Endogenous Rhythmicity of Vertical Migration in a Meiofaunal Population of the Intertidal Collembolan Archisotoma besselsi (Arthropoda: Insecta) Ellen I. Fritts Hopkins Marine Station Stanford University June 10, 1978 Introduction Collembolids comprise an insect order whose members range widely throughout the world, being found in almost every habitat where there is moisture together with decaying vegetation (Bacon, 1941; Chang, 1966). Because of their wide distribution, collembolids have been discussed frequently in the literature. Due to economic factors, large populations inhabiting agricultural soils have received the most attention; far less has been written regarding species found in other habitats. Inter- tidal species, for example, are only occasionally described in the lite¬ rature despite the fact that they also exhibit extremely high population densities, with estimates of as many as 20,000/ m (Davenport, 1903). The meiofaunal component of intertidal collembolid populations has rarely even been mentioned (Cox, 1976; Davenport, 1903), and there is no work on temporal variation in their distribution in the sand. A dis- crepancy in observed activity patterns of surface and meiofaunal elements of a population of Archisotoma besselsi found on a pocket beach at Mussel Point, California, prompted the experiments presented herein. Twenty-four hour field studies conducted on this beach showed high amounts of surface jumping activity concentrated at dawn regardless of tidal conditions; in contrast, core samples taken periodically on the same beach showed evidence of vertical migration related to tidal pheno- mena. Observations through aquarium glass showed rates of migratory move- ment approaching 3 cm/min. Field and laboratory studies revealed a complex pattern of endogenous vertical migration activity having both tidal and diel components. These results are of particular interest when considered in conjunction with work done concurrently on the rhyth¬ mic jumping activity of the adjacent surface population by fellow workers at Hopkins Marine Station (DeLapa, 1978; MoSpadden, 1978). Investigations of temporal variations were complicated by exten¬ sive patchiness in spatial distribution of the population; field studies showed the population to be concentrated along one cóntour representing + tidal feet. In addition, both field and laboratory experiments revealed significant heterogeneity between replicate cores with regard to total numbers of animals present. Materials and Methods I used the coring apparatus described by Cox (1976); fourteen-inch lengths of 3.5-cm diameter schedule-40 PVC pipe were bevelled on one end and marked around their circumference 12" from the lower edge. Treatment with silicon spray lubricant facilitated pushing out samples for analysis; collembolids were never observed sticking to lubricated cylinder walls. In the field, cores were pounded twelve inches into the sand, plugged with a 7 solid rubber stopper, withdrawn, and sealed on the bottom with another stopper to prevent loss of animals and water, and to prevent reception of light cues by the organisms. A twelve-inch depth was chosen and used uniformly throughout the experiments; my ob¬ servations showed the highest concentrations of specimens within this column despite occurrence of collembolids to atleast 18" below the surface. All cores for a particular experiment were moved simultaneously into a constant 18 C room. Here, the sealed cores were separated randomly into batches of n cores and each batch placed upright in a separate container. Originally, n was three but was changed later to 5 in an attempt to minimize variability due to patchiness. At each sampling period (corresponding closely with the external high tide, lowitide, and an intermediate point for most studies), a container of n cores was re- moved from the constant temperature room for immediate processing. To process each core, the bottom stopper was removed and the base of the core set into a sample jar. I would then remove the top stopper and inspect the sand surface within the cylinder to observe any animals active there. I would push out three-inch lengths of the sand column into each of four sample jars, the lowest layer being pushed out first. Jars were rapidly sealed to prevent loss of collembolids. The end of the plunger was rinsed between each usage to prevent contamination from one core to the next. Animals were extracted from each jar by the following technique: To each jar, I added approximately 30 mls of seawater and roughly 5-10 mls of ethanol (95%) to kill the collembolids present. Each jar was swirled gently to dislodge specimens and then decanted into a 150 ml beaker. Washing with seawater alone was repeated twice more and the liquid de- canted into the same beaker. Under a dissecting microscope, the inactive and floating collembola were easily counted. I recorded the number of animals extracted from each layer, then computed the percentage of the core population in each 3" layer. Field studies and collections for laboratory experiments were con¬ ducted on a pocket beach at Mussel Point in Monterey Bay. This beach faces