3. METHODS AND RESULTS.

3.1. METHODS.

The methods used in this work are generally the same as used by Van Zwieten (1989).

In order to estimate trout production in the highlands, the rotenone method was applied for sampling trout, although this method is very difficult to use in the typical trout stream (fast current, steep gradient).

Additional trout specimens were purchased from villagers neighbouring trout streams.

3.2. PHYSICAL CHARACTERISTICS OF TROUT STREAMS.

The lower limit of trout, based on my survey, is 1760 m.a.s.l. This is in accordance with other tropical high-altitude areas, for example in Madagascar, where trout has established in streams above 1700 m (Kiener and Richard-Vinard 1972).

Trout streams in the highlands are typically fast-flowing soft-water streams with rocky and stony riverbeds and clear water; and usually with no macrophytes growing in the streams. Temperatures are within the range of 10–15°C.

The findings in this report are mainly based on trouts caught at the following three sampling stations: Anggura River, Southern Highlands; Kuragamba River, Simbu Province, and Omaigiha Creek, Eastern Highlands. A description of these streams is given in Povlsen (1993a).

3.3. CONDITION FACTOR

Mean condition factor for trout from the three sampling stations are listed in Table 1, together with individual data on standard length, total weight and condition factor.

Data on condition factor of rainbow trouts from Victoria, Australia, revealed that the range of the condition factor varied from approximately 0.8 to 2.0 with the majority being in the range 1.0 to 1.4. A value of 1.25 was adopted as a satisfactory average condition factor for salmonid fishes (Baxter et.al. 1991).

Based on this, the CF-values in Table 2 (minimum=1.15; maximum=2.58; mean=1.60) indicate that rainbow trouts from Papua New Guinea streams are in very good condition.

There are no significant differences between condition factors of trouts from the three sites (t-test; p > 0.05).

3.4 STOMACH CONTENTS.

The stomach contents of rainbow trout caught at the three different sampling sites. are listed in Table 2.

Table 2. Stomach content of rainbow trout from 3 sampling stations in the highlands of Papua New Guinea. %V =percentage volume of that food category of the total volume of food within all stomachs examined. %N =percentage of individuals having that food item within their stomach.

* The content of all stomachs were pooled prior to analysis.

** FPOM = Fine Particulate Organic Matter.

The main food item at all sites is benthic insects, which make up as much as 91.9 % of the food at Anggura River and 74 and 76 % at Kuragamba River and Omaigiha Creek, respectively. Of the benthic insects, larval caddis flies (Trichoptera) are the most abundant prey constituting around 30 % of the stomach content at all three sites. The other major taxa are larval Diptera, Ephemeroptera, Coleoptera and Odonata. Non-insect benthic fauna (e.g. crustaceans and molluscs) were not found in trout stomachs. This is in accordance with Dudgeon (1989), who found that in the Sepik/Ramu non-insect benthic taxa are confined to lower altitudes (below 250 m.a.s.l.).

At Kuragamba River a considerable part of stomach contents was plant matter (16.7%), mainly in the form of terrestrial plant fragments (often 1–2 cm twigs). At Omaigiha Creek plant matter constituted 11 %, whereas at Anggura River it constituted only 2.1 % of the stomach content.

Plant material is often found in large amounts in trout stomachs (Cadwallader & Eden 1982). Plant fragments are probably taken “accidentally” during the voracious feeding behaviour of rainbow trout.

Terrestrial invertebrates seem to play a minor role as food organisms for trout in Papua New Guinea.

No fish were found in the stomachs, although, at least in Anggura River, a native gudgeon species (Eleotrididae) occurs (a juvenile gudgeon was caught during my visit) (see later).

The differences between the three sites probably reflects the difference in availability of food items. The wide Anggura River carries relatively less terrestrial input than the other streams, and terrestrial input (of both plant and animal origin) therefore contribute relatively less to the diet of trouts from Anggura River.

