather g o d s or made into coats, and oil and meat are by-products from d l seals harvested, although in recent years the by-products have not always been utilised. The sealing seasons differ depending on which age or sex is k i n g taken. h p s are generdly taken from mid-July to midOctober, when they are approximately 8- 11 mo old (weaning age), whereas the best time for removing adult bulls is in the fist half of November, before the peak birth season. As a matter of experience, it is found that not only adult bulls are shot but also smaller bulls, some as young as about 6 yr old as judged from their body length. If a cow harvest is to be taken, then 1398

the period causing least disruption to the population is late in the pup sealing season, from September to October, by which time there should be a minimal number of unweaned pups. Associated with bull sealing is disturbance affecting cows with newborn pups and those about to pup. This phenomenon was noted decades ago (e.g. Wand 19551, but it has not been included in previous simulation models (e .g . Shaughnessy and Best 1980) presumably because of its indeterminate nature and its variability depending on the experience of the sealers and the timing of the harvest. If, due to poor management, bull sealing is allowed to extend into the peak pupping season, as has happened on some past occasions, then the consequences can include emigration of females and desertion of the pups by the cows, leading to death of these pups. This may lead to disruption of mating activities and hence reduced pup produeCan. 9. Fish. Aquat. Sei., Vol. 49, 1992


females have virtually no financial value. As far as the males are concerned, the financial value of small mdes is less than that of older males because of the associated smaller mass of genitalia. Therefore, more sf the smaller mdes would need to be killed to achieve the same harvested mass, and this requires a longer sealing season and greater processing time of the carcasses. For these reasons, both mdes and nonbreeding females aged between 1 and 5 yr are thought to be m inviable target for management consideration, and such removals have not been simulated. Seding rates of up to 60%are considered because these are the largest that are likely to be achievable in the field. k g nmcy rate is age dependent but assumed to be independent of bull abundance in the model. For this assumption there are no data, but since no sealing above a 60% rate is investigated, this is taken to be a robust assumption and the results viewed bearing this in mind. In the model, disturbance resulting from sealing operations at a colony causes an increase in pup mortality in the current year and a reduction in births in the subsequent yea. The model output is not scaled to true population size because the results required are merely to compare relative outcome under different management actions. The events occurring at a colony during one year (Fig. 3) are modelled by the following sequence of equations, and the data given in Table 1. Year

Sealing

FIG. 2. Number of pups and bulls harvested since 1900 (data from Wickens et al. 1991). T s date, c ~ w have s only been taken in 1985, 1988, and a few in 1989 and 1990.

tion at the colony the following year, both as a result of fewer pregnancies and the fact that some females do not return but move to another colony.

Model Structure and Parameterisation The model is for a single, spatially unstructured population and is deterministic with a 1-yr time step, separate sexes, and partial age differentiation (ages 0-6 yr). Population growth is density independent which is adequate for the purpose of the model, since there is no evidence of growth limitation, and defining a density-dependent function requires making too many assumptions for which there are no data. Options are included for sealing, disturbance effects of bull sealing, and estimation of food consumption. Natural mortality and mortality as a result of sealing operate independently. It is only possible to exmine the removals of certain individuals from the population for practical reasons involving the availability and selectivity of certain age and sex classes. The possible options are the killing of males (6 yr and older), breeding females, and pups (less than 1 yr old), all of which have been taken in the past from breeding colonies at particular times of year. Killing either nonbreeding females, most of which are between the ages of 1 and 5 yr, or males of the same age is not a practical management option because the sexes are not easily distinguished at this age. If seals of this age were to be taken, the two sexes would need to be taken together. This also presents problems in that these age classes are not concentrated at a colony during any particular perid, so selection of these ages would be time-consuming, and this in turn would cause increased disturbance to the rest of the individuals on the colony. In the case of females, removing these individuals would be done as part of a culling programme, not as a harvest, since

+

Can. J. Fish. Aquae. Sci., Val. 49, 1992

Runs were carried out at a variety of sealing rates. Sealing rate is the number of seals removed relative to the number of pups born. This is done for reasons of practicality because it is only the pups that are censused and therefore only these can be used as a reference point, without making further assumptions about age structure being constant. The number of male and female pups (S, and S, respectively) is reduced by the sealing rate of pups (C,). The number of males in the last age class (S,, + ) is reduced by the number of bulls removed, which is the bull sealing rate (C,) multiplied by the number of pups (S,). The number of cows (Sfi) in each of the sexually mature (i.e. aged 3 yr and older) age classes is reduced by the number of cows removed, which is calculated as the cow sealing rate (C,) times the number of pups times the percentage of females of that age divided by all sexually mature females. We have

Sfi = S, - (C;S ~ fsj,=/ 3C S,)

fori

P

3.

