Vegetatio Vol. ~8, ~-e, pag. 4z-56, I973. STUDIES OF T H E G R A S S L A N D V E G E T A T I O N IN T H E K A W A T A B I SPECIAL R E S E A R C H AREA OF T H E J A P A N E S E IBP Phytosoclological Analysis and Computer Simulation of the Table Arrangement by H. LIETH, M. NUMATA and T. SUGANUMA (University of North Carolina, Chiba University, and Nara Women's University) Keywords: Computertabulation, Phytosociology, Miscanthus sinensis, Grassland, Japan-IBP. L INTRODUCTION In the course of the development of the International Biological Program it soon became apparent that the handling and exploiting of the rapidly growing data files required special evaluation methods. The complexity of the multifactorial models needed to analyze taxa/environment or taxa/taxa relationships created uneasy feelings about the feasibility of the projects proposed and enthusiastically started by so many ecology teams all over the world. It soon became very evident that without a rapid development of computerized models to evaluate the steadily accelerating data inflow it would be impossible to gain the insight into ecosystems one had hoped for during the planning period of IBP. Phytosociological tables are matrices of sometimes enormous size. Handling and sorting such tables require major clerical efforts. The * Contribution from JIBP/PT and CT, and US IBP, Eastern Deciduous Forest Biome, BIOME WIDE STUDIES. This study was supported in part by the special project research "Studies on the dynamic status of biosphere" sponsored by the Japanese Ministry of Education, and in part by the North Carolina Board of Science and Technology and the UNC Faculty Grants Committee. We gratefully acknowledge the assistance of DR. W. C. MOORE,MR. J. S. RADFORD,and MR. R. READERin the execution of the computer analysis. Responsible for theories and execution of computer simulation: H. I,IETH;responsible for field data: M. NUMATAand T. SUGANUMA. 41

final version of a table often conceals other possible viewpoints that could be demonstrated with the same set of species. We try in this paper to demonstrate and discuss some of these problems. We also provide a computer program that eases the clerical effort of tabulation to the point of being almost negligible, once the program is set up in the local computation center. Moreover, we try to make the phytosociotogical tables compatible with other computer models presently developed. In order to make the information stored and displayed by phytosociological tables compatible with other computer models, it is necessary to develop adequate formats and computer routines to handle matrices existing of repetitive investigations of species sets vs. environmental parameter sets. If one can computerize this relationship, one can immediately expect two very important benefits: ~) the compatibility between species lists and functional models within one computer information system, and 2) the immediate access to a large data pool stored in phytosociological or similarly classical ecological work. LmTH & MOORE (I971) developed, therefore, during the last few years procedures which try to ease the evaluation of phytosociological tables, and elaborated data coding and handling procedures, in order to streamline the data pool conveniently for computer use. Our program has been successfully employed for normal phytosociological work by MICHAEL (I969) , WHIGttAM (1971) and BOZEMAN (1971). Our final a i m i s to construct forms of matrices that incorporate the common species lists available from phytosociological investigations into information systems. The adequate labeling of individual species and the organization of information bits, as we demonstrate in the J-IBP grassland table, is an important immediate result, along with the computerized construction of a phytosociological table. Two years ago we asked several colleagues to send us data sets of conveniently manageable sizes to test our computer procedures against real problems, under circumstances in which we could not possibly be influenced intuitively. One reply came from the Japanese IBP Grassland Site. The present paper evaluates the comparison between the two procedures: on one side, the long standing experience of the field ecologist with several years of additional information accumulated; and on the other side, the computer team relying only on the data set given to them and the confidence in having simulated at least the sorting procedures that the field ecologist would apply if he were seeing such data for the first time. Along with this comparison we describe I) a summary of the con42

ditions of the site from which the data were gathered and other information necessary to evaluate the ecological content of the phytosociological tables (for more details on the site see NUMATA et al., 1968), and 2) some of the theory and technique of computerization of phytosociological analyses. II. G E N E R A L I N F O R M A T I O N

ABOUT THE STUDY AREA

a) GEOGRAPHY AND HISTORY

An interim report of the Kawatabi Project (NuMATA, hZUMI & IWAKI, I968 ) described the general situation of the Kawatabi site. The Kawatabi site (I4o ° 15' E, 38° 44' N) is located on the eastern side of the O h - U h mountain range and a hillock district 300-600 m above sea level surrounded by Mt. Kurikoma on the north, the river E-ai on the south, the Onikobe basin on the west and the plain of Osaki on the east. In i884, the Japanese Army Horse Supply was established at this district. Until the end of the Second World War, the semi-natural grassland, which has once been a woodland, was used as the grazing and mowing land for raising the a r m y horses. After the war in 1948, the district was transferred to Tohoku University. Cattle and sheep have grazed regularly ever since, in some part of the grassland. In i967 the special research area for the grassland study of the Japanese IBP (JIBP)/PT and CT was allocated on this experimental farm. A research team of about thirty members headed by M. NUMATA is conducting various studies on producers, consumers, decomposers and environmental components. b) SOIL The soil of the grassland at Kawatabi originates from acidic volcanic ash and indicates the abundance of nitrogen and humus. Soils are generally deficient in calcium, but have adequate phosphorus supply. The phospher absorption coefficient is as high as about i8oo. These characters vary in some degree in accordance with the topographical conditions of the grassland. The humus layer is shallow in the convex part of the landscape while generally deep in the concave parts. The thickness of the humus layer is also related to the grass growth which is good on the east-facing slope and the concave slope areas, but poor on the south-facing slope and on the rim of the ridges. c) CLIMATE

