Management of New Zealand orange roughy fisheries - a deep learning curve

J. Annala^[327], M. Clark^[328], G. Clement^[329] and J. Cornelius^[330]

1. INTRODUCTION

This paper presents two case studies of the management of orange roughy fisheries in New Zealand. When management measures were first introduced for orange roughy in the early 1980s, decisions had to be based on little or no information on the biology of the species or its abundance. New Zealand’s orange roughy fisheries are currently divided into a number of separate fishery management units governed by both Government regulations and industry management agreements. The paper concludes with the fisheries management lessons learned from the New Zealand experience and offers suggestions on how these lessons could be applied to the management of new fisheries for orange roughy and other deepwater, slow-growing fish species.

2. HISTORY OF THE FISHERY

Orange roughy were first taken commercially in New Zealand in 1979. They are caught primarily between 700 m and 1500 m depth. They have been taken in a mix of fisheries on both hills and over flat bottom. Over ten separate orange roughy fisheries have been developed within the New Zealand EEZ since the fishery first began (Figure 1, from Clark et al. 2000). Numerous papers have been published describing the New Zealand orange roughy fisheries (see Clark et al. 2000, Francis 2001, Clark 2001 and references therein).

The early years of most fisheries were characterized by high catch rates (over 50 t a tow in some fisheries) and saturation fishing. These high catch rates lasted only a few years in most fisheries and were followed by sharp declines in CPUE. As the fisheries continued the fisheries on flat bottom areas contracted in both the total area of the fishery and the high catch rate area. Catch rates also declined on hills and serial depletion of hills was observed in some fisheries. This behaviour was observed during the fish-down phase in most orange roughy fisheries when the accumulated, unfished biomass is reduced by fishing.

Quotas were first introduced in 1983 and set at a level of about 23 000 t for all orange roughy fisheries combined. As new fisheries were discovered, the total quota was increased, eventually peaking at about 63 000 t in 1989. Total catches from all areas combined increased rapidly, reaching about 45 000 t by 1985 and eventually peaking at about 57 000 t in 1989. Since the late 1980s total quotas have been steadily reduced as new information has been obtained on orange roughy biological parameters and abundance indices and their stock assessment methods have been refined.

New Zealand legislation requires that stocks be maintained at or above the size that will support the maximum sustainable yield (B_MSY), or be rebuilt to that level if they fall below them. For the nine fisheries for which quantitative stock assessments have been conducted, stock size has fallen below that target level in seven stocks (Figure 2, from Francis and Clark 2005). For all of these seven fisheries, management strategies with reduced catch limits have been implemented to promote stock rebuilding.

3. BIOLOGY

Orange roughy are estimated to be long-lived and slow growing and a number of studies indicate maximum ages are greater than 100 years (e.g. Smith et al. 1995, Tracey and Horn 1999). Estimates of natural mortality (M) are low, 0.04-0.05 (Francis et al. 1992, Doonan 1994). Estimates of age at first maturity are high, at about 30 years (Francis and Horn 1997, Horn, Tracey and Clark 1998). Fecundity is relatively low with the average female producing about 50 000 eggs per spawning (Pankhurst 1988). In addition, recruitment could be low and highly variable (Francis et al. 1992).

The combination of the above characteristics makes orange roughy a relatively unproductive species. Sustainable yields are low at an annual rate of 1 to 2 percent of the virgin biomass and 4 to 6 percent of the biomass that supports the maximum sustainable yield Their predictable aggregation behaviour in time and space for both spawning and feeding makes localized populations vulnerable to overexploitation.

4. BIOMASS ESTIMATES

Four different biomass estimation methods have been used over time. The initial biomass estimates for many fisheries came from random trawl surveys. Commercial CPUE was not used during the early stock assessments because it seemed obvious that it would not be proportional to biomass. Many early fisheries were based on aggregations and it was thought that fishers would be able to maintain high catch rates while the aggregations shrunk and abundance declined. Thus, CPUE was expected to decline more slowly than biomass. However, the opposite may also be true because fishing may disperse aggregations.

