BOSTON PUBLIC LIBRARY 3 9999 06317 730 5 /*/ A PRELIMINARY CLASSIFICATION OF WETLAND PLANT COMMUNITIES IN NORTH-CENTRAL MINNESOTA UNITED STATES DEPARTMENT OF THE INTERIOR FISH AND WILDLIFE SERVICE BUREAU OF SPORT FISHERIES AND WILDLIFE Special Scientific Report-Wildlife No. 168 UNITED STATES DEPARTMENT OF THE INTERIOR Fish and Wildlife Service Bureau of Sport Fisheries and Wildlife A PRELIMINARY CLASSIFICATION OF WETLAND PLANT COMMUNITIES IN NORTH-CENTRAL MINNESOTA By Lewis M. Cowardin and Douglas H. Johnson U. S. Bureau of Sport Fisheries and Wildlife Northern Prairie Wildlife Research Center Jamestown, North Dakota 58401 Special Scientific Repoitt--Wi ldl ife No, Washington, D.(f. • 1973 168 For sale by the Superintendent of Documents, U.S. Government Printing Office Washington, D.C. 20402. Price: 55 cents, domestic postpaid; 40 cents, GPO Bookstore Stock No. 241fJ-00358 CONTENTS Page Abstract ii Introduction 1 Structure of classification 1 Study area 2 Methods 2 Description of the communities 5 Communities without permanent water 5 Communities with permanent water 9 Important plants of the communities 10 Discriminant analysis 10 Conclusions 15 Literature Cited 18 Appendixes 19 ABSTRACT A classification of wetland plant communities was developed for a study area in north-central Minnesota in order to analyze data on waterfowl use of habitat that were gathered by radio telemetry. The classification employs features of several earlier classifications ir, addition to new classes for bogs and lakeshore communities. Brief descriptions are given for each community, and the important plant species are listed. Discriminant function analysis was used for 40 plant species. Seventy-five percent of the stands studied were classified correctly by this technique. Average probabilities of assignment to communities were calculated and helped to identify distinct and poorly defined communities as well as the relationship among communities. ii INTRODUCTION ' The system described here was designed with the primary purpose of placing areas of wetland vegetation into categories that would be meaningful for analysis of data concerning waterfowl use. These data were gathered on a study area in north-central Minnesota by means of radio- telemetry, and the method of analysis required the preparation of maps on which all wetland areas could be divided into discrete units as described by Gilmer et al . (in press). This is not an entirely new classification, but rather a modification of existing systems described by Martin et al_. (1953), Stewart and Kantrud (1971) and Gilmer (1971). Although the classification is restricted both in purpose and in the geographic region to which it may be applied, portions of the system as well as the methodology used to describe the types may have general value in studies of wetlands outside the immediate study area. The classification proposed by Martin et al_. (1953) and used by Shaw and Fredine (1956) is the system that has been most generally used in water- fowl research and management. This system fulfilled one of our requirements in that it can be used to describe areas of vegetation and is not restricted to discrete wetlands. The classes, however, are far too broad for detailed investigation of use of habitat by waterfowl, and many biologists find it of limited value. The classification described by Stewart and Kantrud (1971) has sufficient detail for research but, for us, had the disadvantages of being restricted to the glaciated prairie and of omitting many of the wetland types most important in our area. Their classification is strictly of wetland basins. In the prairies, for which it was designed, the wetland basins tend to be discrete, whereas in our area, basins consist of a complex of plant communities often in association with large lakes and river systems. Structure of Classification The basic unit of our classification is the stand, which we define as an area of wetland vegetation with sufficient uniformity and size so as to be recognizable both on the ground and on aerial photographs. Stands with similar botanical and physical characteristics were grouped into communities in the sense of Curtis (1959:477), "...groups of stands with sufficient characters in common to produce studiable assemblages." Wetlands, such as lakes and rivers, are often large and contain diverse communities that are related only by the fact that they lie in the same basin or channel; therefore, we grouped lake and river communities into wetland complexes when we thought that these complexes would be meaningful for the analysis of waterfowl-use data. In some cases our communities are closely related to, or contained within, wetland types described by other authors (Table 1). We have attempted to indicate some of these relationships through the selection of community names. Table 1. Comparison of communities described in this study with existing wetland classification. Community Stewart and Kantrud (1971) Martin et_ a]_. (1953) (1) Ephemeral (2) Temporary (3) Seasonal-closed (4) Seasonal -patchy (5) Seasonal-open (6) Semipermanent- closed (7) Semipermanent- patchy (8) Semipermanent- open (9) Shrub swamp (10) Hardwood swamp (11) Circumneutral bog - sedge phase (12) Circumneutral bog - shrub phase (13) Circumneutral bog - ericaceous phase (14) Acid bog (15) Softwood swamp (I) Ephemeral (E) ]_/ (1) (IIA) Temporary-fresh (E) CO (2) (IIIA-1) Seasonal-fresh, cover type 1 (E) (3) (4) (IIIA-2) Seasonal -fresh, cover type 2 (E) (3) (4) (IIIA-3) Seasonal -fresh, cover type 3 (E) (3) (4) (5) (IVA-1) Semipermanent- fresh, cover type 1 (E) (3) (4) (IVA-2) Semipermanent- fresh, cover type 2 (E) (3) (4) (IVA-3) Semipermanent- fresh, cover type 3 (E) (3) (4) (5) - (6) - (7) (VII) Fen (alkaline bog) ponds (R) (2) Seasonally flooded basins or flats (I) Seasonally flooded basins or flats (R) Fresh meadows (R) Shallow fresh marshes (R) Deep fresh marshes (R) Shallow fresh marshes (R) Deep fresh marshes (R) Shallow-fresh marshes (R) Deep fresh marshes (R) Open