Meteorological observations from the Kawatabi site are not avail43

able before I968 , but long term records from Narugo-machi close to the area are available and are summarized in Table I. TABL~I Meteorological conditions at Narugo-machi (38o45' N, 14o° 48' E, 295 malt.) close to the IBP site at Kawatabi (38o44' N, I4°° 15' E, 6oo malt.). The annual mean temperature may be about 8.2°C at the Kawatabi site Mean Temperature Jan. Feb. Mar. Apr. May June July Aug. Sept. Oct. Nov. Dec. Annual Mean

--1.9°C --I.1 1.4 8.o 13.8 16.8 21.3 22.5 17.9 11.3 6.2 1.2 9.8

Precipitation 236.2 mm 138.8 156.5 17o.o 134.8 186.5 246.8 283.o 211.7 159.8 198.3 21". 7

Sum Total

2335.1

The duration of sunshine is longest in May. It is shorter in J u n e and J u l y than in May because of the rainy season. The snow falls in winter as deep as I-2 m . The snow melts in middle March on the sunny slope and in late April on the shady slope. The prevailing winds are westerly from October to May and easterly from June to September. The "vegetative season" of the Miscanthus sinensis grasslands is calculated from the date of beginning the height growth to the date of ceasing the height growth. The height growth is effectively promoted by a mean daily temperature above Io°C. The effective accumulated temperature for the vegetative growth of Miscanthus sinensis, the summation of daily mean temperatures over io°C between start and end of height growth, is 15oo to 4ooo°C in J a p a n (NuMATA & MITSUDERA, 1969). The corresponding vegetation period is six to nine months. d) WO~A~ION The Kawatabi site is primarily covered by Miscanthus sinensis grassland which is one of the most representative "tall-grass" type meadows in Japan. Miscanthus is a genus indigenous to Asia ineluding about twenty species. The Miscanthus sinensis grassland is concentrated in the cool-temperate region of J a p a n and is used for mowing. It is mixed with Sasa spp. in the subarctic region and with 44

Pleioblastus spp. in the warm-temperate region (NUMATA I961 ). Miscanthus sinensis is the most common dominant in ordinary succession (i.e. "orthosere" of NUMATA, 1969). The Miscanthus sinensis meadows are usually mown once a year after the heading and have been used for forage, thatch, charcoalbag, mulching, etc. NUMATA and SI-II~ADA conducted a field experiment on the primary productivity of the grassland under different cutting intensities. They traced autecological responses of constittuent species and estimated a stable yield under different management procedures (NUMATA, 1969). It was found that three cuts per year after three years cause the relative summed dominance ratio (SDR) of Miscanthus sinensis to decrease from 12.3 to 8.1, its annual biomass production from 71o g/m z to 75, the weight ratio from 43-3 % to 9.4, the height from 85 cm to 38. Decreasers under mowing are Lespedeza bicolor, Lysimachia clethroides and Miscanthus sinensis. Increasers are Pteridium aquilinum, ArundinelIa hirta, Ca{amagrostis hakonensis, Zoysia japonica, Hydrocotyle ramiflora, etc. Indifferent species are Carex nervata, Carex lanceolata, Viola mandshurica and other less important species. The degree of succession ranges between I oo and IOOO with an average of about 45 ° (NUMATA 1969). In the Miscanthus sinensis grasslands, the number of constituent species is 2o to 7o, S D R of the dominant Miscanthus sinensis is 6o to ioo, its relative S D R °/o to the total S D R values (SDR') is IO to 2o, its fresh weight of the aerial shoots is IOO to 2ooo gr/m ~ (the dry weight is about one third of the fresh weight), its weight ratio to the total biomass is io to 8o °/o, and its height (the maximum height of vegetative organ, not of the heads) is 9 ° to 2oo cm. The area surrounding the Kawatabi grassland site is mostly covered by secondary forests consisting of deciduous broad-leaved trees and one pine. The chief components of the forest are Ouercus serrata, Castanea crenata, Carpinus laxiflora and Pinus densiflora. Plantations of Cryptomeria japonica and of Larix leptolepis are also present. The mountainous district in the northern part of the grassland is covered by a climax forest composed of Fagus crenata. The lowland in the southern part of the area is utilized as cropland. IH. T H E V E G E T A T I O N

ANALYSIS OF THE KAWATABI SITE

A preliminary vegetation map and other physiographical maps showing the humus layer and snow cover were prepared for all interim report (1968) . The same report contains sociological analyses to discriminate vegetation types with dissimilarities large enough 45

to warrant their acceptance as discrete functional or management units. The vegetation types distinguished are shown in Table 2. T h e table is shown in the format used for the computer sort (Table 4), only the species codes used in the other tables (4-6) are arranged against the typed list of full names. Coded species names are highly desirable in computer operations. Complete nomenclature of all species in alphabetical order is repeated in Table 3, as a key to easier evaluation of Tables 4-6. Table 3 also contains the key to converting releve numbers used in all tables to the Kawatabi logbook numbers. A few species names were changed during the revision of this paper. T h e y are listed in Table 3, Section 2. The Japanese IBP working group recognizes mainly four different grassland types on the site. According to the phytosociological classification system as used presently in Japan, the four units belong to the association Iridi-Miscanthetum sinensis. Distinguished are the following units (as demonstrated in Table 2): A-Subass. of Petasitesjaponicus vat. giganteus Variant of Cacalia hastata i Typical variant 2 B-Subass. of Metanarthecium luteo-viride Typical variant Variant of Leibnitzia anandria

i 2

Based on this analysis SU6ANUMA& StrOAWA~A (I97I) have recently completed the vegetation map in the Kawatabi IBP Area. The environmental forces ruling the distribution pattern of the four types are primarily topographical, with soil moisture conditions the principal factor. The spatial pattern of the vegetation types is shown in Figure I. The four grassland units have, expressed in general terms, the following properties: A-I.

