Elwood S. McCluskey and Siu-Ming A. Soong.
Rhythm Variables as Taxonomic Characters in Ants.
Psyche 86:91-102, 1979.

Full text (searchable PDF, 788K)
Durable link: http://psyche.entclub.org/86/86-091.html

The following unprocessed text is extracted from the PDF file, and is likely to be both incomplete and full of errors. Please consult the PDF file for the complete article.

RHYTHM VARIABLES AS
TAXONOMIC CHARACTERS IN ANTS
BY ELWOOD S. MCCLUSKEY' AND SIU-MING A. SOONG* Lorna Linda University, Loma Linda, Ca 92350 A survey of many species of ants in the field indicates a generic difference in the hours of flight and of worker aboveground activity (McCluskey 1973, 1974). Further, species within the same genus may differ consistently (Holldobler 1976; Lkvieux 1977; Whitford and Ettershank 1975). Would the same be true if the habitats of the various species were made identical? The present paper reports a beginning laboratory test of this question in worker ants. Secondly, it uses multivariate analysis to examine rhythm and other laboratory-type variables for their potential as taxonomic charac- ters. A preliminary report was given by McCluskey and Soong (1978).
METHODS AND MATERIALS
The species observed were: Pogonomyrmex californicus (Buck- ley), Pogonomyrmex rugosus Emery, Veromessor andrei (Mayr), and Veromessorpergandei (Mayr) in the tribe Myrmicini of Myrmi- cinae; and Formica pilicornis Emery and Myrrneco~stus mimicus Wheeler in the tribe Formicini of Formicinae. The ants were all collected May 24 and 25, 1973, near Loma Linda or Colton, San Bernardino County, California, at an elevation of about 350 m. Each group of ants was placed in a clear plastic refrigerator dish 9 cm in diameter by 3 cm deep. Sugar water was provided by a cotton- plugged tube hanging down through the lid, and constant moisture by a wick of dental cotton projecting up from water in a like dish nested underneath. A clear plastic (Tygon) tube 8 cm long led to a dish of the same type to serve as a dry arena. The nest dish was totally covered by heavy black paper, while the tube and arena were fully exposed to the light.
'Departments of Biology and Physiology; address for reprints. Department of Biology; present address, 550 Victoria Road, Baguio Villa, Block 46, 3F. Hong Kong.
Manu,tuipt received h.1' the editor July 5, 1979.



================================================================================

92 Psyche [March
There were five of these nest assemblies per species (except only one for M. mimicus), with 30 ants in each. The 26 arenas were lined up side by side, the first replicate of each species, then the second of each, etc. Heating tape was stretched alongside the whole row of arenas so that it was about 1 cm from one side of each arena; it provided constant heat throughout the experiment. The ants were kept in a constant temperature room. The nest dishes averaged 24.3' Ck0.5O. The hot side of the arenas was kept at least 10' (+lo) hotter than this. Twelve hours of fluorescent light (2000 lux) was alternated with 12 hours of darkness (the last 2 nights there was dim incandescent light [I5 lux] to permit observations). The ants were installed May 25 and observations began May 27. Each hour during the light periods of May 27-31, and also during the 2 nights beginning May 30 and 31, the number in the hot half of the arena, in the warm half of the arena, and in the tube were counted visually. In addition, the total number active (moving) in tube and arena were counted.
Except where noted, an arc sine transformation was used, treating the count as a proportion of the 30 ants available (Dixon and Mas- sey 1969). Then the transformed counts for the 3 days May 27-29, when the ants were in the best condition, were averaged for each replicate. Finally the hourly observations were fitted to a 24-hour cosine-curve for determination of the best-fit peak (acrophase) and other rhythm parameters.
Variables evaluated
The fitting method used (Halberg et al. 1972) provides the acro- phase, the level (average), the amplitude, and the sinusoidality (probability of sine shape) of the fitted rhythm. Each of these four parameters (dependent variables) could serve as a potential charac- ter, and each was applied to every one of the following four types of assay (Table 1).
Listed first are the variables related to the total number of ants out (i.e., in tube and arena) at each hour (TOTAL OUT). This would represent choice of an area exposed to alternating light and dark, as well as drier and warmer than the dark humid next dish. The second type of assay was the number of ants in the hot half of the arena (TEMPERATURE). This represents the extreme distance out from the nest dish, and/or the preference or tolerance of high temperature.




