Intermittent streams that cease to flow during dry periods represent more than half of the global river network and are particularly common in arid and semi-arid regions. They are characterized by high spatial and temporal variability in aquatic habitat, forming a shifting habitat mosaic that supports diverse assemblages of native and endemic species. Climate projections for much of the world predict greater variation in precipitation and increasing drought severity, suggesting a need to better understand species’ responses to habitat variability within intermittent stream ecosystems. Here we explored changes in the distribution and abundance of aquatic vertebrates in response to a wide range of annual hydrologic conditions within a Mediterranean-climate intermittent stream in California. We conducted wetted habitat and vertebrate abundance surveys during the dry season for seven years, spanning both extreme wet and extreme dry annual precipitation conditions, and characterized the variation in the size, persistence, and spatial configuration of wetted habitats in relation to the observed abundance and composition of fishes, amphibians, and aquatic reptiles. We characterized the habitat mosaic using a k-means clustering approach and identified three habitat types with distinct size, persistence, and connectivity distributions. We found that some wetted habitat units persisted across all years, regardless of antecedent precipitation, whereas others dried in all but the wettest years. We determined that persistent pools, a subset of wetted habitats present in the stream, support diverse assemblages of native species even during extreme dry conditions, while transient pools act as important habitat for particular species and life stages, including a young-of-year minnow species. Linear mixed models indicated that species’ abundances were influenced by habitat quantity, quality, and connectivity. Our results suggest that intermittent streams can be characterized as shifting habitat mosaics that support unique species and life stages across space and time and play an important role in maintaining regional aquatic biodiversity.

Dataset Collection

Wetted Channel Surveys

We conducted surveys of wetted habitat conditions from 2014 to 2021 (except 2020, due to Covid-19 restrictions) along the 5-kilometer study reach in Coyote Creek (Figure 1b). In 2014, we surveyed at the end of the dry season (i.e., September; Bogan et al. 2019). Beginning in 2015, we expanded the study to include monthly surveys from the onset of disconnection (typically July or August) through the end of the dry season (i.e., until the first autumn rainfall, typically October or November). The timespan within which surveys were conducted encompassed extreme low (2021: 263 mm) and high (2017: 1011 mm) precipitation years, including the final years of a statewide multi-year drought between 2012-2016, capturing the wide range of interannual hydrologic variability observed in the system.

Between 2014-2016, monthly wet-dry mapping surveys were conducted by field crews walking the channel with a handheld Garmin eTrex GPS unit (+/- 15 m accuracy) and recording the coordinates at every point of surface water disconnection and reconnection along the channel. Disconnected reaches were defined as portions of the stream channel where there was a break in continuous surface water connectivity, with surface water patches being at least 1 m in length and > 0.10 meters deep. These coordinates were then used to digitize a streamline in ArcGIS 10.7.1 representing wet and dry reaches at each survey. Between 2017-2021, surveys were conducted by field crews using a Bad Elf GPS (+/- 1 m accuracy) connected to ArcCollector 21.0.4 on a tablet, using the same methods to map the presence of surface water. This latter method automatically digitizes and georeferences the wet and dry reaches along the stream. We used these data to create monthly maps of wetted habitat conditions across all years of the study.

Habitat and Biological Surveys

Along our 5-kilometer study stretch where we conducted wet-dry mapping, we identified five representative 250-meter reaches for more intensive habitat and biological sampling (Figure 1b). We conducted habitat and biological surveys for each wetted habitat unit within each 250-meter reach. A wetted habitat unit was defined as any disconnected surface water patch, which included both flowing reaches and isolated pools, with an area greater than 30 cm2. This size threshold was chosen as habitat patches smaller than 30 cm2 were unlikely to support aquatic vertebrate species. For each wetted habitat unit, presence and abundances of vertebrates were estimated using a rapid assessment approach, in which species abundance was estimated by visually scanning the wetted habitat unit for amphibians, reptiles, and fishes in the water column (see the complete taxa list on Table 1). Water clarity in Coyote Creek is generally high during the dry season, and species can be identified at depths of 2 m (Bogan et al., 2019). We also documented abundances of California floater mussels (Anodonta californiensis) because they are large (up to 12cm), have a life stage dependent upon freshwater fishes, and are of conservation concern. In addition to the visual assessment, we conducted spot sampling in undercut areas and along the benthos using a dip-net to identify benthic-oriented vertebrates. Each species observed was grouped into one of five abundance classes for analyses based on visual estimation: 1-10; 11-100; 101-500; 501-1000; >1000 individuals. Visual estimation of abundance was completed by two observers (typically RAL and SMC). We distinguished between juvenile and adult life stages for some amphibians (Rana draytonii, red-legged frog; Rana boylii, yellow-legged frog) and fish (Hesperoleucus venustus, coastal roach; Catostomus occidentalis, Sacramento sucker; Ptychocheilus grandis, Sacramento pikeminnow). To standardize sampling effort among the units, the time spent sampling was proportional to the surface area of the habitat unit.

For each wetted habitat unit, we collected and compiled a suite of metrics that relate to habitat suitability, including measures of quantity, quality, and connectivity. We measured the maximum length, maximum width, and maximum depth of each wetted habitat unit as habitat quantity metrics. We also calculated a series of connectivity measures for each habitat unit. These include distance to nearest patch, which is the Euclidean distance from the focal habitat unit to the nearest habitat unit, and distance to permanent pool, which is the Euclidean distance from the focal unit to the nearest permanent pool, which are a subset of pools that persisted during the driest conditions surveyed. Distance-weighted area was calculated as the distance to the nearest wetted habitat unit, divided by the area of the focal habitat unit. To characterize the long-term wetted persistence of the focal unit, we calculated the proportion of years in which the focal habitat unit had surface water in the month of September. Beginning in 2017, we additionally measured several physicochemical parameters, including water temperature, dissolved oxygen, pH, and conductivity at each wetted habitat unit.