Search

1. Ecosystems are strongly influenced by land use practices. However, identifying the mechanisms behind these influences is complicated by the many potential pathways (often indirect) between land use and ecosystems and by the long-lasting effects of past land use. To support ecosystem restoration and conservation efforts, we need to better understand these indirect and lasting effects. 2. We constructed structural equation models (SEM) to evaluate the direct and indirect effects of contemporary (2002) land use (agriculture and development) and change in land use from 1952 to 2002 on present-day streams ( n = 190) in Maryland, U.S.A. Additional variables examined included site location, system size, altitude, per cent sand in soils, riparian condition, habitat quality, stream water NO-N and benthic macroinvertebrate and fish measures of stream condition. Our first SEM (2002 Land Use) included the proportions of contemporary agriculture and development in catchments in the model. The second SEM (Land Use Change) included five measures of land use change (proportion agricultural in both times, developed in both times, agricultural in 1952 and developed in 2002, forested in 1952 and developed in 2002 and agricultural in 1952 and forested in 2002). 3. The data set fit both SEMs well. The 2002 Land Use model explained 71% of variation in NO-N and 55%, 42% and 38% of variation in riffle quality, macroinvertebrate condition and fish condition, respectively. The Land Use Change model explained similar amounts of variation in NO-N ( R = 0.72), riffle quality ( R = 0.57) and macroinvertebrate condition ( R = 0.44) but slightly more variation in fish condition ( R = 0.43). 4. Both models identified pathways through which landscape variables affect stream responses, including negative direct effects of latitude on macroinvertebrate and fish conditions and positive direct and indirect effects of altitude on NO-N, riffle quality and macroinvertebrate and fish conditions. The 2002 Land Use model showed contemporary development and agriculture had positive total effects on NO-N (both through direct pathways); contemporary development had negative effects on macroinvertebrate condition. The Land Use Change model showed that contemporary developed land that was forested in 1952 had no effects on NO-N; current developed land that was developed or agricultural in 1952 showed positive effects on NO-N. Forests that were agricultural in 1952 had negative effects on NO-N, suggesting reduced NO-N export with reforestation. The Land Use Change model also showed negative total effects of all types of contemporary developed land (developed, agricultural or forested in 1952) on benthic condition. Developed land that was forested in 1952 had negative effects on fish condition. Forest sites that were agricultural in 1952 had negative effects on fish and macroinvertebrate conditions, suggesting a long-term imprint of abandoned agriculture in stream communities. 5. Our analyses (i) identified multiple indirect effects of contemporary land use on streams, (ii) showed that current land uses with different land use histories can exhibit different effects on streams and (iii) demonstrated an imprint of land use lasting >50 years. Knowledge of these indirect and long-term effects of land use will help to conserve and restore streams. [ABSTRACT FROM AUTHOR]