================================================================================

19791 McCluskey & Soong - Rhythm Variables in Ants 93 The third, POSITION, was calculated as (.80 X total no. out - no. in tube)/(total no. out - no. in hot half of arena). It is an index of how far out into the warm (i.e., nearest) half of the arena one would have to go to encompass 80% of the ants which were out of the nest at the time (50% would not work, because too often the 50% line was down in the tube rather than out in the arena). This position assay may be illustrated by sample calculations: If all the ants were in the warm half of the arena, the 80% point would be 80% of the way out through the warm half. If 20% were in the warm half and 80% in the tube, the 80% point would be 0% of the way through the warm half. If 80% were in the warm half and 20% in the hot half, the 80% point would be 100% of the way out through the warm half. The fourth, ACTIVITY, was the proportion active of the total number out.
The fitting involved above should increase the information con- tent in any given variable. How important is this for taxonomic discrimination? To help answer the question, single "hourly" (i.e., nonfitted) variables were also tested (Table l), each being simply the value. for a particular hour.
Table 1. (see following pages)
Means of variables examined for taxonomic value, averaged over days (May 27-29 except where noted) and replicate nests for each species (abbreviated by initial of genus and species). TOTAL OUT and TEMPERATURE based on arc sine transfor- mation of count; POSITION based on raw count; ACTIVITY based on log of proportion active. Sinusoida1it.1- represented by log of F statistic, Acrophase by clock hour (decimal form: e.g., 15.8 = 3:48 PM; angular avg), and Level and Amplitude by number of ants (except for POSITION, where Level and Amplitude represent distance into arena-see text); these first 4 variables under each type of assay are all derived by least-squares cosine-fitting (Halberg et al., 1972). The next 6 variables under TOTAL OUT are: Shift of Acrophase (no. of h change in acrophase from May 27 to May 29); Rise (h at which count passed up through avg [Level] for the day); Shift uf Rise (no. of h change from May 27 to May 29); Above Average(n0. of h above avg for the day, May 30-31, when observed all 24 h); Light-on Difference (between count immediately after lights on and 2d count several min later, May 28-30); Fluctuation (no. of runs, where "run" = series of hourly values each having same sign of change from preceding value). Corresponding to latter is Fluctuation under POSITION. Finally, under each type are listed the hourly values, unprocessed except for initial transformation and averaging (ACTIVITY 0700 omitted because could not observe all replicates simultaneously at the moment lights came on). F based on 5,20 degrees of freedom (slightly less for some variables with missing observations); * = P<.05, ** = <.01, *** = <.001.




================================================================================

Psyche
[March
Variable
TOTAL OUT
Sinusoidal i ty
Level
Ampl i tude
Acrophase
Shift of acro.
Rise
Shift of rise
Above average
Light-on diff.
Fluctuation
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
POSITION
Sinusoi dal ity
Level
Ampl i tude
Acrophase
Fluctuation
0700
0800
0900
1000
11 00
1200
1300
1400
Species
P.c. Par. V.a. V.p. F.p. M.m.
---




================================================================================

19791 McCluskey & Soong - Rhythm Variables in Ants 95 1500
1600
1700
1800
1900
TEMPERATURE
Sinusoidal ity
Level
Amp1 i tude
Acrop hase
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
1800
1900
ACTIVITY
Sinusoidal ity
Level
Amp1 i tude
Acrophase
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
1800




================================================================================

96 Psyche [March
Several miscellaneous variables are also shown in Table 1, such as "shift of acrophase." They and the fitted variables will be referred to as "processed" variables to distinguish them from the "hourly" ones. Multivariate analysis of variance (MANOVA) and multidiscrimi- nant analysis were used for comparison of species by groups of variables. These methods apply even where variables are highly correlated, which many of those here certainly are. On the basis of known group membership, discriminant analysis emphasizes varia- bles that have a high ratio of among-group to within-group varia- tion. Thus the variables are weighted so as to discriminate optimally between the groups (here species). An introduction to these methods is given by Cooley and Lohnes (1971), and application to behavioral research, by Pimentel and Frey (1978).
Daytime observations
For both species of Pogonomyrmex the total number out rose in early midday and remained high through the afternoon (Figure 1). For the other species the number rose in the afternoon and con- tinued rising through the end of the light period. In V. pergandei and F. pilicornis few ants ever came out as far as the arena; the rhythmic count was made up largely of those in the tube leading to the arena (Figure 1). In V. andrei the proportion in the tube was relatively high at the first and last of the light period. The proportion reaching the hot side of the arena was the greatest in P. californicus.
For most of the analyses, only these daytime observations were used.
24-hour observations
After 3 days very dim lights were left on at night, permitting counts then. It was now 6 days after collection from the field, and some of the ants were dead or else came out into the tubes or arenas less frequently. Nevertheless the patterns for the "day" part of the cycle were similar to before. And now it could be seen that all six species were prominently out in the arenas during the night. That this nocturnal behavior might be explained by the constant high temperature is suggested by field observations. Tevis (1958) found that nocturnal foraging of V. pergandei is rare, but could occur if temperatures are higher than normal; Wheeler and Rissing



