ࡱ>  KLM` bjbjss .*6, SSS82TU6'h.V:hVFVVVdyeDe$$#hj}zd"d}z}z VV b}z pVV}z.D"V"V yS >@sH~ ""elqueee@eee}z}z}z}z6'6'6',S6'6'6'S      Watershed Scale Aquatic Ecological System Stratification for Maine Streams and Lakes Purpose To develop an aquatic ecological system stratifications at the HUC10 scale that can be used in Maine to evaluate the diversity of aquatic ecosystems within the state and provide quantitative landscape data for future analyses. The resultant framework will be an important tool for resource managers to use in conservation planning, and will yield an aquatic ecological framework within which to summarize and communicate aquatic biological information, identify important data gaps, define reference expectations, allocate effort in surveys, and use in other hypothesis formulation and research design. Background on Aquatic Ecosystem Classification Ecological systems (i.e. ecosystems) may be defined as an interacting assemblage of organisms, their physical environment, and the natural processes that affect them. These assemblages recur across the landscape under similar habitat conditions and ecological processes (Higgins et al. 1998). Identifying aquatic ecosystems requires a classification of stream and lake features into recognizable entities or categories. Although a number of nationally recognized terrestrial community classifications exist, the most accepted being the National Vegetation Classification System, currently there is no national or international standard for classifying aquatic communities or aquatic ecosystems (Grossman et al. 1998). Despite the lack of a national aquatic community classification, a variety of aquatic community classifications have been developed at a variety of spatial scales. The classifications generally fall into two broad categories: 1) taxonomic or bio-ecosystem classifications and 2) environmental or geo-ecosystem classifications (Rowe and Barnes 1994); however some classifications combine aspects of both. Taxonomic or bio-ecosystem classifications emphasize biological data and are most often derived from analysis of patterns in species presence or abundance data. By examining these data, assemblages of species that reoccur together can be identified. Examples of bio-ecosystem or taxonomic aquatic community classifications that exist in the northeast U.S. at statewide scales include the Vermont Aquatic Community Classification (Langdon et al. 1998), the New York Heritage Aquatic Community Classification (Reschke 1990), the Massachusetts Heritage Program Living Waters Project (Stuart et al. 2002) and Pennsylvania Natural Heritage Program Aquatic Classification Project (Nightengale et al. 2003). Taxonomic classifications provide excellent descriptive information regarding aquatic species distributions and assemblage structure. By measuring the presence and abundance of taxa at a given location and time, these classifications emphasize the resident current biota and focus on the biotic expressions (taxa) that have resulted from the unique combination of the physical habitat template and the variety of interacting spatial, temporal, biotic, and anthropogenic processes that have operated at the site over time. Biologists and managers often find taxonomic classifications easy to understand and useful in management, such as in biomonitoring, as these classifications depend upon readily identifiable biological entities that can be sampled and monitored at sites. However, taxonomic based classifications have been criticized because research has shown that classifications using strictly biological data or data about one type of organism (such as fishes) rarely represent the complexity inherent in aquatic communities (Higgins et al. 1998, Langdon et al. 1998). For example, taxonomic classifications developed for one group of organisms (e.g. fishes) may not match the distributional and associative patterns in another group of aquatic of organisms (e.g. macroinvertebrates) making it difficult to define a unified aquatic ecosystem classification (Langdon et al. 1998). Stream systems are also extremely naturally dynamic and their biological species composition can vary widely seasonally and over short temporal scales due to changes in environmental or anthropogenic factors. Existing biological classifications of stream communities are also almost always based on data collected from wadable streams which biases their representation of ecological diversity in terms of stream size, gradient, and scale. In addition, historic distribution and abundance data are rarely taken into account, and the future evolutionary potential created by the underlying physical environmental diversity and available dispersal pathways are usually not considered. Taxonomic classifications are also fundamentally impossible or extremely impractical to apply across large areas when no sampled biological data is available for many streams or lakes. Environmental classifications emphasize a stream or lakes relationship to its physical environment and the physical habitat template these features provide. Many aquatic species are limited or show strong correlation in their distributions and assemblage associations to certain physical stream or lake conditions such as water chemistry parameters (ph, alkalinity, conductivity, turbidity, tannins etc.); water temperature; certain physical habitat characteristics of the channel such as form, gradient, and benthic substrate; and the timing, magnitude, and duration of peak, mean, and maximum flow. A variety of landscape and local environmental variables can be used as surrogates in environmental classification to model these ecological processes and habitat attributes (Maxwell et al. 1995, Norton 1982, Higgens et al.1998). These variables often include measures of climate, physiography, bedrock and surficial geology, channel width, depth, and gradient, bed form, and bank conditions (Maxwell et al.1995, Frissel et al 1986, Rosgen 1994). Using GIS, these physical variables can be measured comprehensively and consistently across large areas to inform environmental classifications. Environmental classifications are designed within a spatial and temporal scale hierarchy (see Table 1). A number of environmental classifications recognize a sequential spatially nested hierarchy of classification units from smallest to largest such as pool/riffle system units, reaches, reach systems, stream systems or subwatersheds, watersheds, subbasins, and subzones (Maxwell et al.1995, Frissel et al 1986, Higgens et al 1998). At any point in the hierarchy, the potential capacity or development of a smaller scale system develops within the constraints set by the surrounding larger scale systems. For example, geology and climate factors associated with very large scale subbasins and subzones constrain the development of both physical habitat and biological structure of smaller scale aquatic ecosystems through their large-scale controls on chemistry, hydrology, and sediment delivery (Hawkins et al 2000). The temporal scale or time during which a given type of ecosystem at a given spatial scale is thought to continuously persist will also vary. Smaller spatial levels of aquatic systems, such as a reachs arrangement of pools and riffles, are much more temporally dynamic than larger scale watershed scale systems that are often only significantly altered after major geologic and climate processes that occur over much longer time frames. At any spatial or temporal scale, the variables selected for classification should be those physical entities that