Survival The number of males that are territorial (St) is calculated by dividing the number of sexually mature cows (aged 3 yr and over) by the female aggregation size (A). In order to account for the mortality of territorial d e s (M,), the mortality rate of the males in age class 6 $ (Mm,+) is estimated by apportioning territorial mortality and older natural mortality (Ma) to the number of territorial males and other males aged 6 and older (S,, +), respectively. We have

TABLE1. Parameter values for the model. The notation presented here is subscripted by "m" for males, "j" for females, md "'E" females in the model equations. See text for details on source of data for parameter vdues. Parameter

Notation

Males

Females

Annual pregnancy rate (9%) 3 yr old 4 yr old 5 yr old

Lactating females

for lactating

Source(s) Butternorth and Wiekens 1998 Butternorth and Wiekens 1990 Butternorth and Wickens 1990

+

Annual mortality rate (9%)


~

P

de Villiers and Roux 1992 Butterworth 1998 Butternorth 1990

S

Juveniles and adults TemitoPiial bulls Average mass (kg) 1 yr old 2 yr old 3 yr old 4 yr old 5 yr old 6 yr old

Butternorth and Wickens Butternorth and Wickens Butteworth and Wickens Butteworth and Wickens Butteworth md Wickens Buttemorth and Wickens

+

percentage body mass consumed daily 1 yr old 2 yr old 3 yr old 4 yr old 5 yr old 6 + yr old

1990 1990 1998 1990 1990 1990

Innes et al. 1987 Innes et al. 1987 "Hnnes et al. 1987; "FRI, unpubl. data

Percentage of males at weaning Percentage of year from birth to weaning

David and Rand 1986; David 1987

Number of cows in an aggregation

Rand 1967

Disturbance Culling rate at maximum disturbance Current year Subsequent y e a

Assumed Assumed

Loss in pup production (9%) Guwent yew Subsequent year

Assumed Assumed

Starting number of pups (millions)

The number of seals of each age and sex (Si)grows into the next age class, once it has been reduced by natural mortality (M).The number of seals in age class 6 includes those already in age class 6 f and the seals moving up from age class 5. In the case of pups, the number of 1-yr-olds of each sex is calculated by applying the observed sex ratio (B,). We have

+

tdity rate for tenitorial bulls is calculated from resightimgs of branded animals (Buttemorth 1990). The size of am average female aggregation used as input into the model is estimated from data on counts of a11 females and territorial bulls in different areas of a colony (Rand 1967).

Disturbance

One mortality rate is used for pups for the range of values given in de Villiers and Roux (19921, so the number of pups of each sex is calculated by altering the sex ratio to that observed, which is skewed towards females (Oosthuizen 1991). Mortality rates for older seals are derived from the distribution of known-age females collected at sea for research, and mor-

In the absence of appropriate field data, piecewise-linear disturbance functions for the current (D,) and subsequent year (D,) are assumed, dependent on the sealing rate of bulls (C,) (Fig. 4). Maximum disturbance is denoted by MD, for the current year and MB, for the subsequent year. CD, and CD, are the sealing rates above which disturbance is assumed to be constant for the current and subsequent yews, respectively. If there is disturbance from the previous year and during the current year, their effects are multiplied.

MONTH

COLONY EVENT

MODEL EVENT

AUGUST SEPTEMBER SEALING


OCTOBER -- NOVEMBER

Bull sealing

- DECEMBER

I

Surviving end-of-year ~o~sllation

SURVIVAL

-

DISTURBANCE RECRUITMENT CONSUMPTION

-- JANUARY

OUTPUT

FIG. 3. Temporal sequence of events at a colony and the corresponding sequence in the simulation model.