================================================================================

19791 McCluskey & Soong - Rhythm Variables in Ants 97 HOUR
P. cal ifornic
V. pergandei
RUB.....^^" . . . . . .
Figure 1.
Comparative time patterns during light part of light-dark cycles. Histo- grams based on raw counts averaged over the 3 days (May 27-29) and 5 replicate nests (1 nest for M. mimicus) for each species. Each bar represents total number of ants out at a given observation: bottom part, tube count; middle (shaded), warm half of arena; top, hot half of arena. Dotted line to right of bar represents number active of total number out (omitted for 0700).




================================================================================

Psyche [March
(1975) found it to be common in the summer. McCluskey (1963) reported foraging in V. andrei on warm nights. P. californicus some- times works at night (Cole 1932), and P. rugosus does so when warm (Whitford and Ettershank 1975).
Evaluation of variables as taxonomic characters By virtue of the replicates it was possible to test each variable for species difference by one way analysis of variance, or in the case of phase variables, by the circular distribution test of Watson and Williams (1956) (Table 1). The differences are significant for 42 of the 51 hourly and 18 of the 23 processed variables, suggesting value as species characters. The hourly variables for the last third of the day are more discriminating than those earlier. Thus most of the individual variables distinguish species at least in a general way. As a group would they distinguish between the members of each pair of species? Multidiscriminant analysis was used to answer this (somewhat as Brown and Shipp, 1977, did to evaluate wing morphometric characters).
M. mimicus, with only one replicate, was omitted. MANOVA indicates a significant (P < .01) difference among the other five species, whether based on processed or on hourly variables. There- fore it is appropriate to use multidiscriminant analysis for further examination of the differences. For the set of processed variables, Figure 2 shows the maximum discrimination possible in two- dimensional space. The replicates are well segregated into their respective species. The same was true if based on either of the two stratified subsets of 17 hourly variables tried (not shown). Further analysis is not attempted here because of small sample size. The method was checked by deliberately assigning half the repli- cates initially to incorrect species. The variables then failed to segre- gate the replicates well into the artificial species, or even to show a significant species difference by MANOVA. Later experiment
As laboratory projects in a comparative physiology course, the same species of Pogonom~~rmex and Veromessor were compared simultaneously as before, but the conditions were different. Most notably, the season was October instead of May; and the tempera- ture was cycled between 20' C at night and 30' (35' in hottest part of arena) at midday, instead of being held constant.



================================================================================

19791 McCluskey & Soong - Rhythm Variables in Ants 99 F
F FFF
CANON I CAL WAR IABLE I
Figure 2.
The first two canonical axes, based on all 17 processed variables for which every one of the 25 cases (5 species X 5 replicates) was represented, and for which (phase variables) distribution was across few enough hours to assume linearity. The first axis represents the greatest possible difference between species, and the second; the greatest remaining difference orthogonal to the first. Output from the BMDP7M program, run so as to bypass the stepwise feature (as recommended by Pimentel and Frey, 1978). A = V. andrei; C P. cal(fornicus; F = F. pilicornis; P =
V. pexanclei; R = P. rugosus.
The results again suggest taxonomic differences in rhythm. The cosine-fitted acrophases of number out differed (P < .01) among the four species.
The study conditions were highly artificial-colony fragments (workers only, and in small number), no soil in the nest, tempera- ture high all 24 hours, etc. The fact that the rhythms were neverthe- less striking indicates that at least some of the natural field conditions (e.g., temperature cycle) are not necessary for such behavior.
But the important thing was that the six species were now all under the same conditions. Evidently various species manifest dif- ferent patterns of rhythm even though in identical habitat; or to put it another way, it is not necessarily differences in field habitat that are responsible for the differences in rhythm. For a laboratory example of phase or pattern difference in another insect group, see Hardeland and Slang's (1973) comparison of 40 species of Droso- phila.