Beginning in the early 1980s, egg production surveys were used in three fisheries to estimate absolute abundance (Francis, Clark and Grimes 1997, Zeldis et al. 1997). However, they are no longer used because of the difficulty of estimating the mortality of eggs. While acoustic surveys of orange roughy have been undertaken since the mid-1980s, the technique has only been well-developed enough to estimate absolute abundance since the late 1990s. Acoustic surveys appear promising for fisheries in which most of the stock is in mono-specific aggregations. However, they are of less value for depleted stocks where abundance is low, and for stocks widely distributed over flat bottom as opposed to being in aggregations. Target identification and target strengths of other species become difficult issues to successfully address in these latter situations. In these situations, the large number of other species found in association with orange roughy, most of which have greater target strengths than orange roughy, tend to dominate the acoustic returns of roughy.

5. CASE STUDIES

5.1 Development of fisheries

Two case studies are presented that cover the range of New Zealand’s orange roughy fisheries both in terms of their size and longevity. The fishery on the northeast Chatham Rise is both New Zealand’s largest and longest running orange roughy fishery, beginning in 1979. The fishery on East Cape is one of the smallest and newest of the fisheries, first developing in 1994.

5.2 Northeast Chatham Rise

A retrospective analysis has been carried out for the fishery on the northeast Chatham Rise in ORH 3B (Francis 2001). Table A1 from Francis (2001) with the addition of information on the method of biomass estimation; the stock assessment model used; and the relevant yield estimates, catch limits, and catches are shown here as Table 1. A number of large changes have been made to the data used for the stock assessment and the assessment modelling approach during the history of the northeast Chatham Rise fishery, which have resulted in changes to estimates of biomass and yield. These were as follows.

• Changes to the biomass estimation technique and the way the estimates were incorporated in the stock assessment model - Trawl survey biomass estimates were the only indices used for the northeast Chatham Rise fishery before the 2001 assessment. Before 1989 they were treated as absolute estimates and since 1989 as relative indices. Acoustic biomass estimates (treated as absolute estimates of abundance) and standardized analyses of commercial CPUE were first incorporated in the stock assessment in 2001.

• Changes to biological parameter estimates - Probably the most important changes to biological parameters were the estimates for natural mortality (M), the average age at maturity (A_r) and the age at maturity ogive (S_r). M was assumed equal to 0.10 in the early years of the fishery before any ageing studies of orange roughy had been conducted. This was thought to be conservative given the common value of 0.2 applied to many shelf species. It was reduced to 0.04 to 0.05 in the late 1980s - early 1990s as the results of early aging studies became known. A_r was initially set to five years in 1987 and subsequently increased to 33-34 years in 1994 and then reduced to 30 years in 1997. S_r was initially set to three years in 1991, increased to 9-8 years in 1994 and later reduced to three years in 1999.

• Changes to the stock assessment methods used - Initially, stock assessments were based on treating the random stratified trawl survey biomass estimates as absolute, back calculating the results to obtain an estimate of virgin biomass (B₀), and using MSY = 0.5 MB₀ "relation of Gulland (1971) to calculate the sustainable yield. Since 1989 trawl survey biomass estimates have been treated as relative and stock assessment models have become increasingly more sophisticated. The first "model-based" stock assessment was for the Chatham Rise using a modified de Lury method (Sissenwine 1988). This was followed by deterministic stock reduction models in 1990 (Francis 1990). In 1992 this method was extended to the enhanced stock reduction analysis technique (Francis et al. 1992) that uses stochastic recruitment and allows the fitting of mean length data. The next major step was the development of a Bayesian stock assessment approach for the northeast Chatham Rise in 2001 (Smith et al. 2002). The Bayesian approach was also adopted in a model called CASAL (Bull et al. 2002) that was used for the first time for orange roughy in the 2003 assessment for the East Cape stock (Anderson 2003).

• Changes to model assumptions - There have been a number of changes to model assumptions and model structure through time, which are summarized by Francis (2001). There have also been changes to the areas on the northeast Chatham Rise to which the stock assessment has been applied. From 1986 to 1990 the stock assessments were for the entire ORH 3B management area; however, because most fishing occurred on the Chatham Rise, the stock assessments effectively applied only to that area. Between 1991 and 1995 the assessments were for the entire Chatham Rise, including the Northwest, Northeast and South Rise (Figure 3). From 1996 to 2000 the assessments applied to the Northeast and South Rise combined. In 2001 the assessment applied to the Northeast Rise only. Separate assessments were carried out for the Northwest and South Rise areas, as the Chatham Rise is now assumed to have three separate stocks for management purposes.

