RIVER RESEARCH AND APPLICATIONS River Res. Applic. 21: 245–255 (2005) Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/rra.844 USING COMMERCIAL CATCH STATISTICS TO DETECT HABITAT BOTTLENECKS IN LARGE LOWLAND RIVERS CHRISTIAN WOLTERa* and RONALD MENZELb b a Leibniz-Institute of Freshwater Ecology and Inland Fisheries, 12561 Berlin, Germany Fischereischutzgenossenschaft ‘‘Havel’’ Brandenburg e.G., 14774 Brandenburg, Germany ABSTRACT Thirty-seven years of fisheries records covering a fishing area of 6231 ha in the lower Havel River, Germany, have been analysed to address two issues: (1) detection of the effects of habitat bottlenecks caused by extreme floods and droughts on adult fish assemblage; and (2) evaluating the appropriateness of commercial fisheries statistics in testing the habitat bottleneck concept. Time series analyses of the data were first tested for autocorrelations and then classified using the CHAID (Chi-squared Automatic Interaction Detector) and C&RT (Classification and Regression Trees) algorithms. The significant environmental predictors revealed by these segmentation procedures were used in cross-correlations with various time series of fisheries yields. No single hydraulic flood parameter was significantly correlated with fisheries yield but two drought parameters were. The strongest predictor of native fish species was the minimum discharge during the spawning season (March to June). Although total fisheries yield typically increased in years with low flows, probably because of higher catch efficiencies, the yields of pike and pikeperch significantly decreased in the following two years. By comparison, significantly higher catches of pike, perch, asp, burbot, and ide were recorded in the three years following periods of sustained higher base flows. Our findings suggest that habitat bottleneck effects on fish assemblages have an important role in influencing fish community structure and do require further study. Information on habitat bottlenecks will contribute to the setting of ecological flow requirements and sustainable fisheries in large rivers systems. Copyright # 2005 John Wiley & Sons, Ltd. key words: fish assemblage; large river; lowland river; habitat bottleneck concept; minimum flow; fisheries statistics; time series INTRODUCTION Habitat loss is a major threat to biodiversity (Tilman et al., 1994; Pimm et al., 1995; Casagrandi and Gatto, 1999; Pimm and Raven, 2000), and to fish diversity in particular (Bruton, 1995; Wilcove et al., 1998; Harrison and Stiassny, 1999; Malmqvist and Rundle, 2002). While total habitat loss will result in species extinction (Eldredge, 1994; Casagrandi and Gatto, 1999) the temporary unavailability of essential habitats may significantly affect population dynamics and metapopulation structure (Poff and Allan, 1995; Valentin et al., 1996; Matthews, 1998; Duncan and Lockwood, 2001; Matthews and Marsh-Matthews, 2003). The habitat bottleneck concept was introduced for the first time to evaluate direct impacts of inland navigation on fish (Wolter and Arlinghaus, 2003). Restricted availability or lack of essential nursery habitats due to navigation-induced physical forces ( ¼ habitat bottleneck) significantly impact on fish recruitment in waterways, and therefore may become a main structuring factor for fish assemblages therein. Empirical indications supporting the habitat bottleneck concept were derived from studies of juvenile fish recruitment in a navigation canal (Arlinghaus et al., 2002), of navigation-induced fish displacements and mortality (Holland, 1987; Adams et al., 1999; Killgore et al., 2001; Wolter and Arlinghaus, 2003), and of hydrodynamic impacts on fish due to hydropeaking (Valentin et al., 1996; Liebig et al., 1999; Saltveit et al., 2001). Natural river corridors and floodplains are disturbance-dominated systems and are recognized as areas of physical, chemical and biological interactions between aquatic and terrestrial habitats resulting in high diversity of environmental processes (Tockner and Stanford, 2002; Ward et al., 2002). In river ecosystems, floods and droughts are primary sources of disturbance. However, both may vary in their effects on different fish species and age * Correspondence to: Christian Wolter, Leibniz-Institute of Freshwater Ecology and Inland Fisheries, 12561 Berlin, Germany. E-mail: wolter@igb-berlin.de Copyright # 2005 John Wiley & Sons, Ltd. Accepted 21 July 2004 246 C. WOLTER AND R. MENZEL groups. On the one hand, large floodplains inundated by high floods serve as important feeding areas for fish and nurseries for juveniles (Welcomme, 1979; Poff and Allan, 1995; Schiemer et al., 1995, 2001; Grossman et al., 1998; Jungwirth et al., 2000), while on the other hand fish larvae and juveniles may become washed out because of their low swimming performance (Harvey, 1987; Pearsons et al., 1992; Jackson, 1993; Bischoff and Wolter, 2001). By comparsion, droughts may favour fish species reproducing at low flow conditions (Humphries et al., 1999; King et al., 2003), although the most frequently