The two data sets, 1978 and 1993, contained a combined total of 445 species and 23,365 individuals. The 1978 data set was composed of 228 species and 4,047 individual beetles. The 1993 data set was composed of 357 species and 19,318 individuals. Eighty-eight (39%) of the species in the 1978 data were absent from the 1993 data and 217 (61%) of the species in the 1993 data were absent from the 1978 data.

The experimental design for 1978 incorporated six sites with 42 pit-fall traps and totaled 446 trap units over 19 trap periods (=23 trap units/period). The 1978 data represent the catch from 1610 trap-days (destroyed traps excluded). The experimental design for 1993 incorporated five sites with 28 pit-fall traps and totaled 475 trap units over 17 trap periods (=27.9 trap units/period). The 1993 data represent the catch from 3472 trap-days. In the 1978 study ca. 45% of the traps per period provided no data, whereas in the 1993 study ca. 0.2% of the traps per period provided no data. These differences between the data sets indicate why comparisons between 1978 and 1993 are invalid. In addition to the changes in the ecosystem, the differences in trapping methods introduced a large and unknown source of variation. For example, it is almost impossible to state why a species might have shown different abundances in 1978 to 1993--is it due to some dramatic ecological effect or is it simply due to the different methods employed for each year? In most cases, the presentation of 1993 control data are exclusive of sites 1,3 and 5 because these 1993 sites lacked carcasses. For each analysis the expected outcome under the null hypothesis is presented and the observed outcome is then compared with the expected.

Abundance: Coleopteran

If the null hypothesis were true, one would expect to observe no significant difference between the abundance of beetles in carcass versus control traps. However, the mean number of all beetles per trap unit was greater in the carcass than in the control traps (Table 6) in both 1978 and 1993, and for the combined data. The total abundance of beetles in carcass traps ranged from three to six times greater than the abundance in control traps.

Table 6.Carcass to control comparison of gross beetle abundance. ANOVA of mean number of individuals caught per trap unit between carcass and control traps. Standard deviations are in parentheses. Sites 1, 3, and 5 from 1993 are excluded.

year(s)	carcass mean no. of beetles,	trap units carcass	control mean no. of beetles,	trap units control	F	P<
1978 & 1993	42.5 (73.5)	386	9.4 (11.9)	331	65.829	0.001
1978	12.8 (15)	251	4.3 (4.8)	195	58.448	0.001
1993	97.8 (101.8)	135	16.7 (15)	136	84.429	0.001

Differences between carcass and control data sets are consistent for both 1978 and 1993 across all time periods (Figures 3, 4). The slopes of both years carcass trap beetle abundances (1978=-0.37, 1993=-18.01) were steeper than the control trap beetle abundances (1978=+0.007, 1993=-2.65). This trend was more evident in 1993 (Figure 4).

Abundance: Species

Based on the criteria presented in the Methods section, the abundance of individuals within species was found to be significantly greater in carcass traps than control traps for 37 species of beetle in 1978 and for 42 species of beetle in 1993 (Tables 7 & 8). Eight 1993 and three 1978 beetle species were found to be significantly more abundant in control traps (Table 9). Of the total 385 species tested for association during the study, (1978 and 1993 combined), 57 were found to be carcass associates (Tables 7 & 8) and 11 were found to be control-trap associates (Table 9), leaving 317 species apparently unaffected by the presence of a carcass.

These results can be compared to Bornemissza's (1957). Using the total faunas sampled as n1978=228, n1993=261 adult beetle species for this study and n=47 arthropods for Bornemissza's study, Bornemissza found that 34% of the taxa studied were absent from his control samples, whereas, I found 52% of the 1978 species and 48% of the 1993 species were absent from control samples. Bornemissza found that only 7% of the taxa studied were present in both control and

Figure 3. Distribution of beetle abundance over time, carcass and control trap data from 1978. Data from trap period 11, 12, 13 and 16 were corrected to be comparable with the other trap periods (see Appendix C). See Table 3 for calender dates corresponding to trap periods.

 

Figure 4. Distribution of beetle abundance over time, carcass and control trap data from 1993. See Table 5 for calender dates corresponding to trap periods. Note that the final abundance of the carcass trap data is 12.7 beetles per trap and 5.8 beetles per trap for the control trap data.

