BIOLOGICAL CONSERVATION

Biological Conservation 86 (1998) 207±213

Ecoregions as a level of ecological analysis

R. Gerald Wright a,*, Michael P. Murray b, Troy Merrill b aUS Geological Survey, Biological Resources Division, Idaho Cooperative Fish and Wildlife Research Unit, University of Idaho, Moscow, ID 83844-1136, USA bLandscape Dynamics Laboratory, Department of Fish and Wildlife Resources, University of Idaho, Moscow, ID 83844-1136, USA

Received 14 August 1996; received in revised form 11 October 1997; accepted 24 October 1997

Abstract There have been many attempts to classify geographic areas into zones of similar characteristics. Recent focus has been on ecoregions. We examined how well the boundaries of the most commonly used ecoregion classi®cations for the US matched the boundaries of existing cover mapped at three levels of classi®cation, ®ne, mid- and coarse scale. We analyzed ecoregions in Idaho, Oregon and Washington. The results were similar among the two ecoregion classi®cations. For both ecoregion delinea- tions and all three vegetation classi®cations, the patterns of existing vegetation did not correspond well with the patterns of ecor- egions. Most vegetation types had a small proportion of their total area in a given ecoregion. There was also no dominance by one or more vegetation types in any ecoregion and contrary to our hypothesis, the level of congruence of vegetation patterns with ecoregion boundaries decreased as the level of classi®cation became more general. The implications of these ®ndings on the use of ecoregions as a planning tool and in the development of land conservation e€orts are discussed. Published by Elsevier Science Ltd.

1. Introduction who re®ned physiographic provinces in the US into regions and subregions, based primarily on structural An understanding of the factors that determine where and attributes (Fenneman, 1928, an ecological boundary is located and how it in¯uences 1931). Atwood (1940) [mapped by Raisz (1954)] further our understanding of ecological processes is a funda- re®ned these zones by incorporating human factors into mental issue facing ecologists and land planners (Weins the classi®cation. et al., 1985; Gosz, 1991). It is generally accepted that Other e€orts to divide the 's surface into identi- ecological zones should be hierarchical, with their size ®able zones have been based on climate and vegetation. dependent upon the scale of study, and their boundaries Herbertson (1905) mapped the world into 13 di€erent based on semipermanent components (Bailey natural area regions ranging from polar highlands to et al., 1994). This basis allows recognition of an ecolo- equatorial lowlands. Several other investigators have gical unit regardless of present land use or successional further re®ned this system over the years, both globally status of the vegetation. and for speci®c countries. Merriam (1898) de®ned zones Systems classifying geographic areas into zones con- of natural vegetation and agricultural crops based on taining similar characteristics, so as to better understand climate, and Walter and Box (1976) presented a global their similarities and relationships, date back many plan for classifying also based on climate. decades. Early classi®cations were often based on topo- Potential natural vegetation (Kuchler, 1964) has also graphical contrasts, such as the identi®cation of the been used to classify the US into ecological provinces. major physiographic provinces within the US by Powell Dice (1943) subdivided the North American (1895). Many of the fundamental principles he used into biotic provinces, areas characterized by ``...peculia- were adopted by subsequent scientists in this ®eld of rities of vegetation type, , climax, ¯ora, , study. Foremost among these was Nevin Fenneman climate, physiography, and ''. Canadian resource managers have been among the leaders in developing * Corresponding author. Tel.: 208 885 7990; fax: 208 885 9080; frameworks for land classi®cations based on ecological e-mail: [email protected] criteria that include vegetation, , landform, and