Terrestrial invertebrates have previously been recognised as a very important food source for trouts (Waters 1988), and in general for fishes in lower order tropical forest streams (Lowe-McConnell 1987). Surprisingly, this doesn't seem to be the case in the three streams sampled here. The aquatic insect production in the highland streams of Papua New Guinea may be sufficient to support the trout population. Unfortunately, no data exist on benthic production in tropical high-altitude streams. However, Dudgeon (1989) concluded from his investigations of potential food availability for fishes in the Sepik River that stream benthic communities at higher altitudes are as diverse as those elsewhere in the tropics, and species richness peaked at an altitude of 1800 m.a.s.l. This is further discussed in section 3.6.

3.5. BREEDING HABITS.

Gonadal stage and sex of trouts caught at the three sampling stations are listed in Table 3.

* Caught downstream sampling area.

Nothing conclusive can be said about the breeding habits of rainbow trout in Papua New Guinea. I consider all the trouts caught in Kuragamba and Anggura rivers to be a result of successful natural reproduction (i.e. the last stockings in both areas occurred no later than 1985–86). It seems that spawning occurs in remote, high-altitude streams (above 2300 m.a.s.l.) in very sparsely populated areas. Especially the juvenile trouts caught in Kuragamba River (September 2, 1991), suggest that successful breeding occurs in that area. In support of this, a male and a female with running-ripe gonads (stage 5) were caught downstream the sampling area.

In Omaigiha Creek a male with running-ripe testes was caught by villagers during our visit on August 2, 1991.

It has been suggested that spawning occurs in July - September in Papua New Guinea (Sagom 1989). October - November has also been suggested as the main breeding season, a time when upstream migrations of large trout has been reported (Cadwallader 1991).

In some areas people reported that female trout with eggs were caught all year round (for example, at Komea, Southern Highlands Province).

In temperate parts of the world, variations in stream discharge and temperature have been reported to play a role in upstream movement (and spawning) of salmoniid fishes (Gordon & MacCrimmon 1982). This may also be the case here, although further studies are needed on established trout populations in tropical areas, including Papua New Guinea, to elucidate this matter.

3.6. PRODUCTION OF RAINBOW TROUT IN HIGHLAND STREAMS.

3.6.1. Salmonid production and Production/Biomass (P/B) ratios.

Salmonid production in rivers and streams in temperate regions has been studied extensively and a lot of data exist on annual production and production-biomass ratios, P/B (Waters 1977 & 1988). But currently, no data exist on salmonid production in tropical regions.

In general, annual production depends on the alkalinity/hardness of the water (Waters 1977; Waters et.al. 1990; Whitworth and Strange 1983). In infertile softwater streams in northern (temperate) or mountainous streams annual salmonid production is generally below 60 kg/ha, while in hardwater streams, often in limestone geology, the estimates are between 100–300 kg/ha (Waters et.al. 1990).

The P/B ratio for salmonid streams has been estimated to vary from a low of 0.9- 1.5 to a high of 2–2.4 (Chapman 1978). The highest P/B ratios were in streams in which the winter low temperature did not reach below 6–7°C and the yearly mean was 10°C or more. Under PNG conditions with water temperatures of 10–15°C in the rainbow trout range and no winter low, P/B ratios are probably higher. Chapman stated that under the majority of environments in which salmonids dominate the P/B ratio is around 2. Waters (1988) used a P/B ratio of 1.25, a ratio “commonly reported for stream trout populations”.

Several methods have been applied for estimating secondary production in rivers (Waters 1977). One is the instantaneous growth rate method using the formula P=GB, where P is production (for a given period of time), G is instantaneous growth rate (for the time period) and B is standing stock (during time period).

It follows from this formula that G=P/B.

Consequently, a rapid (though perhaps less precise) method of estimating production is to multiply a “known” P/B ratio by an appropriate measure of standing stock. This is used in the following to estimate trout production in New Guinea streams.

3.6.2. Trout production in streams in Papua New Guinea.

Due to the steep gradients and high water velocities in the altitudinal range where rainbow trout occur in the highlands, it is difficult to undertake thorough sampling with rotenone.

I succeeded in doing two samplings from which I can use the data for estimating biomass/production of trout. In Omaigiha Creek near Goroka, Eastern Highlands Province, the biomass was 69.3 kg/ha, and in Kuragamba River, Simbu Province, the biomass was 46.3 kg/ha.