 

carcass samples, whereas I found that 33% of the 1978 and 35% of the 1993 species were present in both control and carcass samples. When the strength of association is included and the fauna are thus restricted to those species listed in Tables 7 and 8, slightly higher values are obtained: 43% of the 1978 and 50% of the 1993 species were present in both trap types. Bornemissza's results led him to conclude that the control fauna was generally excluded from or actively avoided the carcass region and thus comprised an insignificant portion of the carrion community. My results show a greater overlap between species in control samples and carcass samples. This disparity may result from the differences in scale and taxa studied yet indicates that much remains to be discovered about the role of background biota in carrion communities. To what extent does the carrion food web extend into the ecosystem? How many and to what degree do species traditionally considered to not be saprophagous tie into carrion food webs? This line of inquiry must be further investigated if we are to understand what role carrion plays in ecosystem functioning.

For the data in Tables 7, 8 and 9, it should be noted that in two-category, Chi-square analyses of more than 20 species one expects to see some significant P-values simply due to random chance. For example, if investigating 1000 species, one would expect by chance alone that 50 of those species would display P-values of 0.05 or less, even if the null were true. In this study, the combined 1978 and 1993 data set included 385 species, thus under a true null hypothesis, one would expect 19 species to show P-values of 0.05 or

Table 7. Carcass Associated Beetle Species, 1978. Total abundances of 37 beetle species. See methods for criteria of species inclusion, see Appendix C for bionomic information and see Appendix D for complete species association list.

species	abundance in carcass traps	abundance in control traps	Chi2	P
Thanatophilus lapponicus (Hbst)	467*	1	360.0	0.000
Dermestes talpinus Mannerheim	144*	2	106.4	0.000
Anaspis rufa Say	188	18	102.5	0.000
Creophilus maxillosus (L.)	129*	0	100.2	0.000
Saprinus oregonensis LeConte	133*	2	97.9	0.000
Ptiliid sp. 4	126*	5	84.8	0.000
Trachypachus holmbergi Mann.	201	33	83.4	0.000
Omaliinae sp. 8	84*	0	65.3	0.000
Catops basiliaris Say	78	1	57.9	0.000
Trox sonorae LeConte	70*	0	54.4	0.000
Notoxus serratus LeConte	70	1	51.7	0.000
Necrobia violacea L.	58*	0	45.1	0.000
Staphylinidae sp. 65	66*	3	43.5	0.000
Dermestes fasciatus LeConte	50*	1	36.1	0.000
Staphylinae sp. 7	46*	0	35.7	0.000
Otiorhynchus ovatus (L.)	213*	79	33.0	0.000
Aphodius fimentarius (L.)	42	0	32.6	0.000
Oxytelus sp. 18	37	0	28.7	0.000
Saprinus lugens Erichson	36*	0	28.0	0.000
Staphylinidae sp. 80	28	1	19.1	0.000
Staphylinae sp. 5	24*	0	18.6	0.000
Anchicera sp. 2	23*	0	17.9	0.000
Omosita inversa LeConte	20	0	15.5	0.000
Borboropora quadriceps (LeC.)	19*	0	14.8	0.000
Encimus mimus Fall	19*	0	14.8	0.000
Corticarina cavicollis (Mann.)	22*	1	14.5	0.000
Staphylinidae sp. 63	17	0	13.2	0.000
Staphylinae sp. 4	13*	0	10.1	0.001
Staphylinidae sp. 72	16	1	9.9	0.002
Anotylus sp. 17	10	0	7.8	0.005
Tachinus basalis Erichson	9	0	7.0	0.008
Staphylinidae sp. 64	9	0	7.0	0.008
Cryptopleurum minutum (Fab.)	8*	0	6.2	0.013
Syntomus americanus Dejean	14*	3	4.7	0.030
Xantholininae sp. 43	6	0	4.7	0.031
Apteroloma tenuicorne LeConte	25	9	4.1	0.043
Falagria dissecta Erichson	5*	0	3.9	0.049

* 22 of 37 species that were also found to be significantly associated with carcasses in 1993.

Table 8. Forty-two carcass associated beetle species, 1993. See methods for criteria of species inclusion and see Appendix E for complete species association list. a=elk carcass, b=elk control, c=bison carcass, d=bison control.

