A S S E S S M E N T AND P R E D I C T I O N O F T H E I M P A C T S O F T H E OK T E D I C O P P E R M I N E ON F I S H C A T C H E S IN T H E FLY RIVER S Y S T E M , PAPUA NEW G U I N E A R . E . W . S M I T H and K . G . H O R T L E * Environment Department, Ok Tedi Mining Limited, P.O. Box 1 Tabubil, Western Province, Papua New Guinea
(Received November 1989) Abstract. The Ok Tedi copper mine discharges overburden and ore residues into the Ok Tedi, a tributary of the Fly River. These discharges result in elevated suspended solids and dissolved and particulate associated copper. Analyses of covariance were performed to establish statistical model of the relationships between the mine discharges and fish catches between 1983 and 1988. These models were then extrapolated to predict the affects of future mine discharges on fish catches. The models predicted that if the observed effects were caused by particulate associated copper, the period of greatest impact will be between 1989 and 1993, following which catches should be close to 1988 levels for the remainder of mine life. Some additional catches not included in the data set used to derive the models were found to fit the model predictions well. As the predicted period of greatest impact is short and most species reproduce away from the river channel, the ability of the fish communities to undergo partial recovery after 1991 should be maintained.
I. Introduction
The Ok Tedi copper mine is located on Mt. Fubilan, a foothill of the Star Mountains in the Western Province of Papua New Guinea (Figure 1). Mount Fubilan drains into the headwaters of the Ok Tedi, a tributary of the Fly River ('ok' means river in the language of the local people). The ore body contained gold and copper, with the bulk of the economically recoverable gold in a leached cap over the sulfide copper ore. Construction for the project commenced in 1981, and gold recovery commenced in May 1984. Copper concentrate production began in March 1987, while gold metal production ceased in August 1988. The Fly River system is the largest in Papua New Guinea, and in terms of discharge (mean discharge 6000 m3s-l; Ok Tedi Mining Limited, 1988) it is similar to the Niger and Zambesi Rivers in Africa and the Danube in Europe (source Welcomme, 1985). Being only approximately 1100 km long, with a catchment area of 76000 km 2, in terms of runoff per unit area the Fly River outranks all the worlds major rivers (Holeman, 1968; Welcomme, 1985). This is due to the very high levels of rainfall ranging from in excess of 10000 mm per annum in some parts of the upper catchment to around 3000 mm per year near the coast. The composition of the freshwater fish fauna of the river system is largely determined by its position in the Australasian zoogegraphical zone, although it represents the highest level of diversity for the region. Roberts (1978) listed 105 species from 33 families in the Present adress: KinhiU Engineers Pty, Ltd., 437 St Kilda Road, Melbourne, Victoria 3004, Australia. Environmental Monitoring and Assessment 18:41-68, 1991. 9 1991 Kluwer Academic Publishers. Printed in the Netherlands.
42
R.E.W.
I
SMITH
AND
K. G . H O R T L E
N
I
I iI) 88owo!B
(6>0 sso'"Olfl
oll i
i
-r, I
i
/
I
/
/ I []
I
!I
// ~
d
I
G
/
/
/
/ / /
/
i
// / /o
o
ri O /
.!
n
0
"u
/ I
o~
[] I:1
o
//
i !
0 0
0 tO
0 0
0 u~
0 0
(5~) ~s0,-u0!B
0 tO
o~
-o
o
o
o
o o
o tr)
(O>l)ssowo!8
o o
o
o
~
.....
Illustration of the models for Io{~l biomass caught in tha Fly Rivar at 9
0 j
0'
Di~lPolwd Cu ~llg/L)
st,!
50'
100i
100
D
Fig 8
a
oE 150i
E 150
"~'200
~2oo!
,.. 2.50 ~
~
,_,250
==
300~
300
[]
T..+~,
D
Jl~3
,=;=5
"Paris and symbols a~ for Figure 5
P=r'Liculvte Cu ~ l / ! l )
(d)
Co) 350.
o
DATE
4N~l
(~)
DATE
" IlX)
350
lob
oE 150
"~ 200;
13
300'
300'
,~2.5D.
350'
350'
(=)
>
m
O Z
Z
l;ll '11
,'11 > r O
i-I
56
R.E.W. SMITH AND K. G. HORTLE TABLE IV
ANCOVAR models developed for feeding group catches with significant covariate effects and non-significant interaction effects at the 5% level. Yis the dependent variable, Xis the covariate, N~ is the mean number offish (all types) caught at each site in the year when dissolved copper was close to 5ag/1, mr and s~are the mean and standard deviations for each site of log (total numbers or total biomass + 1) depending on the model, and the other symbols are as described in the text.
Y
X
t
pc
ps
r2
Predators log(100 X N/N,, + 1) log(100 X N / N ~ + 1) (log(N X l)-m~)/s~ log(100 X B/Bmi + 1) log(100 X BIBs, + 1) log(100 X B/Bmi + 1) (log(B + 1)-m~)/s, log(100 X B/B~ + 1) log(100 X B/Bs~ + 1) log(100 X B/Bmi + 1) log(100 X B/B~, + 1) (log(B + 1)-rn,)/s~
P P P P P D D D P D P P
90 180 360 180 180 270 270 360 360 360 360 360
0.249 0.006 0.0001 0.0476 0.0014 0.0001 0.0037 0.0003 0.0004 0.0115 0.0091 0.0288
0.0056 0.0037 0.0001 0.0026 ns 0.0001 0.0001 ns ns 0.0020 0.0013 0.0001
0.280 0.277 0.774 0.317 0.121 0.367 0.399 0.149 0.145 0.340 0.344 0.358
Aquatic Invertebrate Feeders log(100 X B/Bmi + 1) (log(B + l)-m~)/s~ log(100 X B/Bmj + 1)
S S S
90 90 180
0.0014 0.0018 0.0050
0.0001 0.0001 0.0001
0.406 0.479 0.425
Terrestrial Invertebrate Feeders log(100 X N/Nm, + 1) log(100 X N/N~, + 1) log(100 X N/N~ + 1) (log(N + 1)-rn,)/s~ (Iog(N + l)-rni)/si (log(N + l)-mys~ log(100 X B/Bm, + 1) log(100 X B/Bsi + 1) (log(B + l)-mi)/s~
P P P P P P P P P
90 90 270 90 180 270 270 270 270
0.0108 0.0114 0.0142 0.0262 0.0588 0.0281 0.0022 0.0088 0.0253
0.0021 0.0002 0.0001 0.0001 0.0001 0.0001 ns 0.0004 0.0001
0.306 0.384 0.381 0.354 0.341 0.353 0.109 0.411 0.638
Detritivores log(100 X B/B~i + 1) log(100 X B/Bs~ + 1) log(100 X B/B,j + 1)
P P P
360 270 360
0.0168 0.0240 0.0118
0.0010 0.0038 0.0032
0.411 0.383 0.397
3.2. FEEDING GROUPS In contrast with the total catches, the A N C O V A R
analyses for the feeding groups
p r o d u c e d a t o t a l o f 27 m o d e l s w i t h s i g n i f i c a n t m i n e d i s c h a r g e effects a n d n o n - s i g n i f i c a n t i n t e r a c t i o n effects. T h e y a r e listed in T a b l e IV. T h e m o d e l s w i t h t h e l o w e s t p c a n d h i g h e s t r 2 w e r e as follows:
IMPACTS
OF A COPPER
MINE
ON RIVER
57
FISH
Predators (7)
(log(N + 1) - m/s~ = a, - 0.005 • P pc = 0.0001,
ps % 0.0001,
r 2 = 0.77,
rc 2 = 0.05,
t = 360
(8)
log(100 X B/Bmi + 1) = ai - 0.121 X D pc = 0.0001,
ps % 0.0004,
r 2 = 0.41,
rc2 = 0.13,
t=270
Aquatic Invertebrate Feeders log(100 X B/B,~ + 1) = a~ - 0.001 X S pc = 0.0012,
ps % 0.0001,
(9) r 2 = 0.41,
rca = 0.09,
t = 90
log(100 X B/Bm~ + 1) = ai - 0.001 X S** pc=O.O04,
ps 0.05), a n d ** indicates that the O b o sample o f 27 July 1987 was not included in the d a t a set. The residuals were n o r m a l l y distributed ( t indicates p > 0.05, all others p > 0.15). All o t h e r s y m b o l s are as indicated previously. The values o f Nm~, Bs~, B,,~, m , a n d sg are listed in Table II a n d the values o f a, are given in Table III. N o t e t h a t for the aquatic invertebrate consumers, it was suspended solids rather than dissolved o r particulate copper that was found to have a significant effect on the biomass
58
R.E.W.