If C ,

> CD, then i%D, = 0 then Dl

= CD, else Dl = I - (1 - Dl) (1 - MB,) else if D, = 0 then D, = C, elsell, = 1 - (1 - Dl) (1 - C,) If C, < CD2 then D, = (MDdCD,) Cb else 84, = MD,.

Assumptions about bull sealing disturbance me based on observations of the disturbance caused by large but1 harvests at Kleinsee. In 1984, 10 035 bulls were taken, folBowed by 6223 bulls md 6890 cows in 1985 (Wickens et al. 1991). Mortality of pups appears to have increased in the years of the large bull harvests because sealing extended right into the peak pupping season (Best 1990). Aerial surveys showed a decrease from 83 469 pups at the end sf 1983 down to 43 267 and 47 113 at the end of 1985 and 1986, respectively. The removal sf cows in 1985 would immediately reduce the number of births by approximately '9000 hat year, but the decreased pup production is also attributed to reduced number of pregnancies resulting from disturbance during 1984 and, possibly, emigration fmm the-colony. The sealing rate of bulls in 1984 can be calculated as 12% of the number sf pups born at the end of 1983. For the purpose of the model, it is assumed that above particular sealing rates, the mount sf disturbance remains constant whereas below them, it decreases linearly. These disturbance effects can be included or excluded from the model so that the consequences thereof can be evaluated. Recruitment Each cow produces only one pug, so the number of pups (So) is estimated by the sum of the number of females (St) that me multiplied by the proportion sf pups that are either pregnant (Pi) not born (because of lowered pregnancy rates) or die as a result of disturbance (I3

In year following sealing

In yeas sf sealing

.----------------------

Sealing rate ( O h ) FIG. 4. hnctions describing the effect s f disturbance during bull sealing operations on pup production, as assumed in the model.

Age-specific pregnancy rates are estimated from the number of pregnant females in samples of seals collected for research purposes (modified from Butternorth md Wickens 1990). Consumption In the itb age class, males (S,,), lactating cows (Li), and nonlactating cows (Sfi - L,) consume different amounts and are therefore treated separately. Cows with pugs are assumed to be lactating for a portion of the birth year (7') which is taken as 9.5 mo, based on the infomation from David and Rand (1986) and David (1987). The total amount of fish consumed ( F ) is calculated by the following set of equations: 6

i'

=

i-3 6

6

so = iC (pisfi) (1 =3

-

Can. 9. Fish. Aquat. Sci., Vol. 49, 1992

C Ipjsfi (1 - Dl) TI

F =

C [(S,iW,Rd

i =3

+ (Liy,R,) + ((Sfi - LJ WftRfJI I401


Index of number of seals removed PIG. 5. Annual growth rate after 20 yr as a function of an index of the number of seals removed as simulated by the model under different management schemes. The cull rate increases in steps of 10%between 0 md 60%in the direction of the m o w s for the culling of pups (sslid squares), cows (open squares), bulls excluding disturbance effects (solid circles), and bulls including disturbance effects (open circles).

where W , and W,, are, respectively, the mean mass of a male seal and female seal of age i and Wmi, Rli, and Rfi are, respectively, the mnual percentage of body mass consumed by males, lactating females, and nonlactating females. Age-specific body mass values are estimated from seals collected at sea for research (Butteworth and Wickens 1990) with the figure for 6 -t -yr-olds calculated as the masses of ages 6 yr and older weighted by a population age distribution (SEW, unpubl. data). The mass-specific consumption rates for males and nonlactating females are calculated using the biomass ingestion equation in Innes et al. (1987) for marine mammals, m d for lactating females, a rate is assumed (SEN, ungubl. data).