are most general, invariant, and causal for the given frame of time and space (Warren 1979, Warren and Liss 1984, Frissel et al 1986). Classifications utility to planning, reporting, and hypothesis generation should also be limited to the appropriate spatial and temporal scale at which they were derived. See Table 1 by the United States Forest Service for a summary of the assessment units, physical classification variables, associated disturbance patterns, biotic processes, and temporal variability suggested for use in different spatial scales of aquatic environmental classification. Table  SEQ Table \* ARABIC 1: USFS Hierarchical Framework of Aquatic Ecological Units with Mapping Scale, Defining Attributes, and Processes (Maxwell et al. 1995) Mapping ScaleRiverine PatternsPhysical featuresDisturbance patternBiotic processesApprox. time for change/years1:2,000,000Subzones to SubbasinsBasin boundaries, river networks, regional climate, regional geologyTectonics, glacial cyclesSpeciation/ extinction>10,0001:100,000Watersheds, SubwatershedsWatershed boundaries, stream networks, geomorphology, local climateLocal uplift, folding/faulting, flood cyclesGenetic variation1,000-10,0001:24,000Valley SegmentsValley geomorphology, climatic regime, hydrologic regimeValley filling, channel migration, stream incisionPopulation demographics100-10001:12,000Stream ReachsChannel morphology, bed form, materials, bank conditions, woody debrisPeak flows, Sediment transportPopulation dynamics10-1001:1,000Channel UnitsHabitat features, depth patterns, debris patternsHydrolics, Scour and deposition, bedload sortingBehavior patterns1 - 10 Both taxonomic and environmental classifications provide useful approaches to structuring the continuum of aquatic biodiversity patterns that exist on the landscape. Use of one over the other can depend on the availability of comprehensive taxonomic sample data for the entire study area, the desire to continuously classify every aquatic feature (even those without collection sites), the desire to address unknown/unsampled aquatic biodiversity by sampling physical habitat parameters as surrogates for this biodiversity, and the desire to include the unique ecological and evolutionary context of the systems environmental setting in a structured hierarchical manner. Many classifications are beginning to combine aspects of both taxonomic and environmental classifications (Langdon et al 1998, Van Sickle and Hughes 2000, Oswood et al 2000, Waite et al. 2000, Sandin and Johnson 2000, Rabeni and Doisy 2000, Marchant et al 2000, Feminella 2000, Gerritsen et al 2000, Hawkins and Vinson 2000, Johnson 2000, Pan et al 2000). Methods An environmental classification was performed to group Maines HUC10 watersheds into groups of similar planning units that we will refer to as aquatic ecosystem types. The classification is designed at the meso-scale where watersheds are appropriate classification units and where the classification variables include geology, topographic features, connectivity, and local climate patterns. These factors have been shown to drive aquatic ecological and evolutionary processes at this scale (Maxwell et al. 1995, Higgens et al. 1998, Hawkins and Vinson 2000). The classification relied heavily on GIS data, multivariate statistical analysis (TWINSPAN see below), and expert review and refinement of HUC 10 groups. The unit of analysis, variables, and methods used to develop these datasets are presented below. Analysis Units 175 U.S. Geological Survey 10-digit Hydrologic Unit Classification (HUCs) watersheds intersecting Maine were used. These watersheds were chosen for consistency with other aquatics analyses (e.g., ME DEC, MABP) which also use this scale of watershed as planning units. While HUCs do have limitations as freshwater planning units because many HUCs along large river mainstems are not true direct watersheds (Omernik 2003), HUC10s are used extensively as a standard geographic and hydrologic framework for water resource and land resources planning at the meso-scale within the United States. Classification Variables GIS-derived elevation, bedrock and surficial geology, and landform variables were used in the HUC10 watershed aquatic ecological system classification. These variables were chosen as surrogates to represent the geomorphology and local climate variation. Each variable and its relationship to key aquatic ecosystem processes is described below. Primary Classification Variables: 1. Elevation Zones: Elevation zones correspond to local variation in climatic. Varying seasonal temperature and precipitation patterns are particularly important to aquatic ecosystems due to the associated effects they have on key processes and attributes which structure aquatic communities. For example, differences in forest types (e.g. conifer vs. deciduous) which develop in each elevation zone and alter the type of allochthonous organic input to rivers and the types of macoinvertebrates expected at sites. Temperature and precipitation patterns in different elevation zones also result in altered water temperature and flow regime patterns which can limit certain aquatic species distributions. For example, regional species distribution maps show finescale dace and northern redbelly dace occurring only in watersheds of New England dominated by elevation zones greater than 800ft and most commonly in watershed dominated by land above 1700ft (Master 2003). In Vermonts fish community classification, brook-trout-slimy sculpin communities were found limited to elevations greater than on average 980feet, while brook trout only fish communities are typically found above 1400ft (Langdon et al. 1998). A mix of warm and cold water fish species are found in Vermont below elevations of 1160 ft., with nearly all fish communities below 400ft. having purely warm water fish. Characteristic macroinvertebrate communities have also been found correlated with high, mid, and low elevation zones (Langdon et al. 1998). Although it is hard to define exact elevational ranges for aquatic biological communities, the major vegetation zonations in New England were used as surrogate to represent major climatic zonations. These zones have been noted by ecologists since the turn of the century (Anderson 1999), and were used in The Nature Conservancys aquatic classification work (Anderson and Olivero 2003) (Table 2). Table 2: Elevation thresholds and corresponding forest types. Name Range Example of Characteristic forest Very low 0-800 ft Oak, Pine-Oak, Pine-Hemlock, Spruce, Floodplain Low 800-1700 ft Hemlock-N.Hardwoods, N. Hardwoods Moderate 1700-2500 ft Northern Hardwoods. Spruce-Hardwoods High 2500-4000 ft Spruce-Hardwoods, Spruce-Fir Alpine 4000+ ft Krumholtz, Montane Fir, Alpine communities 2. Geology: Bedrock and surficial geology influence the hydrologic flow regime in aquatic ecosystems through their effect on groundwater vs. surface water contribution, stability of flow, water chemistry, sedimentation, stream substrate composition, and stream morphology. For example, certain bedrocks weather to clay with easily erodable textures, high nutrient and high water holding capacity, while other types weather to coarser and more porous soils that are low in nutrients and water holding capacity (Birkeland 1984). Bedrock chemistry is also critical in aquatic systems as bedrock chemistry influences the acid neutralizing capacity of water in stream and lake ecosystems (Norton 1982). Certain stream macroinvertebrates and macrophyte plant species have been noted as particularly limited by water acid