SMITH
AND
K.G. HORTLE
caught. Increased suspended solids levels have been found to adversely affect aquatic invertebrates, particularly benthic forms (e.g. Cordone and Kelly, 1961; Alabaster and Lloyd, 1982) and the macroinvertebrate fauna in the upper Ok Tedi were found to be impacted by increased suspended solids during the construction phase (Ok Tedi Mining Limited, 1983). Note also that models were found for the numbers of predators and terrestrial invertebrate feeders caught. In general the models are similar in form to those for total catches as would be expected. The models are illustrated for Kuambit in Figures 9 to 12. Where the Obo sample of 27 July 1987 was an extreme outlier, the model excluding that sample (marked ** above) was used in the tables and the figures. The predictions of future catches were calculated on the basis of the levels of suspended solids, particulate and dissolved copper predicted by the hydrological and geochemical models using the current mine plan and assuming no waste retention. The levels of mine discharges derived were the predicted mean values for each year, and the predicted catches are plotted at the end of each calendar year in the figures. Some additional catches not used in the generation of the models are plotted in the figures. The models developed for the catches of predators (Figure 9) were based on dissolved copper for biomass and particulate copper for numbers. As the relationship between these parameters is expected to change, the two models predict quite different responses during the remainder of mine life. The model for numbers predicts a loss of predatory species from Atkamba in 1989 and 1990. The fit of additional catches is not good, largely as a result of an unusually large influx of small (400 to 500 mm for length) barramundi, Lates calcarifer, in 1989 (data not presented). As the suspended solids levels in the River System will be relatively constant compared with copper levels over the mine life, the predicted catches for aquatic invertebrate feeders (Figure 10) were less variable than for the other groups. For that group the models predict only slight reductions at the Fly River sites. The additional catches at Kuambit are close to that predicted by the model for the ambient levels of suspended solids. In contrast, the models of best fit for the catch of terrestrial invertebrate feeders (Figure 1l) were for particulate copper rather than for suspended solids. The patterns of change of the future predictions are similar to those discussed for the total catches. The additional catches were generally close to the model predictions for both numbers and biomass caught. It should be noted that some fish species in the Fly River System can switch their food preferences between aquatic and terrestrial invertebrates (Ok Tedi Mining Limited, 1988). The detritivores appear to have been more adversely affected by mine discharges than the other feeding groups, and the models predict losses of the group from Atkamba and Kuambit (Figure 12) from 1989 to 1993, and from Obo in 1989 and 1990. The recent catchers for this group fit the model predictions reasonably well despite the patchiness of past catches of detritivores due to the schooling habit of the dominant species (herrings, Nematalosa papuensis and N. flyensis, and mullet, Crenimugil labiosus and Liza diadema). Both herring and mullet mainly inhabit off-river habitats, and clear water tributaries for C. labiosus, (Hortle, 1986a and 1986b) so the actual impacts to the stocks of these species may be more dependent on conditions in the off-river areas than within the river channel.
0
',,
o
200
o
JANB7
Oo
dAN97
JN4B3
JANB~
o
dN187
o
JNt89
[b
600
800
1000
1200
1400
,3
o
--..:"'"
O
0
9
12
12
15
[]
dAN95
" ...........
dN4113
Dissolved Cu (I~3/L)
6
1 oo -'~'&~'~'~ 0~, .....
1
751
1001
o
,L~N91
(d)
JNt9.~
o
(=)
JAN93
s0
o
o
DATE
JANll I
a
7,5
I00
DATE
dnt89
o
Porticulate Cu (J~:J/g)
400
\",.
o ~
,
dANS~
oo
(b)
O
J,4ND7
18
Fig. 9. Illustration of the models for numbers (model 7) and biomass (model 8) of predatory fish caught at Kuambit. (a) and (c) Model 7 for numbers caught with particulate copper as the covariate. (b) and (d) Model 8 for biomass caught with dissolved copper as the covariate. Symbols as described for Figure 5. The values plotted from December 1989 to 1996 are model estimates based on predicted levels of dissolved and particulate copper.
;
.,.. 75
1ooI
dN183
50'
100'
(,*)
'r
< ~a
O Z
~r
m
0
>
0
> (3 H
60
R.E.W.
SMITH
AND K.G. HORTLE
(=) 150"
'~1o0 m m o
E 0 b3
50
0
0
Oi
o ~
,W~7
a j.qCm
d~l
JA~3
,W~B
,~1r
DATE
(b) 150
~100 m m o
E m
50,
0
0
--=-.'~--~-a._.~___
----'_---
O'
D
o
16o
260
s6o
,.~o
560
660
Suspended 8olida (mg/L.) Fig. 10, Illustration of model 10 for the biomass of aquatic invertebrate feeding fish caught at Kuambit. (a) plotted against time and (b)plotted against the level of suspended solids. Symbols as described for Figure 5. The values plotted from 1989 to 1996 are model estimates based on predicted levels of suspended solids.
4. Discussion Statistical linear models are descriptive by nature, describing relationships that were f o u n d between variables in a d a t a set. Significance testing o f the models does not test causality in the relationships, but simply indicates the likelihood o f such relationships occuring by chance alone. Therefore, using statistical models predictively must be done with caution. Such predictions assume that the observed relationships are indeed causal a n d that no
2si
~
200
800
D
1000
1200
. . . .