0006-3207/98/$19.00 # Published by Elsevier Science Ltd. PII: S0006-3207(98)00002-0 208 R.G. Wright et al./Biological Conservation 86 (1998) 207±213 physiography (Wiken and Ironside, 1977; Government (1968). Kuchler (1964) classi®ed 116 types of natural of Canada, 1991). communities in the continental US de®ned as ``the vegetation that would exist today if human being were removed from the scene and if the resulting plant suc- 2. Ecoregions used in the analysis cession were telescoped into a single moment.'' Omernik's classi®cation was also based on Kuchler's In this paper we focus on land classi®cation systems (1964) vegetation map in conjunction with physio- based on ecoregions. Ecoregions have been de®ned as graphy (Hammond, 1970), land use pattern (Anderson, geographic ``...regions [that] generally exhibit simila- 1970), and soils (USDA, 1957). These measures are rities in the mosaic of environmental resources, ecosys- combined in a way that requires numerous subjective tems, and e€ects of humans..'' (Omernik, 1995). They decisions on the relative importance of the di€erent data are relatively homogenous regions in terms of their layers. Because of this the results are often not ecological systems, organisms and environment. Bailey repeatable (Host et al., 1996). (1983) similarly de®ned ecoregions as ``...geographic zones that represent geographical groups or associations of similarly functioning ecosystems.'' Ecoregions 3. Objectives ``...therefore de®ne broad areas where one can expect to ®nd the same kinds of vegetation and soil associations The broad issue we raise in this paper is whether on similar sites.'' ecoregions provide a useful and appropriate tool for Here we examine and compare ecoregions as de®ned conservation planning. The speci®c question we ask is by the two major proponents of the concept in the US, how well the boundaries of ecoregions, as proposed by Robert Bailey and James Omernik. Both systems were either investigator approximate the extent of mapped developed to facilitate the planning of regional con- vegetation cover types. We further examine the in¯u- servation strategies. We use the ecoregion classi®cations ence that the hierarchical level of vegetation classi®ca- developed in map form at a scale of 1:7 500 000 for the tion has on the concurrence of the correspondence of US in 1976 (Bailey, 1996) and described by Bailey vegetation types within the boundaries of ecoregions. (1980) and modi®ed and published as a map with We hypothesize that there should be `some' congruence accompanying manual by Bailey (1995). Methods used between ecoregions and mapped vegetation types, and in delineating the regions are described in Bailey (1983, that this congruence should increase as the level of 1996). This map de®nes 52 ecoregion provinces for the vegetation classi®cation becomes more general. We fur- US ranging in size from 11 900 km2 to 751 000 km2. ther hypothesize that the ecoregions as de®ned by either We use the 7 500 000 scale Level III ecoregion map of investigator should respond similarly to these tests. We the US published with map supplement by Omernik in recognize in proposing these hypotheses that the ecor- 1987 (Omernik, 1987) with the map slightly revised in egion boundaries were not explicitly based on maps of 1995 (unpublished). To date, Omernik has not pub- existing vegetation and that there are scale issues, i.e. lished a manual describing each of the ecoregions the ecoregion boundaries are mapped at a geographic developed. This map de®nes 78 ecoregions, that range in scale far more coarse than the maps of existing vegeta- size from 15 000 to 330 000 km2. tion. Thus we would not expect complete correspon- Each of the maps is hierarchical in terms of divisions. dence. However, we also consider vegetation to be a Bailey subdivides his map into increasingly detailed good integrator of climate, soils and landform, and thus (smaller geographic areas) of domain, division, province some correspondence would be expected. and section. Omernik (see Omernik, 1995) has subse- We undertook these analyses not to criticize the quently aggregated his original map into three levels, I, ecoregion concept or the delineations of either investi- II and III, also re¯ecting increasingly detailed gator. Rather we did this because individual ecoregions subdivisions. We use the province level classi®cation of have increasingly been used in the US as a basis for Bailey and the Level III map of Omernik in our analysis as organizing and interpreting environmental data for these provided the greatest detail in equivalent versions. inventory, monitoring, and research e€orts (Gallant et Di€erent ways in which ecoregion boundaries can be al., 1995), and because we now have, for the ®rst time, developed are summarized in Bailey (1996). For the detailed regional-scale vegetation cover maps, available ecoregions used in this study, the two investigators used in digital format with which to test our assumptions. di€erent methods to develop the ecoregion boundaries. As we interpret the process, Bailey based his province level ecoregions primarily on Kuchler's (Kuchler, 1964) 4. Methods classi®cation of potential natural vegetation in the US with the boundaries in¯uenced by Koppen's (Koppen, We limited our analysis to the states of Idaho, 1931) climatic classi®cation as modi®ed by Trewartha Oregon and Washington. We used vegetation cover R.G. Wright et al./Biological Conservation 86 (1998) 207±213 209

Table 1 Ecoregions used in the study, and correspondence between them (where possible)

Bailey ecoregions Omernick ecoregions

Name Ecoregion code Area km2 in Name Ecoregion code Area km2 in study area study area