A-2.

B-I. B-2.

46

Grassland vegetation mixed with so-called "'Hochstaudenfluren" in the valley bottoms where the deep snow melts only late in spring, and with a thick humus-rich top soil. Grassland vegetation on the concave lower slopes adjacent to I, where the better water supply causes greater productivity by Miscanthus sinensis. Grassland vegetation on the convex upper slopes which are usually small, dry ridges. Grassland vegetation on the windward slopes, still drier than 3, with the shallow snow cover melting in early spring.

,0..

c~

0

0

Oq

c~ ,<

0

6~

25

50

55

75

50

60

;00 m

c;

I 0 ""

'i 2°

Variant of

(~ lVeibnftzia gmartdria1~ Pioneer stage of Ptnus densiflora commtmity

Cacalio-~lgelicetum urstnae

uteo~viride

;ub. . . . . f M. . . . . thecium{Typical varlet

idf-Misca~d~etum sinensis (Variant of Caca!ia (D ub.... of Petasites /hastata v. ortentalis apon.icusv. giganteus T}q~icatvariant (~

!L..,4 /

NiigOt'O/"~"-

:

",'yomogoto) > o ; o

Akito ~ O"

38

40 °

From the map and the description it becomes clear that the most prominent as well as the most valuable part of the entire research site is covered by vegetation type A-2. General management procedures for the entire site are therefore likely to be based on the needs for this vegetation type. More information is given in the legend to Figure I. IV. T H E C O M P U T E R S O R T I N G a)

THE TABLE

As we pointed out in the introduction, the phytosociological table was used to test a table sorting procedure developed by LIETH & MOORE (I97I). The result of our trial is presented in Table 4. (Consult Table 3 for full species names.) The species and releves are exactly the same as given in Table 2. The sorting of Table 4, however, is based on a computer algorithm that simulates the manual sorting procedure. A raw table that had no resemblance to either Table 2 or 4 was sent by NUMATA to LIETH in I969. This table was used as a data pool to test the simulation procedure. Only part of the simulation was actually done by computer. This, however, is the most important step in elaborating any phytosociological table: i.e. the clustering of species and releves that is needed to select differential or mdicative species, and at the same time group releves with the same indicator species together. Before we analyze differences between Table 2 and 4, we first need to describe the basic concept end procedure of cluster generation. b)

TIKEORY OF CLUSTER GENI~RATION

The phytosociological table is usually treated as a two-dimensional matrix. The first dimension x is the set of releves involved and the second dimension y is the set of species found in different releves. Depending on the sequence of releves and species we can generate arbitrarily x!y! patterns over the matrix. We can demonstrate this conveniently with a re-sort of Table 4. This time we sort so that the chosen sequence of releves and the sequence of species generate diagonal patterns of realized x!y!'s (Table 5). We can also generate patterns arbitrarily and draw pictures with the x!y! combinations at our disposal in Tables 2 or 4. The arbitrary printout in Figure 2 shows, slightly accentuated with lines, the image of the farmer working on our Miscanthus meadow. Table 5 demonstrates alternative dimensions for the releve that 48

might be used for our phytosociological table: each releve has a set of environmental properties printed at the top, and each of these can be utilized potentially to re-sort the releves in 3rd, 4th, or nth dimension. Our sort in Table 5 uses as sorting principle the indication that the releve comes from a valley bottom (v), slope (s) or ridge (r). In this case it is possible to arrange all species from all releves in two diagonal lines. Under ecological conditions such

Fig. 2. I m a g e of a f a r m e r p r o d u c e d b y the same d a t a set used for T a b l e s 2, 4 5 a n d 6. This a r b i t r a r y sort demonstrates the power of a m a t r i x a n d its possible misuse. T h e length of essential lines t h a t accent the farmer's h e a d is c o m p a r a b l e to the length of lines in T a b l e 2.

49

diagonal order is only realized when the releves were selected along one environmental gradient. This is apparently the case in our data set as Table 5 shows, indicating the strong impact of the topography on the species composition. Other diagonal sorts have been presented in the past by ELLENBERG (I950) and many others (see LIETI-I, 1968 , for summary). We are usually confronted, however, with combinations of environmental factors, which alter the number sequence of the releve or species vectors and break the continuous diagonal xy pattern into isolated patches. The matrices of the same data sets then display clusters, cross patterns or wave lines for species/releve (y!x!) combinations. Clustering is usually the first goal of a manual sort and it is therefore our first goal in the computer sort if no environmental information is utilized. Table 6 demonstrates the clusters generated by the computer for the phytosociological table from the Kawatabi site. Technically, these clusters were generated through the application of Lin's algorithm for 3-optimality (LIETI-I & MOORE, 1971 ) . Each phytosociological table contains a certain number of clusters, usually more than one can use for the first order classification of classical phytosociological units. If we can find that clusters are inherent in a phytosociological table we must also take it for granted that certain sets of species are present in one set of releves and missing in others; e.g. i f a given table contains a mixture of 4 different species combinations, and the number ofreleves from each combination is the same, we could expect that the clustering species occur in about 25 °/o of the releves of the table. The clustering procedure must become, therefore, more successful when one restricts the sorting procedure to the species having from 25 To frequency to those with approximately 5 ° °/o frequency. Subsequently, we include in the algorithm application only species having between 2o °/o and 8o % frequency. While the clustering itself may be incomplete, the clusters generated in such manner allow us to select related species/releve groups. From this point it is then simple to find the optimal releve/species clusters and to generate the image of Table 4 from Table 6. C) P R O G R A M O R G A N I Z A T I O N AND CARD DECK FORMAT