1400
,W~7
75
0
o~
~,~50
0
200
o
~
o
JAN85
...... ~89
,.L~91
,
,W,I03
~o -". . . . . . 400 600 800 1000 ParticulateCu O.~g/g)
0-.~..__
,.~t87
(d)
,.L~5
(=)
JN~3
o
DATE
JANgl
o
o
o
DATE
600
0
~
Particulate Cu (~u~/g)
4.00
o
,t~a7
-'~...~'~0"-.r
~
25
~3
=
~50'
v
75'
[b)
,
~
1200
d,~i95
!i!tIttt
.
1400
~A~7
Fig. 11. Illustration of models for the numbers (model 12) and biomass (model 14) of terrestrial invertebrate feeding fish caught at Kuambit as functions of the levels of particulate copper. (a) and (c) Model 12 for numbers caught. (b) and (d) Model 14 for biomass caught. Symbols as described for Figure 5. The values plotted for December 1989 to 1996 are model estimates based on predicted levels of particulate copper.
25
50
75
100
~
z
E
~, 50,
75
100
(=)
O'x
62
R.E.W.
SMITH
AND
K.
G.
HORTLE
(a) 1o
~-~6'
0
(3
oo
E
r~
2
JNt83
JAH85
,I~NB7
JAN89
JN~91
JN~3
JN~I95
.J,~97
1200
1400
DATE
(b) 10'
6'
o
0
E
0
,
o
9
---
0
200
0
0
400
~--- .....
600
D
800
1000
Porticulota Cu ~ g / 9 )
Fig. 12. Illustration of model 15 for the biomass of detritivorous fish caught at Kuambit. (a) plotted against time and (b) plotted against particulate copper9 Symbols as described for Figure 5. The values plotted from December 1989 to 1996are model estimates based on predicted levels of particulate copper. unaccounted factors that may be responsible for the unexplained variation become of overriding importance. The residual variation for all the A N C O V A R models presented in this report did not differ significantly from normality, which indicates that the unexplained variation was due to r a n d o m error, or that any causal factors that were not included in the models were themselves normally distributed for the data set used. Note that a non-causal independent variable which is correlated with the true causal variable would be likely to also demonstrate a significant model, the level o f significance being dependent on the level of correlation. Such a process is likely to have occurred with dissolved and particulate copper in the models presented here as bioassay results have
IMPACTS
OF A COPPER
MINE ON RIVER
FISH
63
indicated that particulate copper from the mine discharges is much less toxic than dissolved copper (48 hr LCso values for Ceriodaphnia dubia were greater than 10000 #g/g for particulate copper and 16 ~tg/1for dissolved copper, Smith et al., 1989). The reason for this is that very little of the particulate copper is available for uptake, with between 7 and 15% of the total concentration in potentially extractable forms (Ok Tedi Mining Limited, 1988). The fish community in the Fly River system has been shown to be very tolerant of high levels of uncontaminated suspended solids (Smith et al., 1989), but the settlement of solids can greatly affect benthic invertebrate populations (Cordone and Kelly, 1961). However, the abundance of invertebrates is very low in the sediments of the Fly River above and below the Ok Tedi junction and in the Strickland River (Hortle, 1987). This presumably results from the instability of the sediments which are transported as bedload by the high current speeds (Ok Tedi Mining Limited, 1988). However, much of the invertebrate biomass in the river channel is composed of isolated aggregations of the filter feeding burrowing mayfly Plethogenesia pallida, which occurs in hard clay banks on erosional bends of the river (Hortle, 1987). This mayfly is an important component of the diet of some aquatic invertebrate feeders (B. Kare, pers. comm.) and a suspended solids model was derived for that group. Recent studies have shown that for species from all feeding groups except predators, specimens collected from the river channel near Bosset had obtained most of their food from offriver and/or terrestrial sources (unpublished data), but few mayfly aggregations exist in that region. Prediction of future events within the observed range of values of the independent variables can be performed with a level of accuracy defined by the confidence levels for the model, given that the system remains essentially the same and unaccounted factors do not change in importance. Prediction of the effects of independent variables outside the observed ranges must assume that the observed relationships hold for the extrapolations which is untestable without further sampling. Therefore, although confidence levels can be computed, they will greatly under-predict the actual level of uncertainty. In fact the assumption of continued linearity of response is generally invalid in biological applications. Even for single species, response patterns are typically unimodal but may be bimodal (Greig-Smith, 1980) over the full range of an environmental effect. For multispecies systems such as that considered here the situation is generally much more complex. Some samples from outside the ranges used to develop the models were obtained in this study, and in general their compliance with the model predictions were reasonable, but the number of additional samples was only four to six and the ranges of the physical variables covered were less than those predicted over the mine life. Therefore, these additional samples do not give a good indication of the level of natural variability nor the pattern of response at the extreme levels predicted for some sites. The model predictions based on predicted levels of mine discharges indicate that in the worst case (i.e. when the covariate is particulate copper) the period of greatest impact will be between 1989 and 1993, following which catches should be slightly less than 1988 levels until the end of mine life, provided the trtsh communities and their food sources can recover from the reductions. The fact that the models generated generally incorporated dissolved or particulate copper rather than suspended solids is of little importance given
64
R. E. W . S M I T H
AND
K . (3. H O R T L E
that the three variables were correlated within the data set (Spearman's rank correlation,p < 0.05 for yearly means). However, it is expected that the relationships between the three variables will change in future as the predominance of sulphide ores increases (Ok Tedi Mining Limited, 1988, 1989). This is reflected in the different predictions of future catches by the dissolved and particulate copper models. The predictions of recovery from periods of reduced fish stocks assume that the necessary recruits will be available. The sources of recruits for many species in the Fly River system are away from the main river channel in tributary streams, in marine locations and in off-river water bodies such as oxbow lagoons, the floodplain and blocked valley lakes (Hortle, 1986b). The mine discharges will not directly affect the tributary streams, and the marine habitats outside the Fly estuary and delta will be impacted slightly if at all. However, at least some of the off-river and estuarine spawning sites will receive elevated copper levels (Ok Tedi Mining Limited, 1989). Limited fish sampling in off-river water bodies has not demonstrated any impacts of mining as yet (Ok Tedi Mining Limited, 1989) and only minor increases in copper concentrations have been detected in the estuary and delta (Ok Tedi Mining Limited, 1988, 1989). As the period of most severe impacts is short, the impacts on the fish stocks in offriver refuges appears to be occurring at a slower rate than in the river channel and fish within the river channel obtain most of their food from offriver and terrestrial sources, the ability to increase stocks after 1991 should be maintained for most species. The maintenance of communities similar to those found in 1988 for the bulk of mine life would require the maintenance of these offriver refuges, which has not been adequately investigated. It would be expected that if maintained, 1988 stock levels could adequately support the current levels of exploitation (6.2 to 31.2% of the potential yield, Ok Tedi Mining Limited, 1988), and would have the potential to recover to pre-mine levels after mine shut down. The modelling approach used in this study has quantified responses of the fish communities downstream of the mine to elevated concentrations of copper and suspended solids. The derived relationships were then be used with caution to predict future effects. Although the demonstration of such relationships does not prove causality, with continued monitoring, the changing nature of the mine discharges will provide information as to which of the chemical parameters tested was or were chiefly responsible for the changes in the abundances of fish. This study has illustrated that statistical models are useful monitoring tools when the complexities of the problems of concern are such that the development of quantitative conceptual models is not practical.