fresh marshes (R) Shallow-fresh marshes (R) Deep fresh marshes (R) Shallow fresh marshes (R) Deep fresh marshes (R) Shallow fresh marshes (R) Deep fresh marshes (R) Open fresh marshes (R) Shrub swamp (E) Wooded swamp (I) Fresh meadows (I) (8) Bogs (E) (7) Wooded swamp (I) 1/ E - equivalent type. R = related type. I - our type included in theirs Study Area Data for this classification were gathered within the Chippewa National Forest on a study area 25 square miles (65 sq. km.) in area and situated 12 miles (19.3 km.) east of Bemidji, Minnesota. The area is glaciated and has till deposits in the north and east and outwash in the south and west (Goltz 1969). Small wetlands are numerous, and the area is surrounded by large lakes. The Mississippi River flows through the southern portion. Domi- nant forest types are pines (Pinus banks iana, £_. resinosa, f_. strobus) , 1/ aspen (Populus tremuloides, P_. grandidentata, P_. balsaminifera), northern hardwood (Acer saccharum> Til ia americana) , "and oaks, (Querelas ell ipsoidalis, Q. macrocarpa) . riydric sites are dominated by black spruce (Pi cea mariana) and tamarack (Larix laricina) . A detailed description of the study area was given by Gilmer (1971 :4-14) . The field work for this study could not have been completed without the help of D. S. Gilmer, I. J. Ball, J. H. Riechmann, R. S. Stott, and T. C. Clodfelter. R. E. Stewart furnished guidance and criticism in developing the classification. We are indebted to D. A. Davenport for assistance with computer programing. V. A. Adomaitis and J. A. Shoesmith performed chemical analyses. P. F. Springer made a critical review of the manuscript. METHODS Stands of wetland vegetation were delineated and mapped from multispectral photography (Cowardin and Myers in press) . Stands of vegetation were then located by means of the photographs, hand compass, and pacing. During the field survey, each plant species within a stand was described in terms of the percent of areal coverage and placed into one of the following categories: u = absent; 1 = rare, a few scattered individuals; 2 = occasional, less than 1 percent; 3 = fairly common, 1 - 10 percent; 4 = common, 11-50 percent; 5 = abundant, 51-100 percent. Plant abundance was described both for the stand as a whole and for the deepest, central portion of the stand. This method has the disadvantages of being partially subjective and somewhat imprecise, but has the decided advantage of speed which allowed a far larger sample of stands to be analyzed than would have been possible if more refined methods such as quadrats were used. All classification was conducted by the senior author during July and August, 1968-1972. The timing of the survey probably introduced some bias by not giving sufficient importance to early - and late-blooming species. In wetland stands surveyed in 1968-70, a water sample was gathered from the surface, preserved with chloroform and chemically analysed at the Northern Prairie Wildlife Research Center. No samples were taken in 1971- 72. 1/ Nomenclature used in this paper follows Fernald (1950) By examination in the field, stands were assigned to communities accord- ing to their physical and botanical characteristics. Species used as indicators of permanence by Stewart and Kantrud (1971) were used whenever possible to help classify communities equivalent to theirs. For community descriptions (Appendices A - 0) , plants were ranked in importance by averaging the abundance categories for each plant in all stands assigned to a community. Only the plants of the deepest, central portion of the wetland were used for ranking. For the purpose of description, character- istic plants appearing in wetlands were limited to those species with average abundance value greater than 0.3. In order to describe the distribution of the common plant species among the communities, an index of abundance was developed. We divided the number of stands in which a species occurred in a given community by the sample size for that community in order to obtain a frequency of occurrence. This frequency divided by the sum of frequencies for all communities furnished an index to the abundance of that species as an indicator of that community. The following hypothetical example illustrates the procedure used for calculating frequency of occurrence and index of abundance: Commun i ty A B C Total Sample size Stands with plant Frequency of occurrence Index of abundance A discriminant function analysis (e.g. Tatsuoka 1970) was applied to the abundance data in order to study several aspects of the classification: (1) how well the communities describe natural associations of species, (2) the relationship between different communities, (3) which plants best separate communities, and (4) which wetlands stands are intermediate between types. Data were processed on an IBM 360 Model 50 computer by means of a Biomedical Computer Program (Dixon 1968). Forty plant species were chosen for use in the discriminant analysis by first eliminating all species that occurred in less than four stands, except where all those stands were in the same community. This reduced the number of plants under consideration from 241 to 129. Of the remaining 129 species, 40 were chosen by determining which species first entered the stepwise discrimi- nant analysis. 