Cascade mixed forestÐcoast ranges section M242A 40 429 Coast range 1 40 950 Cascade mixed forestÐw. Cascade section M242B 97 866 south 4 44 959 Sierran SteppeÐmixed M261D 36 077 Eastern Cascades 9 35332 No equivalent Klamath Mountains 78 15 567 Paci®c lowland 242A 38 588 Puget lowland 2 16 054 N. Rocky Mts M333B 59 204 Northern Rockies 15 111 395 Middle Rocky Mts M332A 127 107 Blue Mountains 11 64 692 Intermountain semi-desert 242 350 231 Snake Basin 12 144 820 Great Plains 331A 17 277 Columbia Plateau 10 90 872 Southern Rocky Mts M331D 10 955 No equivalent No equivalent Northen Cascades 77 30 761 No equivalent Willamette Valley 3 14 745 maps developed for the respective states by the indivi- such as the Wasatch and Unita Mountains that had dual state programs of the GAP Analysis Program only a small portion of their total area in the region. (GAP) (Scott et al., 1993). These maps were created by We used the GRID module of ARC/INFO which is incorporating LANDSAT Thematic Mapper satellite raster based. Both ecoregion and vegetation data layers imagery, aerial photography, existing maps and ®eld were converted from polygon to raster data. The cell reconnaissance (Scott et al., 1993) and compiled at size used in the analysis was 1 km2. The actual vegeta- scales ranging from 1:500 000 to 1:100 000 with a mini- tion layers were combined with each ecoregion classi®- mum mapping unit from 250 to 40 ha. The maps were cation. We assessed each ecoregion separately by digitized into a geographic information system (GIS) calculating the proportion of total area of each cover (ARC/INFO version, Environmental Systems Research type that occurred within a particular ecoregion. This Institute, Redlands ). Actual vegetation was was done for all three levels of classi®cation. We also classi®ed into cover types based on subjective assess- assessed whether one or more vegetation types domi- ments of canopy dominance by one or more species. nated an ecoregion by calculating the proportion of Because each state's GAP program classi®ed vegetation each ecoregion occupied by each vegetation type. This independently, many resulting individual cover types, was also done for all three levels of classi®cation. although very similar in species composition were given a di€erent title by each state (Davis et al., 1995). Therefore, the vegetation classi®cation was subse- 5. Results quently standardized by aggregating similar state plant communities into new standardized multi-state types. To simplify interpretations of the proportion of the Many individual state plant communities were distinct total area of each cover type that occurred within a and were therefore not aggregated. Three hierarchical particular ecoregion, the results were tabulated to show levels of classi®cation were used in the analyses, a com- the number of types that had <20% and >80% of munity or alliance level of classi®cation (Level I), the their total area in each ecoregion. These results for Bai- most de®nitive, consisting of 118 types (Jennings, 1993); ley and Omernik ecoregions are shown in Tables 2 and a dominant species classi®cation (73 types) (Level II) 3. A typical pattern of the proportional occupancy by developed by the authors from the Level I classi®cation; vegetation types in two separate ecoregions is shown in and a formation level (18 types) (Level III) also devel- Figs. 1 and 2. The analysis of the proportion of each oped by the authors and based on structural similarities ecoregion occupied by each vegetation type was tabu- and common growth forms. lated into two levels (<10% and >10% of each ecor- We used the nine Bailey (1996) ecoregion provinces egion) to simplify interpretation. The number of that fall either entirely or partly within the three-state vegetation types that fell into these categories for Bailey area (Table 1). We excluded those ecoregions, such as and Omernik ecoregions are shown in Tables 4 and 5. the California coastal that only had a small por- Over 59% of the total mapped occurrences of Level I tion of their total area in the three-state region. We used vegetation cover types in the Bailey ecoregions occupied the 11 Omernik (1987) ecoregions at the Level III clas- less than 20% of the ecoregion area. Just under 13% si®cation which fell entirely or largely within the three- had >80% of their total area within a given ecoregion. state region (Table 1). We again excluded ecoregions This pattern was observed in all ecoregions except the 210 R.G. Wright et al./Biological Conservation 86 (1998) 207±213

Table 2 Number of vegetation types within three classi®cation levels, with <20% and >80% of the total type area occurring in Bailey ecoregions.