neutralizing capacity (Langdon et al. 1998). Bedrock geology classes for this project were taken from Anderson (1999). Anderson grouped bedrock classes into seven ecologically significant classes due to their texture, resistance, and chemistry properties (Anderson 1999, Table 3). USGS Bedrock geology maps for each of the states in the northeast U.S. were compiled in digital form at a scale of 1:125,000 1:250,000 and reclassified into the seven major bedrock types. The relationship of Andersons seven bedrock types to Nortons four classes of acid neutralizing capacity of the bedrock are also reported (Norton 1982). In Nortons class one bedrock areas of low to no acid neutralizing capacity, acidic precipitation is expected to have widespread effects on aquatic ecosystems. In Nortons class two, medium to low acid neutralizing capacity, we expect effects from acidic precipitation will be restricted to first and second order streams and small lakes. In Norton class three, effects from acidic precipitation will be improbably except for overland run-off effects in areas of frozen ground. In Norton class four, we expect no effect of acid precipitation on aquatic ecosystems (Norton 1982) due to the high acid neutralizing capacity of the underlying substrate. Surficial geology was also used as a classification variable in certain flat areas of New Hampshire because bedrock in these areas was assumed to be deeply buried. For example, deep coarse deltaic or outwash deposits often overlay the bedrock in the flatter pine barrens and sand plains in the northeast and the consolidated bedrock of valleys of pre-glacial lakes may lie under many meters of fine lacustrine or marine clay sediments (Ferree 2003). In these settings, it is the nature of the surficial sediment in terms of texture, compactness, moisture-holding capacity, nutrient availability, and ability to anchor overstory trees in a wind disturbance, that is ecologically relevant; not the nature of the underlying bedrock. Coarse grained (e.g. sand) and fine-grained surficial sediments (e.g. lacustrine or marine clays) for the analysis region were obtained from the 1:1,000,000 scale Quaternary Geology map but the resolution of this data was improved by informing it with our landform map. Our landform layer was compiled at a much finer scale (1:24,000), and we allowed the deep coarse or fine sediments of the USGS dataset to be mapped only on those landforms on which they would naturally be expected to occur such as dry flats. Flats on coarse sediments are assumed to be more permeable to rain due to their larger particle size, providing a more stable groundwater fed hydrology for streams in this sediment. Flats in fine surficial sediments will likely be more flashy as the soils are less permeable due to the fine particle size (Fitzhugh 1999). Table 3: Geology thresholds and corresponding soil and water chemistry attributes. GeologyTypeTextureResistanceAcidityHydrologic FlowNorton Classacidic graniticcoarseHighAcidicunstable stream order 1-3, stable stream order 4-61acidic sedimentary/ metasedimentaryfine to coarsemod (low-high)Acidicstable2acidic shalefineLowAcidicunstable2mod calcareous sedimentary/ metasedimentaryfine to mediumLowNeutralstable3mafic/intermediate graniticmedium to coarseModNeutral to slightly Acidicunstable stream order 1-3, stable stream order 4-63ultramaficFine to mediumModNeutral to calcareousunstable stream order 1-3, stable stream order 4-63calcareous sedimentary/ metasedimentaryfine to mediumLowNeutral to calcareousstable4Note: For crystalline bedrock (acidic granitic, intermediate granitic, ultramafic), the representative area/watershed area drained by 4th and 5th order streams is assumed to be sufficient to include multiple fracture/fault zones that will provide a relatively stable source of groundSurficial deep coarse sedimentcoarseAcidic-neutralstabe5Surficial deep fine sedimentfineAcidic-neutralunstable/flashy6 4. Gradient and Landform: Gradient and landform influence stream morphology, flow velocity, and habitat types due to differences in soil erodability and accumulation, moisture, nutrients, and disturbance history across different landforms. For example, the confined vs. meandering channel morphology of streams differs substantially between mountains and lowland areas due to contrast in the degree of landform controls on channel development and meandering. Lower gradient streams also vary in substrate composition in New England; low gradient streams typically have sand, silt and clay substrates while high gradient streams typically have cobble, boulder, and rock substrates. Changes in stream macroinvertebrate and fish community structure with stream gradient have been noted in a number of biological classifications in the eastern United States (Langdon et al. 1998, Stuart et al. 2003, Reuske 1990). The topographic model developed by M. Anderson and C. Ferree was used to assess the distribution of landforms at the 30m DEM cell size resolution within the analysis area (Ferree 2003). The model has 6 primary units that differentiate further into 17 total landform units. Figure 1: Example of Landforms  The landform model is rooted in an assessment of variation in slope and land position (elevation of a cell in relation to its neighbors). Median slope within watersheds is a variable showing strong correlation to fish species distributions in the eastern United States (Arget et al. 2002). Streams and lakes were included as landforms, however only medium-large streams and lakes were used. Smaller streams were not used in the landform model for two reasons: (1) lack of a consistent stream GIS coverage where the density of digitization is comparable across the state (e.g. some areas of the state were more heavily digitized at the 1:100,000 and 1:24,000 scale making these datasets inconsistent for statewide analysis of patterns in small stream density); and (2) even if we could obtain a consistent small stream coverage, we felt that transforming every 30m pixel under a 1st-3rd order stream into a water landform would eliminate other important landforms and over-represent the stream landform in the landscape given that very few of these small sized streams are actually 30m wide. Because the landform patterns of sideslopes/coves/hills etc. are so closely correlated with the locations of small streams (you can also derive streams from a DEM in a very similar process to how we derived landforms), we felt the most important aquatic landforms to include were the medium and large rivers (>30sq.mi. drainage) and lakes. Medium and large streams were taken from TNCs Lower New England Aquatic Macrohabitat Classification (TNC 2003) which used 1:100,000 US EPA RF3 linework and a 90m DEM to generate the total upstream drainage area of each stream reach. Drainage area was used as a proxy for stream width, depth, and size given the relationship between upstream drainage area and bankfull channel dimensions and flow rate as described for the eastern United States in Dunne and Leopold (1978). Lakes were taken from the 1:100,000 National Hydrography dataset. Analysis Methods Watersheds were grouped into aquatic ecological system groups based on expert review and revision of the clustering patterns that emerged from Two-Way Indicator Species Analysis (TWINSPAN) analysis. TWINSPAN, a multivariate statistical clustering method (TWINSPAN, Hill 1979), is a derisive classification technique where species (e.g. percentages of each input variable in relation to the total watershed area) and sample units (e.g. watersheds) are simultaneously