5. Acknowledgements The authors would like to thank Dr. John Alabaster for suggesting the modelling approach, and Dr. Glen De'ath of James Cook University of North Queensland for advice on the statistical proceedures employed in this report and particularly for suggesting the use of ANCOVAR. This research would not have been possible without the financial and logistical support of Ok Tedi Mining Limited, in particular the Hydrology and Chemistry sections who allowed us to utilise their data, and the technical assistance of
IMPACTS OF A COPPER MINE ON RIVER FISH
65
the staff of the Biology Section - Andy Maie, Yarang Kurtama, Pius Fred, Augustine Mungkaje and Barre Kare. The contribution of David Balloch in setting up the biological sampling programme is also acknowledged. 6. References Cordone, A. J., and Kelly, D. W., 1961: 'The Influences of Inorganic Sediment on the Aquatic Life of Streams', California Fish and Game 47, 189-228. Alabaster, J. J. and Lloyd, R., 1982: Water Quality Criteria for Freshwater Fish', Butterworth Scientific, London, U.K., 361 pp. Boyden, C. R., Brown, B. E., Lamb, K. P., Drucker, R. F., and Tuft, S. J., 1978: 'Trace elements in the Upper Fly River, Papua New Guinea', Fresh Water Biology 8, 189-205. Greig-Smith, P., 1980: 'The Development of Numerical Classification and Ordination', Vegetatio 42, 1-9. Holeman, J. N., 1968: 'The Sediment Yield of Major Rivers of the World', Water Resources Research 4, 737-747. Hortle, K. G., 1986a: 'Survey of the Fish Fauna of the Strickland River at Tiumsinawam with Reference to Sediment Tolerance.' Ok Tedi Mining Limited Report, ENV 86--6, 30pp. Hortle, K. G., 1986b: 'A Review of the Results of Biological Sampling of the Ok Tedi and Fly River Systems, April 1983 to June 19867 Ok Tedi Mining Limited Report, ENV 86--9, 202pp. Hortle, K. G., 1987: 'Studies of the Benthic Fauna of Lowland (Potamon) Localities of the Ok Tedi and Fly River, with Reference to Mining Impacts.' Ok Tedi Mining Limited Report, ENV 87-11,45pp. Maunsell and Partners Pty. Ltd., 1982: 'Ok Tedi Environmental Study. Working Paper 1. Water Quality of the Ok Tedi and Fly River System.' Ok Tedi Mining Limited, 135pp. Ok Tedi Mining Limited, 1983: 'Environmental Monitoring Programme 6 Monthly Report. June - December 1983.' Ok Tedi Mining Limited Report, PA/03/84-8, 71pp. Ok Tedi Mining Limited, 1988: 'Sixth Supplemental Agreement Environmental Study 1986-1988. Final Draft Report.' Ok Tedi Mining Limited, 702pp. Ok Tedi Mining Limited, 1989: 'Supplementary Environmental Investigations. Volume One.' Ok Tedi Mining Limited, 74pp. Roberts, T. R., 1978: 'An Icthyological Survey of the Fly River in Papua New Guinea, with Descriptions of New Species', Smithonian Contributions to Zoology 281, 1-72. SAS Institute Inc., 1985: SAS 9 User's Guide: Statistics, Version 5 Edition, SAS Institute Inc., Cary, USA, 956pp. Smith, R. E. W., 1988: 'Six-monthly Biology Review, 11 October 1987 to 30 March 1988.' Ok Tedi Mining Limited Report, ENV 88-05, 74pp. Smith, R. E. W., Ahsanullah, M., and Batley, G. E., 1989: 'Investigations of the Impact of Effluent from the Ok Tedi Copper Mine on the Fisheries Resource in the Fly River, Papua New Guinea', Environmental Monitoring and Assessment 14, 315-331. Welcomme, R. L., 1985. River Fisheries. FAO Fisheries Technical Paper 262, 330pp.
66
R . E . W . SMITH AND K. G. HORTLE
APPENDIX A. Gill nets used in fish sampling "Mono' indicates net constructed from nylon monofilament, 'multi' indicates nylon multifilament. Mesh size (mm)
Type
Length (m)
Depth (m)
Area (m2)
Number used
25 38 50 63 75 88 100 125 150 150 175
mono mono mono mono mono mono mono mono mono multi multi
40 40 45 40 45 45 45 45 50 25 25
2.3 1.7 2.1 2.8 3.2 3.5 4.2 4.9 6.0 2.8 3.1
92 68 95 112 144 158 189 221 300 70 78
1 1 l l 1 1 1 1 1 2 2
IMPACTS OF A COPPER MINE ON RIVER FISH APPENDIX B List of fish species assigned to each feeding group. Feeding Group
Species list
Family
Predators
Pristis microdon Megalops cyprinoides Thryssa scratchleyi Strongylura kreffti Lates calcarifer Glossomia aprion Lutjanus goldei L. argentimaculatus Oxyeleotris fimbriata O. nullipora (9. lineolatus Arius augustus
Pristidae Megalopidae Engraulidae Belonidae Centropomidae Apogonidae Lutjanidae
Clupeoides venulosus C. papuensis Thryssa rastrosa Arius leptaspis A. acrocephalus A. carinatus Hemipimelodus crassilabris H. macrorhynchus Cinetodus froggatti Nedystoma dayi Cochlefelis spatula C. danielsi Neosilurus ater N. equinus N. meraukensis N. brevidorsalis Neosilurus sp. Oloplotosus luteus Porochilus obbesi P. meraukensis Plotosus papuensis Craterocephalus randi C. nouhuysi Ambassis reticulata Parambassis gulliveri Denariusa bandata Damioides quadrifasciatus
Clupeidae
Aquatic invertebrate feeders
Aquatic invertebrate feeders
Hephaestes roemeri 14. trimaculatus Terapon jamoerensis Amniataba percoides Glossamia trifasciata Acanthopragus berda Nibea semifasciata Glossogobius giurus G. celebius
Eleotridae
Ariidae
Engraulidae Ariidae
Plotosidae
Atherinidae Ambassidae
Lobotidae Theraponidae
Apogonidae Sparidae Sciaenidae Gobiidae
67
68
R. E. W. SMITH
AND
K. G. HORTLE
APPENDIX B (continued) List of fish species assigned to each feeding group. Feeding Group
Terrestrial invertebrate feeders
Herbivores
Detritivores
Species list
Family
G. concQvifrons Mogurnda mogurnda
Eleotridae
Ophieleotris aporos Kurtus gulliveri Cynoglossus heterolepis Anabas testinudineus
Kurtidae Cynoglossidae Anabantidae
Scleropages jardini Arius berneyi Hemipimelodus taylori H. macrorhynchus Melanotaenia splendida M, maccullochi M. oktediensis M. sexlineata Pseudomugil novaeguinea P. gertrudae Iriatherina werneri Zenarchoptenus novaeguinea Toxotes chatareus T. lorentzi
Osteoglossidae Ariidae Melanotaeniidae
Hemirhamphidae Toxotidae
Terapon jamoerensis Pingalla lorentzi Oxyurichthys papuensis Anabas testudineus
Theraponidae
Nematalosa papuensis N. flyensis Crenimugil labiosus Liza diadema Cynoglossus heterolepis
Clupeidae
Gobiidae Anabantidae
Mugilidae Cynoglossidae
(Received November 1989) Abstract. The Ok Tedi copper mine discharges overburden and ore residues into the Ok Tedi, a tributary of the Fly River. These discharges result in elevated suspended solids and dissolved and particulate associated copper. Analyses of covariance were performed to establish statistical model of the relationships between the mine discharges and fish catches between 1983 and 1988. These models were then extrapolated to predict the affects of future mine discharges on fish catches. The models predicted that if the observed effects were caused by particulate associated copper, the period of greatest impact will be between 1989 and 1993, following which catches should be close to 1988 levels for the remainder of mine life. Some additional catches not included in the data set used to derive the models were found to fit the model predictions well. As the predicted period of greatest impact is short and most species reproduce away from the river channel, the ability of the fish communities to undergo partial recovery after 1991 should be maintained.