The program can be utilized by anyone who has both a computer available comparable to the IBM 36o/4o and the necessary PL/I compiler. We are presently converting the program into Fortran IV. Parts of the program have been translated already (LIETH & STEELE, in preparation). The Tables 2 and 4 in this publication have been 5°

executed by means of this new program, using the Arbitrary Printout step. The program source deck contains about 3oo cards o n w h i c h are punched over 6oo PL/I statements. Copies of the listing, along with a previous paper (LmT~ & MooR~, I97 I), may be obtained through the authors. The data deck used for this publication required the transfer of all data from the field worksheets to computer cards, using the coding system specified in the program. The complete deck submitted was organized after the outline below. Data immediately following the last card of the source deck and necessary job control card(s) : A) B)

Title card, free form documentation of dataset processed Parameter card c o n t a i n i n g . . . Number of symbols used in table (integer up to 9)

Col. i - 5 "

Number of copies to be printed (integer) Col. Print original table? (o = no, I yes) Col. Print sorted frequency table? (o = no, I ~- yes) Col. Number of iterations for algorithm table Col. Arbitrary printout? (o = no, i yes) Col. Maximum percent frequency for inclusion in algorithm sort (two-place decimal fraction, between o.oo and I.OO) Col. Minimum percent frequency level (as above) Col. Number of lines skipped between table rows (integer, default ---- I) Col. The symbol card c o n t a i n i n g . . . All alpha-numeric symbols used in the matrix. Total number of symbols read is determined by the number punched first on the parameter card. Col. Corresponding binary sequence of i's and o's. - A given symbol associated with a o (or a 1) will cause a species to be ignored (or used) for all tables when the symbol is used as the species cover symbol (described below). Col. Releve cards (field data) consecutively punched for each releve - - 7 species maximum per card, in the following order: =

=

C)

D)

Col. I-8o

6-1o* I5 2o 2I-25. 3°

31-35 * 36-4 o* 5°

1-4o

41-8o

* All these numbers are right-justified.

51

E) F)

G) H) I)

J) K)

Releve number (right-justified integer), Col. 6-1o Then, for each block of ten columns remaining: Genus name - 4 columns Species name - 4 columns Cover symbol - i column Sociability symbol - i column Nine-card, delimiting end of releve cards (80 9's) Col. 1-8o Species list for special sort (optional), genus and species code as used in field data (8 letters), one card for each species Col. 3-1o Releve list sorted for arbitrary printout (optional), releve number (right-justified integer) Col. 6-IO Nine-card (as in E) Col. i-8o Species list sorted for arbitrary printout (optional), printout genus and species code as used in field data file (8 letters), one card for each species Col. 3-1o Nine-card (as in E) Col. 1-8o End-of-data card, as required by the computation center

Figure 3 shows a sample of a field data card, nine-card, releve number and species list cards in the form required by the program. The "table headings" in Table 2 and 4 were entered in the same form as the individual species. Abbreviations were chosen to fit the otherwise convenient space allocations for genus and species in the field data deck. V. THE

COMPARISON OF THE ANALYSIS PERFORMED FIELD WORKERS AND THE COMPUTER TEAM

BY THE

a) Comments of the field workers As we have already pointed out, we are comparing the results of the table analysis by the field workers who had information not present in the data set with the table analysis of the computer team aided only by their sorting procedure and theory. It is therefore not surprising that differences exist between Table 2, showing four releve groups A-l, -2, B-I, -2; and Table 4 with 3 releve groups A, B and C. In comparing the two tables, NVMATA& SUGANUMA made the following comments: I.

52

Releve i i should be moved from A to B, and 12 from B to C. C should be divided into CI (2, 8, 9, 4, 18 and 19) and C2(I2, 7, 22, 2o and I7). Thus, A, B, C2 and CI in the computer-made

2.

table are A-I, A-2, B-I, and B-2 in the hand-made table, respectively. Regarding the species groups, identification of characteristic species and differential species in the association is most important.

Metanarthecium luteo-viride, Hosta rectifolia, Calamagrostis hakonensis, Haloragis micrantha, Synurus pungens, Polygala japonica, Smilax china, Swertia japonica, and Platycodon grandiflorum in the releve group C of the computer table are the differential species of B in the field team table.

Aster ageratoides v. ovatus, Petasites japonicus v. giganteus, Aruncus dioicus v. kamtschaticus, Lastrea thelypteris, Artemisia princeps, Polygonum cuspidatum, Chloranthus serratus, and Dioscorea tokoro, and Vitis coignetiae and Rubus parvifolius, the differential species of subassociation Petasites japonicus v. giganteus (A-2) are interrupted by Cacalia hastata v. orientalis, Hydrangea macrophylla v. megacarpa, Senecio cannabifolius, Clematis terniflora, and Aralia cordata, all differential species of A-I. Cirsium nipponicum has a low cover value but is an important species in distribution, and is a characteristic species of an association or higher unit. Viola obtusa, Eupatorium lindleyanum, and Inula salicina v. asiatica, Rubia akane and Agrimonia pilosa and Leucosceptrum japonicum, Desmodium oxyphylIum, and Ampelopsis brevipedunculata are less important and are classified as companion species in the field team table. Besides Cirsium nipponicum mentioned above, Iris ensata, and Anaphalis margaritacea v. angustior, Gentiana triflora v. japonica and Aeginetia sinensis are the characteristic and differential species of the association. Hypericum erectum, Carex nervata, Potentilla fragarioides v. major, Solidago virga-aurea v. asiatica and Adenophora triphylla v. japonica, all scattered over the second half of the species sequence, are the character species of a higher unit. Besides Astilbe thunbergii v. congesta and Salix vulpina which are companion species, the remaining species listed above and Metanarthecium luteo-viride are characteristic species of higher units. These differences make it clear that m a n y improvements in the computer program are necessary before the requirements of the phytosociologists are met. b) Comments of the computer team. The similarity of the two tables is evident at first sight. Although the computer table has only 3 groups of releves (Group A, releves 53