species	a / b	c / d	Chi2	P
*Thanatophilus lapponicus (Hbst)	2730 / 0	4958 / 4	7732.9	0.000
*Saprinus oregonensis LeConte	444 / 1	371 / 3	809.1	0.000
*Necrobia violacea L.	52 / 0	553 / 0	609.5	0.000
*Ptiliidae sp. 4	99 / 0	440 / 5	528.1	0.000
*Creophilus maxillosus (L.)	27 / 0	131 / 0	159.2	0.000
*Staphylinidae sp.65	87 / 4	50 / 0	126.4	0.000
*Falagria dissecta Erichson	33 / 1	87 / 4	106.7	0.000
*Staphylinae sp. 4	21 / 0	84 / 0	105.8	0.000
*Dermestes fasciatus LeConte	20 / 0	67 / 0	87.6	0.000
*Anchicera sp. 2	55 / 7	33 / 0	69.7	0.000
Agonum cupreum Dejean	52 /10	37 / 1	61.4	0.000
*Encimus mimus Fall	58 / 2	4 / 0	56.7	0.000
*Omaliinae sp. 8	26 / 0	35 / 2	55.7	0.000
Oxytelus sp.19	7 / 0	48 / 0	55.4	0.000
*Staphylinae sp. 7	24 / 0	27 / 1	48.4	0.000
Sphaeridium lunatum Fab.	1 / 0	44 / 0	45.3	0.000
Ptiliidae sp. 3	8 / 0	31 / 0	39.3	0.000
*Corticarina cavicollis (Mann.)	44 /11	38 / 9	38.1	0.000
Nicrophorus investigator Zett.	13 / 0	20 / 0	33.2	0.000
Sphaeridium scarabaeoides LeConte	0 / 0	32 / 0	32.2	0.000
Ontholestes cingulatus (Grav.)	6 / 0	19 / 0	25.2	0.000
*Borboropora quadriceps (LeC.)	20 / 0	3 / 0	23.2	0.000
*Trox sonorae LeConte	17 / 0	4 / 0	21.2	0.000
Tinotus sp.37	8 / 0	9 / 0	17.1	0.000
Melanophthalma villosa Zimmerman	34 / 5	2 / 4	16.4	0.000
*Saprinus lugens Erichson	8 / 0	8 / 0	16.1	0.000
*Staphylinae sp.	59 / 0	6 / 0	15.1	0.000
Nicrophorus guttula Motschulsky	8 / 0	7 / 0	15.1	0.000
Bryoporus sp.36	1 / 0	16 / 1	14.3	0.000
*Dermestes talpinus Mannerheim	5 / 0	6 / 0	11.1	0.001
*Otiorhynchus ovatus (L.)	196 /139	79 /65	11.1	0.001
Tachinus angustatus Horn	14 / 2	1 / 0	10.0	0.002
*Syntomus americanus Dejean	65 /26	9 /16	9.1	0.003
Nitidula ziczac Say	1 / 0	7 / 0	8.1	0.005
Nicrophorus hybridus Hatch & Angell	5 / 1	9 / 2	7.2	0.007
Heterosilpha ramosa (Say)	0 / 0	9 / 1	6.5	0.011
Tachyporus (T.) sp.25	17 / 8	11 / 5	5.6	0.018
Chrysomelidae sp.42	7 / 1	1 / 0	5.5	0.019
Bembidion dyschirinum LeC.	24 / 7	5 / 7	5.3	0.021
Nicrophorus defodiens Mann.	2 / 0	3 / 0	5.0	0.025
*Cryptopleurum minutum (Fab.)	1 / 0	3 / 0	4.0	0.045
Aleochara verna Say	3 / 0	1 / 0	4.0	0.045

*22 spp. that were also associated with carcasses in 1978.

Table 9. Control Trap Associated Beetle Species. Species significantly associated with control traps. Three beetle species from 1978 and eight from 1993 that were caught significantly more frequently in control traps of more than one site.

year & species	species abundance in carcass traps	species abundance control traps	Chi2	P
1978 species
Calathus ingratus Dejean	20	74	46.8	0.000
Trichodes ornatus Say	28	126	90.8	0.000
Epicauta puncticollis Mann.	6	40	34.9	0.000
1993 species
Listronotus sp.58	8	55	34.7	0.000
Pachybrachis sp.57	0	27	26.8	0.000
Morychus subsetosus Fall	0	13	12.9	0.000
Mordellid sp. 30	1	11	10.9	0.001
Byrrhus americanus LeConte	7	24	9.2	0.002
Canthon simplex LeConte	6	21	8.2	0.004
Microlestes curtipennis Casey	0	6	6.0	0.015
Tachyporus sp.27	8	20	5.1	0.025

less. However, my analysis of species abundance associations with each trap type revealed a total of 147 of the 385 species were observed to display P-values of 0.05 or less. Given a true null hypothesis, the probability of seeing 147 P-values of 0.05 or less with only 19 expected must be very small.

