장음표시 사용
211쪽
Each original sample si . e. 32 samples excepi sor those with very low total abunda iace) was testud by the pro per option os the DIVERSI sol tware IZSΛK l99J) to sit to the truncaled lognormaland to the log- series distributions. This sol tware Contains - among others - routines to calculate jackkni se estimates with confideiace intervals and significance tests os dissere iaces sor severat diversi ly indices. Morco vel . it offers sitiing procedures for the most popular species abundance distribu tions. The truncated lognormal modet provides an S estimate of the total species number. Simulations hased on urn modeis related to the truncated log normal distribution were made
each case l00 simulated recor is species abundarace series) were generaled and the simulated num-her os species was recorde l. Fit si os ali samples with te ad ing relative frequencies similar to that os
212쪽
The estimated number os species calculated by the four methous are summari sed in Table 2. The combino sample os ille summari sed abundances in ullthe 2l samples si om KunsZenim il los is cocled as ΡKS 2l : neglecting one os the2l sumples clue io extreme ly high tolat abundance and willi superdominance os
give the most conservative estimates, heing the nearest to the observed speciesrichness siliat is . a conservative estimation is proposed as best). The 96 dipterous species which were collected in the sield of the V pusZtaV al KunsZent naiklos musibe character istic os the 14y communi ty there io a gi ven extent meastire. There isno ollier GDSoli tes means sor ille assessianent of the estimators hul the iotal numberos species observe l. This is why we must regarii the most conservative est imalion as the best One.
Species os the dipterous fami ly Sphaeroceridae were collected on si cep-rundi oppings with higher probabili ly than the other Hies. So. we selected sphaei Ocerius as the resere iace-group and applied Hodkinsons ' method muttili, multinu X in calculat ing est imal ing) ille number os species by assum ing that ait ille dipterous species are caught by the fame probabili ty as the sphaerocerius.
213쪽
214쪽
215쪽
Is ait the si ies were captured with the fame probabili ty as the sphaeroceridspecies. l. 694'38 - lJ α Si species ut borraspusZta and l. 694'4l l9 α 88 species at Vertesbogiar would be collected sinstead of ββ and 60 species observe l). Again. under ille sanae assumption l. 559φ3J - 2β α 82 species andi. ββ9'38 ε 3l α 90 species would have been collected in Munsgenimii los in l903 and in l094 sin idad of 62 and 69 species observe l) respecti Vely. Hodi insons ' method modis ted illis way is applicabie to extrapolations in Communities where the capture probabili ty os a gi ven taxonomic grotis' is significant ly higher than that os ille resi of the species.
different to XI communiti 'S. Here ni equa is ille Sample Si Ze.
Looking at the data in Table 3. one can come to the conclusion that in therange os abundances belween 339 to 343. estimations based on the communi tyKS 20 and KS 2l are equat ly distant si ona iliose of the M 30β 262 communitys S 20 estimates are higher. those of KS 2l are lower). When numbers arehigher. estimates of KS 2l are closer io those of M30β 262: when lower xvii hnumbers HS 20 estimates are closer to those of M 30β 262.lf the sample si Ze n is smali. ES m) underestimates ille species numbers based on ali the three communities; though willi communi ty KS 20 and 430β 262 the underesti maled values are nol far frona observed values. Since S 930602 is an extreme sample see PAPP l99 ). ii was hetier to leave it out
216쪽
A characteristic t ruit os the frequency structure is the ratio os ille lea lingsreque iacies: those os KS 20. KS 2l and M305262 are l369l4048α n. 2JJ. 5828I968lα 0. 602 and 0. 305. respective ly. Ii appears that ille resulis with this method are sensitive to the exire me ly high frequencies of the sample studi ed sinthe se cases ii overestimates ille species num heri. It must he noted that compara
li vely the hesi estimates are gi ven by the M 30 262 modet and nor the frequency
217쪽
species numbers si ona simulations
When comparing the two combined si eld samples, it is sui prising that is wedi op the sample which gave half of the total abundance, this procedure reduces the total species num her by three; but the communi frequency Structure is liti le
218쪽
species numbers si ona simulations
assected Fig. li. When any os the two combineis siel l samples and simulationmodet M 305262Il00 are plotted on ille sanae scale see Fig. 2) it is obvious ihaithe Ἀ- section os ille former is tess fleel' i. e. the lest supper) hais os that graph is undei posed to that os the truncateis lognormal mode l). This satisfactori ly explans
219쪽
why we had lower observed species numbers in severat Cases than the corresponding simulation resulis. Some ollier correlations based On Our experienCe and On elementary Considerations are as follows For modeis identical or very similar in their sther main characteristi cs; stichas the total abundance, number of species; steepness of the Ἀ- section. relative fre
c) the median is lower in the case of the modet. e ih in k that the propo sed computer simulation method is uses ut forspecies number estimation as weli as for analysis of community frequency structure. Tests by urn mode is with parameters Such as lea ling frequency median Stoepness of Ἀ- section etc., similar to tho se of the fie id communities seem very promising. we are stili only exploring this kind of simulation and analyse their potentials so only a limited number of re mari s is made here.
220쪽
when a series os sumples is taken sor ille sanae si ies; collecting more and more animal individuat s. the imaginary relative abundance graph is sol med si omitte direction os the high ranks and the emerge iace and frequencies os the rare species are added as amplifications. This is why the rank- abundance curve mades rom samples of in adequale ly smali abundances is ne arly linear at the logarithmicscale. and thus rei locis thes under election os res e X eci 'S. li is trivial to say thai the relative position os two species on ille graph is determined by the ratio ostheir abundance. hul it is just as trivial that the addition os one individual io ille doma in os dominant- subdominant species causes change in the seque iace by verylow probabi l ity contrary to that in the doma in of the rare species; noi to mentionille dissere iaces in the shape os ille graph since it is a togarithmic one. This simulation method seems ad vantageous even is compared to the polentials os ES m) estimations. With a gi ven frequency structu re os a Communi ly HS m) calculation gives a single es limate. although by any method os re-sam P
calculate i. At ille sanae time. the proposed simulation based on a gi ven si e quency structure, cuia be made in a discretional number. the numerical resulis os a simulation series mal e ii possit, te to calculate the mean Variance, et C. A, n by
Fig. 3. Relative frequency distribution in the log normal modo is swilli l00 species) used in simulations modet V 480 362 si hic k line) and modet M 305 262 sthin line) on logartihmic scale