2I, I6, iI and 15; Group B, releves 14, 23, I, 6, 5, IO, 3, I3 and 12; and Group C, releves 2, 8, 9, 4, 7, 22, 18, 19, 20 and 17) whereas the field team recognizes 4 groups, the differences are caused basically by the fact that Table 2 was sorted with the additional experience of the Japanese workers. The field team developed its table against a background of hierarchical classification, discriminating various groups of indicator and character species that differ in systematic and geographical significance. A machine sort will never consider this possibility unless the discriminatory properties of certain species are specified in the program. The development of a computer system that would incorporate the inclusion of some 4,ooo to IO,OOOspecies into one phytosociological system is simply a question of investing time and money. Whether a sorting system is the real aim for the future, however, is doubtful. We recognize well the usefulness of the sociologically defined vegetation types, but the usefulness of this method for ecological purposes is only based on the key/lock relation between environment and taxa. The elaboration of ecological gradients indicated by species composition is therefore as valuable as the sociological classification. The separation of species into character or differential species of different hierarchic levels may be detrimental in some cases. The full ecological information inherent in the basic matrix as demonstrated in Table 5 - for example, topographic location and species d i s t r i b u t i o n - is in Table 2 almost as well hidden as in Figure 2, where we abused the power of the twodimensional matrix to construct the picture of the farmer living from the Miscanthus grassland. The phytosociological matrix can be used in various beneficial ways, but can also be misused if employed only arbitrarily. The releve collection presented to us by the field team contained plots with a long listing of species occurring only once in the entire matrix. GOODALL (I969) discusses the probability and significance of such "species tailings". It might be worthwhile for the Kawatabi site workers to further analyze the isolated clusters marked in Table 6. For future incorporation ofphytosociological data into ecosystems models of various kinds, the approach demonstrated in Table 5 is most valuable. For any given ecosystem, the incorporation of the entire biocenosis is important. The relationship of the species composition to the various environmental conditions is one of the most important that can be detected through the phytosociological matrix. It would be very useful for modeling purposes if the parameters 54

given in the table head could be quantified into more basic physical or chemical units. Concluding our evaluation, we would like to mention that the phytosociological table as presented in Table 5 or 6 may be useful for incorporation in an information system as outlined by LIET~ (I969)SUMMARY

I. The paper demonstrates the utilization of a computerized vegetation analysis on the field data gathered at the Japanese IBP Grassland Site. Hand- and computer sorting procedures are compared and discussed. This part of the paper was a test of the usefulness of the computer simulation of the standard sorting procedures of phytosociological work. Tables 2-6 and Figure 2 demonstrate the usefulness as well as the limitations of the computer analysis. The possibility of incorporating phytosociological tables as matrices into computerized ecosystems models is suggested. 2. The vegetation of the J-IBP Grassland Site is described and preliminary evaluations made with regard to environmental parameters (Table 2 and 5, Figure i). Some management influences on the Miscanthus grassland are discussed. The primary above-ground productivity ranges from 30o to 15oo g (usually 4oo to 6oo g) dry matter per m 2. ZUSAMMENFASSUNG

I. Die Arbeit berichtet aber die Analyse einer pflanzensoziologischen Tabelle mit Hilfe eines Komputerprogrammes. Als Rohtabelle wurden die Aufnahmen yon der japanischen IBP Graslandstation in Kawatabi verwendet. Die iibliche Handsortierung der Tabeile wird mit der Komputersortierung verglichen. Dieser Vergleich war Teil eines Testes ft~r die Brauchbarkeit einer yon LIETH & MOORE entwickelten Komputersimulation der klassischen pflanzensoziologischen Tabellierungsmethode. Die Tabellen 2-6 undAbb. 2 demonstrieren Grenzen und Brauchbarkeit der beschriebenen Komputersortierung. Die M6glichkeit der Eingliederung pflanzensoziologischer Tabellen als Matritzen in 6kologische Komputermodelle anderer Art wird erSrtert. 2. Die Vegetation der J-IBP Grasland-Station wird beschrieben und deren Beziehungen zu einigen Umweltsfaktoren diskutiert (Ta55

bellen 2 u n d 5, A b b . i). Die V e g e t a t i o n s d e c k e besteht i m W e s e n t l i c h e n aus v e r s c h i e d e n e n Miscanthus sinensis-Grasland T y p e n . E i n i g e Bewirtschaftungseinfliisse a u f die V e g e t a t i o n s d e c k e w e r d e n b e h a n delt. Die o b e r i r d i s c h e Prim~trproduktivit~it des G r a s l a n d e s v a r i i e r t in d e n G r e n z e n v o n 3oo g bis I5OO g ( y o n 4oo g bis 6oo g u n t e r g e w 6 h n l i c h e n umst~inden) T r o c k e n s u b s t a n z je m ~ u n d J a h r .