demonstrated effects of droughts are declines in population because higher resource competition and predation occur with increasing concentration in the remaining water volume (Canton et al., 1984; Matthews, 1998; Labbe and Fausch, 2000; Schlosser et al., 2000; Lake, 2003; Magoulick and Kobza, 2003; Matthews and Marsh-Matthews, 2003). Impacts of floods and droughts are directly related to habitat bottlenecks, i.e. to the restricted availability of complex habitat structures providing shelter from high currents (Pearsons et al., 1992; Jackson, 1993) and structured aquatic habitats (Magoulick and Kobza, 2003; Matthews and Marsh-Matthews, 2003). For example, in temperate, nutrient-rich lowland rivers only emerged macrophytes and woody debris provide shelter and nurseries for juvenile fish, while submerged macrophytes are naturally absent. These habitat structures may temporarily dry if the water level drops substantially below the long-term mean low water level (MNW). In the lower Oder River, Germany, significantly lower young-of-the-year fish (YOY) densities have been observed following a two-week period of very low flows and in completely dewatered macrophyte stands during the rearing season (Bischoff, 2002). The impacts of very low flows on fish may be pronounced (Matthews and Marsh-Matthews, 2003) especially in temperate rivers with relatively stable base flow conditions. However, studies of the long-term impact of short-term hydraulic disturbances on fish assemblages are limited (Matthews and Marsh-Matthews, 2003). This study aims to detect the effects of habitat bottlenecks causing YOY decrease on the adult fish assemblage. Such an exercise generally requires long-term, large-scale studies of population dynamics, firstly, to compensate for the individuals’ growth period to become adults and for bias from immigrations, and secondly, because significant hydraulic disturbances are comparatively rare events. Thirty-seven years of catch statistics of the fisheries cooperative ‘Havel Brandenburg’ covering a fishing area of more than 62 km2 and nearly 150 river-km with large flushed lakes have been used in this study (Figure 1). This was assumed to be sufficiently large (sensu Fausch et al., 2002) to address possible effects of habitat bottlenecks on the fish community of the Havel River, Germany. The study also evaluates the appropriateness of using commercial fisheries statistics to test the habitat bottleneck concept. METHODS Study area and fisheries statistics The Havel River is a large right-bank tributary of the Elbe River with a length of 325 km and a catchment area of 24 297 km2 (Behrendt et al., 1999). The study area in the lower Havel River is 148 km long and extends from lock Spandau in Berlin to its confluence with the Elbe River. This reach is a regulated inland waterway with numerous flushed lakes and lake-like enlargements and incorporates the 6231 ha (1 ha ¼ 10 000 m2) fishing area of the ‘Havel Brandenburg’ fisheries cooperative (Figure 1). Downstream of Berlin the Havel River has a very low slope (0.002–0.014%) and has an average flow velocity of about 0.1 m s 1 at mean discharge (Naumann, 1995). The hydraulic regime of the lower Havel is represented best by the data logged at the water level gauge Rathenow (Figure 1) (BfG, 1992; Naumann, 1995). The flow regime of the lower Havel River at the Rathenow gauge for the period 1951–1985 is characterized as having a minimum discharge (NQ) ¼ 10 m3 s 1, a mean low flow discharge (MNQ) ¼ 25.6 m3 s 1, mean discharge (MQ) ¼ 95 m3 s 1, mean high flow discharge (MHQ) ¼ 165 m3 s 1, and a maximum discharge (HQ) ¼ 232 m3 s 1. Water levels at the Rathenow gauge for the same period were: minimum water level (NW) ¼ 0.64 m, mean low water level (MNW) ¼ 0.79 m, mean water level (MW) ¼ 1.51 m, mean high water level (MHW) ¼ 2.31 m, and maximum water level (HW) ¼ 2.94 m. Bankfull flow at the Rathenow gauge is equivalent to a water level of 1.85 m and a discharge of 170 m3 s 1 and as a result of water diversions in the lower Havel mean floods (MHQ) are reduced to within channel flow pulses (Tockner et al., 2000). The average number of bankfull flows or spates per year (spate frequency) was 21.6 days for the 1951–1985 period. Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 245–255 (2005) HABITAT BOTTLENECKS IN LOWLAND RIVERS 247 Figure 1. The river and canal network of the lower rivers Elbe and Havel downstream of Berlin (grey). The fishing area of the fisheries cooperative ‘Havel Brandenburg’ is marked in black and the location of the water level gauge Rathenow is indicated by an arrow Instream habitat structures relevant to fish are mainly reed and woody debris formed by roots. These habitat structures become completely dewatered if water levels are 0.5 m below the long-term MW, which occur 64 days per year on average. Extreme water levels, water levels below MNW, occur on average for 5.6 days per year – annual extremes in water level are summarized in Table I. The progeny of fish will be most affected if extremely low water levels occur