Tables 7, 8 and 9 list those species that I have identified as significantly affected by the presence of a carcass. These species are those that would be most likely to be affected by changes in the frequency of carcass production. The silphid Thanatophilus lapponicus (Herbst) was not only the most common beetle collected but was also one of the largest (length= ca. 10 mm). In addition, T. lapponicus larvae were equally abundant on the carcasses, which they were consuming. When considering which species are the more important members of communities, biomass is often an important criterion. Abundance considered without biomass could be misleading (because biomass was not directly measured I discuss only estimated biomass which was obtained by multiplying the estimated mean body length of a taxon by the number of individuals captured). For example, one of the more common species in both 1978 and 1993 was Ptiliid sp. 4 (Tables 7, 8), yet adults of this species are under 2 mm in length and thus represent a small fraction of the biomass. Clearly, the great numbers and large size of T. lapponicus suggests that this species contributes significantly to the movement of energy and nutrients into and out of Lamar Valley carrion communities.

Saprinus oregonensis LeC. (Histeridae) and Creophilus maxillosus (L.) (Staphylinidae) are two large and quite abundant predaceous species in the carrion communities of the Lamar Valley. These species prey on fly larvae and have been considered beneficial due to their impact on fly populations (White 1963). Aleochara verna Say (Staphylinidae), a species found in low numbers in carcass traps during 1993, is a parasitoid of immature anthomyiid and sarcophagid flies and has been documented as beneficial for its control of cabbage and seedcorn maggots (Klimaszewski 1984). In addition to predators that specialize on carrion community members, Tables 7 and 8 list numerous other predaceous beetles that are not documented to occur at carrion. The majority of these species were carabids, such as Agonum cupreum Dej., and their contribution to the carrion community was unexpected.

The species that are present in control communities and are not classified as associates of carrion, such as A. cupreum, are those that most clearly demonstrate the need for the use of control traps when investigating carrion ecology. These species are active members of the control communities and appear to be opportunistically exploiting the carrion resource introduced into their habitat.

The species in the genus Dermestes (Dermestidae) and Trox (Trogidae=Scarabaeidae) are known to feed on dried skin and ligament tissues, and hairs (White 1963). In 1978 the number of adults in the species Trox sonorae Lec., Dermestes talpinus Mann. and Dermestes fasciatus LeC. represented 7% of the number of beetles trapped, whereas in 1993 these three species represented only 0.6% of the number of beetles trapped. I suggest that the near total lack of skin tissues on the carcasses in 1993 might explain this observation. These beetle species and other invertebrates that specialize on the skin tissues of carcasses could experience a significant loss of resources during years in which the demand for carrion by vertebrate scavengers, such as coyotes, was greater than the supply. In 1978 the supply of elk carcasses was so great that the vertebrate scavengers left the carcasses essentially intact--the ideal condition for supporting large populations of skin feeding invertebrates. This insight into the carrion food-web suggests that the ratio of necromass to vertebrate scavengers may be an important variable in the population dynamics of numerous invertebrate species.

Table 10. Total abundances for the combined 1978 and 1993 data in each of two trap types by family or subfamily. The 1993 sites 1,3 and 5 are excluded from the abundance data but not the species counts (which account for two families, Nemonychidae and Lampyridae, containing 1 species each that are not listed here). The families Trachypachidae and Trogidae are included within the Carabidae and the Scarabaeidae, respectively.

	carcass	control	total
Number of trap units	386	331	717
Number of beetles	16418	3106	19524
Number of families	37	31	41
Number of unique families	10	4
family	no. species	carcass	control	Chi2	P
Silphidae	8	8234	13	7023.7	0.000
Histeridae	6	1015	8	847.8	0.000
Ptiliidae	3	708	10	579.1	0.000
Staphylinidae	124	2268	980	334.2	0.000
Cleridae	4	694	130	306.2	0.000
Dermestidae	8	309	6	248.3	0.000
Scraptiidae	3	198	18	124.4	0.000
Lathridiidae	7	239	42	110.2	0.000
Hydrophylidae	12	113	0	96.9	0.000
Leiodidae	14	89	1	73.5	0.000
Scarabaeidae	27	239	67	72.5	0.000
Nitidulidae	6	70	0	60.0	0.000
Cryptophagidae	10	217	66	59.4	0.000
Meloidae	4	34	115	57.7	0.000
Anthicidae	4	218	332	44.6	0.000
Mordellidae	2	6	46	37.4	0.000
Carabidae	57	855	559	25.0	0.000
Byrrhidae	10	14	46	22.5	0.000
Brentidae	1	0	10	11.7	0.001
Chrysomelidae	22	45	71	10.6	0.001
Curculionidae	28	534	379	8.0	0.005
Tenebrionidae	3	79	41	7.0	0.008
Cicindelinae	3	19	5	6.2	0.013
Eucinetidae	1	8	1	4.4	0.035
Agyrtidae	1	26	12	3.3	0.071
Micropeplidae	1	3	0	2.6	0.109
Cerambycidae	2	0	2	2.3	0.127
Melyridae	8	26	32	1.9	0.169
Hydraenidae	2	2	0	1.7	0.190
Clambidae	1	2	0	1.7	0.190
Ptininae	2	2	0	1.7	0.190
Corylophidae	3	2	0	1.7	0.190
Scydmaenidae	3	0	1	1.2	0.280
Lucanidae	1	0	1	1.2	0.280
Buprestidae	2	5	2	0.9	0.350
Throscidae	1	1	0	0.9	0.354
Anobiidae	1	1	0	0.9	0.354
Scolytidae	1	1	0	0.9	0.354
Elateridae	27	131	101	0.6	0.422
Coccinellidae	12	10	8	0.0	0.884
Cantharidae	2	1	1	0.0	0.913