VI. L I T E R A T U R E

BOZEMAN,J. R. 197i - - A sociologic and geographic study of the sand ridge vegetation in the coastal plain of Georgia. Ph.D. Dissertation, U.N.C. Botany Department, Chapel Hill. 216 pp. ELLENBERG, H. 1 9 5 0 - Landwirtschaftliche Pflanzensoziologie, Vol. i, Ackerunkrautgemeinschaften als Zeiger ftir Klima und Boden. Ulmer, Ludwigsburg, I4I p. GOODALL,D. W. 1969 - - A procedure for recognition of uncommon species combinations in sets of vegetation samples. Vegetatio 18: 19--35. LIETI-I,H. 1968 - - Continuity and discontinuity in ecological gradients and plant communities. Botanical Review 34, 291--3 °2. LIETH, H. 1969 - - i n press. Mathematical modeling for ecosystem analyses, UNESCO Symposium on the Productivity of Forest Ecosystems of the World. Brussels, October 1969. LmTH, H. & MOORE, G. W. 1971 - - Computerized clustering of species in phytosociological tables and its utilization in field work. In PATIL, G. P.; PIELOW, E. C.; & ~41ATERS,W. E.; eds., Statistical Ecology, Penn State University Press. I, 403--422. MlCHAEL, J. L. 1969 - - T h e vascular flora of Bullhead Mountain, Alleghany County, North Carolina. M.A. Thesis, U.N.C. Botany Department, Chapel Hill. 52 PP. NIdMATA,M. 1961 - - Ecology &grasslands in Japan. J. Coll. Arts & Sci., Chiba Univ. 3(3): 327--342. NUMATA,M. 1969 - - Geographical distribution and ecology of Miscanthus sinensis. Shin-Noyaku (New Pesticide) 24(2): 8--16.* NIJMATA,M. 1969 - - Progressive and retrogressive gradient of grassland vegetation measured by degree of succession--Ecological judgement of grassland condition and trend, IV. Vegetatio 19(1--6), 96--I27 . NUMATA,M., IIZUMI, S. & IX,VAKI,H. I 9 6 8 - Ecological studies of grassland in the IBP Area for PT and CT at Kawatabi, Japan. NU/CiATA, M. & MITSUDERA, M. 1969 - - Efficient environmental factors to the growth and production of tile Miscanthus sinensis grasslands in Japan. Ecological judgement of grassland condition and trend, V. Jap. J. Bot. 2o (2), 135--151. WFIIOHAM,D. F. 1971 - - An ecological study of Uvularia perfoliata in the Blackwood Division of the Duke Forest, North Carolina. Ph.D. Dissertation. U.N.C. Botany Department, Chapel Hill, 183 pp. * in Japanese 56

T A B LE 2: PhytosOClOlOgical

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . coal )s,flon oE t h e 4 m~sn grassl~d

table showing the s p e e l ~

Suhassoc~at~on of pctasit~ ] a p o ~ , c u s v . g,g~a~.u~ ~arxan* of Cacalia nas~a~a o~l~nta!~s

$ubassocJ&lJon V~i~t

~ypes on lhe Kawatab£

slte

stnensis

[ri~i. M i s c ~ t h e t u m

,f MesanerthecJum

fA £

lu*oo-v,rido

of },~ih~itz~a anmndrt~

The h o d . o f t h i ~ t a b l e

~

was p r o d u c e d by t h e a r b i t i z z y

2

prtnrout

s . e t ~ o n o f o u r puY,ro 71 p r o g r a m . ~ , c Cod*d s p e c i e s names h a v e b e e n r e p t ~ ¢ e d b)

a c>~ea list of f.ll ~ e s The vertical organization of the com~utcr prznTouz r ~ u t r e d the addition of 4 sp~ce Laumm 7 rclcvea 2 5 . 2 8 ] . The horizon~a! gmot~ping IS d ~ a e ~ U I ~h~ ~tand~rd ~)~bol~ ~OT phy~osoclo]ogie~1 R~o~p~ [~*£. ~, Ch* Cx~p). The %o~ llne above (h~ relev~ n,=be~ indieet~s t h e i o t a 3 . u u ~ e r o£ ~ c r t i e a l v e c t o r s i n the e o m p u t e ~ t a b l e , £ n c l u d i n g t a b l e head and a d d i t i o n a l s p e c i e s , D e v i a t i o n s b e t . s e n t h e hum= bet prillred and countable e]~cnt~ are due to minor changes at the tail-end of zhe ?able allot zhe ~c~iew. The d~angeg made dzd nor w a y y s n t a n ~ r u n o n t h e ~om=uccr Coded ~ w o s i n ~h~ t~l~}~ h ~ a A i n ~ ~ e : f o r c o w ; . ~ - h,,ndreai for *opo~raghv. v v a H e y . S = a Dpe, ~ ridge; h~J~ht of vc~rat~v~ is g)ve~ ~. deeimct~-~.

]lei~g of ve£et~t~on

17 IT 1 7

(~)

]ove~ o f eppez l a ~ r (%) . o v e r o~ ~Zddie ~ayer 41~3 Co~,e~ o ~ ~ e ~ i~y~z z~/~

85 I H I s I~ ~o ~o 5 s to

?~ot sl~e ( m ) Topography

s v

to to w q

slope de,,ree

N 10

s s I0 IC

[

.%lu

18 i"

15

I~

I t ' t H 1H ~5 4 0 70

-~ ~o is io ~ ~s s 5 s 2C +,

ts is ss 2s S S

s

<

t = ~* .

.

.

le 13 14 ~? 80 t H [ H so 9 d ~ o ~o 8 0 s { ( ~e ~0 ~ ~ is zs S S $ s

.

.

.

i~ 20 22

2&

t

ii iz ]o i~ t~ ss ~o ~ o ~n ~ s 2 s 2 0 ~ o 4 0 is t o ~e 25 Z5 16 1 6 2 5 1 5

1# ?o ~o

75 %

zs E Is ~

~ %o i ~ zn 1~ so 3c .