5 10 25 40 5 5 5 15 100.0 50.0 20.0 170.0 58.8 29.4 11.8 100. u ■4- RESULTS Description of the Communities Communities without permanent water These communities include all wetlands that can become dry, at least during drought conditions. The communities were separated primarily on water permanence and in many cases are equivalent to wetland types of either Martin et al. (1953) or Stewart and Kantrud (1971) (Table 1). Communities were further separated on the basis of the amount of emergent cover present, a separation similar to the cover types described by Stewart and Kantrud (1971:10-11). Plant species are generally similar to those found by Stewart and Kantrud but there are important differences. For example, wetlands in our area are fresh (specific conductance seldom exceeds 300 ymhos). Also, the area is forested, so tree and shrub vegetation forms an important component of several types. The numbers and total acreage of stands in each community were highly variable, and communities 6 and 7 (Semipermanent - closed, and Semipermanent - patchy) were rare on our study area (Table 2). Chemical characteristics of water samples from stands without permanent water are presented in Table 3. The characteristic plants for communities 1-15 are presented in Appendices A-0. Community 1 . Ephemeral -- Occurs in small depressions where water is held by frozen ground in early spring. The water disappears soon after the thaw in late April or early May. The flora is highly variable and composed primarily of non-wetland species. Community 2. Temporary -- Occurs ir small depressions throughout the study area. Basins are usually dry by late May but are often replenished by heavy rains later in the season. This community is intermediate between ephemeral and shrub swamp. It is also closely related to hardwood swamp which is generally more permanent and has swamp hardwood tree species with crown closure greater than 50 percent. Community 3. Seasonal - closed -- Occurs most commonly in depressions in moraine or till plains but occasionally on outwash. The deepest portion has a closed cover of emergent, shallow-marsh vegetation. Stands usually hold water until late July. The most common plant species are coarse-leaved sedges such as Carex lacustris . -5- Table 2. Number and size of non-permanent wetland stands occurring on a 25-square-mile study area in north-central Minnesota. Community Number Median SIQR 1/ Total Number Stands Size (acres) (acres) Plant Surveys (1) Ephemeral 11 0.96 0.53- 1.38 1.4 6 (2) Temporary 61 0.88 0.49- 1.26 6.7 18 (3) Seasonal-closed 77 1.41 0.76- 2.49 18.5 47 (4) Seasonal -patchy 9 6.U0 2.25- 11.25 7.6 8 (5) Seasonal-open 132 1.78 0.97- 2.48 33.0 63 (6) Semipermanent-closed 5 2.75 1.75- 3.38 1.3 5 (7) Semipermanent-patchy 5 6.00 3.75- 7.25 11.4 5 (8) Semipermanent-open 9 13.00 6.12- 16.25 18.8 7 (9) Shrub swamp 124 11.75 3.83- 35.50 426.6 48 (10) Hardwood swamp 89 1.09 0.59- 1.81 19.6 14 (11) Circumneutral bog - sedge phase 77 33.5 14.0 - 71.0 488.2 31 (12) Circumneutral bog - shrub phase 61 19.00 7.75- 39.25 216. b 23 (13) Circumneutral bog - ericaceous phase 31 11.00 5.25- 20.75 51.8 24 (14) Acid bog 37 18.25 8.75- 39.25 149.9 27 (15) Softwood swamp 44 42.50 21.50- 101.50 438.7 11 Total 772 1890 1/ Semi-interquartile range was used as measure of dispersion because of highly skewed distribution of size measurements. Community 4. Seasonal - patchy -- Occurs in situations similar to Community 3 and is like that community except that the emergent, shallow- marsh vegetation occurs as dense or scattered clumps. Stands usually hold water until early August or mid-August. Community 5. Seasonal ■ open -- Occurs in situations similar to is like these except that in this community Communities 3 and 4 and the center of the stands have little or no emergent shallow-marsh vegetation. Stands usually hold water until late August or early September Community 6. Semipermanent - closed -- Occurs in deeper basins than Communities 3-5 and is more common in moraine than outwash plains. The deepest portion of the stands have a closed cover dominated by deep-marsh species. They usually hold water until late fall and may be wet throughout the year. -6- Table 3. Chemical characteristics of water samples from wetland stands without permanent water in north-central Minnesota. ]_/ Community Sample Size Median pH Median Cond. 2/ Median TDS 3/ Median 0M 4/ (2) Temporary 2 7.5 855.5 223.5 98.0 (3) Seasonal-closed 15 7.2 (6.5-7.7) 80. U (62.3-192.8) 162.0 (90.4-291.8) 63.0 (55.3-129.8) (4) Seasonal -patchy 4 7.2 (6.9-7.5) 88.0 (72.5-105.0) 117.5 (94.5-139.5) 61.5 (40.5-96.5) (5) Seasonal -open 31 7.2 (6.6-7.7) 97.0 (56.3-137.8) 133.0 (94.3-192.0) 55.0 (43.3-98.8) (6) Semi permanent-cl osed 1 7.1 49.0 78.0 48.0 (7) Semi permanent-patchy 3 6.2 41 .0 80.0 46.0 (8) Semi permanent-open 6 7.9 (7.5-8.1) 165.0 (94.3-185.0) 156.0 (101.0-206.8) 45.5 (34.0-66.0) (9) Shrub swamp 9 7.1 (6.7-7.8) 116.0 (108.8-144.3) 187.0 (150.8-259.8) 66.0 (54.8-136.3) (10) Hardwood swamp 6 7.4 (7.3-7.6) 88.0 (71.0-350.0) 127.0 (120.0-223.0) 52.0 (35.0-84.0) (ID Circumneutral bog - sedge phase 16 7.5 (6.8-8.0) 114.0 (64.0-176.5) 160. U (113.0-195.5) 55.5 (37.6-80.0) (12) Circumneutral bog - shrub phase 11 7.2 (5.3-7.7) 150.0 (98.3-204.8) 242.0 (145.3-330.8) 55.3 (117.'o-200.8) (13) Circumneutral bog - ericaceous phase 8 4.7 (4.5-5.2) 53.5 (40.0-86.0) 214.5 (93.5-214.5) 90.0 (57.5-145.5) (14) Acid bog 9 4.4 (4.1-5.3) 71.0 (55.8-88.3) 220.0 (181.8-267.8) 166.0 (106.8-185.3) ]_/ Semi-interquartile ranges for all measurements shown in parentheses, iio range calculated for sample size less than 4. No samples obtained from types 1 and 15. 2/ Specific conductance in ymhos. 3/ Total dissolved solids in ppm. 4/ Organic matter in ppm. -7- Community 7. Semipermanent - patchy -- Occurs in basins primarily in till plain and outwash. Basins are generally larger than those of communities 3-6. Their deepest portions have scattered or dense patches of emergent, deep-marsh vegetation. They usually hold water throughout the year and are dry only under moderate drought conditions. Community 8. Semipermanent - open -- Occurs in medium-sized basins in moraine and outwash and is similar to Community 7 except that the central, deepest portion of the stands has little or no emergent vegetation. The deepest portion of the stand may have submerged, deep- marsh vegetation or may be without vegetation. These stands become dry only under severe drought conditions. Community 9. Shrub swamp -- Occurs throughout the study area in small pockets and depressions as well as adjacent to circumneutral bogs and softwood swamps. Stands comprise parts of large wetland complexes. They have a dense overstory of shrubs