during the rearing period shortly after hatching, when the juveniles depend essentially on shelter and shallow nursery habitats. Later in the year, juveniles may be less influenced when shifting to deeper habitats. Thus, the spawning period for most of the native fish species can be defined as ‘fish relevant’ ranging from March to June (Wolter et al., 1999). Hence, disturbances during this period are critical for fish reproduction. Therefore, additional hydraulic characters were calculated for the relevant fish period only, marked ‘fr’ (Table I). The statistics of the fisheries cooperative ‘Havel Brandenburg’ offered quantitative information on catches of commercially valuable species between 1952 and 1988 (Table II). Non-marketable fish were not recorded and coarse fish not saleable for human consumption were not distinguished and are summarized as feed fish. Fisheries statistics are typically biased because certain proportions of fish were sold unofficially. However, the fisheries cooperative ‘Havel Brandenburg’ use a specific system of catch registration and payment, suppressing this kind of bias in their internal statistics. In the former German Democratic Republic (GDR) all fishery products were subsidized by the government therefore fishermen received higher prices from the official wholesaler than on the public market. To receive these subsidies, all fishermen of the cooperative ‘Havel Brandenburg’ had to deliver their total catch and the cooperative organized its marketing. Thus, each fisherman had a personal incentive to deliver all fish to increase their individual economic benefit. This ensured that the internal fisheries statistics of the cooperative were relatively accurate (Table II). The annual commercial catches were summarized for the entire 6231 ha fishing area and annual fisheries yields (kg ha 1 a 1) were calculated accordingly. Additionally, the total yield was summarized a second time without eel Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 245–255 (2005) Discharge (m3 s 1) Year min max frmin frmax 22 64.1 10.7 10.7 29.5 38.7 27.3 18.5 35.4 45.6 45.6 27.3 10 27.3 11 31.8 12 28 17 11.5 25.5 12.6 24.3 39.1 27.1 14.9 14 24.2 15 16.8 40.4 20 172 22 164 81.2 123 10.7 161 38.5 183 115 171 43.3 126 27.3 117 30.6 151 98.7 201 45.6 192 79.4 207 52.5 178 97.1 227 64.1 137 36.4 127 41 127 41 179 30.2 232 27.5 160 25.5 108 33.5 138 21.7 228 56 167 39.1 216 56.4 225 25.4 152 41.8 139 51 132 26 173 77 269 91.1 220 20.3 171 164 113 116 166 169 126 117 151 201 192 173 178 227 123 127 112 124 150 130 102 110 228 161 216 180 152 122 129 173 212 220 mean frmean 104.3 110.4 73.3 84.0 117.2 95.3 66.0 73.7 104.8 118.9 118.1 110.4 109.0 113.9 81.1 75.9 70.5 87.5 88.7 62.3 81.9 92.3 111.1 112.6 127.5 93.9 78.2 84.7 72.6 90.8 136.8 103.6 111.5 127.7 79.5 81.0 143.7 117.2 86.1 83.4 132.8 136.0 134.5 119.0 151.6 170.5 93.9 84.0 84.7 74.3 91.9 69.5 84.6 77.4 159.3 119.9 138.7 122.1 109.3 88.7 93.1 121.8 152.0 133.4 Observed total duration (number of days) min max frmin frmax mean frmean MNQ 0.76 0.68 0.7 0.76 0.74 0.98 1.02 0.81 1.1 0.74 0.74 0.8 0.88 0.78 0.7 0.78 0.94 0.94 0.78 0.6 0.78 0.63 0.82 0.7 0.82 0.84 0.86 0.87 0.64 0.84 0.93 0.85 0.71 0.72 0.82 0.64 0.68 0.86 0.76 2.26 2.38 1.74 2.61 1.78 2.79 2.55 2.37 2.29 1.8 2.29 2.46 2.36 2.05 1.72 2.16 2.61 2.52 2.62 2.54 2.73 2.23 1.86 1.86 2.47 2.72 2.66 1.63 1.98 2.74 2.34 2.75 2.94 2.18 2.03 2 2.36 2.86 2.68 0.89 0.92 0.84 0.88 0.82 0.98 1.53 0.81 1.26 0.74 0.91 1.69 0.92 0.78 0.81 1.46 0.94 1.24 1 1.44 1.1 0.86 0.9 0.9 0.93 0.94 0.86 0.9 0.7 1 0.93 1.08 0.84 1 1 0.86 1.22 1.28 0.76 2.26 1.97 1.65 2.31 1.76 2.33 2.48 2.36 2.29 1.66 1.7 2.31 2.34 1.85 1.72 2.16 2.58 2.52 2.38 2.42 2.73 1.81 1.86 1.65 1.82 2.15 1.86 1.57 1.68 2.74 2.26 2.68 2.44 2.18 1.75 1.85 2.36 2.6 2.68 1.45 1.25 1.19 1.26 1.31 1.63 1.86 1.58 1.63 1.23 1.37 1.74 1.48 1.26 1.24 1.58 1.73 1.72 1.64 1.67 1.72 1.33 1.27 1.21 1.44 1.48 1.29 1.35 1.49 1.71 1.70 1.89 1.55 1.39 1.38 1.28 1.45 1.82 1.55 1.80 1.48 1.18 1.46 1.20 1.81 2.17 1.68 1.86 1.29 1.29 2.07 1.73 1.42 1.32 1.93 1.93 1.90 1.74 2.14 2.27 1.43 1.36 1.32 1.29 1.50 1.28 1.37 1.32 2.14 1.79 1.97 1.80 1.69 1.41 1.43 1.76 2.04 1.85 3 0 33 14 0 0 0 4 0 0 0 0 8 0 13 0 19 0 30 91 1 20 1 0 0 57 48 3 14 11 0 5 MNW Spate MW-0.5 MW-0.6 frMNQ frMNW frSpate frMW-0.5 frMW-0.6 7 38 39 10 4 0 0 0 0 8 5 0 0 1 7 1 0 0 5 11 1 20 0 12 0 0 0 0 5 0 0 0 19 2 0 14 6 0 2 4 0 0 0 3 1 0 0 0 43 81 54 29 64 0 0 0 7 51 0 0 0 61 0 37 88 0 0 0 3 109 56 92 140 115 153 57 1 0 40 0 114 113 19 29 144 117 24 3 19 63 44 43 90 56 108 57 64 116 36 64 44 4 10 107 90 48 132 60 13 77 41 115 85 103 46 0 0 29 0 75 41 5 8 82 44 7 0 0 37 28 9 64 15 58 14 13 13 14 49 9 0 8 83 40 13 59 25 6 23 3 0 15 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 2 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 0 0 0 0 4 0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 2 1 0 0 0 0 0 0 0 0 30 44 5 29 64 0 0 0 0 0 0 0 0 61 0 37 34 0 0 0 3 59 56 13 12 29 15 39 1 0 16 0 34 21 0 8 26 31 0 3 0 1 0 0 8 13 17 5 8 16 9 25 1 4 0 14 5 3 26 0 0 17 2 0 10 1 34 0 0 9 0 20 0 0 0 17 4 0 0 0 0 0 0 2 1 2 0 0 2 2 19 0 0 0 7 0 0 9 0 0 4 C. WOLTER AND R. MENZEL River Res. Applic. 