I. Introduction
The Ok Tedi copper mine is located on Mt. Fubilan, a foothill of the Star Mountains in the Western Province of Papua New Guinea (Figure 1). Mount Fubilan drains into the headwaters of the Ok Tedi, a tributary of the Fly River ('ok' means river in the language of the local people). The ore body contained gold and copper, with the bulk of the economically recoverable gold in a leached cap over the sulfide copper ore. Construction for the project commenced in 1981, and gold recovery commenced in May 1984. Copper concentrate production began in March 1987, while gold metal production ceased in August 1988. The Fly River system is the largest in Papua New Guinea, and in terms of discharge (mean discharge 6000 m3s-l; Ok Tedi Mining Limited, 1988) it is similar to the Niger and Zambesi Rivers in Africa and the Danube in Europe (source Welcomme, 1985). Being only approximately 1100 km long, with a catchment area of 76000 km 2, in terms of runoff per unit area the Fly River outranks all the worlds major rivers (Holeman, 1968; Welcomme, 1985). This is due to the very high levels of rainfall ranging from in excess of 10000 mm per annum in some parts of the upper catchment to around 3000 mm per year near the coast. The composition of the freshwater fish fauna of the river system is largely determined by its position in the Australasian zoogegraphical zone, although it represents the highest level of diversity for the region. Roberts (1978) listed 105 species from 33 families in the Present adress: KinhiU Engineers Pty, Ltd., 437 St Kilda Road, Melbourne, Victoria 3004, Australia. Environmental Monitoring and Assessment 18:41-68, 1991. 9 1991 Kluwer Academic Publishers. Printed in the Netherlands.
42
R.E.W.
I
SMITH
AND
K. G . H O R T L E
N
I
I iI) 88owo!B
(6>0 sso'"Olfl
oll i
i
-r, I
i
/
I
/
/ I []
I
!I
// ~
d
I
G
/
/
/
/ / /
/
i
// / /o
o
ri O /
.!
n
0
"u
/ I
o~
[] I:1
o
//
i !
0 0
0 tO
0 0
0 u~
0 0
(5~) ~s0,-u0!B
0 tO
o~
-o
o
o
o
o o
o tr)
(O>l)ssowo!8
o o
o
o
~
.....
Illustration of the models for Io{~l biomass caught in tha Fly Rivar at 9
0 j
0'
Di~lPolwd Cu ~llg/L)
st,!
50'
100i
100
D
Fig 8
a
oE 150i
E 150
"~'200
~2oo!
,.. 2.50 ~
~
,_,250
==
300~
300
[]
T..+~,
D
Jl~3
,=;=5
"Paris and symbols a~ for Figure 5
P=r'Liculvte Cu ~ l / ! l )
(d)
Co) 350.
o
DATE
4N~l
(~)
DATE
" IlX)
350
lob
oE 150
"~ 200;
13
300'
300'
,~2.5D.
350'
350'
(=)
>
m
O Z
Z
l;ll '11
,'11 > r O
i-I
56
R.E.W. SMITH AND K. G. HORTLE TABLE IV
ANCOVAR models developed for feeding group catches with significant covariate effects and non-significant interaction effects at the 5% level. Yis the dependent variable, Xis the covariate, N~ is the mean number offish (all types) caught at each site in the year when dissolved copper was close to 5ag/1, mr and s~are the mean and standard deviations for each site of log (total numbers or total biomass + 1) depending on the model, and the other symbols are as described in the text.
Y
X
t
pc
ps
r2
Predators log(100 X N/N,, + 1) log(100 X N / N ~ + 1) (log(N X l)-m~)/s~ log(100 X B/Bmi + 1) log(100 X BIBs, + 1) log(100 X B/Bmi + 1) (log(B + 1)-m~)/s, log(100 X B/B~ + 1) log(100 X B/Bs~ + 1) log(100 X B/Bmi + 1) log(100 X B/B~, + 1) (log(B + 1)-rn,)/s~
P P P P P D D D P D P P
90 180 360 180 180 270 270 360 360 360 360 360
0.249 0.006 0.0001 0.0476 0.0014 0.0001 0.0037 0.0003 0.0004 0.0115 0.0091 0.0288
0.0056 0.0037 0.0001 0.0026 ns 0.0001 0.0001 ns ns 0.0020 0.0013 0.0001
0.280 0.277 0.774 0.317 0.121 0.367 0.399 0.149 0.145 0.340 0.344 0.358
Aquatic Invertebrate Feeders log(100 X B/Bmi + 1) (log(B + l)-m~)/s~ log(100 X B/Bmj + 1)
S S S
90 90 180
0.0014 0.0018 0.0050
0.0001 0.0001 0.0001
0.406 0.479 0.425
Terrestrial Invertebrate Feeders log(100 X N/Nm, + 1) log(100 X N/N~, + 1) log(100 X N/N~ + 1) (log(N + 1)-rn,)/s~ (Iog(N + l)-rni)/si (log(N + l)-mys~ log(100 X B/Bm, + 1) log(100 X B/Bsi + 1) (log(B + l)-mi)/s~
P P P P P P P P P
90 90 270 90 180 270 270 270 270
0.0108 0.0114 0.0142 0.0262 0.0588 0.0281 0.0022 0.0088 0.0253
0.0021 0.0002 0.0001 0.0001 0.0001 0.0001 ns 0.0004 0.0001
0.306 0.384 0.381 0.354 0.341 0.353 0.109 0.411 0.638
Detritivores log(100 X B/B~i + 1) log(100 X B/Bs~ + 1) log(100 X B/B,j + 1)
P P P
360 270 360
0.0168 0.0240 0.0118
0.0010 0.0038 0.0032
0.411 0.383 0.397
3.2. FEEDING GROUPS In contrast with the total catches, the A N C O V A R
analyses for the feeding groups
p r o d u c e d a t o t a l o f 27 m o d e l s w i t h s i g n i f i c a n t m i n e d i s c h a r g e effects a n d n o n - s i g n i f i c a n t i n t e r a c t i o n effects. T h e y a r e listed in T a b l e IV. T h e m o d e l s w i t h t h e l o w e s t p c a n d h i g h e s t r 2 w e r e as follows:
IMPACTS
OF A COPPER
MINE
ON RIVER
57
FISH
Predators (7)
(log(N + 1) - m/s~ = a, - 0.005 • P pc = 0.0001,
ps % 0.0001,
r 2 = 0.77,
rc 2 = 0.05,
t = 360
(8)
log(100 X B/Bmi + 1) = ai - 0.121 X D pc = 0.0001,
ps % 0.0004,
r 2 = 0.41,
rc2 = 0.13,
t=270
Aquatic Invertebrate Feeders log(100 X B/B,~ + 1) = a~ - 0.001 X S pc = 0.0012,
ps % 0.0001,
(9) r 2 = 0.41,
rca = 0.09,
t = 90
log(100 X B/Bm~ + 1) = ai - 0.001 X S** pc=O.O04,
ps 0.05), a n d ** indicates that the O b o sample o f 27 July 1987 was not included in the d a t a set. The residuals were n o r m a l l y distributed ( t indicates p > 0.05, all others p > 0.15). All o t h e r s y m b o l s are as indicated previously. The values o f Nm~, Bs~, B,,~, m , a n d sg are listed in Table II a n d the values o f a, are given in Table III. N o t e t h a t for the aquatic invertebrate consumers, it was suspended solids rather than dissolved o r particulate copper that was found to have a significant effect on the biomass
58
R.E.W.