ranked along a dominant gradient. As a derisive classification technique, TWINSPAN repeatedly divides the correspondence analysis ordination space using a single underlying gradient at each cut. At each successive cut, the previous groups are bifurcated into two more additional groups yielding a final output of 1s and 0s signifying the class breaks. Five TWINSPAN analyses were run for the watersheds using the following input variables: 1. Elevation Zones 2. Geology Types 3. Landforms 4. Elevation Zones, Geology Types, and Landforms 5. Elevation Zones and Geology Types The results of the multivariate analysis were refined by expert review because certain ecologically significant patterns could not be deduced from the multivariate analysis results alone. For example, TWINSPAN can evaluate patterns in similarity or difference in the underlying bedrock and surficial geology, elevation, gradient, and landform percentages within the watershed. However, it could not consider connectivity or adjacency patterns nor weight certain individual input variables more highly based on anything other than their areal extent. Because biological data were also not available for statistical testing, it was important to use expert ecological interpretation to hypothesize the most ecologically significant cluster breaks. This allowed expert knowledge of the diversity and key characteristics of aquatic systems, rather than just the diversity of the underlying data drive the final cluster breaks. The resultant Aquatic Ecological Systems most highly correspond to the raw statewide TWINSPAN Level 2 cluster on Elevation and Geology, with additional breaks input to represent Ecological Drainage Units and some unique coastal connected systems. Results and Discussion Expert review of the TWINSPAN clusters yielded 13 major system types. The Hierarchical relationships, distribution, and defining biophysical characteristics of each group are described in the figures and tables below. Figure 2. Hierarchical Relationships of the Watershed Classifications.  SHAPE \* MERGEFORMAT   Table of Aquatic System Types with their Average Percent of Input Elevation, Geology, and Landform Characteristics  EMBED Excel.Sheet.8  Relationship of Aquatic System Types to Existing Biological and Water Quality Datasets No biological stream or lake classification currently exists in Maine, however qualitative associations between the cluster system groups and other existing biological and water quality data can be made. Although we expect a mosaic of stream and lake habitats to occur within any HUC10 scale watershed, certain types or taxa or certain water quality parameters appear to be more significantly different in different watershed system types. Please refer to draft MRPP results analysis by Peter Vaux for more information. Discussion and Conclusions The rich biodiversity supported by freshwater systems in Maine depend on a natural system of interrelated aquatic ecosystems. Defining meso-scale freshwater ecosystems required the use of watersheds as classification units to appropriately represent the interacting mosaic of freshwater systems at occur at this scale. While terrestrial ecoregions and subsections do a good job of accounting for structural and functional differences in freshwater ecosystems, they do not account for important interactions and compositional differences (species composition and genetic composition) which results from isolation of freshwater faunas largely related to historical and contemporary drainage patterns (Pflieger 1971, 1998). Major drainage systems are analogous to islands embedded within the landscape. Watersheds thus define interacting systems and act as the principle evolutionary and distributional constraint for freshwater organisms at multiple scales (Sowa et al. 2004). Thirteen Major HUC10 Watershed Ecological System emerged to represent patterns in freshwater biodiversity within the state. Differences in types at the highest level are driven by major elevational patterns which are correlated with climate, terrestrial vegetational patterns, and local dominant gradient/landform patterns. Finer splits reveal aquatic system types defined by variation in geology, connectivity, local landform/gradient patterns, and ecological drainage unit. Advantages of the GIS based meso-scale classification technique used included allowing a continuous assessment of all HUC10 watersheds in the state, the ability to measure and integrate multiple input data layers that completely tessalated the landscape, and statistical analysis which highlighted major cluster patterns in the underlying data. Limitations of this technique include the inability to include connectivity/adjacency of watersheds in the statistical analyses, difficulty weighting the input data layers and separate vs. combined cluster analysis results, and difficulty dealing with transitional groups and watersheds with a high diversity or strong internal division/gradient in underlying environmental data layers. A number of other watersheds were also large and diverse in their embedded habitats and/or on transitional elevational or geologic zones making them hard to classify into a single HUC10 group. Despite the classifications limitations, the resultant HUC10 systems provide a structured and ecologically rigorous framework within which to assess aquatic biodiversity patterns within the state. Future research will be necessary to determine the strength of association between the HUC10 system types and measurable aquatic species assemblages. Although comprehensive biological data is not available to test these differences, certain generalized relationships between HUC10 types and freshwater fish or macroinvertebrate biological communities can be postulated and certain measurable differences in biological and water quality sample data support these potential differences. Although correlations between biological assemblages and the HUC10 systems can support the uniqueness of the aquatic biodiversity of a HUC10 system, it is important to remember that the meso-scale watershed systems are not designed specifically to correspond to different aquatic species assemblages. Each HUC10 watershed, given the potential embedded diversity of landscape features, can support a wide variety of specific, finer scale species assemblages, natural community types, and habitat associations. The HUC10 analysis presented here is designed to inclusively represent additional facets of aquatic biodiversity such as unknown or unsampled species, repeating patterns of interacting ecosystems, genetic diversity and/or local populations adaptations to the varying larger environmental setting or isolation, and unique suites of interacting ecological and evolutionary processes that may operate over long time-frames at watershed scales. As biodiversity is increasingly recognized at genetic, population, species, community, ecosystem, and landscape levels of organization (U.S. Congress 1987, Noss 1990), meso-scale watershed based classification and conservation planning is increasingly being recognized as an important conservation planning scale necessary to more fully address and represent aquatic biodiversity. Research in landscape ecology and metapopulation biology within the last decade has also particularly placed new emphasis on the importance of habitat heterogeneity and evaluating interactions over larger spatial and temporal scales. For example, the widely recognized new paradigm for aquatic biodiversity conservation of the riverine landscape is rooted in conserving