covering more than 50 percent of the deepest zone. Permanence is highly variable depending on size and depth, but stands generally become dry by mid-summer. Community 10. hardwood swamp -- Occurs predominantly in small depressions within northern hardwood or aspen forest types. Crown closure is greater than 50 percent over the deepest portion of the stand. Stands often have deep pools that may hold water until late August. Community 11. Circumneutral bog - sedge phase -- Occurs either in large flat basins within till plains and outwash, or on large flat areas bordering rivers and lakes. Stands are found on organic soils and are often highly quaking underfoot. The soil is generally saturated throughout the year and may be flooded in the spring. The vegetation is dominated by fine-leaved sedges such as Carex lasiocarpa. The water has a pH near neutrality. Community 12. Circumneutral bog - shrub phase -- This community is intermediate between Communities 9 and 11 and is often situated between stands of these communities, although it may occur in isolated depressions. The community is similar to Community 11 but has shrubs other than those of acid bogs covering 20-50 percent of the area. The water has a pH near neutrality. Community 13. Circumneutral bog - ericaceous phase -- This type is intermediate between Communities 11 and 14 and is often found as a ring peripheral to acid bog stands. It is also similar to Community 11 but is characterized by small distinct patches and clumps of ericaceous bog plants and Sphagnum spp. moss totaling 20 to 50 percent of the area. The water is more acid than in Communities 11 and 12, because this community is transitional between Communities 11 and 14. Community 14. Acid bog -- Occurs throughout the area, often in associa- tion with softwood swamps. The community is characterized by ericaceous shrubs and Sphagnum spp. moss covering 50-100 percent of the area. Tree cover is less than 20 percent. Community 15. Softwood swamp -- Occurs in low-lying areas throughout the study area. It is typified by coniferous trees covering 20-100 percent of the area. Communities with permanent water The communities described above are often found in association with communities that have permanent water. We have found it useful to group communities having permanent water into wetland complexes that are meaningful for the analysis of wetland use by waterfowl. Lake complexes were classified on the basis of the physical and botanical characteristics of shorelines. River complexes were classified by the ratio of emergent vegetation to water in the channel : bog lake complexes - The shoreline is composed of 80 to 100 percent bog communities (11-14). Intermediate lake complexes - These complexes are intermediate between bog lakes and sand lakes. The shoreline is composed of 50-80 percent bog communities (11-14) or with shores grading from deep marsh to shallow marsh aquatics. Sand lake complexes - The shoreline has less than 50 percent bog communities (11-14). Fifty percent or more of the shoreline is characterized by sandy or boulder-strewn beaches or shallows. Some areas may have steep banks with overhanging brush. River channel complexes - Permanent flowing water in which less than 50 percent of the cross section of the river contains emergent vegetation. Kiver marsh complexes - Permanent flowing water in which more than 50 percent of the cross section of the river contains emergent vegetation. The communities within these habitat complexes have permanent water and were divided on the basis of the dominant species of emergent vegetation. All stands with submerged vegetation were included within Community 16 (open water), because of the difficulty in delineating these stands on aerial photographs. Lakeshore communities were classified on the basis of both physical character- istics and the vegetation within one chain (20.1 m.) of the shoreline. The shoreline classification was modeled after and used for the same reason as the classification described by Gilmer (1971:23) who demonstrated that shorelines should be considered separately from the communities on either side. Communities with permanent water are described below: Community 16 - Open water -- Includes any area that is without emergent and emersed (floating leaved) vegetation at all times during the year. Stands with submerged beds of vegetation were included in this community. Community 17 - Bulrush -- Includes any stands of emergent vegetation dominated by hardstem bulrush (Scirpus acutus). Community 18 - Phragmites -- Includes any stand of emergent vegetation dominated by reed grass (Phragmites communis) . Community 19 - Wild rice -- Includes any stand of emergent vegetation dominated by wild rice (Zizania aquatica). These stands were classified on the basis of the presence of wild rice during the height of its development. For analysis of waterfowl use data in spring, they were classed as open water (Community 16). Community 20 - Emersed aquatics -- Includes any stand dominated by aquatic vegetation with floating leaves, such as Nuphar, Nymph aea, and Potamogeton. Community 21 - Other emergents -- Includes all stands of emergent vegetation which can not be classified as 17, 18 or 19. For uniformity in numbering, shoreline communities begin with number 30; therefore, numbers 22-29 are not used in this classification. The communities used to describe shorelines (Table 4) are self explanatory. The description starts with the stand in the lake and ends with the physical type of shorel ine. Host of the stands in permanent water were classified from aerial photographs, and no plant lists were prepared for communities 16-69 because the complexity and vast number of stands made botanical surveys impractical. Important plants of the communities In order to understand the importance of each species within each community, we examined the indices of abundance for each of the 40 species (Table 5) used in the discriminant analysis. These data demonstrate that some of the most