Table 10 shows that the most species rich family in this study was the Staphylinidae (124 species) and the second most speciose was the Carabidae (57 species). Beetles of these two families were also captured most frequently in carcass traps (Table 10). However, staphylinids are generally very small animals, usually measuring less than 3 mm in length, whereas carabids are more frequently in the 5 to 15 mm range. The family Silphidae, although the most important family in terms of strength of carrion association, was represented by only 8 species. And 99% of the silphids captured belonged to just one of those 8 species, T. lapponicus. In brief, the family Silphidae had the greatest abundance and estimated biomass but a low species richness. The family Staphylinidae had the greatest species richness and the second highest abundance but a lower estimated biomass than either silphids or carabids. And the family Carabidae had the second highest species richness and estimated biomass, but ranked third in abundance.

The families Scarabaeidae (27 species), Elateridae (27 species) and Curculionidae (28 species) comprised the next most species-rich families of the study. The scarabaeids and the curculionids showed significant carrion associations (Table 10). Elaterids and curculionids are phytophagous, which explains why elaterids showed no carcass effect. The carcass effect visible in the curculionid data is due entirely to one species, Otiorhynchus ovatus (L.), (see Tables 7 and 8) a species introduced from the Palearctic (Warner & Negley 1976). This anomalous observation of a phytophagous species occurring more frequently in carcass traps at multiple sites in 1978 and 1993 indicates some aspect of this species' life history is unknown. Perhaps adults of this weevil species are more frequent at carrion because they are obtaining salts or other nutrients. Some lepidopteran adults are known to frequent carrion apparently to imbibe organic compounds to augment their nectar diets (Payne & King 1969).

Another means of testing for differences in abundances between carcass and control samples is to plot the total abundance of each species in carcass traps against the total abundance in control traps. If the null hypothesis were true one would expect to see a strong positive correlation along the line y=x because such a result would indicate that species display equal abundances and variances in both types of traps (i.e. the trap types are indistinguishable based on abundance values for species). Stated another way, the probability of seeing x individuals of species A would be equal for both trap types. However, if the null hypothesis were not true, one would expect to see some deviation from a strong positive correlation along the line y=x.

Figures 5 and 6 display the abundances of species in both carcass and control data sets for 1978 (Figure 5) and 1993 (Figure 6). Both plots show deviation from a strong correlation along the line y=x. Although both plots indicate that many species seem to

display almost equal abundances in both trap types, there are numerous species that show carcass to control abundance ratios which differ from 1:1. These results parallel those of Tables 7, 8 and 9, which indicate that the majority of species (238/385) appear to show nearly equal abundances in both types of traps and that fewer (147/385) show unequal abundances. The data from both years show similar trends. This degree of agreement between data sets taken 15 years apart suggests that similar organizing processes could have been acting during both years.

However, it should be noted that the consistency between the two year's data represented in this form is predicted by the consistency seen in the species rank-abundance plots (Figures 7, 8). Because both years showed the log-normal distribution of many species with low abundances and few species with high abundances the Cartesian plot of abundances must also show similar patterns.

Figure 5. Scatter plot of 228 species by total abundances in carcass and control data sets for 1978. Data were transformed by y=x+1 to allow log transformation. Line: y=x. (Note: many species with equal carcass to control abundance ratios are plotted on top of each other and thus hidden).

 

Figure 6. Scatter plot of 261 species by total abundances in carcass and control data sets for 1993. Sites 1,3 and 5 are excluded. Data were transformed by y=x+1 to allow log transformation. Line: y=x. (Note, many species with equal carcass to control abundance ratios are plotted on top of each other and thus hidden).