Model Results The control of seal population growth rate, reduction in fish consumption, and maximising of number of seals harvested were explored by simulating the sealing of pups, bulls, and cows separately at rates between 0 and 60%. Values are output from the model at the time when the model population is maximum, i.e. directly after recruitment. Outcomes sf the different scenarios are compared after 20 yr of simulation because, with the given model structure, a stable age distribution is achieved after approximately this time period. The model is not scaled to true population size; therefore, the number of seals removed and the amount of fish consumed is a relative scale a d is an index of the number removed and consumption. The efficiency of each scenario is compared. In this context, for the objective variables of population growth rate and seal csnsumption of fish, efficiency is defined as the largest change in each sf these with small increase in sealing rate (cull s r harvest) after 28 yr. In the case of the objective variable, harvest size, efficiency is merely the greatest number that can be removed even after 20 yr. Culling to Reduce Population Growth Rate Under a regime without culling, the mnual population growth rate of the model population is 4.8% (Fig. 5 ) . Based on census data (Wickens et al. 1991), the mnual growth rate of pups at colonies that have undergone no sealing through the census period is in the region of 3%. In the past few decades, pup

numbers of the South African fur seal population have averaged an annual growth rate sf 2.7 9% (Butteworth and Wickens 1 990). Historically, mainly pups and bulls have been removed from the population at highly variable rates, the rate of pup removal having ranged between 45 and B % , averaging 26%, and that of the bulls between 0 and 8%, averaging 2%, in the last two decades (Wickens et al. 1991). The model shows that under the average historical pup sealing rate, the annual population growth rate is 1.796, while that under the historical bull sealing rate is between 3.9 and 4.896, depending on whether disturbance effects are included or not. It must be remembered that the growth rate will have varied annually, depending on the combination sf sealing rate& In addition, the model results in Fig. 5 show population growth rate whereas that measured from census data is the growth rate of thepups only, which, because of the history of sealing, are not necessarily exactly the same. The growth rates are sufficiently similar to that estimated by Butteworth and Wickens (1990) a d from the data in Wickens et al. (1991) to demonstrate that the m d e l output adequately reflects the growth rates found in nature. As expected, population growth rate drops off with the number of seals removed under m y of the scenarios (Fig. 5). However, for pup and cow culls, the number of seals required to be removed to obtain further decreases in population growth declines above a certain cull rate because the population has been reduced after 20 yr and there ax fewer seals from which removals can be made. In the cases sf pup culling and of bull culling including disturbance effects, the results shown after 20 yr may differ from those after a shorter period because of the time lag between the decrease in number of pups and the time to maturity of females (-5 yr) when overall pup production will be reduced. Considering first the case with disturbance effects of bull culling excluded, removing bulls causes little decrease in annual growth rate because rn assuqtion of the model is that pregnmcy rate is independent of the number of bulls. At larger culling rates (>lo%) the removal of cows is most efficient in reducing population growth rate. At a low culling rate of cows, a negative population growth rate can be achieved but slightly less efficiently than when pups are being removed, but a larger culling rate of pups is required to achieve this. If there is disturbance resulting from the culling of bulls, then the removal of bulls may be more efficient than the removal of cows at low cull rates. At bull cull rates above 20% the effects of disturbance are assumed constant, and therefore, with ad&tional removals, the population growth rate is not reduced m y further. At high cull rates of cows, greater reduction in population growth rate is achievable. Thee features associated with cow and bull culling need to be considered. Firstly, when cows are being heavily culled, the model shows that the sex ratio becomes strongly skewed towards males, m d at a 60% cull rate, the population becomes almost entirely male (Fig. 6). Such altered sex ratios have not been observed in nature and many have unforeseen consequences with respect to social structure of the population, and these are not accounted for in the model. Secondly, a long-term undesirable consequence of culling bulls may result from the continued removal of the largest bulls, which will be replaced by smaller bulls. These smaller bulls, being territorial a d therefore having to spend a Iage amount of time on the colony without food, might not have adequate fat reserves and would therefore suffer even greater mortality than that already affecting luge territorial bulls. A consequence of this might also be reduced pregnancy rates, if many sf the small bulls die early Can.

Fish. Aqucmt. Scb., Vob. 49, 1992


Sealing rate (%) FIG. 6. Percentage of males in the population after 20 yr as a function of the cull rate of seals as simulated by the model under different management schemes, which include the culling of pups (solid), cows (dark hatch), bulls excluding disturbance effects (light hatch), and bulls including disturbmce effects (stippled).