Ecoregion code Level I classi®cation Level II classi®cation Level III classi®cation Fine-scale vegetation Mid-scale vegetation Coarse-scale vegetation

# Number Number # Number Number # Number Number Types <20% >80% Types <20% >80% Types <20% >80%

M242A 27 17 2 25 18 0 12 11 0 M242B 72 45 6 50 28 3 17 10 1 M261D 37 26 5 27 21 3 13 10 0 242A 25 14 1 22 15 0 13 11 0 M333B 35 20 0 31 18 3 11 6 0 M332A 62 38 3 45 31 1 15 12 0 242 70 22 30 49 16 17 16 6 1 M331D 14 12 0 10 7 0 9 9 0 331A 24 22 0 20 20 0 11 9 0 Total 366 216 47 279 174 27 117 84 2 %59136210722

Table 3 Number of vegetation types, within three classi®cation levels, with <20% and >80% of the total area occurring in Omernik ecoregions

Ecoregion code Level I classi®cation Level II classi®cation Level III classi®cation Fine-scale vegetation Mid-scale vegetation Coarse-scale vegetation

# Number Number # Number Number # Number Number Types <20% >80% Types <20% >80% Types <20% >80%

1 29 18 3 26 20 0 13 12 0 2 15 10 1 14 10 0 8 6 0 3 15 10 0 13 9 0 11 9 0 4 36 22 1 29 18 0 13 10 0 9 55 45 2 38 32 0 16 16 0 10 41 28 3 30 22 1 14 12 0 11 48 38 1 33 26 0 13 11 0 12 59 21 18 30 13 0 16 8 1 15 44 22 7 34 16 5 13 10 0 77 34 24 1 29 15 1 13 10 0 78 21 14 3 17 1 2 11 10 0 Total 397 254 40 293 192 9 141 114 1 % 64 10 66 3 81 0.7

Table 4 Number of vegetation types within three classi®cation levels that occupy <10% and >10% of the area of each Bailey ecoregion

Ecoregion code Level I classi®cation Level II classi®cation Level III classi®cation

# Number Number # Number Number # Number Number Types types in types in Types types in types in Types types in types in <10% area >10% area <10% area >10% area <10% area >10% area

M242A 27 24 3 25 22 3 12 10 2 M242B 72 70 2 50 48 2 17 15 2 M261D 37 35 2 27 25 2 13 10 3 242A 25 22 3 22 18 4 13 9 4 M333B 35 35 0 31 30 1 11 9 2 M332A 62 60 2 45 42 3 15 12 3 242 70 69 1 49 47 2 16 13 3 M331D 14 10 4 10 6 4 9 5 4 331A 24 23 1 20 19 1 11 8 3 Total 366 348 18 279 257 22 117 91 26 % 95 5 92 8 78 22 R.G. Wright et al./Biological Conservation 86 (1998) 207±213 211

Table 5 The number of vegetation types within three classi®cation levels that occupy <10% and >10% of the area of each Omernik ecoregion

Ecoregion code Level I classi®cation Level II classi®cation Level III classi®cation

# Number Number # Number Number # Number Number Types types in types in Types types in types in Types types in types in <10% area >10% area <10% area >10% area <10% area >10% area

1 29 26 3 26 23 3 13 11 2 2151414104844 3 15 12 3 13 12 1 11 7 4 4 36 35 1 29 28 1 13 11 2 9 55 54 1 38 37 1 16 14 2 10 41 40 1 30 28 2 14 11 3 11 48 46 2 33 29 4 13 9 4 12 59 57 2 30 28 2 16 14 2 15 44 43 1 34 33 1 13 11 2 77 34 32 2 29 25 4 13 10 3 78 21 21 0 17 15 2 11 8 3 Total 397 376 20 293 268 25 141 110 31 % 95 5 91 9 78 22