Although many species classified as saprophagous in the literature were found to display carcass associations, a number of species with no recorded affiliation for carrion also showed significant association (Tables 7,8, Appendix B). The presence of these species, such as many carabids, more frequently at carcasses might be a result of their predaceous feeding habits. Many species are known to be attracted to carrion in order to exploit the concentrated prey resources made available. Species with no known affinities for carrion, such as listed in Appendix B, may be benefiting from the carcasses. Repercussions that might result from changes in the availability of carcasses could impact the populations of these species. Although this study provides only a hint of the entire food web associated with large mammal carcasses in YNP, there are obviously a great many unpredicted links that require detailed investigations to uncover.

Species Diversity:
Abundance & Richness

Under the null hypothesis one would expect to see no difference in species diversity between carcass and control samples. The log series alpha species diversity index was calculated for both the carcass and the control data sets of 1993 and 1978. These results are summarized in Table 11.

Table 11. Log series alpha species diversity index values for four data sets. N=number of beetles, S=number of species.

data set	trap units	N	S	ratio*	alpha
1978 carcass	251	3214	195	2.42	45.69
1978 control	195	833	110	6.77	33.95
1993 carcass	135	13204	212	1.12	35.87
1993 control	136	2273	139	4.50	32.65

* Ratio: (S/N/trap units) x (10000)

The ratio of species richness to total abundance (Table 11) gives an understanding of how many species were captured relative to the number of individuals, adjusted by the number of trap units. These results indicate that the control sample of 1978 caught the most species relative to the number of individuals. The 1993 carcass data set contained the fewest species per individual trapped. This observation is obviously a result of the high abundances of carrion associated species. This conclusion is substantiated by recalculating the ratio without the five most abundant carcass associated species from the 1993 carcass data (S/N(100)= 4.41, N=3480, S=207).

The data used to create the indices presented in Table 11 were those used to create the species rank-abundance plots (Figures 7, 8). According to these diversity indices, the 1978 carcass community was the most diverse of the four data sets. Both year's data indicate the carcass communities were more diverse than the control communities and thus support rejection of the null hypothesis. However, the statistics have not yet been developed that could quantify the strength of similarity or difference between values of this diversity index.

For both 1978 and 1993 the carcass data sets had greater species richness and abundance than the control data sets (Table 11, Figures 7 & 8). Figures 7 and 8 show that although the values were greater for the carcass data sets, the control sets appeared to follow the same underlying distribution. The distribution of all four data sets were principally defined by the presence of many rare species (singletons) and by few abundant species. This distribution closely resembles the log normal pattern of species abundance, which is a distribution that is expected for large, mature, and varied communities (Magurran 1988).

Opinions abound regarding the interpretation of the log normal distribution. There are those who feel this distribution is an artifactual mathematical result with little or no biological importance and there are those who believe the distribution results from and is inherent to underlying evolutionary and ecological processes (Sugihara 1980; Hughes 1986; Magurran 1988). For example, Sugihara (1980) argued that the artifactual explanation was inadequate and proposed an alternative hypothesis. His analysis supported an explanation based on a hierarchical community structure represented by sequentially divided niche space. Sugihara's investigation raised the intriguing possibility that the log normal distribution is a direct consequence of the process of speciation (implying that each successive species that evolves has a smaller portion of the total niche space to use which would translate into fewer resources and lower abundances).

 
Figure 7. Species-rank versus abundance (log), 1978. Carcass and control trap data. Total species in carcass data set=195, in the control=110. Total individuals in the carcass data set=3,214, in the control=833.

 

Figure 8. Species-rank versus abundance (log), 1993. Carcass and control trap data. Data from sites 1,3 and 5 were excluded. Total species in carcass data set=212, in the control=139. Total individuals in the carcass data set=13,190, in the control=2,273.

 

Another, less biologically-oriented explanation, is that the ubiquitous nature of this distribution results from the mathematical tendency for the distribution of errors that each are aggregates of many independent errors to be approximately normal [as described by the Central Limit Theorem, (Box et al. 1978)]. From a biological perspective this tendency is explained by the fact that species have many variables that define their survival (e.g. concentration of O2, temperature, etc.) and that as a variable departs from optimal, the species becomes less abundant. Therefore, because there are many variables there are correspondingly many ways for a species to be uncommon but it is unlikely that all the requirements of a species will be met perfectly--thereby allowing few species to be common (Gustafson, pers. comm.).