21: 245–255 (2005) 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 Water level (m) 248 Copyright # 2005 John Wiley & Sons, Ltd. Table I. Annual hydraulic characters of the lower Havel River recorded at the water level gauge Rathenow (MW-x ¼ mean water level minus x m, fr ¼ fish relevant, March– June, MNQ ¼ mean lowest discharge, MNW ¼ mean lowest water level) 249 HABITAT BOTTLENECKS IN LOWLAND RIVERS Table II. Catch statistics 1952–1988 (annual yield in metric tons) of the fisheries cooperative ‘Havel Brandenburg’ Year Common carp Eel 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 0.67 0.72 0.44 1.63 1.55 1.32 1.99 1.54 3.06 0.43 2.23 4.26 1.63 0.82 0.64 1.64 2.12 1.00 0.58 2.80 1.34 1.05 2.18 3.71 2.82 3.36 3.37 9.88 2.71 2.05 2.17 16.51 22.85 7.15 21.69 16.44 14.29 18.92 24.21 27.08 26.26 23.59 25.07 28.76 36.72 31.73 50.07 43.05 57.37 51.14 49.77 63.80 78.03 73.73 88.43 80.50 66.77 75.91 81.00 80.45 105.64 100.84 103.02 102.74 99.11 90.38 104.80 96.61 94.86 84.11 91.91 109.76 82.11 83.55 Tench Pike 14.48 14.44 12.25 14.13 18.95 16.40 12.47 14.35 9.80 8.93 10.55 11.42 12.17 8.64 12.40 15.40 14.55 13.62 12.10 10.74 9.85 9.69 7.06 6.35 5.91 3.90 4.05 5.08 4.86 3.82 3.43 2.83 3.12 3.38 2.90 4.77 4.47 27.61 36.11 29.99 32.18 37.13 50.28 45.62 44.60 34.00 31.03 49.57 51.94 43.31 34.20 37.40 40.65 36.23 34.16 37.85 42.71 36.82 25.85 23.07 20.21 18.43 13.89 12.66 13.77 13.63 19.66 25.22 24.08 22.68 18.76 14.28 19.80 35.13 Pikeperch Wels Burbot Perch Common Roach Ide bream 15.96 23.12 18.23 16.21 17.03 16.40 17.75 22.18 23.25 14.70 13.38 14.59 12.33 10.45 13.30 22.35 29.89 25.67 24.26 32.42 30.16 20.72 19.51 18.09 20.13 18.69 16.01 16.00 15.85 10.88 12.68 12.75 14.19 18.30 20.62 14.79 16.02 0.50 0.70 0.72 0.62 0.99 0.90 1.11 2.33 2.00 2.88 1.83 2.40 2.26 1.23 1.15 0.93 0.73 0.76 0.58 0.78 0.70 0.77 0.59 0.78 0.77 0.46 0.43 0.67 0.39 0.48 0.28 0.32 0.40 0.58 1.43 0.46 0.33 0.04 0.15 0.09 0.14 0.32 1.28 2.01 11.74 0.14 0.18 0.26 0.20 0.21 0.44 1.16 2.29 2.36 1.42 0.99 2.41 2.21 0.57 1.56 0.80 0.74 0.68 0.61 0.62 0.70 0.88 0.98 0.73 0.85 0.30 0.40 1.12 1.84 3.15 3.35 2.34 3.70 4.68 4.44 5.10 4.27 2.80 4.70 5.36 6.39 6.50 3.12 4.09 5.31 3.62 3.19 3.63 4.66 3.73 2.80 3.29 3.17 2.91 2.53 2.03 1.24 2.16 2.35 2.00 1.76 3.88 1.46 2.59 5.38 4.30 29.96 36.19 35.19 30.79 34.54 40.53 44.25 70.82 89.07 77.59 106.53 148.66 238.65 159.16 168.12 94.30 114.54 106.86 106.50 131.80 37.79 147.49 143.93 112.27 150.05 16.65 79.56 105.54 126.95 108.01 168.51 177.80 133.55 138.45 142.03 133.62 173.82 65.98 84.09 68.70 95.55 124.99 115.87 72.90 50.16 42.75 80.02 101.83 78.30 64.80 48.84 62.48 61.31 44.71 27.21 34.53 49.68 36.03 26.59 32.95 45.81 32.86 29.35 25.92 26.48 21.72 17.98 20.30 21.67 19.37 16.44 18.01 23.86 31.66 Asp Feed fish Total fish 0.01 0.17 1.84 0.04 0.14 0.13 0.18 0.01 0.01 0.01 0.03 0.04 0.01 0.09 0.15 0.16 0.02 0.27 0.05 0.75 0.40 0.12 0.64 0.36 0.93 0.02 0.02 0.04 0.17 0.29 1.27 0.10 0.42 0.51 0.52 0.05 0.54 0.01 0.02 0.40 0.28 1.02 1.83 2.48 0.01 0.01 1.29 2.61 2.27 2.13 3.96 5.27 7.30 10.87 12.88 23.17 0.72 18.32 21.99 10.84 1.64 6.18 20.02 10.38 11.00 33.22 50.46 50.96 242.65 46.67 67.95 115.37 175.75 255.86 251.25 147.99 170.66 166.31 232.10 223.09 248.44 206.50 222.43 190.26 203.20 177.85 223.41 195.11 226.57 271.21 285.30 246.25 284.15 239.36 289.66 357.38 386.81 435.77 323.43 386.20 337.80 336.60 336.86 352.69 398.70 482.80 365.24 384.10 433.41 515.31 451.44 501.37 428.21 451.03 438.87 565.72 579.50 561.36 508.16 561.74 504.34 582.86 (Anguilla anguilla (L.)) and common carp (Cyprinus carpio L.), because in the Havel River both species have only a very sporadic natural recruitment and their commercial catches are directly related to stocking intensity. Statistical approaches Time series statistics were calculated from the difference of the annual yields between consecutive years for all species and total fish catch. The proportion of yield differences to the total yields have been used for standardization, because the data set contained a substantial increase in fishing effort between 1960 and 1962. Beginning in 1960, fishermen replaced cotton nets with lighter, durable nylon nets and constructed larger trap nets with elongated exposure time. This development, completed in 1962, resulted in an increased fishing effort. In the following years the number of fishermen, trap nets and boats remained constant. All time series analyses were first tested for autocorrelations using an independence model that calculates standard errors assuming the underlying process is unbiased and, second, using a partial autocorrelation function which removes effects of correlations at the intervening lags (time periods). The first data analysis was performed on the data set from 1962 to 1988—the period after the increase of fishing effort—using the classification algorithms of AnswerTreeTM 2.0 (SPSS Inc., 1998). The CHAID (Chi-squared Automatic Interaction