Index of number of seals removed FIG.8. Amount of fish consmed by the seal population after 28 yr as a function of m index of the number of seals removed as simulated by the model under different management schemes, both excluding m d including the disturbance associated with bull culling. The cull Bate increases in steps of 80% between 0 and 60%in the direction of the m o w s for the culling of pups (solid squares), cows (opera squares), bulls excluding disturbance effects (solid circles), and bulls including disturbance effects (open circles).

large reduction in fish consumption, but the culling of cows is muck more efficient and results in a greater population reduction above a 80% cull rate. If disturbance during bull removals is assumed in the model, then culling bulls can achieve almost the same reduction in consumption as culling pups or COWS, and it is similarly efficient. Hmesting to Maximise Seal Removals

Year FIG. 7. Change in the length of bulls taken (open squares) in the harvest at KIeinsee (SFRI, unpubl. data) and the number of bulls harvested (solid squares) (Wickens et al. 1991) since bull sealing began there in 1982.

in the breeding season. The mean size of bulls harvested at Kleinsee (SIX%,unpubl. data) decreased during the 8980's (Fig. 7) because smaller bulls were taken to fill the large quotas during this time. In 1988 and 1989 the size of bulls in the hwvest increased, but this was because the small harvest allowed remaining larger bulls to be taken preferentially. Thirdly, the killing of pregnant cows (unavoidable because sf the 1%-mo gestation perid) and of bulls when there is associated disturbance causing pup mortality may be considered to be undesirable on grounds of cruelty. Culling to Reduce Fish Consumption by Seals Simulations of the reduction in seal consumption of fish through the removal of seals give a similar result to that obtained when the objective was reducing the population growth rate (Fig. 8). The major difference is that when bulls we culled, fish consumption decreases whereas in terns of population growth rate in numbers, there is virtually no decline. This results from heavier bulls eating a much lager amount of fish per capita than other seals. Culling either pups or cows will achieve a Can. 9. Fish. Aqwt. Sci., Vol. 49, 1992

The results obtained from the culling simulations can d s o be used to compare the numbers obtained from pup, cow, and bull harvests (see Fig. 5). Assuming no disturbance resulting fiom bull removals, the largest number of animals removed at a set population growth rate can be achieved by harvesting bulls. If there is disturbance resulting from bull removals, then, for a set population growth rate, at low harvest rates (-- lo%), more COWS than pups or bulls may be harvested. At other harvest rates, the largest harvest of seals may be achieved by taking PUPSSensitivity Analysis of Outcomes under Management Actions The results of a 18% increase in pregnancy, survival, and consumption rates from their values in Table 1 on a single run (i ,e. changes which lead to enhanced population size and greater consumption) and likewise to a 10% decrease in these parameter values (i.e. changes which lead to decreased population size and lower consumption) alter some of the conclusions reached above (Fig. 9). If the parameters are altered so that pup production is enhanced, the difference is basically that culling cows causes the largest decrease in population growth rate and consumption and this is done most efficiently. If parameter conditions a e changed to lead to decreases in overall pup production, the same pattern emerges. This results because by increasing survival and fecundity, the breeders in the population are most affected and therefore, when reduced in number, most affect the population. The model population is not scaled to true population size, so annual population growth simply moves up or down the scale when parameter values we altered. No difference is


ENHANCED POPULATION GROWTH AND CONSUMPTION RATES

REDUCED POPULAnION G R O W H AND CONSUMPTION RATES

Index of number st seals removed

Index of number of seals removed

PIG. 9. Sensitivity analysis of annual population growth and amount of fish consumed by the seal population after 20 yr as a function of an index of the number of seals removed. These are simulated by the model under different management schemes when parameter values are altered to either enhance population growth and consumption rates or to reduce growth and consumption. The sealing rate increases in steps of 10% between 0 and 60% in the direction of the mows for the culling of pups (solid squares), cows (open squares), bulls excluding disturbance effects (solid circles), and bulls including disturbance effects (open circles).

detectable in the percentage of males in the population under the different conditions of the sensitivity analysis (not shown).

