Bekæmpelsesmiddelforskning fra Miljøstyrelsen, 82 – Effekt af bekæmpelsesmidler på flora og fauna i vandløb

Effect of Pesticides on Flora and Fauna in Danish Streams

Summary and conclusions

This report presents results from a 2½ year study of the effects of a range of pesticides on benthic microalgal communities and on representative benthic invertebrates from Danish streams. The study was financed by the Danish Environmental Protection Agency and was conducted in collaboration among DHI Water and Environment, Roskilde University, and the Danish National Environmental Research Institute.

The objective of the study was to elucidate the effects of pesticides on naturally occurring algae and invertebrates. The majority of our present understanding of pesticide effects is based on tests conducted under standardised conditions, including exposure to constant concentrations of pesticides and using animals and plants from stagnant water bodies. Such results cannot directly be used to predict effects in streams in which pesticides will tend to occur at high concentrations for short periods of time and because standard test organisms, such as daphnids, are not necessarily as sensitive to pesticides as stream-dwelling species.

The present study focuses on measuring effects of realistic exposure durations, as short as 30 minutes, and examines subsequent (delayed) effects on survival, growth, reproduction and behaviour for up to 30 days post exposure. The pesticides examined in the study (a total of 14) were selected on the basis of their chemical characteristics and extent of use in Denmark. Furthermore, the aim was to ensure that most of the toxic mechanisms of action were represented. We investigated effects on 'natural' communities of benthic microalgae, on populations of crustaceans (2 species), midges, stoneflies (2 species), caddis flies (2 species), mayflies, and snails. We examined effects in different life stages within species, and for one of the species (the midge, Chironomus riparius) we conducted a full life-cycle study. Most of the exposures were carried out in water, however two tests were conducted in which either sediment or food was contaminated.

Combinations of the selected pesticides and study species were chosen on the basis of existing knowledge about the pesticides' toxicity and mechanism of action. Therefore we tested primarily herbicides in tests with algae and primarily insecticides in tests with invertebrates, whereas fungicides were tested with both taxonomic groups.

The experiments with benthic microalgae showed that the herbicides isoproturon and metribuzine inhibited photosynthesis at concentrations in the range 0.3 – 1 µg l^-1. These concentrations are in the range shown to occur in several Danish streams for isoproturon and for pesticides belonging to the triazine group (i.e., atrazine, terbuthylazine, simazine) that have a toxicity similar to metribuzine. It is therefore very likely that these herbicides can influence primary production in a few streams. However since the effects were fully reversible, and since the duration of increased concentrations is typically 2 – 10 hours in small streams, we conclude that direct effects on microalgae and indirect effects on invertebrates are minimal.

The herbicide pendimethalin affected algal communities at relatively low concentrations. Green algae were particularly sensitive, and effects on their biomass intensified even after transfer into clean water following exposure to the lowest test concentration (10 µg l^-1). In comparison, mortality of the crustacean, Gammarus pulex, first occurred at concentrations in the mg l^-1 range, and then only after continuous exposure for several days. Pendimethalin is one of the most widely used herbicides in agriculture, but measured concentrations and the percentage of samples in which it has been detected are relatively low due to mandatory buffer zones along streams and due to a low mobility. In any case, we estimate that pendimethalin may cause effects on benthic algae because it probably accumulates in stream bottoms where benthic microalgae are exposed, because it is relatively persistent, and because effects are long-lasting. There are, however, no measurements of pendimethalin in stream sediments that support this suggestion.

The fungicide azoxystrobin was only weakly toxic to algae, whereas concentrations resulting in mortality of the most sensitive invertebrate, Gammarus pulex, were markedly lower (LC₅₀(4-14 days) = 3-10 µg l^-1). Brief exposures (0.5 – 2 hours) resulted in delayed mortality, but only if exposure concentrations were unrealistically high (2.5 mg l^-1). On the other hand, feeding activity in G. pulex was significantly reduced even after short (0.5 hour) exposures to 20 µg l^-1. In comparison, breakdown of contaminated leaves was not affected. Azoxystrobin was first introduced to agriculture in 1998, and consequently there is little information on its occurrence in Danish streams. The occurrence of azoxystrobin in concentrations around 20 µg l^-1 in Danish streams is not likely despite the fact that its mobility is relatively high and its degradation very slow. However, since azoxystrobin is new on the market and has not been included in comprehensive monitoring programmes, realistic environmental concentrations are unknown.

