Developments in Earth & Environmental Sciences, 9 B. Bolton (Editor) r 2009 Elsevier B.V. All rights reserved DOI 10.1016/S1571-9197(08)00412-6 Chapter 12 Use of Changes in Fish Assemblages in the Fly River System, Papua New Guinea, to Assess Effects of the Ok Tedi Copper Mine Andrew W. Storey1,Ã, Markson Yarrao2, Charles Tenakanai3, Boga Figa3 and Jessica Lynas1 1School of Animal Biology (M092), The University of Western Australia, Crawley, WA, 6009, Australia 2Environment Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, W.P. Papua New Guinea 3Livelihood Programs Department, Ok Tedi Mining Limited, PO Box 1, Tabubil, W.P. Papua New Guinea 12.1. Introduction The Fly River in Papua New Guinea is one of the largest rivers in Australasia (mean annual discharge B6,000 m3 sÀ1). It has a catchment area of 76,000 km2 and flows a distance of over 1,200 km from its source in the central highlands of New Guinea to the Gulf of Papua. Much of the catchment, particularly in the upper reaches, consists of dense primary tropical rainforest, while in the middle and lower reaches open savannah forest, swamp forest, and seasonally inundated grasslands predominate. Although the upper catchment extends to altitudes of more than 3,500 m, the majority of the drainage basin is low-lying and flat, to the extent that the port of Kiunga, which is 800 river km from the coast, is only 20 m above sea level. The combination of low topography and high rainfall has resulted in a broad floodplain with extensive shallow lake systems, occupying an area of 4.5 million ha, making it the largest wetland system in the country. The wetlands ÃCorresponding author. Tel.: (618) 6488 1482; Fax: (618) 6488 1029; E-mail: [email protected] (A.W. Storey). 428 A. W. Storey et al. of the Fly River are highly productive and play a vital role in the ecology of the river system. The area is sparsely populated, with an average human density of one–two persons per square kilometer. Prior to 1960, the highlands area had little contact with the outside world, and the river downstream was relatively pristine, with no mining or logging, and only low-scale commercial fishing for barramundi. Following the discovery of the Mt Fubilan ore deposit, scientists began surveying the aquatic fauna of this previously unstudied, remote river system (Boyden et al., 1978; Roberts, 1978; DPI, 1979, 1980; Robertson and Baidam, 1983). These studies revealed that the river system has the most diverse freshwater fish fauna in the Australasian region, and is known to support over 115 freshwater and marine vagrant species (Roberts, 1978; Maunsell and Partners, 1982; Allen, 1991; Coates, 1993; Swales et al., 1999). Of these, 17 are endemic to the Fly Basin, and over 30 are known only from the Fly and one or more of the large rivers in central-southern New Guinea (Roberts, 1978). The fishes of the Fly River basin are characterized by the large individual sizes of some species, an abundance of endemic fishes, and the presence of species that are poorly represented in other parts of the world, particularly the ariid and plotosid catfishes. In most other ways, the composition of the freshwater fish fauna is largely determined by its position in the Australasian zoogeographical zone (Roberts, 1978; Coates, 1993). Given the uniqueness of the fish fauna and concerns over the potential impact of a mine on its headwaters, environmental monitoring of the Fly River for the proposed mine commenced in 1981 (Maunsell and Partners, 1982). Monitoring began with an impact assessment which involved an expeditionary survey of water quality, fish communities in different habitats and metal levels in biota (Maunsell and Partners, 1982). Soon after, Ok Tedi Mining Ltd (OTML) implemented an extensive biological monitoring regime that commenced during the mine construction phase in 1983 and included all river reaches, from the headwaters to the delta and into the Gulf of Papua (Wood et al., 1995; Swales et al., 1999). An important aspect of this program was monitoring fish populations, partly in recognition of their value as a tool for assessing anthropogenic impacts, and as an indicator of ecosystem health and productivity (Fausch et al., 1990; Harris, 1995). It is also because the fish fauna forms an important subsistence food source for village communities along the river (Hortle, 1986). Due to this reliance on fish, the main concern of the State of Papua New Guinea was that this resource had to be protected (OTML, 1988, 1990). As a result, monitoring tended to concentrate on numbers and weight of fish available at locations throughout the river system. This emphasis has continued through mine life, and has influenced sampling methods, data collected, and how analyzed. Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine 429 There have been three operating stages in mine life to date. Stage One (the treatment of gold ores by cyanide extraction) commenced in 1984. Stage Two commenced in mid-1987 and consisted of the gold extraction circuit running in parallel with a flotation circuit to produce copper concentrate. Stage Three commenced in mid-1988 when the gold circuit was decommissioned (and all use of cyanide ceased) and the mine became a copper concentrate producer. Additional infrastructure was completed in 1989, enabling the mine to treat 80,000 t of ore per day at peak production. The next major change was in 1998 when a dredge was deployed on the lower Ok Tedi to dredge the channel with the intention of alleviating bed aggradation in the lower Ok Tedi and Middle Fly. Currently, approxi- mately 50 Mt per annum (pa) of waste rock and 30 Mt pa of tailings are discharged directly into the Ok Tedi and its tributaries, and approximately 15 Mt pa of sediment are dredged from the lower Ok Tedi and deposited in ‘‘cells’’ on the adjacent floodplain. The original mine plan was to operate with riverine disposal of waste rock, but storage of tailings in a dam in the upper catchment. However, the tailings dam collapsed during construction, and to allow the mine to proceed, the State permitted riverine disposal of detoxified tailings. The Ok Tedi mine subsequently received notoriety in the mid-1980s when the cyanide detoxification system failed on several occasions, releasing tailings with elevated cyanide levels into the Ok Tedi, resulting in the death of fish, turtles, and crocodiles for approximately 100 km downstream. Since the closure of the gold extraction circuit (1988), the mine has operated with a copper flotation circuit, which recycles the flotation chemicals. As such, the tailings have no elevated chemicals, but are elevated in a suite of particulate and to a lesser extent dissolved metals including Cu, Pb, Zn, Cd, As, and Fe, compared to average crustal abundance (Bolton et al., 2009). Monitoring the effects of the Ok Tedi mine on the river system has continued to present, principally to document and understand impacts resulting from the mode of operation of the mine (Pickup and Cui, 2009). Relationships between changes in fish stocks in the main river channel and mine waste discharges have been reported (Smith et al., 1990; Smith and Hortle, 1991; Smith and Morris, 1992), as have impacts to the fish fauna of riverine and floodplain habitats (Swales et al., 1998, 1999, 2000). Since these publications, additional data have been collected and further analyses conducted. This chapter describes temporal variations in fish catch and assemblage composition, and updates previous assessments of species diversity and biomass from sites downstream of the mine (OTML, 1994, 1995, 1996; Swales et al., 1998, 1999, 2000). 430 A. W. Storey et al. 12.2. Sampling and Analyses Although monitoring only commenced during mine construction, providing limited baseline data, the sampling method adopted at that time has been maintained throughout the project, providing a standard sampling approach, now covering almost 25 years. This provides an immensely valuable time series, unlikely to be matched from any other river in the world, let alone a large, tropical river system. Such intensive sampling is known to yield a high number of species and more accurate and precise estimates of richness (Cao et al., 2001; Hughes et al., 2002; Reynolds et al., 2003; Kennard et al., 2006). Methods were based on those used by Maunsell and Partners (1982) using a standard set of 13 gill-nets, ranging in stretched mesh size from 25 to 175 mm (Table 12.1). The selection of gill nets as the main method was a compromise over what was readily available and could be used and maintained in a remote region, and what was effective in the conditions and across habitats. It is also a sound, simple, and technically robust method to obtain catch per unit effort (CPUE) data, ideally suited to a remote location where failure of more technical equipment (i.e., electrofishers or hydro- acoustic gear) can result in many months delay. Seine netting, rotenone, trapping, and electrofishing were also used selectively in certain habitats to supplement catches; however, these data are not included in the current Table 12.1: Gill nets used in the Standard Gill Net Set used by OTML since 1983, giving stretched-mesh size of each net, line type, number set, dimensions, and area. Stretched-mesh size and type Number in Length (m) Depth Area net set (m) (m2) 1v (25 mm) Monofilament 1 40 2.3 92 1½v (38 mm) Monofilament 1 40 1.7 68 2v (50 mm) Monofilament 1 45 2.1 95 2½v (63 mm) Monofilament 1 40 2.8 112 3v (75 mm) Monofilament 1 45 3.2 144 3½v (88 mm) Monofilament 1 45 3.5 158 4v (100 mm) Monofilament 1 45 4.2 189 5v (125 mm) Monofilament 1 45 4.9 221 6v (150 mm) Monofilament (6M) 1 50 6 300 6v (150 mm) Multifilament (6C) 2 25 2.8 70 7v (175 mm) Multifilament (7C) 2 25 3.1 78 Fish Assemblages in the Fly River System and Effects of the Ok Tedi Copper Mine 431 analyses as they can confound CPUE data across sites and times.