connectivity and interactions over larger watershed scales (Schlosser 1991, 1995a, 1995b, Schlosser and Angemeier 1995). In this model, fish movement is critical in transporting different life stages across landscape scale areas to occupy patches of critical habitat required to fulfill their life history. The model notes the inherently patchy distribution of habitat features in aquatic systems at an intermediate scale and the necessity of fish to interact over and often travel long distances to reach habitat patches required to complete their life history. These life history events can include spawning, feeding, rearing, refugia from disturbance, overwintering areas, or the colonization and recolonization movement necessary maintain metapopulations. Schlosser and Angermeier (1995) also noted importance of the protecting key ecosystem processes that operate at intermediate to large scales such as water volume, flow rate, flow timing/stability, and flooding which creates and maintain the mixture of habitat patches needed and regulates complex interactions between terrestrial and aquatic systems. These key flow processes are expected to vary in type and intensity in different meso-scale aquatic watershed system types due to the controls geomorphology and climate have on these processes (Maxwell et al. 1995). Planning at the HUC10 scale should incorporate many of these larger scale processes and conservation priorities. Conclusion In conclusion, the HUC10 system classifications are suggested as a useful framework within which to make conservation planning decisions. They provide a priori hypotheses regarding how large-scale suites of environmental features directly or indirectly influence the aquatic biota of Maine. Although future testing against biological datasets will improve the ability to distinguish the aquatic biodiversity represented within and across these HUC10 system, the initial Aquatic Ecological System types provide an ecologically rigorous, structured and geographically comprehensive framework within which to consider aquatic biodiversity. Quantitative information on the distribution and abundance of physical landscape attributes of each watershed can also now be easily queried and compared to review given watersheds similarity or difference to each other. Given that a relatively small portion of the total land and water base will be devoted to biodiversity conservation in the near future, using this HUC10 meso-scale watershed ecosystem framework should help managers more efficiently identify aquatic ecosystem types which may be more threatened or underrepresented in conservation areas managed exclusively or primarily for the long-term maintenance of populations of native species and natural ecosystem processes. By focusing on HUC10 watersheds as units of assessment, managers will also become more comfortable considering entire watersheds as the focus for conservation strategies rather than focusing only on species, reaches, or only instream features. Given the dynamic connectedness of aquatic ecosystems at multiple scales and critical terrestrial-aquatic ecosystem linkages and interactions, this approach will hopefully lead to more comprehensive, representative, and ultimately successful aquatic conservation planning. BIBLIOGRAPHY Allen, J. D. 1995. Stream Ecology: Structure and function of running waters. Kluwer Academic Publishers. Dordrecht, The Netherlands. Angermeier, P.L. and Schlosser, I.J. 1995. Conserving aquatic biodiversity: Beyond species and populations. American Fisheries Society Symposium 17: 402-414. Angermeier, P.L, and M.R. Winston. 1998. Local vs. regional influences on local diversity in stream fish communities of Virginia. Ecology 79(3):911-927. Argent, D.G., J.A. Bishop, J.R. Stauffer, Jr., R.F. Carline, W.L. Myers. 2002. Predicting freshwater fish distributions using landsacpe-level variables. Fisheries Research 1411:1-16. Birkeland, P.W, 1984. Soils and Geomorphology. Oxford University Press, New York. Dunne T. and L. Leopold. 1978. Water in Evironmental Planning. Freeman and Company, New York. Fausch, K.D., C.E. Torgensen, C.V. Baxter, H.W. Li. 2002. Landscapes to Riverscapes: bridging the Gap between Research and Conservation of Stream Fishes. BioScience 52(6):483-498. Fels, J. and R. Zobel. 1995. Landscape position and classified landtype mapping for the statewide DRASTIC mapping project. North Carolina State University. Technical Report VEL.95.1 to the North Carolina Department of Environment, Health, and Natural Resources, Division of Environmental Management. Feminella 2000 Ferree, C.F. 2003. Ecological Systems. The Nature Conservancy. Unpublished Report. Frisell, C.A., W.J. Liss, C.E Warren, and M.D. Hurley. 1986. A hierarchical framework for stream habitat classification: viewing streams in a watershed context. Environmental Management 10(2): 199-214 Gerritsen, J., M.T. Barbour, and K. King. 2000. Apples, oranges, and ecoregions: on determining pattern in aquatic assemblages. Journal of the North American Benthological Society 19: 487-496. Grossman, D.H., D. Faber-Langendoen, A.S. Weakley, M. G. Anderson, P. Bourgeron, R. Crawford, K. Goodin, S. Landaal, K. Metzler, K.d. Patterson, M. Pyne, M. Reid, and L. Sneddon. 1998. International classification of ecological communities: terrestrial vegetation of the United States. Volume 1. The National Classification System: development, status, and applications. TNC Arlington, Virginia, USA Hawkins, C.P, R.H. Norris, J. Gerritsen, R.M. Hughes, S.K. Jackson, R.K. Johnson, and R.J. Stevenson. 2000. Evaluation of the use of landscape classifications for the prediction of freshwater biota: synthesis and recommendations. Journal of North American Benthological Society 19(3):541-556. Higgins, J.V., M. Lammert, M.T. Bryer, M.M. DePhilip, and D.H. Grossman. 1998. Freshwater Conservation in the Great Lakes basin: Development and Application of an Aquatic Community Classification Framework. The Nature Conservancy, Great Lakes Program, Chicago, IL. Higgens, J., M. Lammert, M. Bryer, T.FitzHugh, M. DePhilip, P.Gagnon, R.Smith. 2002. Draft Standards for the Freshwater Components of Ecoregional Planing. The Nature Conservancy. Unpublished report. 16pp. Hill, M.O. 1979. A FORTRAN program for arranging multivariate data in an ordered two-way table by classification of the individuals and attributes. Microcomputer Power Version 3.12, Ithaca N.Y. Hunt, D. 2003. St. Lawrence Ecoregional Aquatics Classification. The Nature Conservancy. Unpublished report. Jackson, D.A. and H.H. Harvey. 1989. Biogeographic associations in fish assemblages: Local vs. regional processes. Ecology 70: 1472-1484. Johnson, R.K. 2000. Spatial congruence between ecoregions and littoral macroinvertebrate assemblages. Journal of North American Benthological Society 19:475-486. Johnson, L.B. and S.H. Gage. 1997. Landscape approaches to the analysis of aquatic ecosystems. Freshwater Biology 37:113-132. Johnson, L.B, C Richards, G.E. Host and J.W. Arthur 1997. Landscape influences on water chemistry in Midwestern stream ecosystems. Freshwater Biology 37:193-208. Langdon, R., J. Andrews, K.Cox, Fiske, S., Kamman, and N., Warren, S 1998. A Classification of The Aquatic Communities of Vermont. Prepared by The Aquatic Classification Workgroup for The Nature Conservancy and the Vermont Biodiversity Project. Marchant, R., F. Wells, and P. Newall. 