west and is composed of well-sorted granitic sand, fine to medium in grain size. Use of graded Tyler sieves showed representative sand samples to be characterized by the following regime of size classes: #10-10%, #12-11%, #20-14, +35-26%, and 60 or smaller-9. In sev- eral experiments requiring sand devoid of organic matter, commercial sand of grades 20 and 35 was mixed in a ratio of 2:1 to approximate natural conditions. A preliminary survey of the distribution of organisms from lower low water up to the permanent vegetation (at +11 feet above MLLW) revealed two things. First, the meiofaunal population was highly localized along a contour corresponding to +4 féet' tidal height. This area,: later referred to as"station 18', consisted of points located 18 meters seaward of the vegatation/beach interface. Underwater coring seåward of the swash zone showed sustained high concentrations of animals in the sand at station 18, despite a rising tide. This phenomenon is not mentioned in the literature on intertidal collembolid species. Collec- tion of cores for subsequent laboratory studies of endogenous behavior were always made at this station. The second major finding résulting from preliminary field studies was that of marked differences in micro-scale distribution. Replicate cores taken simultaneously at any one level of the beach invariably showed a significant degree of heterogeneity (Rx C tests, Sokal and Rohlf (1969); G values of 39.6 and 173.22 in two determinations, n-5 each time). Laboratory experiments showed this patchiness phenomenon to persist even when animals were presented with a homogeneous environ- ment of clean commercial sand kept under constant conditions. A five¬ gallon translucent white plastic container was painted black, then filled with moist sand (8% by weight). A similar but unpainted container was inverted and sealed to the first. Members of the experimental popula- tion jumped off of a piece of filter paper which was introduced and later removed; this method avoided bias due to localized addition of dead or- ganisms. Diffuse dim lighting and constant temperatures were maintained for twenty-four hours prior to sampling. Studies of porosity explored this environmental factor as a possible cause for patchy distribution of the field population; in order to ana- lyze rates of water-flow through a column, sand from four sibling samples was recombined, dried, and mixed thoroughly to simulate as closely as possible porosity properties of the original core. Sand was poured to a depth of 40 cm in lengths of 1.7 cm-diameter glass tubing to which screen ad been affixed at the bottom. Columns prepared in this manner were then suspended from a ring stand and wetted. Replicate rates of flow were recorded for each column and the means of the results compared by t-tests. Effects of an altered diel cycle on endogenous vertical migration were tested in one experiment conducted May 24-26. Cores receiving a 3-hour delayed day/night cycle were capped with inverted translucent disposable 150-ml tri-lipped beakers. Two cans containing five cores each were placed in a cardboard box and subjected to illumination of 300E during 14 hours of imposed daylight; the cores were left undis- turbed for 49 hours prior to the first of two sampling periods. In a similar experiment employing a 3-hour delayed tidal cycle, "high" tide consisted of letting 11 C seawater stand within the PVC pipe to a depth of one inch above the sand for approximately 64 hours. "Low tide conditions were imposed by draining the water out the bottoms of the tubes through phytoplankton screening placed over a hole in the bot- tom stopper. Water was reinstated 64 hours later (at the next "high" tide) by affixing a pinch clamp to flexible tubing emerging from the stopper hole. Due to problems inherent in patchiness of the population under study, statistical analyses were extremely important. In conjunction with t-tests, linear regressions were used to adjust for a trend toward smaller core populations with increasing time between time of collection and processing; while this phenomenon suggested escape of collembolids from cores kept in the laboratory, spot checks found no evidence of this. With single classification analyses of variance (used with f-tests), I used log transformations of percentages in order to normalize the data. Cox (1976) has shoun this mothod to be acceptable in studies of another intertidal collembolid Anurida maritima. Significance of changes seen c in distribution within columns as a function of time or as a function of variations between cores was checked using Rx C contingency tests of the actual counts found. Results & Discussion All results indicate components of heterogeneity quite separate from, and in addition to, temporal variations. Coring showed both heavy concentrations of animals along a contour 18 meters from the perma¬ nent vegetation and high variability between replicate cores. In combi¬ nation with this, core populations showed variable distribution within the sand column in apparent response to a complex endogenous rhyth. While patchiness in collembolid populations has been mentioned (Butcher. Snider, and Snider, 1971; Chang, 1966; Cox, 1976; Joose, 1966), the extent and intricacy of this puzzling phenomenon has not been documented. Observations of high concentrations of animals at station 18 were frequently coupled with discovery of a dense one- to two-inch layer of decaying vegetation in the 9"-12" stratum. Since decaying organic matter is known to be a primary food source for many species of collembolids (Bacon, 1941; Britt, 1951; Fenton, 1947), attraction to food seems likely as a cause for patchiness of thesé organisms at Mussel Point. Cccasionally in samples taken from station 18, up to forty or fifty per- cent of the animals found in the deepest layer are less than two-thirds the size of the remainder present. Of these smaller animals, some ex- hibit the bluish black coloration of adults (Maynard, 1957) while the smallest organisms display white coloration characteristic of juveniles (Essig, 1958). Frequently, large concentrations of juveniles occur in samples from the lowest layer which simultaneously show high amounts of organic matter; this suggests that reproduction may occur in decaying wrack buried here. Observations of another intertidal collembolid Anurida maritima show that it carries out most of its reproductive cycle in pockets of buried decaying vegetation (Joose, 1966). An explanation for the apparent increase in numbers of smaller animals with depth may be due to microhabitat structure changes inherent in greater depth; specifically, diminishing pore space may limit the depths reached by collembola of different sizes just as it does for the meiofuana in general (Christiansen, 1964). Laboratory experiments testing possible effects of substrait struc¬ ture on patchiness were revealing but did not show positive correlation between the two. Studies of porosity versus number of animals extracted showed no significant difference between cores which had shown very high populations and those which had not. In experiments ranging in length from one-half to three hours, animals inhabiting 3" layers of sand col- lected in the field showed no migration into clean commercial sand placed adjacent in the cylinders to the natural sand. Simultaneous field and laboratory experiments (n-7 for each) showed that, like those in the field, animals exposed for twenty-four hours to a homogeneous environ- ment of commercial sand also exhibited significant patchiness (indices of dispersion: X =73.8 and X =22.8 for lab and field studies, respec- tively; p2.001 in both cases). The implication is strong that biolo- gical interactions are in part responsible for the patchiness displayed by both'the laboratory and field populations in this experiment. Studies of microhabitat moisture differences supported previous work showing differential distribution of collembolids in response to this environmental parameter (Bacon, 1921; Butcher, Snider, and Snider, 1971; Christiansen, 1964; Ford, 1937; Joose, 1966). In a comparative experiment, station 15 displayed uniforaly and significantly higher per- centages of animals in the lowest layer than was found from station 18 (px.O1, t-test for percentage); animals higher on the beach might be 10 found deeper in the sand due to more favorable moisture conditions nearer the water table (Hedgpeth, 1957). Another expariment employed strong artificial lights 14.5 hours per day for two days prior to sam- pling; dåtå from two sampling periods 34 hours apart (n-5 each time) revealed over sixty percent of the animals to be located in the lower two layers at each sampling. The sand in these cores had dried out (probably due to a slight elevation in temperatures caused whenever'the artificial lights were activated); an apparent downward migration in response to dessication and elevated temperatures was indicated. Findings of extreme patchiness in the beach population meant that patterns of temporal variation are superimposed on and perhaps often obscured by the variability in macro- and micro-distribution. Standar- dizing temporal parameters inherent in the sampling process was critical for observation and characterization of endogenous temporal variation. Cores which were processed immediately after removal from the substratum gave percentage values which did not differ significantly from those ob- tained when samples were sealed and placed under constant laboratory conditions for thirty minutes prior to processing; see Table I for data. An experimental sampling frequency of approximately 34 hours was used for most of my experiments; sampling more frequently did not greatly increase resolution of trends in periodic behavior. Figure 1B shows the results of such a study. The lines shown are drawn through data corre- sponding to those collected on May 4 & 5 (Figure 14). The curves, when drawn through data points separated by the same 34-hour interval, show striking similarities; the other points, while adding scatter, nonethe- less support the illustrated trends in endogenous vertical migration activity. Note that the tidal cycle occurs at the same time of night 11 in each case. For May 4th and 5th, elevation of values for percentage in the bottom