Detector) procedure Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 245–255 (2005) 250 C. WOLTER AND R. MENZEL segmented the data set into homogenous subgroups (F-statistics) according to the total annual fisheries yield, and selected in a second step the best predictor variable to explain this segmentation from all hydraulic characters summarized in Table I. The recursive segmentation procedure C&RT (Classification and Regression Trees) was then used to validate the CHAID results. At each classification level, this binary procedure divided the target variable into two homogenous subgroups, with increasing homogeneity in the new subgroups. The procedure stopped when complete homogeneity was reached. The C&RT segmentation became inefficient when the same predictor was selected twice. The strongest predictors revealed by AnswerTreeTM algorithms were cross-correlated to the various time series of fisheries yields, yield differences and standardized yield differences of the whole data set. Crosscorrelations were used to analyse retarded effects that become evident with more than one lag delay, which is particularly important for time series related to fisheries yields. Possible habitat bottleneck impacts on juveniles may be expressed with a few years’ delay because commercially valuable fish require a certain growth period to reach marketable length. In the study area, average growth periods between three and six years have been observed for the recruitment of commercially important species in the catch: pike (Esox lucius L.) three years; pikeperch (Sander lucioperca (L.)) four; perch (Perca fluviatilis L.) and tench (Tinca tinca (L.)) five; and common bream (Abramis brama (L.)) six years (unpublished results of age determinations). These calculations were performed using the SPSS software package (SPSS Inc., 1999, release 9.0.1). RESULTS Annual fisheries yield increased between 1952 and 1988 (Figure 2) and this was significantly autocorrelated to the yield of the previous year (partial autocorrelation coefficient 0.721  0.16 SE). By comparison, the yield difference to the previous year was significantly autocorrelated to the third lag (partial autocorrelation coefficient 0.355  0.162 SE), i.e. to a yield difference three years ago. The first CHAID classification of annual fisheries yield revealed a tree with a single segmentation level (Figure 3). The homogenous subgroups were best predicted by the number of days with ‘fish relevant’ minimum discharge (frMNQ-Days) explaining 42.61% of the total variability. Low discharges during the spawning season were associated with an improved fisheries yield. The recursive C&RT algorithm explained 90.4% of the total variability (Figure 4) and the first-level segmentation was best predicted by the lowest observed ‘fish relevant’ discharge (frQmin), which improved the homogeneity of subgroups by 48.9%. According to this segmentation, increasing fisheries yields were directly related to low flow conditions during the spawning season—a result similar to that of the CHAID classification. The second-level segmentation improved the homogeneity of subgroups by another 26.7% and was best predicted by MQ. Moderate discharges, slightly below the long-term average MQ, Figure 2. Development of total annual fisheries yield (1) in the study area between 1952 and 1988 and the variability of commercial catches expressed as difference of annual yield to the previous year (2) Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 245–255 (2005) HABITAT BOTTLENECKS IN LOWLAND RIVERS 251 Figure 3. CHAID classification tree of homogenous subgroups according to the total annual yield of the years 1962–1988 predicted by the number of days with minimum discharge during the rearing season of fish (frMNQ-Days). N is the number of years Figure 4. Most consensus C&RT classification tree of homogenous subgroups according to the total annual yield of the years 1962–1988. The significant predictors were frQmin ¼ lowest discharge during the rearing season of fish, MQ ¼ annual mean discharge, MW-0.5-Days ¼ number of days with water levels at least 0.5 m below the long-term MW. N ¼ number of years were associated with increased fish yield compared to those in a period with higher flow pulses. The third segmentation level was predicted by a possible habitat bottleneck, the number of days with completely dewatered instream habitat structures (MW-0.5-Days). Long-term dewatering significantly impacted on the fisheries yield. Two further segmentations predicted by the number of days with highest discharge (HQ-Days) and again by frMNQ-Days were rejected by a cross-validation using 25 randomly selected subgroups as not significant. Cross-correlations were analysed between the discharge parameters predicted by the classification algorithms (frMNQ-Days, frQmin, MQ, MW-0.5-Days, frMW-0.5-Days, frQmax, HQ-Days, frSpate) and fisheries statistics. Parameters of possible flood disturbance (frQmax, HQ-Days, frSpate) were not at all significantly correlated with fisheries yield, and only two of the drought characters (frQmin, frMW-0.5-Days) (Figures 5, 6). Habitat bottlenecks due to dewatered instream habitat structures resulted in significantly decreasing yields of pike and pikeperch in the next one and two years respectively (Figure 5). However, in the respective years with very low discharges the fisheries yields typically increased (Figures 3, 4). In contrast, a hydraulic year with a higher base flow indicated by Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 245–255 (2005) 252 C. WOLTER AND R. MENZEL Figure 5. Cross-correlation function between the frequency (number of days) of fish relevant water levels more than 0.5 m below the long-term average, considered as habitat bottlenecks, and the annual yield of pikeperch (Sl, above) and pike (El, below). Thin lines mark the confidence limits. Lag numbers correspond to years high frQmin values resulted in significantly increased catches of pike, perch, asp (Aspius aspius (L.)), burbot (Lota lota (L.)), and ide (Leuciscus idus (L.)) within the next one to three years (Figure 6). The most common fish species in the study area, common bream (Abramis brama (L.)) and roach (Rutlus rutilus (L.)), were neither significantly impacted by very low flows nor by high water levels. DISCUSSION Impacts of extreme floods and droughts on fish larvae and juveniles are well documented (Canton et al., 1984; Harvey, 1987; Pearsons et al., 1992; Jackson, 1993; Labbe and Fausch, 2000; Schlosser et al., 2000; Bischoff and Wolter, 2001; Magoulick and Kobza, 2003; Matthews and Marsh-Matthews, 2003). However, testing of the habitat bottleneck concept with respect to its effects on the level of fish assemblages has been limited. This partly results from inappropriate analysis of fisheries statistics. In contrast to the significant impacts of extended periods of drought on fish production (e.g. Welcomme, 1986; Laë, 1995), possible effects of short-term disturbances by droughts or floods have been insufficiently reflected in catch statistics. Strong autocorrelations of total fisheries yields have resulted mainly from the controlled annual increases in fish production, by raising feed fish yields. Small common bream and roach contribute substantially to feed fish, thus, questions remained unresolved as to whether both species were affected by hydraulic disturbances or simply reflected systematic increased market forces. Empirical evidence of habitat bottlenecks for juvenile fish recruitment during low flows was derived from a YOY survey in the lower Oder River, where Bischoff (2002) observed significantly lower juvenile fish recruitment associated with very low flows in June 1998. Low flows as habitat bottlenecks seemingly contradict the low flow recruitment hypotheses by Humphries et al. (1999). Results presented here suggest both Australian and European Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 245–255 (2005) HABITAT BOTTLENECKS IN LOWLAND RIVERS 253 Figure 6. Cross-correlation function between fish relevant minimum discharge (frQmin) and the annual yield of pike (El), perch (Pf), roach (Rr), and other species (burbot, ide and asp). Thin lines mark the confidence limits. Lag numbers correspond to years rivers may contain species more adapted to high and low flows respectively, with the latter rather benefiting from droughts (Matthews, 1998; Puckridge et al., 1998; Humphries et al., 1999; King et al., 2003; Matthews and MarshMatthews, 2003). Possible impacts strongly depend on timing and duration of the hydraulic disturbance (Humphries et al., 1999; Bischoff, 2002). In the Havel River, significant recruitment limitations have been observed for pike and pikeperch, both top predatory fish, caused by extremely low water levels. Surprisingly, these limitations were expressed in the catches one year earlier than expected according to their average recruitment time (Figure 5). Thus, it seems possible that extremely low water levels impacted one-year-old specimens (in their second year after a first overwintering, 1þ fish) much more than YOY. Unfortunately, in large European rivers the survival of 1þ fish has not been studied in detail because YOY were commonly considered as the more sensitive indicators of ecological integrity of large rivers (e.g. Jungwirth et al., 2000; Bischoff, 2002). YOY constitute a significant proportion to the diet of 1þ pike (Hart and Connellan, 1984) and 1þ pikeperch (Frankiewicz et al., 1999). If extremely low water levels cause significant lower YOY densities, as Bischoff (2002) has shown, a cascading effect and food shortage for the 1þ predatory fish cannot be excluded. However, further experiments have to be performed to verify this. In contrast to the impacts of very low water levels, higher instream flows are considered beneficial for fish (Tockner et al., 2000; Schiemer et al., 2001; Tockner and Stanford, 2002). Higher within-bank water levels are directly related to improved habitat availability and ecotone connectivity (Schiemer et al., 1995, 2001; Ward et al., 1999; Jungwirth et al., 2000; Tockner et al., 2000; Tockner and Stanford, 2002; Pusey and