common species, such as Carex lacustris, are not particularly valuable for classification because they occur in almost all communities. Discriminant Analysis Discriminant analysis was developed by Fisher (1936) in order to pro- vide a rigorous basis for using quantitative data for taxonomic purposes. It presupposes a correct assignment of individuals to groups, and determines how the groups can best be distinguished on the basis of a set of measure- rents. It is most useful (Gower 1969) in situations where distinctions between groups are not clear-cut and where gradation between groups may occur. This is the usual situation with the classification of plant com- munities. -10- Table 4. Communities used to describe shorelines of lakes One Chain (20.1 m.) Community Strip in Lake Shoreline 30 Water - river mouth 31 Water - sand or gravel beach 32 Water - deep marsh aquatics - shallow marsh aquatics 33 Water - bog mat 34 Water - sand bar - shrub swamp 35 Water - sand bar - pond 36 Water - overhanging - brush 37 Water - steep bank 38 Water - rock and boulder 39 Water - residential 40-49 Bulrush - (same as shown for types 33-39) 50-59 Phragmites - (same as shown for types 30-39) 60-69 Wild rice - (same as shown for types 30-39) For the purpose at hand, wherein a classification has already been assumed, discriminant analysis offers several advantages: (1) Being a multivariate technique, it makes fuller'use of the data than do univariate approaches such as indices of abundance (Table 5). Moreover, it takes into account the differing precision of variables and also interrelationships among them. (2) By applying the discriminant functions to the data on which they are based, one determines whether or not each observation is correctly classified by the procedure, floreover, a "probability" is calculated which reflects the likelihood of that observation belonging to each of the groups. The obser- vation is assigned to the group for which its probability is greatest (Table 6). This resubstitution of the original data into the model is useful for several purposes. The most immediate use is to point out observations that may contain obvious errors. A quick examination of those observations that the method assigned to a group very distinct from the group to which it belonged often reveals a transcribing error in the data. We applied discriminant analysis to our raw plant data and field-checked all stands that were misclassified by the method. In this way we detected several recording errors and several cases where important species had been omitted from stand descriptions. These errors were corrected, and the analysis was rerun. -11- oc\j^j-cr»o->c\JcriLnc\Jcn<— cMaiiDcsjr^^-ojN^cvjmfOO'^^uDaiorOi — lo ■ — coco^i — <^r lo lo cm co cm cm < — ou~> ■ — ix) rv id cm " — ■ — lo fOi — i — ■ — ^- cm cm i — coo ■ — < — CO i — co lo i — i — «=3- cmo co r-* C\J cm co i— o r^. Lor-.ro lo ■ — r^- o c*~> « — ld cm ^ rococo oo n LO i — LO i— LO LO cm r- C\J i — CO r— LO co lo r-. o co en lo en cxi CO <3- ^ ID CO CO CO i— CO r-* lo <3- O ^O CO CM CO LO i — «3- r^~ lo -=d- LO C\J CO 00 COO LOIC LO LO CM CM co I — CTt LO p^ i — *d- en r-^ -=1- cm LO cr> cm co o ^3- ■=1- co co ■3- 10 r^ co co en CM i — i — CM CO CO . — CO CO CM CM CO CM i — CO CTt CM CO CO CT> LT> "3" CO <3" CT) CTI CM LO <3- LO COi i — CM > — LO CO LO •=& CO CO Ci >* <* r— LD CO CO i — CO >* CM CM I— CO COCM LO CO i — COi — COi CO CO I— <* r~ co o <^j- co lo co ^t- CTI CM CO r^ lo lo LO LO ID LD1* LO o «=r o •^" l LO LO LO >3- id > E in o : >, o id i — •!—(/) O- (V -o c i- u c m n3 IO Q _ o E o a. 3 _ ES-S-C-"-C713El04->U 3iOiaroS-3CLO-i-c:rO 4-> ■— E .a id s- u -o •■- ,— d> +J ID W XO l«tJ«lr- X X X X r-r OJ D-r 3 3 1U Q) (U OJ _3L.UCD)C-(JL.L.S-L, cl| ct id -(— -r— fb i — dJ id id id id jacn/i S- >> X id E Q- o -o s_ ai en LO "O •i- to I— C CO fO n3 -i- 3 O f— O" ro id •.- s- 4-> O LO LO _0 3 o i- S- ID 3 cn-i- O ID i- 0) Ed) Q- fT3 U ° « ^ i — fo i — - 00 ro aj c •r- CO ID 3 01 •1 — N E dl IO J- dJ O CO C c •i- S- ■o ■i— E +-> ■o O o • i — " 3 -E d) <0 r - C3 L0 >r— s_ •1 — E a. E S- S- +-> c O 4-> 3 O E O ID ID . — (D CO E CO >) E IO n- ID (0 ID i— i — S- CO 3 1 '1 — C o S- o 1 — o •1 — cn S- C IO 3 o ■i — I — JZ i — ID 4- Dl L0 o o o s_ d> C7) •l- CL E , — i — i — u o O ■ 3 o Q_ n3 E +J "i— ■r— D- cu o ■■- OJ M- ■r— C C cx-o L0 ■1— • i — 3 S- C i — ID s- u o +-> E -i- S- 0) o o LO ID d) 00 to s- E IO (G ID i — +j CO ID S- 3 +J | > +J +J i — 4J s- r— io u cr > LO i — • 1 — S- -t- •i- ID ID dl cu E O ■ I— 01 d) 1— £ -r - 3 ro -r- CO -a D-:j= C i— fO 3 a C7> 3 O E > -o 3 +-> U S_ C ai +j 3 ID CO o O E C a o o cz m IO -,— ID C r — X X LO Ol ID t/> E E ai id a i — i— jn ID CL E 3 +J 4-> J= d) £ s- o-+-> fO E 3 S- ■i- O o ID ID •!- o- >> i — ■ i — o O CL S- CI- t—1 oo => E CO z Lj_ O Q. o_ CO CO cn CO 1— =1 CO Q o. CO Q -12- >> to c +-> c ) 4- +-> i — o (/) aj V) s- ro S- , — o o o >> +J -a ■a ■ 1 — cu , — c c 01 3 o- ■i — E ■1— LL- E in O t/> C o =t ■r— r— cnj cm n oo o o ' — aiiDcouicomo co i^ ' — iDy3o**f^r^coii)ixtcocoo cooor^coro-iLnr-.co^3-i — ro <3- i — . ■ — sd" CO ■* I — CO CNJ CM CVI , i— r-Wr-W ■3- ro ld ooli— ro ro i— cm in] ro lt>|cni cm cnj i— ro "=1-1 i— ro u->|lt> cxi <* LT> cnj ro ro ,— <^- cm ro ro cm ■*l -a QJ cu -a +-> o ai s- s_ o o ai S- B cxi ro -^r \n co i^ co ch o ■ — CM CO >* LT> •13- (3) If, as was done in the present study, a stepwise algorithm is used to calculate the discriminant functions, the order in which a variable enters into the model relates to the importance of that variable in separating the groups. This relationship is less exact when the groups are widely dis- parate in sample size because a variable which distinguishes a group with few samples may enter later than a variable which distinguishes a well -represented group. Also, the ordering of variables can be used to ascertain which variables are superfluous to the classification. These may then be omitted from further data gathering (Norn's and Barkham, 1970). We used this technique in reducing to 40 the sample