SMITH
AND
K.G. HORTLE
caught. Increased suspended solids levels have been found to adversely affect aquatic invertebrates, particularly benthic forms (e.g. Cordone and Kelly, 1961; Alabaster and Lloyd, 1982) and the macroinvertebrate fauna in the upper Ok Tedi were found to be impacted by increased suspended solids during the construction phase (Ok Tedi Mining Limited, 1983). Note also that models were found for the numbers of predators and terrestrial invertebrate feeders caught. In general the models are similar in form to those for total catches as would be expected. The models are illustrated for Kuambit in Figures 9 to 12. Where the Obo sample of 27 July 1987 was an extreme outlier, the model excluding that sample (marked ** above) was used in the tables and the figures. The predictions of future catches were calculated on the basis of the levels of suspended solids, particulate and dissolved copper predicted by the hydrological and geochemical models using the current mine plan and assuming no waste retention. The levels of mine discharges derived were the predicted mean values for each year, and the predicted catches are plotted at the end of each calendar year in the figures. Some additional catches not used in the generation of the models are plotted in the figures. The models developed for the catches of predators (Figure 9) were based on dissolved copper for biomass and particulate copper for numbers. As the relationship between these parameters is expected to change, the two models predict quite different responses during the remainder of mine life. The model for numbers predicts a loss of predatory species from Atkamba in 1989 and 1990. The fit of additional catches is not good, largely as a result of an unusually large influx of small (400 to 500 mm for length) barramundi, Lates calcarifer, in 1989 (data not presented). As the suspended solids levels in the River System will be relatively constant compared with copper levels over the mine life, the predicted catches for aquatic invertebrate feeders (Figure 10) were less variable than for the other groups. For that group the models predict only slight reductions at the Fly River sites. The additional catches at Kuambit are close to that predicted by the model for the ambient levels of suspended solids. In contrast, the models of best fit for the catch of terrestrial invertebrate feeders (Figure 1l) were for particulate copper rather than for suspended solids. The patterns of change of the future predictions are similar to those discussed for the total catches. The additional catches were generally close to the model predictions for both numbers and biomass caught. It should be noted that some fish species in the Fly River System can switch their food preferences between aquatic and terrestrial invertebrates (Ok Tedi Mining Limited, 1988). The detritivores appear to have been more adversely affected by mine discharges than the other feeding groups, and the models predict losses of the group from Atkamba and Kuambit (Figure 12) from 1989 to 1993, and from Obo in 1989 and 1990. The recent catchers for this group fit the model predictions reasonably well despite the patchiness of past catches of detritivores due to the schooling habit of the dominant species (herrings, Nematalosa papuensis and N. flyensis, and mullet, Crenimugil labiosus and Liza diadema). Both herring and mullet mainly inhabit off-river habitats, and clear water tributaries for C. labiosus, (Hortle, 1986a and 1986b) so the actual impacts to the stocks of these species may be more dependent on conditions in the off-river areas than within the river channel.
0
',,
o
200
o
JANB7
Oo
dAN97
JN4B3
JANB~
o
dN187
o
JNt89
[b
600
800
1000
1200
1400
,3
o
--..:"'"
O
0
9
12
12
15
[]
dAN95
" ...........
dN4113
Dissolved Cu (I~3/L)
6
1 oo -'~'&~'~'~ 0~, .....
1
751
1001
o
,L~N91
(d)
JNt9.~
o
(=)
JAN93
s0
o
o
DATE
JANll I
a
7,5
I00
DATE
dnt89
o
Porticulate Cu (J~:J/g)
400
\",.
o ~
,
dANS~
oo
(b)
O
J,4ND7
18
Fig. 9. Illustration of the models for numbers (model 7) and biomass (model 8) of predatory fish caught at Kuambit. (a) and (c) Model 7 for numbers caught with particulate copper as the covariate. (b) and (d) Model 8 for biomass caught with dissolved copper as the covariate. Symbols as described for Figure 5. The values plotted from December 1989 to 1996 are model estimates based on predicted levels of dissolved and particulate copper.
;
.,.. 75
1ooI
dN183
50'
100'
(,*)
'r
< ~a
O Z
~r
m
0
>
0
> (3 H
60
R.E.W.
SMITH
AND K.G. HORTLE
(=) 150"
'~1o0 m m o
E 0 b3
50
0
0
Oi
o ~
,W~7
a j.qCm
d~l
JA~3
,W~B
,~1r
DATE
(b) 150
~100 m m o
E m
50,
0
0
--=-.'~--~-a._.~___
----'_---
O'
D
o
16o
260
s6o
,.~o
560
660
Suspended 8olida (mg/L.) Fig. 10, Illustration of model 10 for the biomass of aquatic invertebrate feeding fish caught at Kuambit. (a) plotted against time and (b)plotted against the level of suspended solids. Symbols as described for Figure 5. The values plotted from 1989 to 1996 are model estimates based on predicted levels of suspended solids.
4. Discussion Statistical linear models are descriptive by nature, describing relationships that were f o u n d between variables in a d a t a set. Significance testing o f the models does not test causality in the relationships, but simply indicates the likelihood o f such relationships occuring by chance alone. Therefore, using statistical models predictively must be done with caution. Such predictions assume that the observed relationships are indeed causal a n d that no
2si
~
200
800
D
1000
1200
. . . .
1400
,W~7
75
0
o~
~,~50
0
200
o
~
o
JAN85
...... ~89
,.L~91
,
,W,I03
~o -". . . . . . 400 600 800 1000 ParticulateCu O.~g/g)
0-.~..__
,.~t87
(d)
,.L~5
(=)
JN~3
o
DATE
JANgl
o
o
o
DATE
600
0
~
Particulate Cu (~u~/g)
4.00
o
,t~a7
-'~...~'~0"-.r
~
25
~3
=
~50'
v
75'
[b)
,
~
1200
d,~i95
!i!tIttt
.