TABLE 1
Chatham Rise Management Data

* T(A) = trawl survey indices used as absolute biomass estimates, T(R) = trawl survey indices used as relative abundance indices, A(A) = acoustic survey indices used as absolute abundance indices, CPUE = standardised CPUE analysis relative abundance indices.

** dSRA = deterministic stock reduction analysis, sSRA = stochastic stock reduction analysis, BAY = Bayesian stock assessment model. See text for details.

*** MSY = Maximum Sustainable Yield, Y = Yield, C = Current Annual Yield.

Discussion

Stock assessments have been carried out for the northeast Chatham Rise fishery since 1985. However, as described above, there have been many changes to the models used and the data inputs over time as more knowledge has been gained about the species and the fishery. What have been the impacts of all these changes?

The retrospective analysis of Francis (2001) shows a consistent reduction in the estimates of virgin biomass through 1989, which he attributed to the short time series of data in the earlier years (Table 1). However, since 1990 there has been no consistent trend that can be attributed to the longer time series of data available for the assessment. Inspection of the changes to the model used and the data inputs and assumptions does not reveal any single change that lead to the reduction in the estimates during the earlier years. It is most likely that these reductions were due to a combination of factors acting in complex ways.

Between 1987 and 1988 the estimates of B₀ more than halved because of a change in how biomass was calculated from the trawl surveys (no school height adjustment was applied). In 1989 the trawl survey biomass estimates were changed from relative to absolute, the M estimate halved from 0.1 to 0.05, and A_r increased from 6 to 20 years. All these changes made little difference to the estimates of B₀. However, the change from the use of Maximum Sustainable Yield (MSY) as the reference yield, where MSY = 0.5 MB₀, to the use of Current Annual Yield (CAY)^[331] resulted in a substantial reduction in the yield estimate, from 20 350 t in 1988 to 8 000 t in 1989, while the estimates of virgin and current biomass changed little between those years. Between 1989 and 1991 the estimates of B_current decreased from 33 percent of B₀ to 10 percent of B₀. This appears to have been mainly caused first by catch levels being substantially greater than the production from the stock and second, by the effect of sampling error in the trawl survey estimates. The change from a deterministic to stochastic stock reduction analysis in 1992 made little difference to the estimates of biomass and yield.

The biggest change to the yield estimates resulted from a reduction in the estimate of B₀ from over 900 000 t in 1987 to about 400 000 t in 1988, coupled with a change to the yield estimation technique from the use of MSY = 0.5 M B₀ to the estimation of CAY from the deterministic stock reduction analysis and a reduction in the estimate of B_current from 33 percent of B₀ in 1989 to 10 percent of B₀ in 1991. This reduced the estimate of sustainable yield from 46 750 t in 1987 to 2 200 t in 1991. As a result of the reduction in biomass and yield estimates in successive stock assessments during the late 1980s and early 1990s, the catch limit was reduced from 38 300 t to 8 000 t over the six year period from 1988-89 to 1994-95. Since 1997 the area for the stock assessment has been reduced to the northeast Chatham Rise only, so biomass and yield estimates and catches and catch limits are not comparable to those for earlier years.

In the 2001 assessment the estimate of B_current increased to 44-45 percent of B₀ due mainly to the inclusion of the acoustic biomass estimates and CPUE indices in the models. The model estimates indicate that the stock is currently above the B_MSY target and some model trajectories suggest that it may never have been below B_MSY.

In summary, for the Chatham Rise fishery, as more and better information was obtained a number of positive steps were taken in the management of the fishery.

• The fishery was progressively subdivided into a number of management units that better reflected what is known about the biological stocks in the area.

• Catch limits and catches were progressively reduced during the fishing down phase of the fishery.

• It appears that the stock size has been maintained at a level near or above the target level of B_MSY.

5.3 East Cape

Several hill complexes in the East Cape area (Figure 1) of ORH 2A (North) were first fished commercially in 1994. The fishery has occurred mainly during the June winter spawning period. Over time a number of critical changes have been made to the data used in the stock assessment and the assessment modelling approach that resulted in changes to estimates of biomass and yield (Table 2). These are described as follows.

• Changes to the biomass estimation technique and the way the estimates were incorporated in the stock assessment model - Before 2000 the only biomass estimate used in the stock assessment was an absolute estimate of abundance from an egg production survey carried out in 1995 and incorporated into the assessment for the 1996-97 fishing year. This estimate was reviewed the following year and revised downwards. Relative biomass indices from a standardized analysis of commercial CPUE were first incorporated into the assessment in 2000. The base case assessments for 2000 and 2003 used the CPUE analysis only and did not incorporate the egg production biomass estimate.