layer together with reduced values in the top layer is probably weather-related; high surf conditions prevailed when samples were collected for the May 4 study whereas conditions were much calmer prior to the May 20 experiment. Complete results of the May 4 & 5 study are shown in Figure 2. The trends seen suggested a predictive model of temporal movement in the meiofaunal collembolid population: an increase in the percentage of animals in the top layer is seen at low tides while a decrease is ob- served for high tide; a similar but inverse relationship to tides is seen for the lowest layer. In this graph, the middle layers show an inverse relationship to each other, and the movements in these layers do not correspond well to known exogenous cycles. In all studies, mi¬ gration activity shown by these middle layers displayed much higher var- iability than that shown by either the top (0-3") or the bottom (9-12") layers. Therefore, in subsequent figures only data from the top and bottom strata are illustrated and discussed. Despite the fact that data from most 24-hour experiments appeared by inspection to fit the proposed hypothesis of migration patterns : correlated to tidal phenomena, statistical tests sometimes cast doubts. For example, performing a linear regression followed by pair-wise t-tests on data collected May 4 & 5 shows differences between mean high tide and low tide values within each layer to be statistically insigni- ficant. Similarly, when data from this study is combined with that from four others conducted between May 9 and May 25, log-transformed means compared by t-test again fail to show significance between all the high and low tide values obtained; this is probably due to higher variability shown in patterns of migration during daylight hours. 12 When percentages from the bottom layer in each of five studies are analyzed with respect to high tides occurring during the night (See Figure 13), statistical significance is readily demonstrated. Here, pair-wise t-tests of log-transformed means show expectations of random occurrence of p £.02 (f-16) between valuas for the preceding low tide versus the nocturnal high tide; similarly, expectations of random occur- rence show p 2.03 (f-18) for the high versus the subsequent low tide, whether that low occurs in darkness or during the first hours of daylight. while statistical analyses verify the probability of a tidal component in endogenous patterns of migration, elements of a diel component may also be present. The similarity in shape of Figure 3-F to Figures 3 .-A,B,C,D,& E suggests a rise in percentage of animals occupying the lowest layer after the onset of darkness, independent of high tides. Additionally, all cases except 3 -D and 3 -E exhibit a decline in values of the lowest layer shortly before dawn. The 5/24 and 5/25 studies (Figures :3'-D & 3 -E) were part of a 53-hour investigation in which I was sampling only every six hours. Hence, I was not sampling at a frequency which would have revealed low pre-dawn values had they occurred. Thus, Figure 3 shows that darkness appears to have a pre- dictable and additive effect on the tidal component of an endogenous vertical migration pattern. Laboratory studies undertaken to separate these components consisted of imposing three-hour delayed light and tidal cycles to separate batches of cores maintained independently under constant temperature conditions. Two batches of five cores weré subjected to altered tidal conditions for 494 hours before sampling; results showed that this regimen was insuffi- cient to produce migration behavior different from that expected in an 13. undisturbed field population. Cbservations of ten cores subjected to altered light/dark conditions for 493 hours prior to sampling proved in- conclusive; unlike normal samples, these cores showed no apparent change in the percentage of animals found per layer from the first to the second (and last) readings. Core dessication may have caused death. Figure 4 presents results of a 24-hour study conducted May 16 & 17 which yielded statistically significant but anomalous behavior in the surface layer. Single-classification analyses of variance (Sokal and Rohlf, 1969) reveal that there is a significant change in the means ob- served for the top layer through time (pX.05). Inspection of the graph affirms this; not only are peaks in activity of the top layer seen at low tides, but similarly dramatic peaks appear at high tides. Note that peaks seen at low tides are significantly higher than peaks at high tide; pair-wise t-tests for percentage showed the difference be- tween the means of the high tide and low tide peaks to be expected as random with p .01. This study shows rhythmic migration behavior which is exceptionally well correlated with tidal phenomena; however, peaks in surface layer activity at high tide contradict trends expected by the predictive model. The réasons for this anomalous bimodal pattern are unknown. Such a phenomenon was not seen in any other experiment; a study conducted three days later (on May 20) showed a return to previously seen activity patterns (See Figure 14). Figure 5 shows the