Intermountain Semi-desert area (242) where 43% of the 10% of the types had >80% of their total area within a types fell in the latter category and 31% of the vegeta- given ecoregion. As with the above analysis, there was tion types occupied less than 20% of the ecoregion area. one exception, the Snake River Basin (12), where 31% Excluding this one ecoregion, only 5% of the mapped of the types fell in the latter category and 36% of the occurrences occupied more than 80% of the given ecor- vegetation types occupied less than 20% of the ecor- egions. This same pattern held true when the higher egion area. This ecoregion is roughly equivalent to Bai- levels of classi®cation were analyzed (62% and 10%, ley's Intermountain Semi-desert ecoregion (Table 1). and 72% and 2% respectively at Level II and Level III). Excluding this ecoregion, only 6% of the mapped In the Omernik ecoregions, almost 64% of the total occurrences occupied more than 80% of the given ecor- mapped occurrences of vegetation types at Level I egion. Again, as with the above analysis, this same pat- occupied less than 20% of the ecoregion area. Whereas tern held true when the higher levels of classi®cation were analyzed (63% and 3%, and 81% and 7% respec- tively at Level II and Level III). Over 95% of the total vegetation types mapped in the Bailey ecoregions at Level I occupied <10% of the area of the ecoregion. No types occupied more than 60% of the area of a given ecoregion. This pattern held for all ecoregions and for all levels of classi®cation. Similarly, over 95% of the total vegetation types mapped in the Omernik ecoregions at Level I occupied <10% of the area of the ecoregion. No types occupied more than Fig. 1. Proportional occupancy of the Cascade Mixed Forest Ecor- 55% of the area of a given ecoregion. This pattern again egion Province (M424A) of Bailey et al. (1994) by Level I vegetation held for all ecoregions and for all levels of classi®cation. types. A total of 27 types are found in this ecoregion.

6. Discussion

The results of our analyses are remarkably similar for the two ecoregion classi®cations. This is surprising con- sidering that the two classi®cations are by no means identical, and in the case of Great Plains Palouse (331A) and Columbia Plateau (10) or Middle Rocky Moun- tains (M333B) and Blue Mountains (11) they di€er markedly in location. We calculated the proportion of a Fig. 2. Proportional occupancy of the Blue Mountains Ecoregion (11) of Omernik (1987) by Level I vegetation types. A total of 48 types are given vegetation type within an ecoregion to examine found in this ecoregion. the ®delity between cover type and ecoregion boundary. 212 R.G. Wright et al./Biological Conservation 86 (1998) 207±213