Capmgsari~apnof ERectivewess of Management P$ctisns The simulations suggest that, under the model assumptions, and accounting for the results of the sensitivity analysis, different strategies should be adopted for managing the seal pop-

ulation depending on whether the objective is to reduce seal population growth rate, reduce fish consumption, or mmirnise the seal harvest. The effectiveness of the management actions relative to the three objectives is s u m m ~ s e din Table 2, based on the management actions that are considered feasible and practical. Effectiveness is measured by combining the success with which the management action achieves the desired result and the efficiency with which it is achieved. Hn the case of

TABLE2. Summary of model results with respect to the t h e e management objectives and alternative management actions. Management objective

Management action Pup removal COWremovala Bull removalb with no disturbance Bull removalb with disturbance'

Reduce population growth rate

Reduce fish consumption

Maximise seal Rawest

Success Effectiveness level Efficiency (rating based (El on S and E)

Success Effectiveness level Efficiency (rating based (S) E) ~ g Sl m d E )

Success Economic Effectiveness level Efficiency value (rating based (S) (E? (V) onS,E,agldV)

Medium Medium High High Low Low Medium High -

-

1 4

High High Low

Medium High Low

3 1 4

Medium Medium Medium Low Low Low High High High

2 4 1

2

High

High

1

Low

3

3

Low

High

-

"But sex ratio severely altered - Desirable? bBut contiglued removal of bulls causes reduced pup pmduction - Desirable? "But pup mortality increased because of disturbance - Acceptable? Cart. J . Fish. Aquat. Sca'., Vob. 49, 1992


maximising a hmest, the effectiveness is further weighted by an index of the relative economic value of the products. Specific economic values for each of the seal products are difficult to obtain, paffticulaly as they vary depending on the political climate of the time and the acceptability of their sale (Wickens et a%.1991). Therefore, an index of their importance commercially is most appropriate and is indicated in Table 2 as being indicative of the situation at the present time. Management actions are ranked according to their effectiveness. If there were no disturbance during bull removals, then the most effective strategy for achieving either a reduction in seal population growth rate or reduced fish consumption is the removal of cows. However, the severe skewing of the sex ratio under a cow culling regime may be ecologically undesirable and could lead to destabilisation or extinction. If the objective is to maxirnise the number of seals taken, them the removal of bulls is most effective. The effectiveness sf a pup h m e s t is secondary to that of bulls, both in tems of the number taken and the economic value of the products. Not only is the removal of cows ineffective in tems of numbers removed, cows also currently have a low per capita economic value, and therefore the removal of cows is clearly a poor choice with respect to maximising harvest. If there is disturbance during bull removals, then the most effective action for obtaining a reduction in fish consumption may be the removal of bulls, and this action may also be quite effective in reducing seal population growth rate. However, the reduction is caused partly by increased mortality of pups which is not an acceptable means because it permits deliberately poor management, and it may not be desirable to remove large bulls continually. Two of the management goals, achieving a reduced growth rate of the seal population and reducing fish consumption by the population, are compatible whereas the remaining goal, achieving the most profitable h m e s t from the seal population, is mot compatible with the first two. Since management policy is likely to attempt to account for some combination of these three goals, this may best be achieved, not by removing one sex or age class exclusively, but by removing animals from a combination of sex and age classes, the proportions varying depending on the relative weight assigned to the three goals.

Acknowledgements The authors would like to thank all those who participated in the Benguela Ecology Programme workshop in 1987 for providing input and ideas and the reviewers for their constructive comments. Financial support for the first author was received from the South African National Council for Oceanographic Research through the Systems Analysis Project of the Benguela Ecology Programme, and logistical support was given by the Zoology Department of the University of Cape Town. Both of these are gratefully acknowledged. Ms C . Schr6dinger is thanked for her input during preparation of the manuscript.