1400
~A~7
Fig. 11. Illustration of models for the numbers (model 12) and biomass (model 14) of terrestrial invertebrate feeding fish caught at Kuambit as functions of the levels of particulate copper. (a) and (c) Model 12 for numbers caught. (b) and (d) Model 14 for biomass caught. Symbols as described for Figure 5. The values plotted for December 1989 to 1996 are model estimates based on predicted levels of particulate copper.
25
50
75
100
~
z
E
~, 50,
75
100
(=)
O'x
62
R.E.W.
SMITH
AND
K.
G.
HORTLE
(a) 1o
~-~6'
0
(3
oo
E
r~
2
JNt83
JAH85
,I~NB7
JAN89
JN~91
JN~3
JN~I95
.J,~97
1200
1400
DATE
(b) 10'
6'
o
0
E
0
,
o
9
---
0
200
0
0
400
~--- .....
600
D
800
1000
Porticulota Cu ~ g / 9 )
Fig. 12. Illustration of model 15 for the biomass of detritivorous fish caught at Kuambit. (a) plotted against time and (b) plotted against particulate copper9 Symbols as described for Figure 5. The values plotted from December 1989 to 1996are model estimates based on predicted levels of particulate copper. unaccounted factors that may be responsible for the unexplained variation become of overriding importance. The residual variation for all the A N C O V A R models presented in this report did not differ significantly from normality, which indicates that the unexplained variation was due to r a n d o m error, or that any causal factors that were not included in the models were themselves normally distributed for the data set used. Note that a non-causal independent variable which is correlated with the true causal variable would be likely to also demonstrate a significant model, the level o f significance being dependent on the level of correlation. Such a process is likely to have occurred with dissolved and particulate copper in the models presented here as bioassay results have
IMPACTS
OF A COPPER
MINE ON RIVER
FISH
63
indicated that particulate copper from the mine discharges is much less toxic than dissolved copper (48 hr LCso values for Ceriodaphnia dubia were greater than 10000 #g/g for particulate copper and 16 ~tg/1for dissolved copper, Smith et al., 1989). The reason for this is that very little of the particulate copper is available for uptake, with between 7 and 15% of the total concentration in potentially extractable forms (Ok Tedi Mining Limited, 1988). The fish community in the Fly River system has been shown to be very tolerant of high levels of uncontaminated suspended solids (Smith et al., 1989), but the settlement of solids can greatly affect benthic invertebrate populations (Cordone and Kelly, 1961). However, the abundance of invertebrates is very low in the sediments of the Fly River above and below the Ok Tedi junction and in the Strickland River (Hortle, 1987). This presumably results from the instability of the sediments which are transported as bedload by the high current speeds (Ok Tedi Mining Limited, 1988). However, much of the invertebrate biomass in the river channel is composed of isolated aggregations of the filter feeding burrowing mayfly Plethogenesia pallida, which occurs in hard clay banks on erosional bends of the river (Hortle, 1987). This mayfly is an important component of the diet of some aquatic invertebrate feeders (B. Kare, pers. comm.) and a suspended solids model was derived for that group. Recent studies have shown that for species from all feeding groups except predators, specimens collected from the river channel near Bosset had obtained most of their food from offriver and/or terrestrial sources (unpublished data), but few mayfly aggregations exist in that region. Prediction of future events within the observed range of values of the independent variables can be performed with a level of accuracy defined by the confidence levels for the model, given that the system remains essentially the same and unaccounted factors do not change in importance. Prediction of the effects of independent variables outside the observed ranges must assume that the observed relationships hold for the extrapolations which is untestable without further sampling. Therefore, although confidence levels can be computed, they will greatly under-predict the actual level of uncertainty. In fact the assumption of continued linearity of response is generally invalid in biological applications. Even for single species, response patterns are typically unimodal but may be bimodal (Greig-Smith, 1980) over the full range of an environmental effect. For multispecies systems such as that considered here the situation is generally much more complex. Some samples from outside the ranges used to develop the models were obtained in this study, and in general their compliance with the model predictions were reasonable, but the number of additional samples was only four to six and the ranges of the physical variables covered were less than those predicted over the mine life. Therefore, these additional samples do not give a good indication of the level of natural variability nor the pattern of response at the extreme levels predicted for some sites. The model predictions based on predicted levels of mine discharges indicate that in the worst case (i.e. when the covariate is particulate copper) the period of greatest impact will be between 1989 and 1993, following which catches should be slightly less than 1988 levels until the end of mine life, provided the trtsh communities and their food sources can recover from the reductions. The fact that the models generated generally incorporated dissolved or particulate copper rather than suspended solids is of little importance given
64
R. E. W . S M I T H
AND
K . (3. H O R T L E
that the three variables were correlated within the data set (Spearman's rank correlation,p < 0.05 for yearly means). However, it is expected that the relationships between the three variables will change in future as the predominance of sulphide ores increases (Ok Tedi Mining Limited, 1988, 1989). This is reflected in the different predictions of future catches by the dissolved and particulate copper models. The predictions of recovery from periods of reduced fish stocks assume that the necessary recruits will be available. The sources of recruits for many species in the Fly River system are away from the main river channel in tributary streams, in marine locations and in off-river water bodies such as oxbow lagoons, the floodplain and blocked valley lakes (Hortle, 1986b). The mine discharges will not directly affect the tributary streams, and the marine habitats outside the Fly estuary and delta will be impacted slightly if at all. However, at least some of the off-river and estuarine spawning sites will receive elevated copper levels (Ok Tedi Mining Limited, 1989). Limited fish sampling in off-river water bodies has not demonstrated any impacts of mining as yet (Ok Tedi Mining Limited, 1989) and only minor increases in copper concentrations have been detected in the estuary and delta (Ok Tedi Mining Limited, 1988, 1989). As the period of most severe impacts is short, the impacts on the fish stocks in offriver refuges appears to be occurring at a slower rate than in the river channel and fish within the river channel obtain most of their food from offriver and terrestrial sources, the ability to increase stocks after 1991 should be maintained for most species. The maintenance of communities similar to those found in 1988 for the bulk of mine life would require the maintenance of these offriver refuges, which has not been adequately investigated. It would be expected that if maintained, 1988 stock levels could adequately support the current levels of exploitation (6.2 to 31.2% of the potential yield, Ok Tedi Mining Limited, 1988), and would have the potential to recover to pre-mine levels after mine shut down. The modelling approach used in this study has quantified responses of the fish communities downstream of the mine to elevated concentrations of copper and suspended solids. The derived relationships were then be used with caution to predict future effects. Although the demonstration of such relationships does not prove causality, with continued monitoring, the changing nature of the mine discharges will provide information as to which of the chemical parameters tested was or were chiefly responsible for the changes in the abundances of fish. This study has illustrated that statistical models are useful monitoring tools when the complexities of the problems of concern are such that the development of quantitative conceptual models is not practical.