of plant species to be included in the analysis. (4) It is possible to locate observations which are intermediate between groups by considering the probability of assignment to each group. If, for a particular observation, the probabilities associated with the groups are each about 0.5, that observation would represent an intermediate between those groups. Table 7 illustrates a wetland stand intermediate between Community 9 (shrub swamp) and Community 12 (circumneutral bog - shrub phase). A comparison of the plant species present with those presented in Appendices I and L illustrates the intermediate nature of the stand. Note that Calamagrostis canadensis is ranked high in both communities. Typha lati folia is absent in shrub swamp but ranked highly in circumneutral bog - shrub phase. Corn us stolon if era and Caltha palustris are represented in shrub swamp but absent from circumneutral bog - shrub phase. (5) Another useful concept arises from averaging those probabilities of classification across all observations in a group. This will suggest which other groups are closely related to that particular group. This procedure was used to illustrate the relationship between plant communities (Figure 1). If the average probability that the observations in a particular group are assigned correctly to that group is near 1.0, it follows that the group is internally consistent and, on the basis of the variables measured, represents a natural grouping. Note the consistency within communities 6, 10, 13, 14, and 15 (Figure 1). The fact that the model was able to assign 74.8 percent of the stands to their proper communities also demonstrates that the communities described are consistent with the groupings of plants occurring in nature. It should be pointed out that if the same model were used to classify additional wetlands, we would not expect an equally high proportion of correct classifications. The average probabilities demonstrate the relationships between types and identify those types that are poorly defined. Community 15 (softwood swamp) was the best defined; 100 percent of the stands were classified correctly with the use of 40 plant species. Community 7 (semipermanent- patchy) was the most poorly defined; only 40 percent of the stands were -14- Table 7. Plant abundance values for species in a stand intermediate between Community 9. Shrub swamp and Community 12. Circumneutral bog - shrub phase. 1/ 2/ Species Ab undance Mean Ab undance Mean Abundance in Stand Shrub Swamp Circumneutral Bog • - Shrub Phase Calamagrostis canadensis 5 2.3 2.5 Alnus rugosa 3 3.0 1.7 Carex aquatilis 3 0.5 0.4 Carex rostrata 3 0.6 0.7 Gal ium trifidum 3 0.6 0.8 Salix discolor 3 2.8 1.7 Typha lati folia 3 - 1.7 Dryopteris thelypteris 2 1.4 0.8 Cornus stolonifera 2 0.7 - Caltha palustris 2 0.4 - Salix gracilis 2 2.7 2.6 1/ Discriminant analysis assigned probability of 9 as 0.34 and probability of 12 as 0.65. 2/ Only plants from the list of the 40 used in discriminant analysis are shown for this stand. correctly classified. This is probably due in part to the small sample size for this community. The cover types (closed, patchy, and open) are relatively difficult to separate on the basis of the plant abundance. This is as would be expected because these types are closely related (see Figure 1). Communities with shrub cover, Community 9 (shrub swamp) and Community 12 (circumneutral bog - shrub phase) are also related (Figure 1). CONCLUSIONS The classification presented here can be used for most wetland communities in north-central Minnesota. It is based on limited data from a small area, and extension to other areas will require modifications and additions. An attempt has been made to adopt the most effective features of existing classifications in order to develop a system that can be used for analysis of radio-telemetry data on waterfowl habitat use. The communities in Minnesota are related to those described by Stewart and Kantrud (1971) on the glaciated prairies. One major difference between the areas is the fact that the wetland communities in our area vary only slightly in salinity, and therefore the subclasses as used by Stewart and Kantrud are unnecessary. There is considerable variation in pH among types, -15- z LU Z to to < >- Z> O o 1 2 | 3 | m 4 - 1 . 5 , 6 1 7 8 1 9 1 10 -1 11 1 _ 12 1 13 1 14 -1 15 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 AVERAGE PROBABILITY (HEIGHT OF BAR) OF ASSIGNMENT TO COMMUNITY Figure 1. Average probability of assignment of stands to each of 15 plant communities. Probabilities were determined from discriminant function analysis of plant abundance for 40 species. •16- and the acid bog communities represent the most acid conditions. Although the cover type as used by Stewart and Kantrud is a meaningful concept, cover types in our area are more closely correlated with permanence than in the prairies; the more open wetlands are the most permanent. Discriminant function analysis proved to be a useful tool for the analysis of plant abundance data, aiding in understanding the relationships among types, and obtaining a quantitative estimate as to which species are most useful as indicators of wetland plant associations. iJorris and Barkham (1970) also found discriminant analysis to be a useful analytical tool for studying plant communities. We believe that the quantitative multivariate approach to wetland classification is logical and promising, and we suggest that further data should be gathered over wide areas with maximum variation in ecological types in order to develop wetland classifications that are simple enough to be useful yet reflect the complexity and interrelationships of wetland plant communities. -17- LITERATURE CITED Cowardin, L. M. , and V. I. Myers, (in press). Uses of remote sensing for the identification and classification of vegetation in a forested region. J. Wild! . Manage. Curtis, J. T. 1959. The vegetation of Wisconsin - an ordination of plant communities. Univ. of Wisconsin Press, Madison. 