2000. Assessment of an ecoregion approach for classifying macroinvertebrate assemblages from streams in Victoria, Australia. Journal of the North American Benthological Society 19:497-500. Maxwell, J.R., C.J. Edwards, M.E. Jensen, S.J. Paustian, H. Parrott, and D.M. Hill. 1995. A Hierarchical Framework of Aquatic Ecological Units in North America (Neararctic Zone). General Technical Report NC-176. St. Paul, MN: U.S. Department of Agriculture, Forest Service. Norton 1982 Oswood, M.W., J.B. Reynods, J.G. Irons, and A. M. Milner. 2000. Distributions of freshwater fishes in ecoregions and hydroregions of Alaska. Journal of the North American Benthological Socity 19:405-418. Pan, Y, J.R. Stevenson, B.H. Hill, and A.T. Herlihy. 2000. Ecoregions and benthic diatom assemblages in mid-Atlantic Highland streams, USA. Journal of North American Benthological Society 19(3):518-540. Pflieger, W.L. 1989. Aquatic community classification system for Missouri. Jefferson City, MO: Missouri Department of Conservation, Aquatic Series No. 19. Poff, N.L. and Allan J.D. 1995. Functional organization of stream fish assemblages in relation to hydrological variability. Ecology 76:606-627. Rabeni, C. F. and K. E. Doisy. 2000 The correspondence of stream benthic invertebrate communities to regional classification schemes in Missouri. Journal of the North American Benthological Society.19:419-428. Reschke, C. 1990. Ecological Communities of New York State. New York Natural Heritage Program. NYS Department of Environmental Conservation Latham, NY. Richards, C, L.B. Johnson, and G.E. Host. 1996. Landscape-scale influences on stream habitats and biota. Canadian Journal of Fisheries and Aquatic Sciences 53(Suppl. 1)295-311. Richards, C. R.J. Haro, L.B. Johnson, and G.E. Host. 1997. Catchment and reach-scale properties as indicators of macroinvertebrate species traits. Freshwater Biology 37:219-230. Rosgen, D.L. 1994. A classification of natural rivers. Catena 22: 169-99. Sardin, L., and R.K. Johnson. 2000. Ecoregions and benthic macroinvertebrate assemblages of Swedish streams. Journal of North American Benthological Society 19:462-474. Schlosser, I.J. 1991. Stream fish ecology: a landsacpe perspective. BioScience 41(10)704-712. Schlosser, I. J. and P. L. Angermeier. 1995. Spatial variation in demographic processes in lotic fishes: conceptual models, empirical evidence, and implications for conservation. American Fisheries Society Symposium 17:392-401 Schlosser, I. J. 1995. Critical landscape attributes that influence fish population dynamics in headwater streams. Hydrobiologia 303:71-81. Van Sickle, J., and R.M. Hughes 2000. Classification strengths of ecoregions, catchments, and geographic clusters for aquatic vertebrates in Oregon. Journal of the North American Benthological Society 19:370-384. Waite, I.R., A.T. Herlihy, D.P. Larsen, D.J. Klemn. 2000. Comparing strengths of geographic and nongeographic classifications of stream benthic macroinvertebrates in the Mid-Atlantic Highlands, USA. Journal of North American Benthological Society 19(3):429-441. Warren 1979 Warren, C.E., and Liss, W.J. 1983. Systems classification and modeling of watersheds and streams. Unpublished report, Department of Fisheries and Wildlife, Oregon State University, Corvallis, Oregon. 193pp.  HUCs (Hydrologic Unit Code) are drainage basins that are generally referred to by the number of digits in the code the more digits, the finer scale the drainage basin. HUC 10s have been recognized by NH State Agencies as a useful unit for setting conservation priorities. Eighty-one HUC 10 watersheds in New Hampshire were used in this analysis, ranging in size from 52 to 362 square miles, and averaging 152 square miles.      PAGE 1  DATE \@ "MM/dd/yy" 08/23/06  FILLIN \* MERGEFORMAT  Page  PAGE 1 2 11 10 9 8 7 4 1 5 3 6 13 12 DI\bj 1 abc_ӿ}kYk#ht}h k5CJOJQJ^JaJ#ht}hJ5CJOJQJ^JaJ&ht}hJ56CJOJQJ^JaJ#ht}hJ6CJOJQJ^JaJ ht}h bht}hJh h"4[5hKh/ ht}h kjht}h 0JU ht}hF!\hXb hJ5 ht}hJ h/5ht}hJ5ht}hJ5\1\]^_aj a#/L#' d`gd%dd`d $da$gdXb $da$6h#y",./\ Nf*Cgu-7`ir &¾ԷɳЯԷԨШШԡԝ ht}h%% ht}hwBht*hKh\ ht}h% ht}h#hXb h"4[ ht}h2Hhid ht}h%H ht}hK+h# ht}hJ#ht}hJ5CJOJQJ^JaJ#ht}h%H5CJOJQJ^JaJ5&.8{7XZ{ =AFO  t 0!1!""H#J#K#t####o$$$$&&뻴ƭƭʭ򻦻ƟƦ»ٻʻƻƻʻʻ ht}h ht}hK+ ht}h ht}ht* ht}h k ht}hJhXb ht*h\hK ht}h). ht}h bh"4[ ht}h_ ht}hwB ht}h%% ht}h%<&&'$'%''''''''%()(q(r(((t)u)))=*>***}+~+,,%-N-----.<.D.V..//0$0벥zrht}hJ5 ht}h\h\ h\h\ ht}hF!\hKOJQJaJht}hJhht}hJOJQJaJht}hJ5B*aJhphht}h_5aJht}h8p5aJmHnHujht}hJ5UaJht}hJ5aJhr ht}hJ,''r((((((( d$Ifd`((( )O)i))>55555 d$Ifkd$$IfֈT X 6$rb t04 a))))))*5kd$$IfֈT X 6$rb t04 a d$If*0*=*>*G*W**5kd$$IfֈT X 6$rb t04 a d$If*******5kd$$IfֈT X 6$rb t04 a d$If*C+b+v+}+~++5kd$$IfֈT X 6$rb t04 a d$If++++ ,,,5kd$$IfֈT X 6$rb t04 a d$If,,000$0W3X3g3555.7/7Q7>>?J?K????$a$gd $da$  & Fdgdvd $da$gdrdd`$0G0Q0R0[0b0y000000^1e1f1i111111111122Y2s222222222233X3f3g3j3k3ѻѰѩѩѢњ|hYh.B*phhYhYB*phh.hJ5h.h.5 ht}h?a ht}hM ht}hR3hid ht}hmIhY ht}hQhr ht}hJhR3Vh\ ht}h> ht}h h[ ht}h). ht}hF!\,k3n33333333,4.4148494:4?4@4E4I4N4V4W4r4u44444444555555555555566!6*6j666666666ӯת׽׽ ht}hmI ht}h5Gh'h5 h'5 ht}h hYhYhR3Vh' ht}h ht}hMhY ht}hp h[h. ht}h_ ht}hJhYhJB*ph8667-7.7/7Q7T7}77777 888 8'88888888889999#9S9W99999999999:: :H:w:úɴîɢɮɖÖɮɮɢâɴht}h'B*aJhph h'YaJ h haJ h0VaJ hR3VaJ h UaJ haJ hHaJht}h'aJ hKaJht}hJaJht}h5GaJhHhJ6CJaJ ht}h5G ht}hJhR3V ht}hv0w:{::: ;;;I;J;O;P;Q;U;V;[;\;s;;;;;;;<<<<<<<<<======>#>Y>i>k>ًvpj hR3VaJ haJht}hJaJhR3VB*aJhphhpB*aJhphh0VB*aJhphhh'YaJhhhaJhhh<aJhhHB*aJhphh<B*aJhphhB*aJhphht}h'B*aJhphh'YB*aJhph)k>o>>>>>>>>?(?)?I?J?K?]?^?????????????@@@@@3@4@5@F@G@ʺ|tlatlatht}hCJaJhJCJaJh CJaJhR3VCJaJht}hJCJaJ ht}hhJh ht}hJ h h hJ5aJ h 5aJht}hJ5aJ h aJhhMaJht}hMaJht}hJaJ hR3VaJhh<aJ hpaJ h<aJ&??@@4@5@r@s@t@CHN O!O-O5O@OHOXOeO$If` $ !a$$a$G@t@u@@@@@@@0AAAOAPAAAB0BWBrBBBB1CSCCDDDDBDGDLDMDDDDDD E/E^EgEùézssihh65aJ h<5aJhh<5aJhh5aJht}hM5aJht}h5aJ hBo95aJ hp5aJht}hJaJht}h_5aJht}h65aJht}hv5aJht}hmI5aJht}hJ5aJht}h5aJht}hJCJaJ)gEmErEEEEEEEF FFFF0F4FAFDFFFFFeGGGGHHHbHHHHHHHHII2IIFJUJ_JLL$L3L4LALfLnLƼƫƫƫơἡڡhBo9hBo95aJ hBo95aJht}h(-5aJht}hJ5aJ h5aJht}hM5aJht}hmI5aJht}h?a5aJht}h_5aJ h<5aJht}h65aJhh65aJhh<5aJ3nLMNNNNNO O!OeOfO}O~OOOPP$P%P:P;PvPwPPPPPPP QQ(Q)QwQxQQQQQRR0S1SvSwS2UAU V WWWWWX$X%XXļĸļļĸļĸļĸĴļĸļĸļļļļȭȩȢĢjht}hJU ht}h?mh U ht}hAhR3Vh<hCJaJh ht}hJht}h?m5aJht}h?a5aJht}hM5aJht}hJ5aJhBo9hBo95;eOfOvO}OOOOA;;;;;$Ifkd$$Iflvֈ w3N644 la]p<OOOOOP8222$Ifkd$$Iflֈ w3N'644 la]p< $$Ifa$PPPPPP2kd<$$Iflֈ w3N'644 la]p< $$Ifa$$IfP$P(P/P8P:P $$Ifa$$If:P;PgPvPzPPPA;;;;;$Ifkdy $$Ifl;ֈ w3N'644 la]p<PPPPPP8222$Ifkd $$Ifl'ֈ w3N'644 la]p< $$Ifa$PP Q QQQ2kd $$Iflֈ w3N'644 la]p< $$Ifa$$IfQ(Q,QBQuQwQ $$Ifa$$IfwQxQQQQQQA;;;;;$Ifkd0 $$Iflֈ w3N'644 la]p<QQQR8/ $$Ifa$kdm$$Iflֈ w3N'644 la]p< $$Ifa$RRSSS(S.S0S $$Ifa$$IfFkd$$Ifl644 la]p 0S1SNSSSUSdStSA;;;;;$Ifkd%$$Iflֈ w3N'644 la]p<tSvSwSxS W833dkdb$$Iflvֈ w3N'644 la]p< $$Ifa$ WXXaY^_M< & Fd`gdifR & FdEƀF`gdN & FdEƀfgd U d`gd U>XXXXXXXXY/YLYXYYYaYbYgYhYmYYY Z Z ZZ-Z1Z[ [8[:[;[<[?[@[G[[[[[[[\\Z\`\\\]]]] ^^^t^u^^^^^___ͻܫܫܥ h&laJ h[aJ haJht}hvH*aJht}h_aJht}hAaJht}h?maJ haJht}h5GaJht}hvaJ hifaJ h<aJ h UaJ>_______```M`d`q`````````$a%aFcGcKczccccccccccd d̲̿wmwccccch OJQJaJhYOJQJaJh2eOJQJaJht}h?mOJQJaJht}h&lOJQJaJh&lh&lOJQJaJh&lOJQJaJht}hAOJQJaJht}hvOJQJaJht}hJOJQJaJht}hJ6 ht}hJ h UaJht}h5GaJht}hvaJ%__`Gcccccd-dghhhiiijj dgd dgd d`gd, $da$gd,d` & Fgd `gd?md d,d-d[dbdtddddddddeDeeeeeeeeeeVfsfxfffffgMgQgdgsguggܸܫܞܔܞzm`ht}hrcOJQJaJht}hAOJQJaJht}hLc=OJQJaJht}hJzOJQJaJh&lOJQJaJht}h?JOJQJaJht}hJOJQJaJht}h_OJQJaJh,OJQJaJht}h?mOJQJaJht}hAOJQJaJht}hKmOJQJaJhKmOJQJaJ%ggggh!h1h2hlhhhhhhh i!iiiiiiiiiiҾȱ}ume]R]R]Jmh&lCJaJht}hCJaJhCJaJhCJaJh:CJaJhq6CJaJh,CJaJht}hJ5CJaJh 5CJaJht}hK+5CJaJht}hJCJaJht}hKmOJQJaJh:OJQJaJhMQOJQJaJhKmOJQJaJhAOJQJaJht}h?JOJQJaJht}hAOJQJaJijjjjjjjjj j!j"j#j$jSjTjjjjjjjjjɺܱ}}odSoJhz5CJaJ jh-hz5CJUaJjZ.F hzUVjh-5CJUaJhgMo5CJaJh-5CJaJht}hrOaJj#h:h:UaJhh 5CJaJh5CJaJj(#h:h:CJUaJ%jh:5CJUaJmHnHuh:CJaJjh:CJUaJhCJaJhifCJaJj j!j#jjjj k d^gdf<KdEƀNfgd  & FdgdrO dgd: dgd!jjk kIkOkkkklmmmmm$m/m0m1mjmvmmmmt?tNtStdtht uuu]uuu:v=vvvvvvvvvKwҿҴҩҩҞҋҀumғbWbh~rhCJaJh~rhcCJaJhKmCJaJh~rhJ;CJaJh~rh5,CJaJh^7CJaJh~rhFXCJaJh~rh_CJaJh~rh}RCJaJh~rhc4CJaJh$.CJaJh~rh~rCJaJh~rhXCJaJh~rCJaJhc4CJaJht}hXCJaJh#3CJaJ!Kwywzwwww5x6xXx{xxxxxxx yy1y3y4y5yyyyyyzzzz)z:z_zbzɾɳɾ|unugu`u`uYuYu ht}hL2 ht}h?4 ht}hFX ht}hc ht}h@*hKmht}hL2CJaJhQCJaJh#3CJaJhYCJaJhKmCJaJhvNCJaJht}h OCJaJht}h?4CJaJht}h!