If the distributions of the species in the communities sampled were log normal then these communities were not sampled to completion, (i.e species remain undiscovered). This conclusion is based on the "Veil Line" concept of Preston (1948), as described by Magurran (1988). Preston's veil line defines the boundary in community ecology data sets at which, usually, the largest category is composed of singletons. Most community ecology data sets show only the portion of the curve to the right of the mode when plotted, which is a function of singletons being the greatest category (Magurran 1988). In exhaustively thorough sampling, the veil line is pushed to the left of the mode to reveal the complete bell-shaped curve of the log normal distribution. It seems logical that, with continuous sampling of a fauna finite in size, the number of singletons in the data will decrease as subsequent specimens are collected. Eventually, there should be very few species that are collected only once and these species would have a higher probability of being "tourist" or non-resident species. Of course, the more mobile the taxa under investigation are, the greater the number of "tourist" species expected.

Species Occurrence

Both the 1978 and 1993 control data sets shared ca. 40% of the species found in carcass data sets (Table 12).

Table 12. Shared species matrix. The percent of species in common between data sets listed horizontally are presented. Total shared species counts are listed in parentheses. Sites 1, 3, and 5 are excluded from the 1993 control data.

	1978 car. (195)	1978 con. (110)	1993 car. (212)	1993 con. (139)
1978 car.	1.00(195)	---	---	---
1978 con.	0.40( 77)	1.00(110)	---	---
1993 car.	0.44( 85)	0.36( 39)	1.00(212)	---
1993 con.	0.30( 58)	0.30( 33)	0.43( 90)	1.00(139)

The similar results for both year's data sets indicates it may be possible to use the species richness of a carrion community to predict the species richness of the local control community. If there is a consistent correlation in the species richness or diversity between carcass and control communities, the sampling of carrion communities may provide a quicker and cheaper means to investigate a region's fauna.

The Morisita-Horn similarity index produced a value of 0.313 for the comparison of the 1978 carcass and control data and a value of 0.054 for the comparison of the 1993 carcass and control data. When this index compares two random samples taken from the same fauna it produces a value of 1.0 and correspondingly produces a value of zero for two samples taken from totally different faunas (no species shared).

This comparison demonstrates that for both year's data the carcass fauna differed from the control. Under a true null hypothesis, comparisons between carcass and control faunas would have produced similarity coefficients close to 1.0. The 1993 data show a value close to zero, indicating these faunas are unquestionably different. Whereas the similarity coefficient for the 1978 data is comparatively large.

The similarity analysis of traps within and between trap types at a site provided a view of the variation in faunal samples not seen when individual traps were pooled (Figure 9). The most obvious pattern, which corroborates the pooled data analysis, shows the between trap similarities (carcass traps vs. control traps) to average far lower than the within trap similarities (carcass vs. carcass, etc.). Most of the clusters of four traps, regardless of type, showed great variation--indicating that one or more of the four traps placed within 2 m of each other produced samples differing from its neighbors. Only site 2, 1978 carcass and site 2 and 7, 1993 control sampled nearly homogeneous faunas (Figure 9). Unfortunately, the source of the great variation in the samples from trap clusters is unknown.

 
Figure 9. Trap similarity. Each vertical bar represents the spread of the similarity index values for the comparison of the pooled samples of single traps within a year. For each catagory (Carcass, Control and Carcass vs. Control) the bars are read from left to right representing 1978 sites 1,2,3,4,5, and 1993 sites 2 and 7.

 

---

87

Species Richness

Table 13 shows the observed and estimated number of species for each of four data sets. The ratios of estimated to observed richness indicate that the data sets differ slightly in the characteristics used to calculate the estimated richness. That is, had each ratio been identical then the percentages of singletons would have been identical for each data set as well. However, the differences were not so great to change the relative ranking of each data set's richness from greatest to least.

Table 13. Species Richness Estimates. The control 1993 data set includes only sites 2 and 7.

	total 1993	total 1978	carcass 1993	carcass 1978	control 1993	control 1978
rank	1	2	3	4	5	6
observed	261	228	212	195	139	110
estimated	353	316	284	274	206	167
ratio (est/obs)	1.35	1.39	1.34	1.41	1.48	1.52

Table 14. Comparison of carcass to control species richness. ANOVA of mean number of species caught per trap unit by trap type. Standard deviations are in parentheses. Sites 1,3, and 5 from 1993 are excluded.

year(s)	carcass trap units	control trap units	mean no. species, carcass	mean no. species, control	F	P<
1978 & 1993	386	331	9.7 (8)	4.5 (3.5)	120.260	0.001
1978	251	195	5.9 (4.9)	2.6 (1.8)	82.562	0.001
1993	135	136	16.6 (8)	7.1 (3.7)	157.022	0.001

Figure 10. Species discovery curves from carcass and control trap data, 1978.

