HYDROLOGICAL PROCESSES Hydrol. Process. 24, 2269–2276 (2010) Published online 24 March 2010 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/hyp.7615 Can the distribution of headwater stream chemistry be predicted from downstream observations? J. Temnerud, 1,2 * J. F¨ olster, 2 I. Buffam, 3 H. Laudon, 4 M. Erlandsson 2 and K. Bishop 2 1 Swedish Meteorological and Hydrological Institute, Research Department, Norrk¨ oping, Sweden 2 Department of Aquatic Sciences and Assessment, Swedish University of Agricultural Sciences, Uppsala, Sweden 3 Department of Zoology and Center for Limnology, University of Wisconsin-Madison, Madison, Wisconsin, USA 4 Department of Forest Ecology and Management, Swedish University of Agricultural Sciences, Ume˚ a, Sweden Abstract: Small streams with catchment areas <2 km 2 make up the majority of all stream length and are of great ecological importance. Surveys of first and second order streams reveal great spatial and temporal variability in the water chemistry of these headwaters, but their assessment presents a serious challenge since systematic, representative data are usually only collected in larger streams and rivers. Using low flow synoptic survey data from seven mesoscale Swedish catchments, this study tests the hypothesis that downstream monitoring data can be used to predict key features of the distribution of chemistry in headwater streams [median and interquartile range (IQR)]. Three ecologically relevant analytes were tested: pH, acid neutralizing capacity (ANC) and total organic carbon (TOC). For all seven catchments, the outlets (36–127 km 2 ) were considerably less acid with lower TOC than the median of the headwaters (<2 km 2 , N D 19–45). Among catchments, headwater median and IQR were positively correlated with the value at the outlet, for all three analytes. A univariate general linear model (GLM) was used to predict the headwater chemistry distribution for each catchment from its outlet chemistry, using the relationship established with the other six catchments. Headwater median pH and IQR of ANC were well predicted by a single downstream sample [median adj. R 2 ¾ 0Ð7, normalized root mean squared error (NRMSE) <0Ð7]. Other response variables were not as well predicted, with median adj. R 2 ranging from 0Ð08 to 0Ð48, and NRMSE up to 1Ð1. A minority of models were significant at ˛ D 0Ð05, in part due to the limited availability of catchments with such extensive survey data. However, the clear trends observed suggest that with additional model development, downstream chemistry could ultimately provide a valuable tool for characterizing the range of chemistry in the contributing headwaters. Copyright  2010 John Wiley & Sons, Ltd. KEY WORDS headwater; spatial scale; boreal stream Received 14 September 2009; Accepted 15 January 2010 INTRODUCTION The importance of headwaters as a resource for biodiver- sity and human welfare is increasingly recognized (Lowe and Likens, 2005; Bishop et al., 2008). One of the rea- sons is that headwaters make up most of the watercourse length and hence provide a large proportion of water and solutes to downstream locations (Person et al., 1936; Leopold et al., 1964). In Sweden, for example, streams with catchment size <2 km 2 make up approximately 80% of the total length of all perennial watercourses (Nisell et al., 2007). Studies of the biota in headwaters have also found them to be important for biodiversity, in part because of species endemic to headwaters (Meyer et al., 2007). It is widely known that variability in water qual- ity changes with catchment size, typically with small watercourses showing the highest variability in space (Wolock et al., 1997; Temnerud and Bishop, 2005) and time (Nagorski et al., 2003; Buffam et al., 2007). Sig- nificant efforts (e.g. Hutchins et al., 1999; Smart et al., * Correspondence to: J. Temnerud, Research Department, Hydrological Unit, Swedish Meteorological and Hydrological Institute, Folkborgsv¨ agen 1, SE-60176 Norrk ¨ oping, Sweden. E-mail: Johan.Temnerud@vatten.slu.se 2001; Likens and Buso, 2006) have been made to quan- tify the variability of headwaters. One of the most notable recent studies is the US EPA’s ‘Wadeable Stream Assess- ment’ (WSA) (US European Protection Agency, 2006). The WSA was a statistically valid survey of the biologi- cal condition of small perennial streams at the continent scale based on 1 : 100 000 scale maps. But since many first and second order streams are not found on maps of this scale, headwaters are likely to be considerably underrepresented even in that survey. In spite of these recent efforts, headwater stream ecosystems are not systematically documented, and it is rare to find either the current status or degree of human influence satisfactorily quantified in the multitude of headwaters (Gergel et al., 1999; McGlynn et al., 2004). This observation led Bishop et al. (2008) to describe headwaters as ‘Aqua Incognita’. The absence of systematic coverage of headwaters results in part not only from a bias in environmental monitoring towards larger watercourses, but also from the sheer magnitude and complexity of the problem. These challenges do not make the need to better characterize headwaters any less urgent. For instance, the European Union Water Framework Directive clearly states that all waterbodies should be considered in evaluating the status of the environment. Copyright  2010 John Wiley & Sons, Ltd.