• Changes to biological parameter estimates - Probably the most important changes to biological parameters were those for the average age at maturity (A_r) and the age at maturity ogive (S_r). (M has always been set equal to 0.045 for the stock assessments for this fishery.) A_r was initially set to 33-34 years in 1996 and subsequently reduced to 30 years in 1997 and then 26 years in 2000. Sr was initially set to 9-8 years in 1996, and later reduced to 4 years in 1997 and then 3 years in 2000.

• Changes to the stock assessment methods used - The deterministic stock reduction model of Francis (1990) was used for all assessments through 2000. The CASAL model (Bull et al. 2002) was used for the 2003 assessment.

TABLE 2

East Cape. Biological parameter estimates

* Egg = egg production survey used as an absolute abundance estimate, A(A) = acoustic survey indices used as absolute abundance indices, CPUE = standardised CPUE analysis relative abundance indices.

** dSRA = deterministic stock reduction analysis, CASAL = CASAL model. See text for details.

*** C = Current Annual Yield.

¹ Biomass estimate from acoustic survey not accepted for stock assessment.

² A separate catch limit for the East Cape hills did not exist in 1993-94 and it was included in overall TAC of 6 666 t for ORH 2A.

5.4 Discussion

The absolute abundance estimate from the 1995 egg survey was revised from 40 000 t in 1996 to 29 000 t in 1997 and was the main reason for the decrease in the estimates of biomass and yield from the stock assessments for those two years. In 2000 relative abundance indices from a standardized CPUE analysis were incorporated into the stock assessment for the first time. The base case assessment used only the CPUE indices and provided substantially reduced estimates of biomass and yield compared with the previous assessment in 1997. The alternative case used both the revised egg survey estimate and the CPUE indices. However, the six years of CPUE data dominated the single absolute biomass estimate from the egg survey in the alternative case, and the estimates of biomass and yield were similar between the two cases. Changes to the biological parameter estimates appear to have had little impact on the model estimates of biomass and yield.

The only major change in 2003 was the use of the CASAL model for the first time. The base case again used the CPUE data only. The estimate of current biomass increased from 14 percent B₀ for the 2000 assessment to 24 percent B₀ for the 2003 assessment and the CAY increased from 130 t to 370 t. The 2003 model results suggested that the stock size had been rebuilding over the previous few years.

As more and better information was obtained for this fishery, and the early catch limits and catches were determined not to be sustainable, positive management interventions were made with the catch limit being reduced from 2 500 t in 1998 to 200 t in 2001.

6. SEAMOUNTS META-ANALYSIS

A meta-analysis of data from orange roughy fisheries on seamounts and hills in the New Zealand region was carried out to determine if information from fishing on these features can be informative in predicting possible population size for new fisheries on these features (Clark, Bull and Tracey 2001). Physical attributes and catch data for orange roughy fisheries for 77 hills and seamounts were analysed as independent variables against the minimum orange roughy population size estimated from the historical level of catch taken from these features. These date were evaluated to determine if they were useful predictors of the likely safe catch level from newly developed hills and seamounts. It was concluded that data on the physical features of hills and seamounts can be informative in predicting approximate possible stock size of orange roughy on previously unfished features and can provide useful guidelines for the management of these fisheries.

7. MANAGEMENT OF DEEPWATER FISHERIES IN NEW ZEALAND

What are the main features of New Zealand’s Quota Management System (QMS) that have provided the framework for managing New Zealand’s deepwater fisheries?

• A fundamental, underlying feature of New Zealand’s QMS is the expectation that a rights-based fisheries management system provides the proper incentives to quota holders to maintain their asset value through long-term sustainable management

• The main management objective is to maintain stocks at or above the level that will support the Maximum Sustainable Yield (B_MSY). Where stocks are below that target level, they must be rebuilt to at least the target level.

• A flexible management approach has developed within the framework of the QMS. The major examples of this flexibility in the deepwater fisheries have been (a) the establishment of area catch limits, often around specific features such as hill fisheries, within the Total Allowable Catch (TAC) to allow for fine-scale management and (b), the implementation of an adaptive management programme in new fisheries (e.g. ORH 1) that allows catch limits to be set at experimental levels, followed by a monitoring programme, to "test" what catch levels may be sustainable in the long-term

• TACs and sub-area catch limits are set to attain the target biomass level. During the fishing down phase TACs and catch limits may be set at levels higher than the long-term sustainable yield until the target biomass is reached. For stocks that have been reduced below the target biomass level, TACs and sub-area catch limits are set at levels below long-term sustainable yield levels to promote stock rebuilding.