 

Figure 11. Species discovery curves from carcass and control trap data, 1993.

 

Figures 10 and 11 show the accumulation of species richness as a function of the accumulation of trap units. These curves show that a constant difference is maintained in species richness between carcass trap samples and control trap samples across the season of trapping for both 1978 and 1993. Note also that the curves have not reached plateaus, thereby indicating species remain uncollected within the communities sampled. The drop in the slope towards the end of sampling is a standard observation associated with the end of a season (M.A. Ivie & L.L. Ivie, pers. comm.).

Principal Components
Analysis

For this analysis four control variables were employed: year, site, trap type (carcass/control) and trap direction (N,S,W,E). The assemblages of species and their abundances from each of the total 72 traps pooled over all trapping periods of both years constituted the observations. Each of 73 species that were represented by 30 or more specimens were included in the analysis.

The first three principal components accounted for 49.1% of the variation (Table 15). Figures 12 and 13 show the observations plotted against the first and second principal components. Figure 12 shows two clearly separated groups demonstrating that the year of collection was a strong influence on the data. The clustering was especially evident for the 1978 data. The 1978 observations are tightly clustered because their variances were small in comparison to the 1993 variances. When this plot is compared to Figure 13, which shows the carcass influence, a distinct pattern is also visible. Among the 1978 observations there are two separate and distinct, but closely placed groups constituting the carcass and control traps of 1978. The 1993 observations show no overlap between the carcass and control points, but place the control traps far closer together than the carcass traps. Analysis of the sites (Figure 14) demonstrates that the 1993 site 7 carcass trap fauna was most distinct from the 1993 site 2 carcass trap fauna, with the remaining sites showing less variance and falling between these two sites. The powerful influence of the contrast between sites 2 and 7 may result from differences in vegetation (site 7 had more grasses, site 2 had more sagebrush) and other site-specific factors. However, the intriguing possibility remains that some of the differences could have resulted from the different carrion source taxa--a bison at site 7 and an elk at site 2. This possibility is reduced by the comparison of the control fauna observations at each site. The site 7 control fauna separates clearly from the remaining control faunas, indicating a difference exists even without the carcasses that is most likely due to the habitat type.

Table 15. Eigenvalues defining the first six principal components for the combined 1978 and 1993 data set, based on a correlation matrix of the abundances of 73 species in 72 traps.

principal component:	1	2	3	4	5	6
Eigenvalues	16.69	10.29	8.88	4.67	3.88	3.15
Percent of variance	22.86	14.09	12.16	6.40	5.32	4.32
Cumulative percent	22.86	36.96	49.11	55.52	60.83	65.15

Interpretation of the principal components was fairly straightforward. The first principal component separates both the 1978 data from the 1993 and separates the carcass traps from the control traps. Thus, it appears that the differences between the years constitute as significant an influence on the data as the influence of the carcasses on the biota. The second principal component identifies the 1993 site 7 carcass species as distinct from the remaining fauna, and contrasts the 1993 site 2 carcass species. The third through the fifth principal components identify the control traps of the 1993 sites 3, 7 and 5 respectively. It should be noted that no evidence was found to indicate that the direction of a trap (N,S,W,E) contributed any significant nor consistent variation to the data.

These results corroborate the other analyses showing the effect of the carcasses on the beetle fauna, but also, and most importantly, provide information on the relative strength of that effect. The first principal component simultaneously represented the effect of year and carcass, thereby demonstrating that when the year effect is ignored, the greatest contributor of variation is the carcass. As stated earlier, because each year's data are independent tests of the same hypothesis the direct comparison of differences between the years is irrelevant to the study. Thus, the first principal component clearly demonstrates that of the data collected within each year, the most significant biological influence was the presence of a carcass.

Figure 12. The 72 pooled samples of the 1978 and 1993 data sets scored on the first two principal components. Coded by year, 1:1978, 2:1993.

 

Figure 13. The 72 pooled samples of the 1978 and 1993 data sets scored on the first two principal components. Coded by carcass: 1, control:2.

Figure 14. The 72 pooled samples of the 1978 and 1993 data sets scored on the first two principal components. Coded by site number.

 

 

---

Introduction | Materials and Methods | Results and Discussion | Conclusions & Suggestions for Future Research | References Cited | Appendices