+CJaJh~rh!+CJaJh~rhFXCJaJh~rh(CJaJh~rh?4CJaJ"bzX{d{q{z{{{{{{{{ ||$|7|?|[|h|k|z|{||||||| } } }} ~&~i~j~k~~~~~Á?HS༴ht}h{hht}h{hht}hSNhht}h@*hht}hFXh ht}h{ ht}h_ ht}h{h^7hY ht}h?4 ht}hFX ht}hL2 ht}h( ht}h@*hQ:ScW)*+7Xa678Cøyqf[P[fhYhYCJaJhYh^7CJaJhYhKCJaJh#3CJaJhQCJaJht}hYCJaJhYCJaJ ht}hK ht}hchYhcCJaJhYh@*CJaJhYhSNCJaJhKmhKm5hKmhKm5h hKmh h@*hhYh{hht}h{hht}h@*hht}hSNhʌ،ٌbcVW ÐĐgdq # p@ P gdqgdq d`gdY $da$gdKmCajqr~ˉ׉؉9Š%1JTalx؋ۋGɌʌˌϹ힓ړړړړ퓈}vrhq hq5\hYhwSCJaJhYhq CJaJhYh CJaJhYhJWCJaJhQCJaJhCJaJhYhcCJaJhYh{CJaJhYh@*CJaJhYh{CJaJh#3CJaJhYhKCJaJhvNCJaJ,UV !vwxyz{>?gdqdgdq # p@ P gdq # p@ P dgdqv%,Uc567 019:;<TU[ӻӻӦӗ}hO*~B* CJmHnHphuhB* CJphjhB* CJUphhB*CJ aJ phhO*~mHnHujhUhO*~jhO*~Uh&hjh0JUht}hJ5\ hq6hqOJQJhqhK+<=bcԟ՟?@ѡҡgdq # p@ P gdq@ArstLMTU! # p@ P 0^`0gdq  !gdqgdq # p@ P gdqUabc456fghj $7$8$H$a$gd: !$$a$$da$gdq[\bcdefghjknorsuvxy{|~ÿêêht}hJ5\hJh:)h8 Gh:B*CJOJQJ^JaJphh&hhB* CJph"h0JB* CJmHnHphuh0JB* CJph jh0JB* CJUph$jknorsuvxy{|~ 7$8$H$gd: $7$8$H$a$gd:$da$,&P/ =!"#$% $$If!vh5r5b5555#vr#vb#v#v#v#v:V  t05r5b555544 $$If!vh5r5b5555#vr#vb#v#v#v#v:V  t05r5b555544 $$If!vh5r5b5555#vr#vb#v#v#v#v:V  t05r5b555544 $$If!vh5r5b5555#vr#vb#v#v#v#v:V  t05r5b555544 $$If!vh5r5b5555#vr#vb#v#v#v#v:V  t05r5b555544 $$If!vh5r5b5555#vr#vb#v#v#v#v:V  t05r5b555544 '$$If]!vh555555#v#v#v#v#v#v:V lv6,535N5555/ /  / a]p<;$$If]!vh555555#v#v#v#v#v#v:V l6,535N55559/ / / /  a]p<;$$If]!vh555555#v#v#v#v#v#v:V l6,535N55559/ / / /  a]p<;$$If]!vh555555#v#v#v#v#v#v:V l;6,535N55559/ / / /  a]p<;$$If]!vh555555#v#v#v#v#v#v:V l'6,535N55559/ / / /  a]p<;$$If]!vh555555#v#v#v#v#v#v:V l6,535N55559/ / / /  a]p<;$$If]!vh555555#v#v#v#v#v#v:V l6,535N55559/ / / /  a]p<;$$If]!vh555555#v#v#v#v#v#v:V l6,535N55559/ / / /  a]p<y$$If]!vh5#v:V l6,5/ / a]p ;$$If]!vh555555#v#v#v#v#v#v:V l6,535N55559/ / / /  a]p<;$$If]!vh555555#v#v#v#v#v#v:V lv6,535N55559/ / / /  a]p<Dd O'ZZ<  C A2B e;21]b`!B e;21]b.f0aH0x |UULᆄs[@RQd9H@ g2"*(,"E. " (-xp- f{fLC&d/&{]UÙΘS1YO%hڬ~< 32z f%;@zSsrhKOI_E˝fV}Yo!q XKo3zpj:9ݍX,ɍ1=;t*hsZnܵY o~V Sl'k/V>?V/')IimI*G#gYˆ8(8Lu;nY-k Ggve'4 N;)+kL^3BuI*ꔶfmX"Ю$.݈Eq{!6/v/IґQI8V j a'HZ$ߞ>.#K.DЈ<%r@ud`ɖ_>o^vTzgnJg+ۘK Gڥ#]RTuLS3UӥܤkxE+/ Ue"xi,OPjHzˊơZ_ҙD(3lVow|vIn!Wj= ZPKji-$i'i1m6H['io F>NӵüvߩYVYjvZ;D(hSDemBah8 Dy}x;q܎S'sVZV*Uh*:j96Wb[ S 8Txu?LRj >&L5 T,\j5רo!~Ϩ[P m 4PG@ŬD 88YX %atV0 ZDHtHp \TJʆ:WoՇ+:$蹹:gJWXU$Wd (7@Lr(<-3)$L9r6Q.G\ 0A~r; n"+;Q& h+4$sPAs|+b;|">F.6P{!b%K-V,WI"1Vg9"`hJ0A$~3"H7Fh)0J]CPDMQVM(X 'JyGܥ4UGRUEWyN}ůUsWA[LV TE}-|G6<[5U=~TI|S 2uQg{_U}~V5u.lEl.b+a:{0F0Tv#Ўk ~Dv,$hW=/nV$zQsg7+=X<;T|t6|TrTp%'] e]o|pBF`^^0ˡ*&笔B|Z(v\NygR&Ne0ꋍvo)a};ڔMװc2c6)N=׋踗jگLj4X^7':?^B~b/)~ٜ6+822&$:3D"͇Z45'2?WJJolwѵDfTk/]Cd6eӍX&tCV,̓P.Jz=ͣ蓨ꥬ,ϴ䛢_63`#l.#%=T.oew~8:H0!#'PH^ ,e$$O%y$dd` Qz'uϴ ŦTWLu! )Ѹ۟[dj?|O#,ɟ7#ɵiX? Q/F_xZX+UHOHk բX%k.ZOrk.fk:>iMǬ|mekn0o `Ax~q7T6fcVlA {Sڇ{t'Bҁt-&b +5Q3Q ^44OEo82njpng$_Pz>c;26}d<5#^6Ëx ˌ sQ44j gpl&(aP0N5ت2Wƻ ޘUmcbUqB*뫟J{<_'A<{.#O}PKnWwerH֧ȥ DOz?K>ʇr^]Nenȑ:C_{F w?iї{'D7E?,:5VQE$@K"^}box7_7K! h/jd_eQ[զ🴞֙ЮǵֈRn6߬Uj]$$ZS(ư^\>G>EfE5}ĕO~ڄ#Tz{P8Kw)=䵋AdIJRWA&XA#(^JV顸bR#unK'_^桷;)MQ%).ҟd^w /%'㞸b4?N^ǔ:ČP4)& fJ*vJKaVމ1Wl#Yu2+'_^R4$Pүcxk4ͬhqHj㊍ EMAMs;89xH,I@JpP:lmtW48!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠW>8!(?agꨠw>8tmgov6Py?ؓ }38_ L?D%uT0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3ea > ]U0o3eExƖ:F}_[^iWjz|%KO `Du/*~.(((WR>ºu/*~.((((((((((((((((((((((((((((O$ ]Ur x7_!+(((( o~1P$"Xd`A񦣯j<͵{>\P۲gfvF8|F<a gǛc236}팪UqѺ'<9]=!gT+3+jG? 6MO\3y m(2ހ H/Ok ki 28G Dn1QUy^ĴԹz}EyOOMi~L:ur%eU8\ms7?A&,#RH|HH8:`wc$XZ9dY"' # R8 "((((((((((((((((((((((((((((((((]KJ ꫕Կ{((]KJ ꫕Կ{((?|O5ޯEӾ[Ğ_t̪I׿Ex[[MOmWy;~1cz-BHu->_&LMH3)8#Q^A/]u KnS$c`0Vnqk o|yn{|Yx+Wz1^W<* Rp4E|?_: iHRaMuYޡ@D[ᶥjt_i/?HvXv;uk6>!,m&u1,R>;EhPE?}P@ɺ#TL<Iq3}1EŹ>Y%W"ra@YbLyQxQ_";;w iq$Ab+45(`Rkrی2.{A@]W>?OxNHԌ:Eշq!ԿXnG:q{2TWߴ?dn6}km ;fE8 4A?>>zNomy~]F2c$rN{^@Q@Q@W~տ|OOEߵyKgcۦ:EQEQE'Y脮_K<`[/PEPEPEP^u\ "/k0X; #1ݕr@kɢi%W bxI&O]Ro_ψ;<+Llr,8 ĄvzRWcJ^hmk$.T[BQd'F 'ю-B4fvO΍ OiSȫ?Zg[Ms75'YCrO߳x<]d$(v!G5g|0G vs=.BG-+kdc|"6ϫ\Ivr9 ?ޠL^<񿁯]$2Z nG4 yux#m2liX?M t|:ЈLz( sN?2şaa2jA-KkKP|# }ב7?u[^h7kI x`)2k~*.Mã,Vᶙn Io@><|AӠtÚM3ƹhB,bq*]ɹBʞf'v2 P,lgl,dKg=1<j'}ak:^ǎ{69'(^z_$J_"z4ψvNy{L,?p ÕżJxp?;r{z_~&xRҾ>u $xFҴe i|`s܀z"Kvm,u-&d|#*Pr xᏄ޷m[s @O9V#޼ >3LOG5}i xi+=~B_zWyjR5oq$dyh @GU7-o|3ik`^+ȣMj(F~c;?M x~ mCqUa*.᱓Ah8Ƞ7,w?ig>:?'}R_t6BXx@>7F:, 1\98|uOZ-.\pU@>a¼EMN4Gp&!S<8yײ||MVg{j;GpY#*C`nSr1KރE|)&j7zR6#s#>UX۸]ύI᭴ ARX JB[LFq#5>ƏTCgPJI3׫B|MnUw-!0_B%܆\37?{3Zl,tm9K^(a!)88YgDy Q+p~5䟳Ƒ?Q'ïV̳w/6Lvs;D_~NӾ-_լ/r2~UPBeJA8Sm4&Sc%XKjO7!Q _ZC04c*J)lF 88⸏5w8<^!6f 21eBہ^\V]p:+f^ NLv6 q3w7 LUVt|n+ɩaQѓqZ}ןO[4MUIKo@+Pς47ۨ9NHkkiK9n~,C)k}uJ7^Xέ]ϫ謯 xWF|hCyn 8}+V#((( = ( ( ( ( S>U}s\ZWcm<6YOOݤ$cGW̟̺u&x%#y uF㞣;׿~ɿMOa~~~QKFQ mW7Fku yO°' [c͖=ɮW~Οk;)i Lf%*-nb*SSo6$yOxYl4شvh籓l$.37d`2x徇1 eS_ʭՍƬUI%2_ GXgAff8+WJFo !%m .Ir1ӽK*Ҵ_{tјח$o}ğ~v75ddm2<]̹\|#DrD-rqT>eGc)QH>۶F99<9CxMmXGl"(?k$𮋨VvPEi I'"xdO?)xW1xŞ(v_ܠ@׶#(`3z<%3iIS%nDO*g~KFՇ֣cA}6- N"A^> LJ?i-[F{{BI-it*\ (Fm{ڷ>nnlc)UGB=N89n_&.xG>43I[sv$w;6F 8cOpW|}'ڜЦOfNB }~4PzſZƛkij$ss 3(P2rǡB/GnZPį<Y$k54#LJu[;ؾ[HFAEs_oMg/|TF`V=ʭ"+oB܍=NZ&z/2k9 uGG2ˆđ8OQ@5湧~^> Rn~]j.vl$`kҿh6W1*_v4>.b' ^Ey]X|дsOp)=-kN2{M+>>No SxzȀJ'h!AeuYH>w(BŭcMҵ?ovy(9cЏ4+=_z -ͫZH]HT9%kVcW˫ ]xÓLe6A#2ʰG>8sg4> rAo[pp A9TPEPEPEP+Ig e~,olBWU@Q@Q@]gk5Am 4+HFK1<$_(k:|p;xìE=.vIT1*Iil~xpgV `ߏ1#~`EPxgZ~jwGkR#{|qgW=zviNע+:n1.>bʼe8縩iF`@ 8"IZ̖EBeScrcX$fܲN:RqK^o@]6 E3G5 O2!;2 N>+ijNbhڽT6Kdr/NY@ g /Aךڅ֧KKʐn.#g1gNO5UaEPEPEPEPE[T۽BO*i6ڊdxӌ\?W~P6kc,) A@[oFc#s>4xG~mJR'fӯ"3)l1*P q~2/2oiqXCo>GEÁ09(f߲KyVn eb?;cw WP}_GjlgrX%} |TVd p*9buMr; )sĺ65^\O:xh%(͂t[mm 0i^V q"2H @:Uo_4(lxM5&[E R%pҹ㊅ڊWOn8[qcs.*cֶ $W¾qviaij|7/,,j:d.І;w89GGG_ru~ xoHÙ Zy/ӎz XM-z_#]SU?eK[IX`\ Y q^"Xn#+Hdh| J=ս6Ii%ۨQ'y$rNNkS{[Mk$Id",qŋNS}|ߛ9rFgm^j]M1iNю19==kJ]k]AUwdjULEr;8$0zxIIau7PrE jKX]>ꚳ#ꬤ)imtί,Q+FUHy|.hbڡnBҺXx88l$RޫA:^D爥Q H'?BjP- 3%[{$%A6 97+3; ?4i5O!_˞phO8ܠ _ *miuw]hs+((((((((((((((((((WR>ºu/*~.(((WR>ºu/*~.(,xx7湭 =[|fcA>[E|% U]geMӿfx-Kƞ1 Ju42 "p2=/U>2y-u!C ;yN .'ʓ">Ƣ鷰j:uﶹ&"PEPEPE|*> |:e[SOϽ{;Nm۹3]QEQEQEQEQEQEr x7_!+W?D%uTQEW?Ʒ &/pdB_c@$ |[Fis/<<}/#G{!das溳)IE]+! tOĞ ~$x05}L! GU,>PI %^+˜vu<[;n90ʥ$i28zU젻14wIGG(ˑR5]dGlGYxC Hk~J55}n擿/=X|QSkfx-`ܷ9Fx8yͿiy7qD0,ܐN:}x'd5+D}-׶%b(i^lh?l\/}ڧc2B~Or-k^%ejFác߲^_fo{;.@q㜟j~xW NN|>H9;^JO_ x- wg~W=0O'QqvgEb ( + (E >kc$) $@[oFc#xg?k6?6? xrNX3nܣ?yIqֵKOW>'4 rj:m&G;pNY 2N8^i~Go/o3;jG}4\c=~l|'/#ռZҲ7Lc 0[ T> x?HKa#Cw t jL``JJj^>tc)=KMñFm."cr91G gxo3]xTҼkɩϾ;SɎ4 4xm.3ioSSn٠,O B3p]9쬼++CaxpbuS +6n-?ЅKo5o2)5 -0]29C VYЏiLĄ˸(†(82IּCsxotFE)2 a{$ѥ[IgYpSA FwJʠ9'Fz :Qa+E(NQw[Y;=oٷ%E*[;孷HFR0Tr⢤[y|`Ji=:_-Mkv;V(6K;L玬GT !CqnO 1jvDAdv=nޗ< bj5k`Ҭ9Y~Q'9CJ(sYVBk2$dd{dS謹$ܷ"[Xj[2CX|,쭃JIA;^yyK:3( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( u/*~.+WR>€:( ( u/*~.+WR>€: |?$IIT@L`kBX$O sڌZb<# mܸܨ͜h  u/ Xdbݒ C%Bbk_ 7k+7j+}PO&03coC^o SRuR[asS_ jjuh dh> G𷄥oZl88P33=2 gi߳Ə.xW/qwF7'KzW?KH^;4.=t^OdZw{ŗkqew-(YZf&GGUf}G틫\4,"~U#~"+R V*?e!>*+{-aA\jB10ʜA#y9>u, |cآwʭq,HbP(;<(vgWn3Z08ЭuƟ>ͤiEkZp@*r@'i{.imWVw$h2/3RYύmeDr\z|,=09 ڗPjeqIBB[@(\6O,WG;Vu4﬇.^0$% 1&.Kw{7E2\>x~1J44fc}N@f/l˥xs>^y;f.=YozτUK öMooi!93|`v5 xO5ےb*} ]czqgKEqtIo,|DNH@?_Zu06WWQ(aw6v#*A<=Ǿ3NPY$H%|sJ*@qŞ KC]Adː;ܹd$Tg¿!&c\s7)#؊oUMzR?)7ynۑ1_k?>._*6>9O?֙^eG@[Ώ{\u\ZC-ϱE{|g}OkQÚf2@E+5j[MxwKt'gW"}Edww'yƲ6V# n(`+tۍĞ0]m7.>ؚo-SΩim 3tT[K@F8\jkjuYn曷qu2/j k <-^mkDl e9Hesݑ30_hW_%7A\t?cj;x^<|1]wdxЂA+;c sGVGev?