Can. J. Fish. Aquar. Sci., V01. 49, 1992

References ANONYMOU~ . 1881. Report on the Benguela Ecology Programme workshop on seal-fishery biological interactions. Rep. Benguela Eeol. Prog. S. Afr. 22: 65 p. BEST,P. B. 1996). Annex 12. Departures from sealing quotas and recommendations, 1980-1989, p. 105-106. Hn Report of the subcommittee appointed at the request of the Minister of Environment Affairs and of Water Affairs, to advise the Minister on scientific aspects of sealing. Ministry sf National Education and Environment Affairs, Cape Town, South Africa. 112 p. BEVERTON, R. J. H. 1985. Analysis of marine mammal-fisheries interaction, p. 3-33. Hn J. R. Beddington, R. 9. H. Beverton, and D. M. Lavigne [ed.] Marine mammals and fisheries. George Allen 8 Unwin, London. B ~ R W O R TD.H S. , 1990. Annex 4. Estimates of non-juvenile survival rates, p. 62-64. Pn Report of the subcommittee appointed at the request of the Minister of Environment Affairs and sf Water Affairs, to advise the Minister on scientific aspects of sealing. Ministry of National Education and Environment Affairs, Cape Town, South Africa. 112 p. B U ~ E R ~ ~ OD. R TS., H ,D.e. DUWY,P. B . BEST,AND n%. 0. BERGH.1888. On the scientific basis for reducing the South African seal population. S. Afr. 9. Sci. 84: 179-188. BUTTERM.~ORTH, D. S., AND P. A. WICKENS.19%. Annex 2. Modelling the dynamics of the South African fur sed population, p. 33-61. Hn Report of the saatxommittee appointed at the request of the Minister of Environment Affairs and of Water Affairs, to advise the Minister on scientific aspects of sealing. Ministry of National Education and Environment Affairs, Cape Town, South Africa. 1 12 p. DAVID.J. H. M. 1987. The South African fur seal, Arctocephalus pusillus pusileus, p. 65-7 1. In J. P. Croxall and R. L. Gentry [ed.] Status, biology and ecology of fur seals. Proc. Int. Symp. Workshop, Cambridge 1984. NBAA Tech. Rep. 5 1. 1989. Seals, p. 287-302. In A. I. L.Payne and R. I. M. Crawford [ed.] Oceans of life off southern Africa. Vlaeberg, Cape Town, South Africa. DAVID,J. H.M., AND R. W. RAND.1986. Attendance behaviour of South African fur seals, p. 126-141. In R. L. Gentry and G . L. Kooyman [ed.] Fur seals: maternal strategies on land and at sea. Princeton University Press, Princeton, Nb. DE VILLBERS, D. J., AND I-P. Roux. 1992. Mortality of newborn pups of the Cape fur seal (Arcaocephaluspusillus pusillus) in Namibia. S . Afr . J . mar. Sci. (In press) INNU,S., D. M. LAVIGNE, W. M. EAIPLE, AND K. M. KQVACS. 1987. Feeding rates of seals a d whales. I. Anim. Ecol. 56: 115-130. B o s a ~ u a zW. ~ ~ PI. , 1991. General movements of South African (Cape) fur seals Arctocepkalus pusileus pusillus from analysis of recoveries of tagged animals. S. Afr. J. mar. Sci. 11: 21-29. OOSTHUIZEN, W.H., AND J. H.M. DAVID.1988. Non-breeding colonies of the South African (Cape) fur seal Arctocephalus pusilhspusillus in southern Africa. S. Afr. Div. Sea Fish. Invest. Rep. 132: 17 p. RAND,83. W. 1955. Reproduction in the female Cape fur seal, Arctocephalus pusillus (Schreber). Roc. Zool. Soc. h n d . 124: 7 17-740. 196'9. The Cape fur seal (ArctocepfmIuspusillus). 3. General behaviour on land and at sea. S. Afr. Div. Sea Fish. Invest. Rep. 60:39 p. SHAUGHNESSY, P. D. 1979. Cape (South M c m ) fur seal, p. 37-48. Pn Mammals in the seas. F A 0 Fish. Ser. 5(2). 1985. Interactions between fisheries and Cape fur seals in southern Africa, p. 119-134. In JJ. R. Beddington, R. J. H.Beverton, and D. M. Lavigne [ed.] Marine mammals and fisheries. George Allen & Unwin, London. SHAUGHNESSY, P.D.,AND P. B. BEST.1980. A discrete population model for the South African fur seal, Arctocephalus pusildus pmiklus, p. 163- 176. Pm Maaaamals in the seas. FA0 Fish. Ser. 5(4). WICKENS, P. A., J. H.ha. DAVID,P. A. SHELTQN, AND J. G.FIELD.1991. Trends in harvests and pup numbers of the South African fur seal: implications for management. S. Afr. 9. mar. Sci. 11: 307-326.