5. Acknowledgements The authors would like to thank Dr. John Alabaster for suggesting the modelling approach, and Dr. Glen De'ath of James Cook University of North Queensland for advice on the statistical proceedures employed in this report and particularly for suggesting the use of ANCOVAR. This research would not have been possible without the financial and logistical support of Ok Tedi Mining Limited, in particular the Hydrology and Chemistry sections who allowed us to utilise their data, and the technical assistance of
IMPACTS OF A COPPER MINE ON RIVER FISH
65
the staff of the Biology Section - Andy Maie, Yarang Kurtama, Pius Fred, Augustine Mungkaje and Barre Kare. The contribution of David Balloch in setting up the biological sampling programme is also acknowledged. 6. References Cordone, A. J., and Kelly, D. W., 1961: 'The Influences of Inorganic Sediment on the Aquatic Life of Streams', California Fish and Game 47, 189-228. Alabaster, J. J. and Lloyd, R., 1982: Water Quality Criteria for Freshwater Fish', Butterworth Scientific, London, U.K., 361 pp. Boyden, C. R., Brown, B. E., Lamb, K. P., Drucker, R. F., and Tuft, S. J., 1978: 'Trace elements in the Upper Fly River, Papua New Guinea', Fresh Water Biology 8, 189-205. Greig-Smith, P., 1980: 'The Development of Numerical Classification and Ordination', Vegetatio 42, 1-9. Holeman, J. N., 1968: 'The Sediment Yield of Major Rivers of the World', Water Resources Research 4, 737-747. Hortle, K. G., 1986a: 'Survey of the Fish Fauna of the Strickland River at Tiumsinawam with Reference to Sediment Tolerance.' Ok Tedi Mining Limited Report, ENV 86--6, 30pp. Hortle, K. G., 1986b: 'A Review of the Results of Biological Sampling of the Ok Tedi and Fly River Systems, April 1983 to June 19867 Ok Tedi Mining Limited Report, ENV 86--9, 202pp. Hortle, K. G., 1987: 'Studies of the Benthic Fauna of Lowland (Potamon) Localities of the Ok Tedi and Fly River, with Reference to Mining Impacts.' Ok Tedi Mining Limited Report, ENV 87-11,45pp. Maunsell and Partners Pty. Ltd., 1982: 'Ok Tedi Environmental Study. Working Paper 1. Water Quality of the Ok Tedi and Fly River System.' Ok Tedi Mining Limited, 135pp. Ok Tedi Mining Limited, 1983: 'Environmental Monitoring Programme 6 Monthly Report. June - December 1983.' Ok Tedi Mining Limited Report, PA/03/84-8, 71pp. Ok Tedi Mining Limited, 1988: 'Sixth Supplemental Agreement Environmental Study 1986-1988. Final Draft Report.' Ok Tedi Mining Limited, 702pp. Ok Tedi Mining Limited, 1989: 'Supplementary Environmental Investigations. Volume One.' Ok Tedi Mining Limited, 74pp. Roberts, T. R., 1978: 'An Icthyological Survey of the Fly River in Papua New Guinea, with Descriptions of New Species', Smithonian Contributions to Zoology 281, 1-72. SAS Institute Inc., 1985: SAS 9 User's Guide: Statistics, Version 5 Edition, SAS Institute Inc., Cary, USA, 956pp. Smith, R. E. W., 1988: 'Six-monthly Biology Review, 11 October 1987 to 30 March 1988.' Ok Tedi Mining Limited Report, ENV 88-05, 74pp. Smith, R. E. W., Ahsanullah, M., and Batley, G. E., 1989: 'Investigations of the Impact of Effluent from the Ok Tedi Copper Mine on the Fisheries Resource in the Fly River, Papua New Guinea', Environmental Monitoring and Assessment 14, 315-331. Welcomme, R. L., 1985. River Fisheries. FAO Fisheries Technical Paper 262, 330pp.
66
R . E . W . SMITH AND K. G. HORTLE
APPENDIX A. Gill nets used in fish sampling "Mono' indicates net constructed from nylon monofilament, 'multi' indicates nylon multifilament. Mesh size (mm)
Type
Length (m)
Depth (m)
Area (m2)
Number used
25 38 50 63 75 88 100 125 150 150 175
mono mono mono mono mono mono mono mono mono multi multi
40 40 45 40 45 45 45 45 50 25 25
2.3 1.7 2.1 2.8 3.2 3.5 4.2 4.9 6.0 2.8 3.1
92 68 95 112 144 158 189 221 300 70 78
1 1 l l 1 1 1 1 1 2 2
IMPACTS OF A COPPER MINE ON RIVER FISH APPENDIX B List of fish species assigned to each feeding group. Feeding Group
Species list
Family
Predators
Pristis microdon Megalops cyprinoides Thryssa scratchleyi Strongylura kreffti Lates calcarifer Glossomia aprion Lutjanus goldei L. argentimaculatus Oxyeleotris fimbriata O. nullipora (9. lineolatus Arius augustus
Pristidae Megalopidae Engraulidae Belonidae Centropomidae Apogonidae Lutjanidae
Clupeoides venulosus C. papuensis Thryssa rastrosa Arius leptaspis A. acrocephalus A. carinatus Hemipimelodus crassilabris H. macrorhynchus Cinetodus froggatti Nedystoma dayi Cochlefelis spatula C. danielsi Neosilurus ater N. equinus N. meraukensis N. brevidorsalis Neosilurus sp. Oloplotosus luteus Porochilus obbesi P. meraukensis Plotosus papuensis Craterocephalus randi C. nouhuysi Ambassis reticulata Parambassis gulliveri Denariusa bandata Damioides quadrifasciatus
Clupeidae
Aquatic invertebrate feeders
Aquatic invertebrate feeders
Hephaestes roemeri 14. trimaculatus Terapon jamoerensis Amniataba percoides Glossamia trifasciata Acanthopragus berda Nibea semifasciata Glossogobius giurus G. celebius
Eleotridae
Ariidae
Engraulidae Ariidae
Plotosidae
Atherinidae Ambassidae
Lobotidae Theraponidae
Apogonidae Sparidae Sciaenidae Gobiidae
67
68
R. E. W. SMITH
AND
K. G. HORTLE
APPENDIX B (continued) List of fish species assigned to each feeding group. Feeding Group
Terrestrial invertebrate feeders
Herbivores
Detritivores
Species list
Family
G. concQvifrons Mogurnda mogurnda
Eleotridae
Ophieleotris aporos Kurtus gulliveri Cynoglossus heterolepis Anabas testinudineus
Kurtidae Cynoglossidae Anabantidae
Scleropages jardini Arius berneyi Hemipimelodus taylori H. macrorhynchus Melanotaenia splendida M, maccullochi M. oktediensis M. sexlineata Pseudomugil novaeguinea P. gertrudae Iriatherina werneri Zenarchoptenus novaeguinea Toxotes chatareus T. lorentzi
Osteoglossidae Ariidae Melanotaeniidae
Hemirhamphidae Toxotidae
Terapon jamoerensis Pingalla lorentzi Oxyurichthys papuensis Anabas testudineus
Theraponidae
Nematalosa papuensis N. flyensis Crenimugil labiosus Liza diadema Cynoglossus heterolepis
Clupeidae
Gobiidae Anabantidae
Mugilidae Cynoglossidae