The fungicide fenpropimorph was, like azoxystrobin, only weakly toxic to microalgae, and there was full recovery of both photosynthesis and biomass following transfer to clean water. In the NOVA program fenpropimorph has been found in 2-4% of the samples taken, with a median-maximum concentration of 0.07 µg l^-1 (DMU 2001, 2002). On this background we consider that effects of fenpropimorph on algal communities are unlikely.

The fungicide prochloraz was moderately toxic to Gammarus pulex with effect concentrations falling sharply with increased exposure duration (LC₅₀(24 hours) = 22 mg l^-1, LC_{50,14 days} = 0.6 mg l^-1. There was no acute mortality at realistic exposure durations (< 24 h) up to concentrations of 10 mg l^-1. On the other hand we observed delayed toxicity following brief exposures (0.5 - 1 h) to very high concentrations ( > 5 mg l^-1). Prochloraz has not been detected in Danish streams by the NOVA program, and together with the high effect concentrations we consider it unlikely that this fungicide affects the invertebrate fauna in Danish streams.

Toxicity of the insecticide dimethoate was low for all the invertebrates tested with LC_{50,96 h} values of 5 mg l^-1 or higher. In the crustacean Gammarus pulex, toxicity increased with increasing exposure duration, but even after 14 days the LC₅₀ was greater than 1 mg l^-1. There was no effect on Gammarus' breakdown of alder leaves even after exposure to 2 mg l^-1. Considered in relation to the very low percent detection and measured concentrations in Danish streams, we consider therefore that the effects of dimethoate are not very likely.

The insecticide pirimicarb was the least toxic of the pesticides tested. Exposure to concentrations as high as 5 mg l^-1 resulted in no toxicity in five species even after 14 days. Compared with the low percent detection (4%) and very low measured concentrations (maximum 0.026 µg l^-1), we consider that effects of pirimicarb are very unlikely in Danish streams.

The toxicity of the insecticide diflubenzuron was investigated in five different invertebrates. We only found acute (96 h) mortality at the highest test concentration (Gammarus pulex: 0.2 mg l^-1; other species: 3-5 mg l^-1). Insecticides in the benzoylphenylurea group act specifically on growth/molting in insects (and presumably crustaceans), and effects of short-lived exposures would not be expected. In Denmark insecticides of the benzoylphenylurea group are not used in agriculture, and therefore the risk of transport to streams is very small.

The pyrethroid insecticide esfenvalerate was by far the most toxic of the investigated pesticides. Standardised acute tests showed an LC_{50,96 h} value for Gammarus pulex to occur at the same level as in daphnids (ca. 200 ng l^-1), and the acutely toxic exposure concentrations were markedly lower than observed for the other invertebrates tested. During longer term exposures (14 days) the LC₅₀ reduced to a value of approx. 30 ng l^-1. Even brief (0.5 – 1 h) exposures to low concentrations (50-200 ng l^-1) resulted in subsequent effects on survival, growth/development, reproduction and feeding activity.

Investigations of invertebrate drift following exposure to esfenvalerate showed effect concentrations between 0.2 and 20 ng l^-1 with Gammarus pulex as clearly the most sensitive of the tested species. Esfenvalerate had no effect on drift of the two caddis fly species tested.

Neither esfenvalerate nor other pyrethroids have been detected in the NOVA program, however, under spate conditions in Danish streams concentrations up to 660 ng l^-1 have been measured. The increased drift at a nominal concentration of 0.2 ng l^-1 is much lower than the highest measured concentration, but also much lower than the detection limit for chemical analysis. It is well known that pyrethroids and other pesticides with a high particle affinity under natural conditions occur primarily bound to particles. It has also been shown that the particle-bound fraction is less toxic than the dissolved. This makes it difficult to predict effects of pyrethroids on the invertebrate fauna in Danish streams on the basis of chemical analyses. We believe, however, that with such low effect concentrations even after brief exposures and the widespread effects on vital processes such as survival, reproduction and behaviour, pyrethroids present the greatest pesticide threat to invertebrate fauna in Danish streams.