• Fisheries management is supported by sound, high-quality scientific research and stock assessment, and a comprehensive enforcement programme that for the deepwater fisheries are largely paid for by the fishing industry.

• Fisheries management, research and stock assessment is largely supported by the fishing industry through purchase of additional research (e.g. acoustic surveys and at-sea biological data collection), stock assessments and assistance with Ministry contracted research (e.g. provision of catchers vessels for biomass surveys).

8. DISCUSSION

8.1 Approaches to stock assessment

Francis (2001) suggested three possible approaches to stock assessment (and the subsequent implications for provision of management advice) in data limited situations with a short time series of biomass estimates. These were

i. say that we don’t know where we are and that we need to wait for more data before providing stock assessment advice

ii. use the mode of the distribution of B₀ (rather than the mean or median) in the provision of assessment advice and

iii. constrain recruitment to be deterministic, which underestimates the true uncertainty but allows the early provision of advice. This was Francis’ (2001) preferred alternative.

From a fisheries manager’s perspective, the first alternative of doing nothing is not acceptable. We have to use the best information that is available, even when it is incomplete, and make decisions based on that information. If that information is subsequently proved incorrect, then we have to be prepared to change our decisions.

8.2 Lessons learned

What lessons have been learnt from the past 20 years of research, stock assessment and management of New Zealand’s orange roughy fisheries?

i. Because of their low productivity, sustainable yields from orange roughy fisheries are estimated to be low, at an annual rate of 1 to 2 percent of the virgin biomass, B₀, and 4-6 percent of B_MSY.

ii. Because of the aggregating behaviour of orange roughy, particularly on hill features and during the spawning season, it is easy to overestimate the unfished biomass.

iii. The major scientific challenges have been to obtain reliable estimates of orange roughy life history parameters and stock size in order to estimate the yields required to move the stocks from B₀ to B_MSY and maintain them at that level. These challenges have required the development of innovative scientific and technological solutions to obtain estimates of biomass and productivity of these fish, which live at 800 m to 1 500 m depth, are seasonally mobile and are difficult to survey. Significant advances in biomass estimation and modelling techniques have been made in the late 1990s for the larger fisheries. Moreover, using the results of quantitative analyses such as the seamounts meta-analysis and qualitative comparisons with similar existing orange roughy fisheries, it is possible to make approximate estimates of likely unfished stock size for new fisheries. The challenge remains for managers and scientists to continue to refine these methods and to apply them to the smaller fisheries.

iv. Given the combination of low long-term sustainable yields and difficulties in accurately estimating initial stock size, it is difficult to accurately specify a time stream of future catches and catch limits that will result in an orderly fishing down phase to achieve the target biomass. In New Zealand this has resulted in the B_MSY target being exceeded in many fisheries and the subsequent need to rebuild stocks back to B_MSY. However, it is possible to set appropriate catch limits to prevent stocks from being reduced below target levels, at least for large stocks such as the northeast Chatham Rise, given a good scientific basis for research and stock assessment and a management framework that supports sustainable management decision-making.

v. The major fisheries management challenge is to use information on possible stock size and our knowledge about the low productivity of the species to devise an orderly fish-down strategy that satisfies both the desire for high initial catch levels and the need to ensure that the target biomass level is not exceeded. The use of fine-scale catch limits and the adaptive management programme, as is done in New Zealand, provides fisheries managers with two mechanisms that support this approach.

These lessons learnt from orange roughy can be applied to fisheries for other deepwater, slow-growing fish species. Such species would include black oreo, smooth oreo and black cardinalfish. These three species all form aggregations in New Zealand waters and elsewhere for spawning and feeding, often over seamounts and hill features and are often caught in large quantities by bottom trawling. Given that the longevity and productivity for all three species are similar to that for orange roughy, the same management approach as that proposed for orange roughy for setting catch limits for new fisheries of these species can be applied.

9. ACKNOWLEDGEMENTS

The authors would like to thank Chris Francis for his constructive comments on an earlier version of the manuscript and for his permission to use Figure 2.