657 pp. Dixon, W. J. 1968. ed. Stepwise discriminant analysis, pp 214a - 214t j_n BMD Biomedical Computer Programs. Univ. of California Press. Berkeley. 600 pp. Fernald, M. L. 1950. Gray's manual of botany. American Book Co., New York. 1632 pp. Fisher, R. A. 1936. The use of multiple measurements in taxonomic problems. Ann. Eugenics 7:179-188. Gilmer, D. S. 1971. Home range and habitat use of breeding mallards (Anas platyrhynchos) and wood ducks (Aix sponsa) in north-central Minnesota as determined by radio tracking. Ph.D. thesis. Univ. of Minnesota. 142 pp. Gilmer, D. S., S. E. Miller, and L. M. Cowardin (in press). Analysis of radio-tracking data using digitized maps. J. Wildl. Manage. Goltz, G. E. 1969. Soils of the Chippewa National Forest. U. S. Dept. of Agric. For. Serv., Eastern Region. 14 pp. Mimeogr. Gower, J. C. 1969. A survey of numerical methods useful in taxonomy. Acrologia 11 (3) : 357-375 . Martin, A. C, in. Hotchkiss, F. M. Uhler and W. S. Bourn. 1953. Classi- fication of wetlands of the United States. U. S. Fish Wildl. Serv. Spec. Sci . kep. Wildl. 20. 14 pp. Morris, J. M. , and J. P. Barkham. 1970. A comparison of some Cotswold beechwoods using multiple-discriminant analysis. J. Ecol . 58(3) :603-619 . Shaw, S. P., and C G. Fredine. 1956. Wetlands of the United States. U. S. Fish and Wildl. Serv. Circ. 39. 67 pp. Stewart, R. E., and H. A. Kantrud. 1971. Classification of natural ponds and lakes in the glaciated prairie region. U. S. Fish and Wildl. Serv. Resour. Pub! . 92. 57 pp. Tatsuoka, M. M. 1970. Discriminant analysis: the study of group differences. Inst, for Personality and Ability Testing. Selected topics in advanced statistics 6. Champaign, Illinois. 57 pp. -18- Appendix A. Characteristic plants of Community 1. Ephemeral Species Mean Abundance Agropyron repens ]_/ 1.7 Calamagrostis canadensis 0.8 Maianthemum canadense 0.8 Fraxinus nigra 0.8 Populus tremuloides 0.8 Trifol ium repens 0.8 Equisetum sylvaticum 0.7 Aster sp. 0.7 Ulmus americana 0.7 Poa pratensis 1/ 0.7 Impatiens capensis 0.5 Cory! us cornuta 0.5 Taraxacum officinale 1/ 0.5 Lactuca canadensis 0.5 1/ Plants listed as indicator for type by Stewart and Kantrud (1971) ■19- Appendix B. Characteristic plants of Community 2. Temporary, Species Mean Abundance Calamagrostis canadensis ]_/ 1.1 Equisetum sylvaticum 0.8 Iris versicolor 0.8 Scirpus pedicellatus 0.8 Carex sp. 0.8 Dryopteris thelypteris 0.7 Scutellaria epilobii folia 0.7 Populus tremuloides 0.7 Carex lacustris 0.7 Carex vesicaria 0.6 Salix gracilis 0.6 Alnus rugosa 0.6 Spiraea alba 0.6 Polygonum lapathifol ium 0.5 Salix discolor 0.5 Si urn suave 0.4 Polygonum coccineum 0.4 Glyceria boreal is 0.4 Rumex mexicanus 1/ 0.4 1/ Plants listed as indicator for type by Stewart and Kantrud (1971) -20- Appendix C. Characteristic plants of Community 3. Seasonal-closed, Species Mean Abundance Carex lacustris 3.9 Carex rostrata 1 .6 Potentilla palustris 1.4 Calamagrostis canadensis 1.1 Polygonum cocci neum 1/ 0.7 Scirpus pedicellatus 0.0 Typha 1 atifol ia 0.5 Salix gracil is 0.5 Si urn suave _]/ 0.5 Alnus rugosa 0.4 Cicuta bulbifera 0.4 Scutellaria epilobi ifol ia 0.4 Salix discolor 0.4 Glyceria grandis 1/ 0.4 1/ Plants listed as indicator for type by Stewart and Kantrud (1971) Appendix D. Characteristic plants of Community 4. Seasonal-patchy. Species Mean Abundance Carex lacustris 3.0 Potentilla palustris 1.6 Calamagrostis canadensis 1.3 Carex lasiocarpa 1 .1 Carex rostrata 1 .0 Polygonum cocci neum 1/ 1.0 Glyceria boreal is 0.9 Utricularia vulgaris 0.9 Si urn suave 1/ 0.9 Salix pedicellaris 0.8 Alopecurus aequalis 1/ 0.5 Typha lati folia 0.5 Bidens sp. 0.5 Ranunculus flabellaris 0.5 Lemna minor 1/ 0.4 Typha glauca 0.4 Bidens frondosa 0.4 Polygonum amphibium 0.4 Potamogeton gramineus 0.4 Alisma trivale 1/ 0.4 ]_/ Plants listed as indicator for type by Stewart and Kantrud (1971) 22- Appendix E. Characteristic plants of Corrmunity 5. Seasonal-open Species Mean Abundance Si urn suave ]_/ 2.2 Glyceria boreal is 1 .8 Carex rostrata 1 .2 Carex lacustris 1 .1 Polygonum cocci neum ]_/ 1.0 Lemna minor ]_/ 0.7 Carex atherodes 1/ 0.6 Alopecurus aequalis ]_/ 0.5 Alisma triviale 1/ 0.5 Polygonum amphibium 0.5 Potentilla palustris 0.4 Potamogeton foliosus 0.4 Calamagrostis canadensis 0.4 ]_/ Plants listed as indicator for type by Stewart and Kantrud (1971) -23- Appendix F. Characteristic species of Community 6. Semipermanent-closed Species Mean Typha latifol ia 1/ Si urn suave Equisetum fluviatile Lemna minor 1/ Carex lacustris Carex rostrata Sparganium chlorocarpum Utricularia vulgaris Glyceria boreal is Sal ix gracilis Bidens sp. Glyceria grandis Polygonum cocci neum Alnus rugosa Sparganium minimum Eleocharis palustris Potamogeton gramineus Acorus calamus Spirodela polyrhiza Lysimachia thyrsiflora Aral i a nudicaul is Carex lasiocarpa Alisma trivial e Scirpus pedicel latus Abundance 4. 6 3. 0 1 . o 1 . ,8 1. ,2 1, .2 1. .2 1 , .0 1 .0 0 .8 0. .8 0 .8 0 .8 0 .b 0 .b 0 .6 0 .6 0 .6 0 .4 0 .4 0 .4 0 .4 0 .4 0 .4 1/ Plants listed as indicator for type by Stewart and Kantrud 1971 ) . -24- Appendix G. Characteristic plants of Community 7. Semipermanent-patchy. Species Mean Abundance Si urn suave 2 .6 Typha latifolia 1/ 2.4 Lemna minor 1/ 1 .8 Glyceria boreal is 1 .4 Potamogeton natans 1.4 Carex lacustris 1 .0 Typha glauca 1 -0 Sphagnum sp. 1 .0 Carex sp. 0.8 Eleocharis palustris 0.8 Ranunculus flabellaris O.b Sagittaria latifolia 0.6 Salix discolor 0.6 Salix gracilis 0.6 Hypericum virginicum 0.6 Potentilla palustris 0.6 Riccia fluitans ]_/ 0.6 Dryopteris thelypteris 0.6 Call a palustris 0.6 Carex comosa 0.6 Dulichium arundinaceum 0.6 Nuphar variegatum 0.6 Potamogeton foliosus 0.6 Acorus calamus 0.4 Carex rostrata 0.4 Alopecurus aequalis 0.4 Mentha arvensis 0.4 Epilobium leptophyllum 0.4 Polygonum coccineum 0.4 Scirpus pedicellatus 0.4 Carex stipata 0.4 Glyceria grandis 0.4 Iris versicolor 0.4 Salix amygdaloides 0.4 Salix pedicellaris 0.4 Eleocharis obtusa 0.4 Spirodela polyrhiza 0.4 Polygonum amphibium 0.4 1/ Plants listed as indicator