@TUY @eY\ Xk_h+/^$YxKQ^y+$3qN+O_>'Cu|3׮t[090oaW^O֖>՜\D>έN`w!ec<]WxEzo4ѡPє:Wא>(W?D%uU'Y脮 ('guo.a.6BBƙs\y"E \]x[Qd,ML?p@]nswWLgI.ƖDݵ8B8&>-vO&5u&b+FA(TIb2Q\U)>DtS1-2IV2K HU ȗI2w-s4o>Fqh|9'R}24r1f. %⻷ đ8)DR31W=AhNTvhSVzaƙr]$ַ@r gh$[ӋfRwqGW1^7ÚLp%a#BEld[WkVaEt[=]'cc0.A>^7$2<ÓTpr<`1lÑ?y\-7}84Ж'>aݝc:4Go7b4<]ho>>kkG%J)t%hP:/gk{y#ѕR$?0/ ;NvfAp85^k_iNBA!7WisžkusaJ|ѐcicJ.HU*m̠G2N:wJī8mXrx5ГۦSETQEVW,4CNۯ7I)UPK?ƹcfi:wLz"Fzӵai~0wiL[ h嫖0%xu0bqddkKWJoK~O;=N ^e E v7@&gC=_L O!f,W9gٴM:-FM:* r@)$즰--Te۴G!R<^Pgp}+0fUykBi[iO'> zoiiLVeD¾NqVcZ^i:ZO#_]m31qҭ@7q'<֥qoY8|X'g>3m)jTM{eCEnxW 3.rs-ad>`1,HyQ\[AsH6gjPi)uOM=}Ӽ_sYK5~:^ R$dWϿwoh=[d-͝Fhz\2w3HJ+UA9`+iѫҫVv/?y*񕿏< kҧu`ySF~ek)]m|A2SJ8&eef1I# [N+ ^&iRyWЬW#l)ʑ#oլmߎ J(K ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( ( u/*~.+WR>€:( ( u/*~.+WR>€,CӵmcΛ٭/-)n]?͕V -ֹO?&k5mgnZ{m2cQA` L V-CWv:֙+bkt3*w+(#<ױQ@/ֿoˇRDIEէ ۆ)NF21/OitKh*fl`14uGr?,sTߌ3+(nuRV_,vuɦk"G3/rPֽ>u?O?" 3 y^C(Gߜg—k{Uk?;{ff`s]x_Ɨop/mx9WPzeAשƟ|[K}xuoO%wơT9BrM}'Eq_ ,'7]0h否Tqu'mOv]ź_"m<`V3";\v_D@+ \7x>Ϳ{ߊ-9%xli/FnSr3 @7Kގ<1{_oJ]61i{Wi𭾇hЙ%nv`]=W3:?9& o2)c;d@ד 騠?|}eO[Z#DD{=?եٯRԵu/)W-o JҎZi0s"kHxQc(jCX2jA2Xۊ}eotJ -u9mK[X*B99z~A6>(ֺ֯h-"_ݡ(m!UGwg=@=o>𦓢[`am>fNO^]Kֻ|cl%@< mݑʒ)!Q>E|!qɣG ӢǨ3[v:׾x{GcPllXbL =OrzܚТ>g|CGuIL/H|̀ԕ@`A8MNcm#\Y-Q7pNv8#y>hմеvnTfӡ |fA{i<-o2\I甀 N$Pª?a}M>c[ܳ,cd%A=h+υ/G-׶~w3`F港UZ4,/iؿǟvmWQ@Q@Wo _x᮱*[hoo<-Y1ds{u|)}?YmǶfh z05@|oiV'Gޮ\.j]XkM3xK,Y[u`I8aA!Kpŀ=| M3CFK'A\ׁ~Ӛ]D+I2 [k+YLg V5k*?iZlznO>)q3yW/+m?+2G>hOdovBNeB.|ca#ctYWJqᡥi%Gnn$Q@i..Fm;Z[^nq+F \|1ѫik??< ;Bj͑'=Z2~x IH=-ZdbdQvu9'n>>"Ծx &7Lh2H_J~<@' k>$hˋ +d9#'j*g}((O$ ]Ur x7_!+[٭gv#Я eRPUjN9| h #bx3±0Χw{*W?tUyO%qX֩7.yP[DA8*j+:d*X;v<*J(wCr>T9gץ<ą9'1!~n:InG:ǿU$F=oAI隒\a0uk1۰2X[rHcVw)T 5{lmd zRJcPJ2F}k|<9=cz?kԅuUNՊIn4˟"GMen* gsAR9h4xv06z-4OMu]=,*5V]fM3q;qFLPpMf6Wj€p~۳eg$,x;ZCnیEy'|+?Lowp]"kw\=z`g|cohey~f918? ה|n#=(G)GI$Rer"d^-snW[~6jSۧ^kiX$.NRO$I$QpG~CD #p5sܦ޷}|v&t$c.~t?iccX ci]G`GAXa';ʤiX+;Ng/n{!2-峆lvbқsi&tڔ86 QZ[{"L*"z̒kܥI;VJ5nۛߩQIF7om[C~4&HfT%w++qy4;x2Yb8q#9# pC|=2z?ͮT_ RK^8d9=N&{8nmdyLz|8v. U'KI/z^k@ڕt `rKItV q_w yǯzբlPVH ƅdY.uΧem7=$힙q*dA%UŘpN{ ]528iЕp rqҔtZ-YPdNԒUo8]3\0,@۞f 7=N#<>O? |U(UvyH<8B)%} ߇Wz}|F H ZM`vU6Cm0J8TP0Rhѐ7g}^V&8 VMX>;"eKy7 ܄q= נҦK*yX7~;w5o~S/mԤ٥thSp7}/[kejf*}zgj+B;SNxQK23:y:z>*R䨬mƤy>d2 `= xG]u 7Z_nc5r*&]JpT޽,$fh$m۲W&hI>9֍7[,ۙ"̌0@v㿡[G|?,V:<2³@xS%y8*yrE6˙u6K"5T1H,v62WGvfϤ]Gf̱KolҡV9\/*O;{n-_sp@|:J/I޾V^-ms~>.}#4moL֡tk '0ܽTx p?)ivRZxnM_:?(sd x;\6S@ݴǃdp+ؼ3i=\Zۭ `Wh%H FEarL:$o|5jV_8? x%ٮȑf$N냏j/VqzaXr4гX\ AyAyE|kŸƜi oWEo:iaaMneWu =j*w1n 0ϧx$#wkS^DLo`LirʱmrA?{)cFE}Y~?YyN;38h~坋 zr? mim"H J*!w5H Qpla$Qrl/#$fKF&~ʄ\%voKMj;tkQ{yYU%*v1ljKo/F{{>r'JAAa^x|i Mmm7$Yyf* H˺gm'Hm@pY6+~SL8\L&k8'ZWmۧfsR|[\mU91Œ1h[r2Fqoo<yV35O6u>[^G+ʌm$T,<AxXi}o6U^1TAsx{KGGn::w٧}7ף*xI,Bf5 3ÿWBkpU Ѫx0Oן|64}C>X =&ZDPIp@quz2+#V+!O֫RYﭷ}=Պ5QC70:dExP‰Ip,A*L(挤?482n%W;wȤ'󨵨eUF2A6`0>R2OYQQB\紹6#lu |`t$g䞘5zf<]of y3ghH;2ޫӊOZu+di%s@YN'@%8~Τ䌃Ќ{؞Va#O|/yyUⵀ0$GZiZI)Hm/Mb86\:⻰W]/.Fw7+E~?YPP īBpN|aK!v9]+:ꜹk (aEPEPEPEPEPEPEPEPEPEPEPEPEPEP\%O}uU_T=`]OGPUEPEP\/4mgO$5u)&NG'q]r?S?O(?]U?O(?]U?O(?]U?O(?]U?O(?]U?O(?]U?O(?]U?O(?]U?O(?]U?O(?]U?O(?]U?O(?]Ue&|U9#B9%$vQ0NĒEG+cU{??XCğQjz *#5??XCğQjz \ /'==J0Eӫ^޾lɟrX$G~?Y_XCğQjz}NO d<vΌ/xCUiDŽl_:oi|Hd<=Ί^W \`g5YVi>T̾pw9_@'z=# xK_jwܸe#9PjS1|uwAۏ[yZ[c.pܠv.g9Vu/\xaҏP+܃z4g[lpG9%Q#aU^iL+.@e2:|A_AS͌{ӽ^yW+u1{}Y57;J2׬ōa'sk +3JWiNFA?λw[<ؽ$`Rq <ࢅ!il״[ywyϵ~~Y1yE(Qikwi8omSg.Mz6OߥׯDdix巷MVX WbVU$8FvGB˭Bԭmn& "m>Z f~8{ [];J+2–*B=? yqq^oB46gue?,+}A41 o_ҖGʍ&?ƤDTQs^:"xXq0,GK!yh) JIV[9ۻ##$]|B{ |^XSwv "qQRJ-)ZS.WV#d3cEC EA@E$u{ D$a<3Z$# / `2_*A,L18= |*5=MMZ=h7aҫj.ez՚ nwz~qm&KZUURJKEQ@yLY;xc»etVF 29Wj&l$@ -#% uu>Ԓu2G!9 F:8`1]_֏z3AJNIY;[cIx]tK9 _i I*A'RN>/BΨSI1݊Z_ b IV 0ЍÃ=_ŝ>5`vHP&>A1ܸ-23?15wzƣ{nr M_iru<>WR^n^wt<S6N é]ڻH-4ʹ`;ܒc8QxWĒԺƚ͎>eabmqܩSzcs N8u;ΤeF]zEr{,~7<7 7eڈ d gXT!Wa6!d0G tێO,5/k6QW6nsld2H_0oRrVpԥ:Oj2RW ,ٺ9"IdV(N&ylIt31彻U Kٮ\F`:9=uzY\nFfxy3*OLj(S|=U^G~T4h4Bڻq y^N%.ORX澔M;QEQEQEQEQEQEQEQEQEQEQEQEQEQEQE_T=`]OGWU\%O}uTQEQE_T=`]OGWU\%O}uTQEQEQEQEQEQEQEQEQEQYך_܆n}*'deZ(ǚlq⍥p)m$n?JXh~c=;ᴞ\mV۩OjTVps$ڴKƑ.*SOI%dQE ( ( ( ( ( ( ( ( u[;SfؓYU2MniT$˲+վ5FoٵO9ڀ=w¼υn?Ok26yK#*,95j7\#ίÚUivPyP"FՉ|&9i<ןSWs;QPPT@cRG_֫j2.,do*LFHsV` 2~m 1[:nS[^MJN=kOWVH𕋩.!$<4j(JO3]˛o]m* 9O r7 F*| 0d3(xK<ܮڮ~^y +4{Q^QE?o`TlzȒ).|A#y;䎌F:>J=NQEfEPEPEPEPEPEPEPEPEPEPEPEPEPEP\%O}uU_T=`]OGPUEPEP\%O}uU_T=`]OGPUEPEPEPEPEPEPEPEPE29wd=U;#*գF<Q, qO5 :<5.j->jSK/m[~($I &;+q;;TW6 tL#RzE]FKoauӌub%i" .qGeU袺IYd)i;cR֥䝸:  {eRs zZDoF386J9ju$pb ( ( ( ( ( ( ( ( ( ( ;Y& hQY\*F2Y$mi'Xx'2WT"3`򯱘`O& oG2o syDZEReۊxvE뵛2EЬt5qik Vl7N9$Ʋ.7Փ&u(dA"98׍@q5=A /'I( qSoq.Z4q,cc޳|QXۉ5k(U^PB}#g ^GMND6(qsj3<+v.s`Mb m͈R6o+y_ş {KnuC,cSʌq2AFqi&M{ i,4l$"g'kf NFHFCl $׏|ees%Α= ]iЕA7*OX|3 )Ԏm;IY'M[VƯ[zxǺUx Ӽ)2U<$s#h.asf+ 3[<^qWHewBJo-' cqں3F(r|o֍jM#m6>|Ӣ] G> A`@SWſ#QE%{㿊ԟMkWs$8ns⷇uhf7r4]nqdxq)m7}ȶguN0r k2i:kc>0$L~[,6 _<%R \{]8o+'VyyJN6vv~gM+wM̠ BghpI @=<Z牉Trŷveݖ9 ;EvVpqDsƒшq[LG#4*[ >c8 Jt);ޖtzߵr|}שBUFȻW :^zKe$rD{cǧm"!kSQiqN.)gyGߎgPnm _5//7f-5W%{~롡MEPGP}jvVODnH8< RGL'R[2Ϙı|JYX?Sn  2[%%v=,QUeO +CՠAj\l@-8-RM1x=;ӧyro_˒7J;Y %ߧ[d<{sNs튭kKq-jčZ|6FJ{[Z[b8aTvO w ȥC+) ]ƺui7[m1A$$_mvxmSW5Gt(ms0,>qԂO:2bO㶳<75̆5޺yA#7F8 x3I5$&/.m!dRgs_cĘXgٺ-F{u]޶V'GUץ^]Y* ϒSr[ ѾAd!['krb_ 4:R&[պ{KyD.򾌯dW?jZwQ-֓U ܬjym厅d5zJ .s7*QmX(fEPEPEPEPEPEPEPEPEPEPEPEPEPEP\%O}uU_T=`]OGPUEPEP\%O}uU_T=`]OGPUEPEPEPEPEPEPEPEBːG}j'deV(ǚlQKuG?2P\(3g'>\;nV)ʹQ ITv(:Buʅz0U˴QEa+!yI+63 TcJ/VGEIM4֌Pyr]3ɥZ.PWFdڜս+Lip i#ҬZۥ{SzM\+6ݏg !v"@G`1KE`(((((((((((_~֧61ˬ,P6p+[óZ:1C!;D',w3pr7"_T׵?6+KF e?01^\xa҇VWk`K쑞J5)vmQdc}ٯ*0uEĤ6UYTF'f[<`2q+P[kwgIU%.;|b lw.9c*7 gZf\jyWDc* 2A$S1sB+ѭR6$q~Z.T*{HJ/݋|lEoo<:EEIUIWH'[5ܚVG&dlS8<# quxĚ{x̲- Rp`_wX|EyW\G19F)CvUKZxe[f%()ȯqA v PơBO3ޮmuc)t_*%J6AEWj2vgSWmʨfc:7~S-Lm&8kX [J)3&KJ|֌ev 8ǭjJ2c{zZֶ^(L#F` u 2mTk2M{^vlA MMđ4{3%* BDm褝I1RZ}C8$syTz (aX>(hJhڗ ʜa`X+ E\5EVdfLS^I$3jry]\O.n ;l Nv0wr=Cp< }Qf5ͣoKu8^jKfOruWFWPE6vDAdv=2(0"~BY>q!5a)p]rH8*)jVDUu{%9X$iXwqw{W~5Hܢkp˾4I#6*d|:ދuMWĶ$aqd_D^:3aO^?EHzo5;rR(cgPp7']T"?w:;_]&лVc$+R9`4`vR ?#IiRjnnv3<=]m̱ʣdi#+0eaA"mlu! o:#BqCq0i^^U-F٠6 y1Ut 3=Bo~-/ɻ+uK-Ku[|E̒p@?2FBqeY^8J7 m~Vк,īxхfa)AѼeB=*H('i8Ns>)r&_ך|H+oީjZn&DjWO6^X|]’22 lWv)ǿW:~ i8(F %Sm4bNAFā#vjx⸶9$l]H #c9rb)¤UIhug݇"J( ><S9ak_HW?׵+ѡ4zo +cp((((((((((((((WR>ºu/*~.(((WR>ºu/*~.((((EQEQEdl('q II5^_*?xW9b+ƄIfo.X9{*[afD#i`LV]I%Gc %9M*u=):ƅ܀NWb*)g!Tw'u3H@QݞHnQHPxI5gd1Kݽ>Уj%Z*ne k fbnE~~0HMmṔ-I(1XX ׬OyN#H¢HUcW~𞵪1n$F2Iq.Ymkl58Us}_Nc*TRk=c.=+A@w!<>c԰YGGH-yF{, 2H<uBvrT3?wxN+{v XVޏuj:bkkgPJ8OU}藠䗉PYR@rJU ez pRr{eouo0:RKugm&Nh"2HԹ;?\gyMv;T >pN1VUJYUjwj5M;X "P5-t٧GH1%A.q5G~$|<ͬ^GjJ)(qPN=GxΗ/6۴᷶L2`,p#񶽼$a>jѦ{ɾ:լb K^/N>3D"iNA0pN^KiyfF?3IUgԒMkiŵ]$I$ 0SI=)9qX<5N[i-)y*G7y1.7dpG8*ā>8B^*[u+9~P{ 暍] (SLmO4яrK`O .#$dg';xdy(cRUT Icw_3qMG˚_TeR~e Qrccy8NO߳~ɜxgmgQP e!_239j]C)= |GZw( 8grėRW~USY_4۲$MpD~Yu5iX-)7G {EHH7$2x$`0k:RGz:n+]W N\N3}\}^`=i_YxzV]g*pZwX5HHB50y)6Qn@5KkQԡc_2_|?ӵ [aKK˵@n^tVKx%[۔NIJ;6do@ )F!OkZIv(D72w ff?N[P":0G/i/Z蔮M<%`OVrݘaYt+G˕e^ 'ozҬՕok+!=!;AxSlV4` ,sn`3XeE)Mrb. ԩʼnU>RKr^̳5M 9>Iܥ%t>IdGTUu*@# Le%*1O/bz8sxo /g!?,v;i6>h׃ i;H57DiآYJCǘ 0#Ͳ.# LZۅo[>~DTs_ez_~kWH?V2]\ْ $<NӀq_KWğugD!J2 0H_mם<L$)*_)j QYEPEPEPEPEPEPEPEPEPEPEPEPEPEP\%O}uU_T=`]OGPUEPEP\%O}uU_T=`]OGPUEPEPEPEP\/bE9ǽ(sfYc>SyBP^H忷AvI~:U])#xH+JSi`;~ڻF Ώ#Pfl#_4*铲J)&.G(hym,;]OoZRrwgm:qX+ )QEQEQEQEQEQEQEQEQEQEQEQEQ^G|CX4Y^zg(6k~Anjp [Fis/<<}/#G{!das溳z2~YqY_|3%cF΋D&n\#;UA[NreFROp I?j)(Y~ͥm}OŚd$4rT؃=DwHp3=p c9?sEI5(;=Df$ӵ%1ɇc*<8듀E^p9T~g+W9FNI=JQ^[ݷHfO22 9Si}fڥܖ Dž\q gg54iεHӦ&Kz$i&myll56Dۛ"IJw%H یv5 7[;<`MD:7FI92mo_ּKoWlq";pB.0`"bڰ(ViSI&[woOF.I-TPH߾ZIx11?` 헩ջQ UsSx Ht׳yPڣfrA 9(13ɨC̖p7 1M.w׵9/з0xsO9iV!~AݎI2C7G?AG5f(˾NF7`=5eK;FmFvsVCPpN9 _Ud_Y);'5|IRZU5]kR$ "F^$+W!y2Ͼ}j+nM'?ZүĴj/-ZڂiJ+Ccd@rN"K4r23u#˓*+[ F[V*r'I,pΧ=;o-F/+PĮ?")`l2X}fln'8<&&>cƣ!fxn9.^Ijb~,]Ԭ|-5mӑc\u#sqxkN!kx{1,{nc\,RkLdM-}W=z׬%L9UR(A7'/nmP8f9 Kk^H!*hv???yW]WiOV%`p܌WH'w8Ӱʑxz}Or7w"l<ժ1u1U9M]t0a*.oZ=8*0v|Y㜎Fj֍[ֱ^m8rS#g*":Ɵ)P6xKa ں=KK]Љn`aˡ AA'k-iҺ)VN<)-Img b߼Rv/jqjZƇsy5\I&Q='ga;t'ڬ7|6Kr,dgbe•> #1)B2䏘1#5