10. LITERATURE CITED

Anderson, O.F. 2003. CPUE analysis and stock assessment of the East Cape hills (ORH 2A North) orange roughy fishery for 2003. New Zealand Fisheries Assessment Report 2003/23.

Bull, B., R.I.C.C. Francis, A. Dunn & D.J. Gilbert 2002. CASAL (C++ algorithmic stock assessment laboratory): CASAL user manual v1.02.2002/10/21. NIWA Technical Report 117. 199 pp.

Clark, M. 2001. Are deepwater fisheries sustainable? - the example of orange roughy (Hoplostethus atlanticus) in New Zealand. Fisheries Research. 51: 123-135.

Clark, M.R., O.F. Anderson, R.I.C.C. Francis & D.M. Tracey 2000. The effects of commercial exploitation on orange roughy (Hoplostethus atlanticus) from the continental slope of the Chatham Rise, New Zealand, from 1979 to 1997. Fisheries Research. 45: 217- 238.

Clark, M., B. Bull & D. Tracey 2001. The estimation of catch levels for new orange roughy fisheries on seamounts: a meta-analysis of seamount data. New Zealand Fisheries Assessment Report 2001/75.

Doonan, I.J. 1994. Life history parameters of orange roughy: estimates for 1994. New Zealand Fisheries Assessment Report Document. 94/19.

Francis, R.I.C.C. 1990. A maximum likehood stock reduction method. New Zealand Fisheries Assessment Report Document. 90/4.

Francis, R.I.C.C. 2001. Stock assessment for 2001 of orange roughy on the northeast Chatham Rise. New Zealand Fisheries Assessment Report 2001/41.

Francis, R.I.C.C. & M.R. Clark 2005. Sustainability issues for orange roughy fisheries. Bulletin of Marine Science 76(2): 337-351

Francis, R.I.C.C. & P.L. Horn 1997. Transition zone in otoliths of orange roughy (Hoplostethus atlanticus) and its relationship to the onset of maturity. Mar. Biol. 129: 681-687.

Francis, R.I.C.C., M.R. Clark and P.J. Grimes 1997. Calculation of the recruited biomass of orange roughy on the northwest Chatham Rise using the 1996 Graveyard Hill egg survey (TAN9608). New Zealand Fisheries Assessment Report Document. 92/29.

Francis, R.I.C.C., D.A. Robertson, M.R. Clark & P.J. Grimes 1992. Assessment of the orange roughy 3B orange roughy fishery for the 1992-1993 fishing year. New Zealand Fisheries Assessment Report Document. 92/4.

Gulland, J.A. 1971. The fish resources of the ocean. Fishing News (Books) Ltd., Surrey, England. 225 p.

Horn, P.L., D.M. Tracey & M.R. Clark 1998. Between-area differences in age and length at first maturity of orange roughy (Hoplostethus atlanticus). Mar. Biol. 132: 187-194.

Pankhurst, N.W. 1988. Spawning dynamics of orange roughy, Hoplostethus atlanticus, in mid-slope waters of New Zealand. Env. Boil. Fish. 21: 101-116.

Sissenwine, M.P. 1988. A quick method of estimating absolute biomass from a time series of relative biomass and catch data. New Zealand Fisheries Assessment Report Document. 88/3.

Smith, A.D.M., A.E. Punt, S.E. Wayte, P.J. Starr, R.I.C.C. Francis, T.K. Stokes, R. Hilborn & A. Langley 2002. Stock assessment of the northeast Chatham Rise orange roughy for 2001. New Zealand Fisheries Assessment Report 2002/25.

Smith, D.C., G.E. Fenton, S.G. Robertson & S.A. Short 1995. Age determination and growth of orange roughy (Hoplostethus atlanticus): a comparison of annulus counts with radiometric aging. Can. J. Fish. Aquat. Sci. 52: 391-401.

Tracey, D.M. & P.L. Horn 1999. Background and review of ageing orange roughy (Hoplostethus atlanticus, Trachichthyidae) from New Zealand and elsewhere. NZ J. Mar. Freshw. Res. 33: 67-86.

Zeldis, J.R., R.I.C.C. Francis, M.R. Clark, J.K.V. Ingerson, P.J. Grimes & M. Vignaux 1997. An estimate of orange roughy, Hoplostethus atlanticus, biomass using the daily fecundity reduction method. Fishery Bulletin. 95: 576-597.