for type by Stewart and Kantrud (1971) -25- Appendix H. Characteristic plants of Community 8. Semipermanent-open. Species Mean Abundance Lemma minor 1/ 1 .9 Sparganium chlorocarpum 1.6 Spirodela polyrhiza 1.3 Sagittaria latifolia 1.3 Potamogeton foliosus 1.1 Glyceria boreal is 0.9 Glyceria grandis 0.9 Carex lacustris 0.7 Eleocharis palustris 0.7 Potamogeton gramineus 0.7 Utricularia vulgaris ]_/ 0.7 Salix gracilis 0.7 Potamogeton natans 0.7 Lemna trisulca 0 .7 Polygonum amphibium 0.7 Si urn suave 0 .6 Ranunculus flabellaris 0.6 Carex comosa 0.6 Typha glauca 1/ 0 .6 Acorus calamus 0.4 Phalaris arundinacea 0.4 Carex atherodes 0.4 Typha latifolia 0.4 Scirpus pedicel latus 0.4 Potamogeton illinoensis 0.4 1/ Plants listed as indicator for type by Stewart and Kantrud (1971) -26- Appendix I. Characteristic plants of Community 9. Shrub swamp. Species Mean Abundance Alnus rugosa 3.0 Sal ix discolor 2.8 Salix gracilis 2.7 Calamagrostis canadensis 2.3 Carex lacustris 2.2 Potentilla palustris 1.8 Dryopteris thelypteris 1.4 Betula pumila 1 .1 Polygonum coccineum 0.8 Scutellaria epilobiifolia 0.8 Cicuta bulbifera 0.7 Cornus stolonifera 0.7 Carex rostrata 0.6 Lycopus uniflorus 0.6 Galium trifidum 0.6 Iris versicolor 0.6 Campanula uliginosa 0.6 Carex lasiocarpa 0.5 Si urn suave 0 .5 Carex aquatilis 0.5 Lysimachia thyrsi flora 0.4 Caltha palustris 0.4 Carex sp. 0 .4 Impatiens capensis 0.4 -27- Appendix J. Characteristic plants of Community 10. Hardwood swamp, Species Mean Abundance Fraxinus nigra 3.0 Si urn suave 1 .8 Carex intumescens 1.4 1)1 mus americana 1 .3 Carex rostrata 1 .0 Equisetum sylvaticum 0.9 Carex lacustris 0.5 Polygonum coccineum 0.4 Salix discolor 0.4 Salix gracilis 0.4 Acer rubrum 0 .4 ■28- Appendix K. Characteristic plants of Community 11 . Circumneutral bog sedge phase. Species Mean Abundance Carex lasiocarpa 3.4 Typha latifol ia ]_/ 1.7 Calamagrostis canadensis 1.7 Potamogeton gramineus 1.5 Potentilla palustris 1.4 Polygonum coccineum 1.4 Salix gracil is 1.3 Carex lacustris 1 .3 Campanula uliginosa 0.9 Carex aquatilis 1/ 0.9 Utricularia intermedia 0.8 Dryopteris thelypteris 0.7 Utricularia vulgaris 0.7 Carex rostrata 0.6 Lysimachia thyrsiflora 0.5 Galium trifidum 0 .5 Phragmites communis 1/ 0.5 Salix discolor 0 .5 Rumex orbiculatus 0.4 Cicuta bulbifera 0 .4 Sagittaria lati folia 0.4 Acorus calamus 0.4 Hypericum virginicum 0.4 1/ Plants listed as indicator of fen as described by Stewart and Kantrud (1971). -29- Appendix L. Characteristic plants of Community 12. Circumneutral bog sedge phase. Species Mean Abundance Salix gracilis 2 .6 Carex lasiocarpa 2.5 Calamagrostis canadensis 2.5 Potentilla palustris 2.0 Typha latifol ia 1.7 Alnus rugosa 1.7 Salix discolor 1 .7 Carex 1 acustris 1 .7 Betula pumila 1 .3 Dryopteris thelypteris 1.1 Lysimachia thyrsiflora 0.8 Galium trifidum 0.8 Salix Candida 0.7 Campanula uliginosa 0.7 Carex rostrata 0.7 Polygonum coccineum 0.7 Salix pedicellaris 0.7 Hypericum virginicum 0.7 Scutellaria epilobifol ia 0.6 Epilobium leptophyllum 0.6 Scirpus pedicellatus 0.5 Carex stricta 0.4 Asclepias incarnata 0.4 Lycopus uniflorus 0.4 Cicuta bulbifera 0 .4 Rumex orbiculatus 0.4 Carex stipata 0.4 Carex aquatil is 0.4 Utricularia vulgaris 0.4 ■30- Appendix M. Characteristic plants of Community 13. Circumneutral bog ericaceous phase. Species Mean Abundance Chamaedaphne calyculata 3.2 Carex lacustris 2.4 Sphagnum sp. 2.0 Carex lasiocarpa 1 .7 Potentilla palustris 1.3 Calamagrostis canadensis 1.2 Betula pumila 1 .1 Drepanocladus sp. 1 .1 Sal ix pedicellaris 1 .0 Carex oligosperma 1.0 Kalmia polifolia O.b Vaccinium oxycoccus 0.8 Salix gracilis 0.8 Scirpus pedicellatus 0.8 Eriophorum spissum 0.8 Alnux rugosa 0.7 Pinus banksiana 0.6 Andromeda glaucophylla 0.5 Typha lati folia 0.5 Betula papyri fera 0.4 ■31- Appendix N. Characteristic plants of Community 14. Acid bog. Species Mean Abundance Chamaedaphne calyculata 4.6 Sphagnum sp. 3.7 Betula pumila 2.1 Pinus banksiana 1 .6 Larix laricina 1 .5 Andromeda glaucophylla 1.3 Kalmia poli folia 1 .3 Vaccinium oxycoccus 1.3 Picea man ana 1 .2 Ledum groenlandicum 0.9 Carex oligosperma 0.9 Eriophoroum spissum 0.9 Carex lasiocarpa 0.9 Alnus rugosa 0.8 Salix pedicellaris 0.8 Drepanocladus sp. 0.8 Carex lacustris 0.7 Calamagrostis inexpansa 0.5 Betula papyri fera 0.4 Sarracenia purpurea 0.4 Calamagrostis canadensis 0.4 ■32- Appendix 0. Characteristic plants of Community 15. Softwood swamp. Species Mean Abundance Sphagnum sp. Picea mariana Ledum groenlandicum Larix laricina Alnus rugosa Smilacina tri folia Carex disperma Circaea alpina Betula pumila Carex aquatil is Carex lacustris Dryopteris spinulosa Caltha palustris Potentilla palustris Chamaedaphne calyculata Abies balsamea Cornus canadensis Scutellaria epi 1 obi i folia Ribes hirtellum Cornus stolonifera Viola sp. Viburnum trilobum Rubus canadensis Vaccinium angustifol ium Galium asprellum Dryopteris thelypteris Equisetum sylvaticum Aralia nudicaul is Carex stipata Fraxinus nigra Galium trifidum Ulmus americana Salix discolor Salix pedicellaris Carex projecta Calamagrostis canadensis Polygonum coccineum Lysimachia thyrsi flora Impatiens capensis Salix Candida Rubus pubescens Glyceria grandis 3.1 3.0 2.9 2.8 2.8 1.5 0.8 0.8 0.8 0.8 0.7 0.7 0.7 0.7 0.7 0.7 0.6 0.6 0.6 0.5 0 0 0 0 0 0 0 0.5 0.5 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 -33- ■& U.S. GOVERNMENT PRINTING OFFICE : 1973 O-506-879 As the Nation's principal conservation agency, the Department of the Interior has basic responsibilities for water, fish, wildlife, mineral, land, park, and recreational resources. Indian and Ter- ritorial affairs are other major concerns of this department of natural resources. The Department works to assure the wisest choice in managing all our resources so that each shall make its full contribution to a better United States now and in the future. UNITED STATES DEPARTMENT OF THE INTERIOR FISH AND WILDLIFE SERVICE BUREAU OF SPORT FISHERIES AND WILDLIFE WASHINGTON. D- C. 20240 POSTAGE AND FEES PAID U.S. DEPARTMENT OF THE INTERIOR INT 423