Arthington, 2003). This has been demonstrated to a certain degree in this study (Figure 6). In particular pike and the typical river fishes ide, asp and burbot benefited from wet years with higher base flow and higher minimum discharge. These findings indicate that a higher minimum discharge would sustain fish diversity as well as fisheries. In regulated rivers this ‘ecological flow’ could be comparably easy to manage. In summary, the evaluation of the effects of habitat bottlenecks on the fish assemblage level requires further investigation. In particular the survival of 1þ fish and possible recruitment deficits of YOY have to be addressed Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 245–255 (2005) 254 C. WOLTER AND R. MENZEL in further studies. Furthermore, analysing fisheries statistics in consideration of environmental variability may prevent overfishing and serve to develop sustainable fisheries yields. ACKNOWLEDGEMENTS We thank Craig Boys and Martin Thoms (CRC for Freshwater Ecology, University of Canberra), whose profound comments on earlier drafts improved this paper substantially. REFERENCES Adams SR, Keevin TM, Killgore KJ, Hoover JJ. 1999. Stranding potential of young fishes subjected to simulated vessel-induced drawdown. Transactions of the American Fisheries Society 128: 1230–1234. Arlinghaus R, Engelhardt C, Sukhodolov A, Wolter C. 2002. Fish recruitment in a canal with intensive navigation: implications for ecosystem management. Journal of Fish Biology 61: 1386–1402. Behrendt H, Huber P, Opitz D, Schmoll O, Scholz G, Uebe R. 1999. Nährstoffbilanzierung der Flußgebiete Deutschlands. Texte des Umweltbundesamtes 75: 1–288. BfG—Bundesanstalt für Gewässerkunde (ed.). 1992. Gewässerkundliches Jahrbuch der Deutschen Demokratischen Republik. Abflußjahr 1989. BfG: Berlin. Bischoff A. 2002. Juvenile fish recruitment in the large lowland river Oder: Assessing the role of physical factors and habitat availability. Aachen: Shaker. Bischoff A, Wolter C. 2001. The flood of the century on the River Oder: effects on the 0þ fish community and implications for flood plain restoration. Regulated Rivers: Research & Management 17: 171–190. Bruton MN. 1995. Have fishes had their chips? The dilemma of threatened fishes. Environmental Biology of Fishes 43: 1–27. Canton SP, Cline LD, Short RA, Ward JV. 1984. The macroinvertebrates and fish of a Colorado stream during a period of fluctuating discharge. Freshwater Biology 14: 311–316. Casagrandi R, Gatto M. 1999. A mesoscale approach to extinction risk in fragmented habitats. Nature 400: 560–562. Duncan JR, Lockwood JL. 2001. Extinction in a field of bullets: a search for causes in the decline of the world’s freshwater fishes. Biological Conservation 102: 97–105. Eldredge N. 1994. The Miner’s Canary: Unraveling the Mysteries of Extinction. Princeton University Press: Princeton. Fausch KD, Torgersen CE, Baxter CV, Li HW. 2002. Landscapes to riverscapes: bridging the gap between research and conservation of stream fishes. BioScience 52: 483–498. Frankiewicz P, Dabrowski K, Martyniak A, Zalewski M. 1999. Cannibalism as a regulatory force of pikeperch, Stizostedion lucioperca (L.), population dynamics in the lowland Sulejow reservoir (Central Poland). Hydrobiologia 408/409: 47–55. Grossman GD, Ratajczak RE Jr, Crawford M, Freeman MC. 1998. Assemblage organization in stream fishes: effects of environmental variation and interspecific interactions. Ecological Monographs 68: 395–420. Harrison IJ, Stiassny MLJ. 1999. The quiet crisis: a preliminary listing of the freshwater fishes of the world that are extinct or ‘missing in action’. In Extinctions in Near Time, MacPhee RDE (ed.). Kluwer Academic/Plenum Publishers: New York; 271–331. Hart PJB, Connellan B. 1984. Cost of prey capture, growth rate and ration size in pike, Esox lucius L., as function of prey weight. Journal of Fish Biology 25: 279–291. Harvey BC. 1987. Susceptibility of young-of-the-year fishes to downstream displacement by flooding. Transactions of the American Fisheries Society 116: 851–855. Holland LE. 1987. Effect of brief navigation-related dewaterings on fish eggs and larvae. North American Journal of Fisheries Management 7: 145–147. Humphries P, King AJ, Koehn JD. 1999. Fish, flows and flood plains: links between freshwater fishes and their environment in the MurrayDarling River system, Australia. Environmental Biology of Fishes 56: 129–151. Jackson DC. 1993. Floodplain river fish stock responses to elevated hydrological regimes in unimpacted stream reaches and stream reaches impacted by clearing, dredging and snagging. Polish Archive of Hydrobiology 40: 77–85. Jungwirth M, Muhar S, Schmutz S. 2000. Fundamentals of fish ecological integrity and their relation to the extended serial discontinuity concept. Hydrobiologia 422/423: 85–97. Killgore KJ, Maynord ST, Chan MD, Morgan II RP. 2001. Evaluation of propeller-induced mortality on early life stages of selected fish species. North American Journal of Fisheries Management 21: 947–955. King AJ, Humphries P, Lake PS. 2003. Fish recruitment on floodplains: the roles of patterns of flooding and life history characteristics. Canadian Journal of Fisheries and Aquatic Sciences 60: 773–786. Labbe TR, Fausch KD. 2000. Dynamics of intermittent stream habitat regulate persistence of a threatened fish at multiple scales. Ecological Applications 10: 1774–1791. Laë R. 1995. Climatic and anthropogenic effects on fish diversity and fish yields in the Central Delta of the Niger River. Aquatic Living Resources 8: 43–58. Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 245–255 (2005) HABITAT BOTTLENECKS IN LOWLAND RIVERS 255 Lake PS. 2003. Ecological effects of perturbation by drought in flowing waters. Freshwater Biology 48: 1161–1172. Liebig H, Cereghino R, Lim P, Belaud A, Lek S. 1999. Impact of hydropeaking on the abundance of juvenile brown trout in a Pyrenean stream. Archiv für Hydrobiologie 144: 439–454. Magoulick DD, Kobza RM. 2003. The role of refugia for fishes during drought: a review and synthesis. Freshwater Biology 48: 1186–1198. Malmqvist B, Rundle S. 2002. Threats to the running water ecosystems of the world. Environmental Conservation 29: 134–153. Matthews WJ. 1998. Patterns in Freshwater Fish Ecology. Chapman & Hall: New York. Matthews WJ, Marsh-Matthews E. 2003. Effects of drought on fish across axes of space, time and ecological complexity. Freshwater Biology 48: 1232–1253. Naumann A. 1995. Hydrographische und hydrologische Charakteristik. In Die Havel, Landesumweltamt Brandenburg (ed.). Studien und Tagungsberichte 8: 11–14. Pearsons TN, Li HW, Lamberti GA. 1992. Influence of habitat complexity on restistance to flooding and resilience of stream fish assemblages. Transactions of the American Fisheries Society 121: 427–436. Pimm SL, Raven P. 2000. Extinction by numbers. Nature 403: 843–845. Pimm SL, Russell GJ, Gittleman JL, Brooks TM. 1995. The future of biodiversity. Science 269: 347–350. Poff NL, Allan JD. 1995. Functional organization of stream fish assemblages in relation to hydrological variability. Ecology 76: 606–627. Puckridge JT, Sheldon F, Walker KF, Boulton AJ. 1998. Flow variability and the ecology of large rivers. Marine and Freshwater Research 49: 55–72. Pusey BJ, Arthington AH. 2003. Importance of the riparian zone to the conservation and management of freshwater fish: a review. Marine and Freshwater Research 54: 1–16. Saltveit SJ, Halleraker JH, Arnekleiv JV, Harby A. 2001. Field experiments on stranding in juvenile Atlantic salmon (Salmo salar) and brown trout (Salmo trutta) during rapid flow decreases caused by hydropeaking. Regulated Rivers: Research & Management 17: 609–622. Schiemer F, Zalewski M, Thorpe JE. 1995. Land/Inland water ecotones: intermediate habitats critical for conservation. Hydrobiologia 303: 259–264. Schiemer F, Keckeis H, Reckendorfer W, Winkler G. 2001. The ‘‘inshore retention concept’’ and its significance for large rivers. Archiv für Hydrobiologie Suppl. 135: 509–516. Schlosser IJ, Johnson JD, Knotek WL, Lapinska M. 2000. Climate variability and size-structured interactions among juvenile fish along a lakestream gradient. Ecology 81: 1046–1057. Tilman D, May RM, Lehman CL, Nowak MA. 1994. Habitat destruction and the extinction debt. Nature 371: 65–66. Tockner K, Stanford JA. 2002. River flood plains: present state and future trends. Environmental Conservation 29: 308–330. Tockner K, Malard F, Ward JV. 2000. An extension of the flood pulse concept. Hydrological Processes 14: 2861–2883. Valentin S, Lauters F, Sabaton C, Breil P, Souchon Y. 1996. Modelling temporal variations of physical habitat for brown trout (Salmo trutta) in hydropeaking conditions. Regulated Rivers: Research & Management 12: 317–330. Ward JV, Tockner K, Schiemer F. 1999. Biodiversity of floodplain river ecosystems: ecotones and connectivity. Regulated Rivers: Research & Management 15: 125–139. Ward JV, Robinson CT, Tockner K. 2002. Applicability of ecological theory to riverine ecosystems. Verhandlungen der Internationalen Vereinigung für Theoretische und Angewandte Limnologie 28: 443–450. Welcomme RL. 1979. Fisheries Ecology of Floodplain Rivers. Longman: London. Welcomme RL. 1986. The effects of the Sahelian drought in the fishery of the Central Delta of the Niger River. Aquaculture and Fisheries Management 17: 147–154. Wilcove DS, Rothstein D, Dubow J, Phillips A, Losos E. 1998. Quantifying threats to imperiled species in the United States. BioScience 48: 607–615. Wolter C, Arlinghaus R. 2003. Navigation impacts on freshwater fish assemblages: the ecological relevance of swimming performance. Reviews in Fish Biology and Fisheries 13: 63–89. Wolter C, Bischoff A, Tautenhahn M, Vilcinskas A. 1999. Die Fischfauna des unteren Odertales: Arteninventar, Abundanzen, Bestandsentwicklung und fischökologische Bedeutung der Polderflächen. In Das Untere Odertal. Auswirkungen der periodischen Überschwemmungen auf Biozönosen und Arten, Dohle W, Bornkamm R, Weigmann G (eds). Limnologie aktuell 9. Schweizerbart: Stuttgart; 369–386. Copyright # 2005 John Wiley & Sons, Ltd. River Res. Applic. 21: 245–255 (2005)