For both ecoregion delineations and for all three vege- compensating factors, e.g. soil conditions may modify tation classi®cations, there was essentially no con- the apparent e€ects of climate. In fact, the criteria used gruence. Most cover types had a small proportion of to classify GAP vegetation maps are based largely on their total area in a given ecoregion (<20%), and as a the interpretation of digital imagery and aerial photos, result they occupied several di€erent ecoregions. There but also incorporated elements such as elevation, slope, was thus no support for the hypothesis that the bound- and aspect which should, in part, take into account aries of ecoregions would approximate the boundaries compensatory factors. We further recognize that one of of the vegetation types. the factors used in Omernik's ecoregion designation was As a result of the ®rst analysis we looked at the pro- land-use pattern and that this could in¯uence the degree portion of an ecoregion occupied by given vegetation of congruence between ecoregion and vegetation types. This would tell us if one or more vegetation types boundaries. However, this was not a heavily weighted dominated an ecoregion and thus might provide a criterion and we also excluded existing agricultural meaningful de®nition to its boundaries. This was not lands from our analyses. the case in either ecoregion system or at any level of We hypothesized that the level of congruence of vegetation classi®cation. The results, again remarkably vegetation patterns with individual ecoregions would consistent among the two ecoregion classi®cations, increase as the level of vegetation classi®cation became showed virtually all mapped vegetation types occupied more general. We found the opposite. The proportion <10% of a given ecoregion, and no one type or types of the total types within a given ecoregion was the low- dominated any ecoregion. In almost all cases where a est at the highest level of classi®cation (Level III). In high proportion of a vegetation type fell within a parti- fact at this level of classi®cation, most of the types were, cular ecoregion, it still occupied a small proportion of on the average, found in each ecoregion (76% for Bai- the total area of the ecoregion. For example, eight types ley, 75% for Omernik). Conversely, at Level I, 34% and had 100% of their total area within Omernik's Snake 31% respectively of the types were, on the average, River ecoregion (12), but together they occupied only found in each ecoregion. 8% of the area of the ecoregion. Similarly, 13 types had We had expected our analyses to show a di€erent >95% of their area in Bailey's Intermountain Semi- picture and were surprised at the results. We do not desert ecoregion (242), but they cumulatively occupied argue that ecoregion boundaries should be exactly only 4% of the area. coincident with boundaries of vegetation cover types. Our results show that the patterns of existing vegeta- However, since vegetation plays a dominant role in tion mapped at varying levels of resolution do not cor- in¯uencing occupancy by vertebrate and invertebrate respond to the boundaries of the ecoregions we species (Edwards et al., 1996) some congruence is desir- analyzed. We cannot extrapolate beyond the ecoregions able if ecoregions are to be used as a conservation of the north-west. Both ecoregion delineations were planning tool. based, at least in part, on potential vegetation mapped by Kuchler (1964) not existing vegetation as we mapped it. Kuchler's map is at a scale of 1:7 500 000, comparable 7. Implications for conservation programmes to the ecoregion maps but much coarser than our exist- ing vegetation maps. We do not believe at this stage, The issue we raised was whether the ecoregion scale however, that the lack of congruence can be solely of analysis is both useful and appropriate. We fully attributed to scale di€erences. Kuchler's map is also recognize that the matter of utility is contingent on the quite detailed in the three-state region we studied, iden- question being asked. The ecoregion has been used as a tifying 17 di€erent types of potential vegetation. This primary level of classi®cation in examining where to site compares favorably with the number of types mapped new national parks (Wright et al., 1994) and where to at our formation level. We intend in the future to com- designate additional wilderness areas (Merrill et al., pare our map of existing vegetation with that of Kuch- 1995) in Idaho. Likewise the Conservancy, a ler. However, that was not the object of this paper. private non-pro®t land conservation organization in the In reviewing this paper, Bailey (pers. comm.) com- US has proposed the use of Bailey's (Bailey, 1995) mented that meaningful comparisons of vegetation dif- province level ecoregions as a guide for its land acquisi- ferences across ecoregions can only be made if atypical tion and conservation programs (Williamson, 1996). vegetation types are eliminated from the analysis. We The results presented here may raise some questions agree and do not consider any of the vegetation types over the utility of this approach and in fact have caused used in this analysis to be atypical (Jennings, 1993). He the senior author (R. Gerald Wright) to rethink recom- further commented that the low correspondence mendations made in an earlier paper that ``The eco- between vegetation and ecoregion boundaries may be region is the most appropriate unit of spatial analysis... that the criteria used to classify the vegetation types were to de®ne gaps in the protection of biological.. applied uniformly over the area without considering resources..''