12 17 18

B-Iz is i~ to eo oo 14 ~o ~o s o ~o 7 e

i5

I0 E Io ~z

~s ~s 40 N E N i o to i5 zz ~o ~

s

~

8

o

7~

zo so zo zo 2 5 2 5 ~ 0 2S

s

I ~ W io z o 2~ ~

~

B--2 I~ II I~ ~s Be Bs so ~o ~o eo 9 5 ~o

~5 5

to

]o

W i~ i o ~ o is ~0 S4 St i~

.

*2

+Z +

~2 ~ ;2 • 2 +2

~2 * t ~ t

+2 +z

*z

s Petaslte~ j a p o n ~ c u ~ v. ~Ipante~

+Z

polyg~m eu~pixtatum Aru~c~s 4 i ~ e u ~ ~. ~aeteeb~t~cue O~ior~nt~us serrates

v~tgs c~gaeg~a~

Rydraagea ~crophylla

v . ~gacarpa

[

Cle~a~is

~er~iflora

1

Metm~a~he~i~m |~te~-vi~£ae sstor~Sis m ~ = ~ a : ~ a

+~

+~

Y :

] ca~a~g~oscls

+z

+~

h~one~£~

+z

+]

~# : , * . . . .~ . t~

+ +z

Pop~l~ aleboldi£ p~rsassi~ ~alus~ris s~p~o~os~esi~ ~h£nens{s

. +2

+z + -

-

t

~+ 5~ S5 ~

e~e~e~m

aquil~num

~z + 2 + ~ + 2 * • Z +2 Z Z • +

g~-slm~hi~ clechroide~ D~s~orum s m i l a c i ~ ~s#ece~ ~ex

1

÷2

$~ +2 +Z -~

+2 +2 ".Z

+

lz

~2 lz

+Z iz z ~. *

+ 2 + z ~2 + 2 ~ Z +2 +2 +a z2 ÷

12 ~]

12 lZ *2 7~, i~

+2 42

blcolv~ f. acu~Ifo~i(

tz

*

Z3

S3 S~

+Z +Z • ~ +z +1 *z

~caber

kr~i~

i +z Iz +2 +2

Japonica

+

IS

+Z iz * z *z

*z

+z

potentill~

£re~ia~

+?

+z

+~

•

t

[zz +

13

+~

[~

* *z + ]" *z is ,z +? +z +2 *z :

.

.

.

.

.

.

.

.

.

.

.

.

c~re~ ~ e r w ~

• 2 22

3~ * 2 +z

~6 +z

÷ 2 6% 33

lal~x ~ i p z n a In~la s a l l a i ~ v , Epip~et£~

asla~ca

~bergi,~

K~bia akane

Am~lopsis

brevipe~unculata

Ach£11ea ~iblr£e~ sya=ang~z p a ~ i a ~

~z

.o +z

pedicularls resupzna ra F r u n e ~ l a ~ I g a r i s v, l£taci~ Angelica ~rs~nn pleetra~ghuS c~b~

2~

Lapor=ea 5algeria

Z~ySla 3apontca

Viola verecundm +. Cmlastrug

~o. 7: n ~ . 27: I~o. i4;

~o. i: ~ . ZO:

~o. 2"

Calamagr~gt~s

~Ige ....

mrblc~larus

~.

l

TABLE 3 .,is

USED IN

TABLg

2,

Key t o cc,mp~rJson o f s p ~ i o s ~nd r ~ l e v c listings in ' { a h ] e 2 and r~ble~ 4 - 6 . The B.le~t~r ~bbreviati~ms o9 the species names, as seen ill Tables 4-6, arc su),tBd in alphabotlcal order. The fr'w chgngss in names ne¢c~ssary ~fter the rBview a r e indicated in Section 2 of the t a b l e . The ¢onverslon of the rOlevo n ~ b e r s used in all lahles ZO the stand ll~,bers in ~he logbooks o£ the Kaw&tabl site workers ~s gi ven in S~ctlo.

5,

ACHI ~ I ~ I

Ach£ilea ~ l h i r i c a

~EGI SIN~

A e g i ~ e ~ a sinensis

pARN P A t U

AGRI P l L O A M P E 5R~V ~M~P M A R ~

Aari.on~ pilosa Ampelopsi~ brevipe~nculata A~phal~o m~gari£~¢e~ ~. angus~i~ anget~c~ ~ra~ Aralia cordata artemisia japonica A r t e m ~ s ! a prlace=~ A~nc~ aloie=s v. kam~chatCc~ i~s~dinel]a biota Aster a~er~oides v, ovatus Aster scaber As~ilbe thu~bergil v. n ~ e s t a ~ h m e r i a nrlc~spi~ C B c a l ~ h ~ R t a ~ = v. orlen~al~s Galamagros~is epiEe%o~ G~la~gros~i~ hako~ensis

PAT~ $ C ~ a p~DI ~65~ p~T~ J A P ~

A~GE URSI

aRak C O R D AETE JAPO ~ T E PRIN ARUN OIOl ARiel HIRT AST5 AGER

~STE S C A B A $ ¥ [ THUN BQ~M T R I C CAC~ H&~T CALA ~ N I G CAt A H A K n

P a r ~ a ~ g i a paluatri~ Pa=rinia scablosaefol£a Pedi~ularis r e s ~ p £ ~ a ~et~i~as ~poni&~s v. glga~£e~s PI~F I£RN ~inellia t e r s ~ a PEAT G R A N Pla~ycodon grandifloru~ p[A~ T [ P U Platan~hera tipul~idea V, ~ippomi=~