================================================================================

100 Psyche [March
Many location and activity variables were tested in the six species for their potential as taxonomic characters, including cosine-fitted rhythm parameters as well as simple hourly values (Table 1). For most of them the species difference was significant. By contrast with the fitted-rhythm and other processed variables, the hourly variables taken singly can indicate only the level (of activity or location) at a given hour. In the aggregate they become more meaningful. For example, scanning the hourly P values in Table 1 indicates one time of day to be more important than another in distinguishing species. Or more formally, a multivariate approach can compare in terms of the whole hourly pattern. MANOVA con- firmed the overall species difference, and multidiscriminant analysis made it possible to discriminate each species from the others. It should be noted again that the purpose here is to examine these variables as taxonomic characters, rather than to develop a classifi- cation.
Nearly every one of the processed variables was abstracted from all the hourly values of a given type of assay, and hence should have high information content. So it is noteworthy that the groups of hourly variables seemed to distinguish species about as well as the group of processed ones. However, if some of both kinds were included in the same analysis, the most highly weighted were processed variables. most highly weighted were processed variables. How might the list in Table 1 be compared with catalogues of ant behavior, such as that of Wilson (1976)? In the first place, the varia- bles reported here were population rather than individual phe- nomena, Secondly, many of them are derived rhythm parameters (e.g., acrophase or amplitude) of certain items of behavior, rather than the behavior itself. The hourly variables, on the other hand, do represent behavior more directly (e.g., the number active or the number in the hot location, at a given point in time). Any compre- hensiveness of catalogue (e.g., for choice of heat or light or location in the nest assembly) would require more habitat complexity, and preferably field studies as well. But such comprehensiveness is needed to facilitate behavioral comparisons at species or higher levels (Wilson 1976).
In summary, this study suggests a variety of rhythm-related char- acters that might be observed in laboratory or field, whether single hourly values, composite hour-by-hour patterns, or fitted rhythm parameters.




================================================================================

19791 McCluskey & Soong - Rhythm Variables in Ants 101 In the field the time of day of certain types of behavior distin- guishes various species and genera. Would the same be true if their habitats were made identical? And could rhythm-related variables in general be regarded as taxonomic characters? Groups of workers of Pogonomyrmex californicus, P. rugosus, Veromessor andrei, V. pergandei, Formica pilicornis, and Myrmecocystus mimicus were studied simultaneously in constant temperature and alternating light and dark. A hot dry area and a dark humid area were available in each nest for choice at all hours. This permitted different types of position or activity assays and the calculation of several cosine- fitted rhythm parameters for each. These as well as simple hourly values served as variables. For most of the variables tested there was a significant species difference. With multidiscriminant analysis it was possible to go beyond this general species difference and segre- gate the replicates of each species from those of other species; either the hourly or the more highly processed variables were adequate here. This study suggests a variety of rhythm-related characters that might be observed in laboratory or field. The data on which this report is based are largely from the second author's M.A. thesis research (Loma Linda University, 1975). We thank Andre Francoeur for determining F. pilicornis and Roy Snel- ling, M. mimicus; Glen McCluskey and Duane Zimmerman for programming; Paul Yahiku and Richard Pimentel for statistical help; and Leonard Brand, Conrad Clausen, David Kissinger, Gerald Scherba and Paul Yahiku for reading earlier drafts of the manus- cript. Computer time was supported in part by NIH, Grant RR- 00276. Also used were BMDP programs developed at UCLA with NIH support.
BROWN, K. R., AND E. SHIPP.
1977.
Wing morphometrics of Australian Luciliini (Diptera: Calliphoridae). Aust. J. Zool. 25765-777.
COLE, A. C.
1932.
Notes on the ant Pogonomyrmex californicus, Buckley (Hym.: Formici- dae). Entomol. News 43: 1 13-1 15.




================================================================================

102 Psyche
[March
COOLEY, W. W., AND P. R. LOHNES.
1971. Multivariate data analysis. Wiley, New York. DIXON, W. J., AND F. J. MASSEY, JR.
1969.
Introduction to statistical analysis, 3d ed. McGraw-Hill, New York. HALBERG, F., E. A. JOHNSON, W. NELSON, W. RUNGE, AND R. SOTHERN. 1972.
Autorhythometry-procedures for physiologic self-measurements and their analysis. Physiology Teacher l(4): 1 - 1 1. HARDELAND, R., AND G. STANG.
1973.
Comparative studies on the circadian rhythms of locomotor activity of 40 Drosophila species. J. Interdisc. Cycle Research 4:353-359. HOLLDOBLER, B.
1976.
The behavioral ecology of mating in harvester ants (Hymenoptera: For- micidae: Pogonomyrmex). Behavioral Ecology and Sociobiology 1:405- 423.
LEVIEUX, J.
1977. La nutrition des fourmis tropicales: V.-Elements de synthkse. Les modes d'exploitation de la biocoenose. Insectes Soc. 24:235-260. MCCLUSKEY, E. S.
1963.
Rhythms and clocks in harvester and Argentine ants. Physiol. Zool.