. (Wright et al., 1994, p. 212). In fact, based R.G. Wright et al./Biological Conservation 86 (1998) 207±213 213 on the results presented here, ecoregions may not be the Gosz, J.R., 1991. Fundamental ecological characteristics of landscape most appropriate level of strati®cation for developing boundaries. In: Holland, M.M., Risser, P.G., Naiman, R.J. (Eds.). land protection plans, as least in the north-western . Chapman and Hall, New York, pp. 8±30. Government of Canada, 1991. The State of Canada's Environment. states. Canada Communication Group-Publishing, Ottawa. Hammond, E.H., 1970. Classes of land-surface form. In: The National Atlas of the United States of America. US Geological Survey, Acknowledgements Washington, DC, pp. 62±63. Herbertson, A.J., 1905. The major natural regions, an essay in sys- tematic . Geographical Journal 25, 300±312. This work was supported by the GAP Analysis Pro- Host, G.E., Polzer, P.L., Mladeno€, D.J., White, M.A., Crow, T.R., gram of the US National Biological Service. J. Michael 1996. A quantitative approach to developing regional Scott, Mike Jennings, Robert A. Bailey, and three classi®cations. Ecological Applications 6, 608±618. anonymous reviewers provided thoughtful reviews Jennings, M.D., 1993. Natural terrestrial cover classi®cation: and helpful suggestions on earlier versions of the Assumptions and de®nitions. Gap Analysis Technical Bulletin 2. Idaho Cooperative Fish and Wildlife Research Unit, Moscow. manuscript. Koppen, W., 1931. Grundriss der Klimakunde. Walter de Gruyter Co, Berlin. Kuchler, A.W., 1964. Potential Natural Vegetation of the Con- terminous United States. American Geographic Society Special References Publication 36, New York. Merriam, C.H., 1898. Life zones and crop zones of the United States. US Dept. . Bull. Div. Biological Survey 10, 1±79. Anderson, J.R., 1970. Major land uses. In: The National Atlas of the United States of America. US Geological Survey, Washington, DC, Merrill, T., Wright, R.G., Scott, J.M., 1995. Using ecological criteria pp. 158±159. to evaluate wilderness planning options in Idaho. Environmenal Atwood, W.W., 1940. The Physiographic Provinces of North Amer- Management 19, 815±825. Omernik, J.M., 1987. Ecoregions of the conterminous United States. ica. Ginn and Co., Boston. Bailey, R.G., 1976. Ecoregions of the United States (Map). US Annals of the Association of American Geographers 77, 118±125. Department of Agriculture, Forest Service, Ogden, UT. Omernik, J.M., 1995. Ecoregion: A framework for managing ecosys- Bailey, R.G., 1980. Description of the Ecoregions of the United States. tems. George Wright Society Forum 12 (1), 35±50. Powell, J.W., 1895. Physiographic regions of the United States. US Dept. Agriculture, Miscellaneous Publ. 1391, Washington, DC. Bailey, R.G., 1983. Delineation of ecosystem regions. Environmental National Geographic Society Monographs 1 (3), 65±100. Management 7, 365±373. Raisz, E., 1954. A Landform Outline Map of the United States and Adjacent Parts of Canada and Mexico. Harvard University Press, Bailey, R.G., 1995. Description of the Ecoregions of the United States, 2nd edn. US Dept. Agriculture, Miscellaneous Publ. 1391, Cambridge, MA. Washington, DC. Scott, J.M., Davis, F., Csuti, B., Noss, R., Butter®eld, B., Caicco, S., Bailey, R.G., 1996. Ecosystem Geography. Springer, New York. Grooves, C., Ulliman, J., Anderson, H., Wright, R.G., 1993. Gap analysis: A geographic approach to protection of biological diver- Bailey, R.G., Jensen, M.E., Cleland, M.T., Bourgeron, P.S., 1994. Design and use of ecological mapping units. In: Jensen, M.E., sity. Wildlife Monographs 123, 1±41. Bourgeron, P.S. (Eds.). Ecosystem Management: Principles And Trewartha, G.T., 1968. An Introduction to Weather and Climate. Applications, Vol. 1. US Forest Service General Tech. Rep. PNW- McGraw-Hill, New York. US Department of Agriculture, Agriculture Experiment Station. GTR-318, pp. 95±106. Davis, F.W., Stine, P.A., Stoms, D.M., Borchert, M.I., Hollander, (1957) Soils of the North Central Region of the United States. A.D., 1995. Gap analysis of the actual vegetation of California 1. North Central Region Publ. 76, Bulletin 544. Madison, WI. The southwestern region. Madrono 42, 40±78. Walter, H., Box, E., 1976. Global classi®cation of natural terrestrial ecosystems. Vegetation 32, 75±81. Dice, L.R., 1943. The Biotic Provinces of . University of Michigan Press, Ann Arbor. Weins, J.A., Crawford, C.S., Gosz, J.R., 1985. Boundary dynamics: A Edwards, T.C., Deshler, E.T., Foster, D., Moisen, G.G., 1996. Adequacy conceptual framework for studying landscape ecosystems. Oikos 45, 421±427. of wildlife relation models for estimating spatial distributions of terrestrial vertebrates. Conservation Biology 10, 263±270. Wiken, E.B., Ironside, G., 1977. The development of ecological (bio- Fenneman, N.M., 1928. Physiographic divisions of the United States. physical) land classi®cation in Canada. Landscape Planning 4, 273± Annals Association of American Geographers 18, 261±263. 275. Williamson, D. (Ed.), 1996. Conservation by design: a framework for Fenneman, N.M., 1931. Physiography of the Western United States. McGraw-Hill Co, New York. mission success. The Nature Conservancy, Washington, DC. Gallant, A.L., Binnian, E.F., Omernik, J.M., Shasby, M.B., 1995. Wright, R.G., MacCracken, J.G., Hall, J., 1994. An ecological eval- Ecoregions of Alaska. Professional Paper 1567, US Geological Sur- uation of proposed new conservation areas in Idaho: evaluating proposed Idaho national parks. Conservation Biology 8, 207±216. vey, Washington, DC.