P£EC T R I C POLy C U S P POLY JAPO P[lLy ODOR POPU SI~ 5

eCtra~ ~ r e o ~ p u . o!ygonQm cuspi a um Polyg a~a japonica Poly~onatum odoratum v. pl~rifolia Pop~lus sieholdii ~ R A G ~ete~tilla f l ~ g a r i ~ d ~ v. ma]o~ ~EV ~oten£il[a fre)~iana V U L G Prunella~Igaris w lilaeina AQU 1 Pteridlum aquilinum

~OT[ @OTE pRUN

pT ER

CAR~ L~NC Carex ]~nceelata CAR£ R [ R V ¢arex ~ v a t a ~LA Ofl~ Celastr~s o=bicolatus C ~ t O S~RR Chloranth~s serra~us CIRS dhPC] ~irsi~m japonicum O I R ~ NI PP Cirsi~m n~ppon~c~ CL~ T£~ C l ~ t i ~ ter~ifl~ra

OUER SERR RANU JAPO RU~ ~KAN RUBU pALM RUBU PARV ~ t ~ VULP

~uncul~ jap~iaic~ flubia ~kane E u h D s p a i ~ C u ~ v~ coptophyllus ~ubusparvifolius S ~ i X vulpine S~ P C L M sesa p a l ~ t ~ SCEP TERN Botzychium cer~atum SENE C A N N $enecio cannab~folius SIPH C N I N SiphenDste~ia chinensls SmIL 81FL $~il~x c h ~

CLIN MICRo m i t r e d ~ Y N A PANI Gyanchum p~niculatum D[SM OXYP Desmodium ox~phyllum OlOs TOKO 01o~corea t o k o r o OISp S~lh DJsporu~smilaclnum E P l P THUN ~plpacti~ t h u n b e r z i l COP4 CHIN Eupatori0m chinens¢ v, simpiiclfolium £UPA LIND ~upatorium lindleyan~m G ~ ~RIF ~ n ~ £ a n a ~ r i f i o r a v . japgni~a H A L O RICR ~ a l o r a g t s m l c r a n c h a

STRU NIPO

H £ R ~ LONG H@$T REC]

THE S C H [ N TILl M ~ X I

SeCT VIRq3 S o l t d a ~ s v i = ~ a - a ~ m

SPOD SIBI ~Wp~ J A P O

VITI ¢O]O VII I FIE1

Viti~ noi~netla~ ViCi~ [icifolla v. loDa~a

V I T I FLEX Vi~is flezuosa

VIOk MANO Vlol~ mandshu~Ica V ~ O L O B T U Viola ohtu~a VIOL SP Vi~la ~ VIOL VFRE Viola ec~nda ZOy S JAP~ goyaia japonica ENO£ . . . .

Ixeri~ de.tara

BULR L a p o r t e h bulbifer~ THEL Lastrea ~heiyp~eri$ ANAN L e l b u ~ t z i a anatWr~g

~[CO

v. hypoleucum ~ e s i u m chinemse Tilla m~ximowi¢~ia~

WEIG HORT ~¢ei~ela h o r ~ e . ~ i ~ Yl~U 9IbA viburr~m d i l a t a ~

INIIL SALI Inula saiCci,,~ ~. ~ i a =a IRIS ENSA Irls ensa~a v. spolltanea

IXFR D E N T

v . asidE%ca

Spodiopogon slbiricus s~echnum nippo~icum Swerti~ japonica

SY~D PU~G S y n u ~ p ~ g e n ~ TWAL M I ~ U T}lal~ctrum m l n ~

}~erminium longicrure Hosta r~c~ifolia HYDR ~ACR ~ydrangea (racrophyliav. megacarpa HYDR PA~[ ~iyerangee ~antcal~ta H y D R RAMI ~ydr~co~y]~ ~amiflo~a H Y P E ~REC })ypericum erectur0 tl

LAPO LAST L~IB LESP

Q~ercuaserreta

Lea~edez~ bicolol f. ac~tifolia

I ~ C JAPO Leuc~scep{~um japon~m~m L[LI ~URA IAli~m auratu~ LIPA KUMO Lipazi~ k~mokiri tvsI

CUE;

MEYI L U T E M[$C SI~E M[JHL JAPi]

2)

Lysimaehi~ clenhroide~

Meta~arti~eci~m lute-v/ride ~iscanthus ~inensis ~uhlenbergla japonica

DIS{]~P~NCIES

9E~EEZ

USED

3)

~ELD~

NO,

O~

lWE

THE

ORIG~L

I N ORIGIIq~L

~I,EVE

US~-0 FOR

TABL~

NA}~

~iE

IN

COMP U ~ R

IASL~

Omitted Labia~le ~p.

SrRU

NIPO

B l e c h ~ m nlpponicum

TABLES

VS.

X

S~l 592 593 59~ 596 5~7 599

I ii iz 13 ]4 15 16 23

ARE

5LW~

~Tmm

. . ~01 602 603 604 622 ~23 628

~0.

.

or

.

~L~IEVE$ USF~D IE

T~m

.

ORIC~

. la . 19 20 21 22 Z3

SORT

AND

~

2

MICR

c0~u~E~

I[~%~BE~ K C E ~ I N ~

~W

I~BL~

L%IN

"°

3 5 6

RAW

~IE~D

.

C0~IP~T~R TABL~

TA~Lm

62~ 627 629 631 832 63~

ORGA~IZA~I£g.

F~NAL

TABLE

~

~

~

÷~

~

~

~

~

~

~

o ~ i~!~!

~

o~

÷

÷

~÷

....

~

~

~

~ ~

; ~ ; ~

~;

;t

~ii

~

÷

+

~-~

~.

! !iiii i!i i

~ ~iiiii ~i~i~ ~~i!i!!ill~ i~

~

~

~,~

oTM

~,~.-~~~°~~