Linking fish assemblages and spatiotemporal thermal heterogeneity in a river-floodplain landscape using high-resolution airborne thermal infrared remote sensing and in-situ measurements

Abstract

River floodplains are heterogeneous landscapes composed of a mosaic of aquatic and terrestrial habitats. While flow has frequently been considered as a master variable that controls aquatic biodiversity and species behavior in river floodplains, little is known about thermal heterogeneity and its effect on aquatic biota in complex landscapes. However, quantifying thermal patch dynamics at the river floodplain scale remains a challenging task. In this study, we applied airborne thermal infrared (TIR) imagery to characterize spatial thermal heterogeneity under high (March 2010) and mean (July 2010) flow conditions in a 347.5 ha lowland river floodplain (Oder River, Germany). Concurrently, we electro-fished all major floodplain water bodies (main channel, side channel, and permanently connected ponds) to identify the composition of associated fish assemblages. In addition, we deployed temperature loggers from March to July across the entire range of floodplain water bodies to assess seasonal temperature patterns at 20 min intervals. Under both flow conditions, the TIR imagery revealed a complex mosaic of thermal patches across the floodplain. Cumulative degree-days and average and maximum temperatures were the main variables that thermally differentiated individual water bodies. The water body types delineated based on spatial and seasonal thermal signatures also contained distinct fish assemblages. Distinct temperature gradients at the floodplain scale and both within and across distinct water body types were resolved with airborne thermal data, providing a fully spatial and concurrently available temperature mosaic. A delay in response of fish distribution to a strong thermal gradient recorded in spring, that has never been predicted or observed elsewhere, could be assessed by the use of TIR imagery. This study demonstrates the potential of airborne remotely sensed TIR imagery as a non-invasive method for detecting and quantifying spatial heterogeneity and ecologically relevant temperature gradients in complex landscapes, thereby facilitating the “up-scaling” of ecosystem processes to the landscape scale and expanding current understanding of fish behavior.

Introduction

High-resolution satellite and airborne remote sensing are rapidly evolving tools for accurately quantifying the structure and dynamics of complex river landscapes, a prerequisite for improving our understanding of ecosystem processes and biodiversity at various scales (Goetz et al., 2008, Johnson and Host, 2010, Marcus and Fonstad, 2008, Mertes, 2002; and references therein). Remote sensing sensors include (i) multispectral imagers (e.g., CASI, GeoEye, Ikonos, QuickBird) for quantifying suspended sediment concentrations, chlorophyll, turbidity, and water depth; (ii) laser scanners (e.g., LIDAR) for the generation of digital topographic maps and for water surface classification and delineation; and (iii) thermal infrared cameras (e.g., FLIR) for measuring surface temperatures.

River floodplains are heterogeneous landscapes composed of a shifting mosaic of interconnected aquatic and terrestrial habitats. The composition, configuration and degree of hydrological connectivity of these habitats determine biodiversity and ecosystem processes (Tockner and Stanford, 2002, Ward, 1998). Moreover, they determine fish assemblage diversity (Schomaker and Wolter, 2011, Van den Brink et al., 1996, Welcomme, 1979), fish reproduction and juvenile recruitment (Bischoff, 2002, Bischoff and Wolter, 2001, Grift et al., 2003), as well as fish production (Welcomme, 1979).

Temperature is a master variable that influences physical, chemical, biological, and ecological processes in aquatic ecosystems (Caissie, 2006, Magnuson et al., 1979, Webb, 1996; and references therein), triggers the dispersal of ectothermic organisms such as fish (Buisson et al., 2008, Tiffan et al., 2009), and determines the behavior and survival of fish (Buisson et al., 2008, McCullough et al., 2009, Pörtner and Farrell, 2008). In complex river floodplain mosaics temperature is inherently highly variable in space and time (Caissie, 2006, Tonolla et al., 2010), and, therefore, difficult to quantify using conventional in-situ methods.

Thermal infrared (TIR) imagery has been successfully used to determine the spatial heterogeneity of stream and river temperatures (Cristea and Burges, 2009, Faux et al., 2001, Torgersen et al., 2001), to identify areas of groundwater-surface water interactions (Deitchman and Loheide, 2009, Loheide and Gorelick, 2006), to determine thermal mixing dynamics and velocity fields (Andrews et al., 2011, Cardenas et al., 2011), to calibrate and validate stream temperature models (Cristea and Burges, 2009, Loheide and Gorelick, 2006), and to monitor the success of restoration projects (Loheide and Gorelick, 2006, Shuman and Ambrose, 2003). Moreover, TIR imagery has been employed to identify warm and cold-water refuges that are critical for the survival of many biota, including fish (Madej et al., 2006, Torgersen et al., 1999, Torgersen et al., 2006), as well as for linking thermal with microbial distribution patterns (Dunckel et al., 2009). However, up to now, most studies have used TIR imagery at the micro- and meso-habitat level, or for mapping the temperature along river corridors. So far, we are only aware of two studies that have used TIR imagery to quantify thermal heterogeneity at the landscape scale (Alpine landscapes: Scherrer & Körner, 2010; braided river floodplains: Tonolla et al., 2010).

The main aim of the present study was to quantify the spatiotemporal heterogeneity of temperature (during two different flow conditions) at the river-floodplain scale by applying aerial TIR imagery together with in-situ temperature recording using conventional loggers. The second aim was to evaluate the potential link between thermal heterogeneity and the structure of fish assemblages at the river floodplain scale. We hypothesized that ectothermic fish strongly respond to high temperatures, especially in spring. It was further hypothesized that given a strong correspondence between in situ measurements and TIR imagery, the latter can be used to predict spatiotemporal heterogeneity in fish assemblage structure.