Using those figures as mean standing stock (assuming no significant variations throughout the year) and assuming a P/B ratio of 1.25 as suggested by Waters (1988) (which might be an under-estimate for PNG-conditions) gives a production estimate of 86.6 kg/ha/year in Omaigiha Creek and 57.9 kg/ha/year in Kuragamba River.

Of course, this is very rough estimates based on uncertain assumptions. But they give an indication of the range of potential production in the highland streams of Papua New Guinea.

Neves et.al. (1985) estimated annual production of rainbow trout at 36 kg/ha in an Appalachian stream in Virginia, USA. They found that for older trouts (2 years and older) production was negative during winter. Other estimates for annual production of rainbow trout (as listed in Neves et.al. 1985) range from a maximum of 132 kg/ha in Bothwell's Creek, Ontario, Canada, to a minimum of 24 kg/ha in Lemhi River, Idaho. As Neves et.al. point out, these varying estimates of rainbow trout are not directly comparable because of differences in species composition and physicochemical variables among the streams.

A comparison between temperate and tropical environments is even more problematic. However, an important factor for production in tropical streams is without doubt the lack of a winter-low, which is seen in temperate streams. As a result, the growth rate (P/B-ratio) is probably higher under tropical conditions. Since I have used a “temperate” P/B-ratio in my calculations, the estimates of rainbow trout production are conservative.

According to Coates (unpublished data), the Sepik/Ramu catchment has 148.9 km² of streams suitable for fish production in the altitudinal range of 1800–2800 m.a.s.l. (the expected altitudinal range of rainbow trout in Papua New Guinea). This gives a total potential trout production of 1074 tons/year in the Sepik/Ramu catchment alone.

Most of the streams where trout populations could establish in Papua New Guinea are low-fertile softwater streams with alkalinities ranging from 30 to 70 mg/l (Petr 1983). In low-fertile, softwater streams in temperate regions, benthos production has often been reported to be much too low to support the levels of trout production commonly reported (Waters 1988); a phenomenon known as the “Allen Paradox”. An obvious explanation of this is that trout exploits other food sources than benthic invertebrates. Waters (1988) suggests that trout, more than other stream-dwelling fish species, appear to feed on terrestrial surface-drift.

In this study, investigation of stomach content revealed that terrestrial invertebrates play a minor role in the diet of trout in Papua New Guinea. Aquatic insects constitute the dominant part of the diet.

According to Dudgeon (1989) stream benthic communities in the Sepik River seem as diverse as those elsewhere in the tropics. At an altitude of 2990 m.a.s.l. species richness had not declined, and it peaked at an altitude of 1800 m.a.s.l. Although benthic production was not estimated, it seems that the production of benthic invertebrates may be sufficient to support the estimated production of trout.

Allochthonous/terrestrial invertebrates may be a supplemental food source (especially after rain showers) for trout in tropical environments, but may not be as important as for trout in temperate environments. The “Allen Paradox” may be a phenomenon restricted to temperate regions, although further investigations on benthic production in tropical high-altitude streams are needed to elucidate this.

Neves et.al. (1985) compared available production estimates of rainbow trout with estimates of brook trout production. They found that rainbow trout is more productive than brook trout (brook trout production estimates ranged from 3.1 kg/ha/year to 19.3 kg/ha/year). Rainbow trout has higher growth rates and fecundities, a larger maximum size and greater tolerance of flooding and water temperature variations (Neves et.al. 1985). Due to the very limited data available on production of non-salmonid fishes, it is difficult to compare production of rainbow trout with other coldwater fish species (i.e. the Himalayan cyprinids recommended for stocking Sepik/Ramu, see later). Shrestha (1990) noted that the production of Mahseer, Tor putitora, in natural waters in Nepal is in the range of 12–18 kg/ha/year. No calculation method was given. Data on other coldwater species are very limited.

In conclusion, rainbow trout is one of the most productive salmonid species in low-fertile, cold-water streams and is suitable for rivers and streams at altitudes above 1800 m in Papua New Guinea.