results of an experiment conducted May 16, when high tide corresponded almost exactly with dawn. Note that this study was actually conducted the day before that shown in Fizure 4; as in the latter, surface activity increases rather than decreases for condi- tions of high tide. Anomalous trends in activity of the top layer in these temporally contiguous experiments might suggest that simultaneous occurrence of high tide and dawn temporarily destabilizes expected rhythms of endogenous activity. Interestingly, MeSpadden (1978) observed that when high tide and dawn corresponded during a series of free-run experiments on the periodicity of jumping activity in A. besselsi, a radical phase shift was associated with the day on which "crossover" of the diel and tidal cycles occurred. 15 Summary In a meiofaunal population of the intertidal collembolid Archisotoma besselsi found on a pocket beach at Mussel Point. California: (a) An endogenous rhyth of vertical migration was de- monstrated and a predictive model of location in the sand- versus tidal condition was developed. (b) Tidal patterns with diel influences were implicated as major components of this endogenous rhyth. (c) Gross patchiness was exhibited by the population and was attributed in part to characteristics of the sand column. (d) The majority of the population was localized at approximately + feet above MLLW and was found to remain at varying depths within the sand even when completely inundated by the tides. (e) Exposure to approximately forty-nine hours of a 3-hour delayed tidal regime was insufficient to change expected trends in the migration pattern. (f) Suggestion was made that simultaneous occurrence of dawn and high tide temporarily destabilizes expected activity rhythms. 16 Acknowledgement I'd like to direct my thanks in three directions. First, to the faculty at Hopkins Marine Station for being there en mass in a cold, driving 4 a.m. rain when there were no 'bolids; their dedication and enthusiasm were an inspiration throughout the quarter. Secondly, I'd like to give Robin Burnett heartfelt thanks for all his help with: stats and direction; I'll never wonder about carrots and sticks again. And to Michelle, who livened up those long early morning hours in the lab, a big grin. Literature Cited Bacon, G. A. 1941. The distribution of Collembola in the Claremont-Laguna region of California. J. Ent. Zool. 7:137-179. Britt, W. W. 1951. Observations on the life history of the collemolen Achorutes armatus. Trans. Am. Microscop. Soc. 70; 119-32. Butcher, J. W., R. Snider, & R. J. Snider. 1971. Bioecology of edaphic Collembola and Acarina. Annual Review of Entomology 16: 249-283. Chang, S. L. 1966. Some physiological observations on two aquatic Collembola. Trans. Am. Microsc. Soc. 85(3): 359-71. Christiansen, K. 1954. Bionomics of Collembola. Annual Review of Entomology 9: 147-73. Cox, J. L. 1976. Sampling variation in sandy beach littoral and nearshore meiofauna and macrofauna. U. S. Army Corps of Engineers, Tech. paper F76-14. Davenport, C. B. 1903. The Collembola of Cold Spring Beach with special reference to the movements of the Podu- ridae. Cold Spring Harbor Monograph II. Delapa, M. 1978. Jurping mhythmicity n acollembolid population on a Central California beach ≈ the effect of tem- perature on jumping activity. Essig, E. 1958. Insects & mites of western North America. New York: The MacMillan Co. Fenton, G. 1947. The soil fauna. J. Animal Ecol. 16:76-93. Ford, J. 1937. Fluctuations in the natural populations'of Collembola and Acarina. J. Animal Ecol. 6: 98-114. Joel W. 1957. Sandy beaches. p. 587-608. In Hedg- Hedgpeth, peth, J. W. (Ed.) Treatise on marine ecology and paleoecology. Vol. 1. Geol. Soc. Am. Mem. 67. Joose, E. N. G. 1966. Some observations on the biology of Anurida maritima. 2. Morphol. Cekol. Tiere 57: 320- Maynard, Elliott A. 1951. A monograph of the Collembola of springtail insects of New York State. New York: Comstock Publishing Co. 18 Literature Cited (continued) MoSpadden, Michelle M. 1978. An analysis of the endogenous rhythms in jumping activity in an intertidal col- lembolid population on the Central California coast. Sokal, R. R. and F. James Rohlf. 1969. Biometry. San Fran- cisco: W. H. Freeman and Co. Table Legend Table I - A comparison of data gathered using immediate versus 30-minute delayed processing of cores. N and X values shown are percentages. TABLE I Station 18 Analyzed Imediately 30-minute Delay Layer 2 X N 1 X Ni N. 9-3" 54.7 56.0 55.4 24.5 51.9 38.2 3-6 40.0 31.2 35.6 36.0 27.9 31.9 6-9" 39.6 21.2 30.4 5.3 12.8 9.1 Station 19 Analyzed Immediatelv 30-Minute Delay Layer 0-3" 28.1 28.2 3-6 36.6 36.7 6-9" 35.2 35.3 — Statistics P( 1.28 .25 56 .62 2.15 08 TABLE I Analyzed Immediately 30-minute Delay Station 18 Layer N N. 9-3" 54.7 56.0 55.4 24.5 51.9 38.2 3-6 40.0 31.2 35.6 36.0 27.9 31.9 6-91 5.3 12.8 9.1 30.4 39.6 21.2 Station 19 Analyzed Immediately 30-Minute Delay Layer 0-3" 28.1 28.2 3-6" 36.6 36.7 6-9" 35.2 35.3 Statistics PC — 1.28 56 2.15 08 Figure Legend Figure 1 - Replicate studies showing recurrence of endogenous patterns in the top and bottom layer when tidal and diel conditions repeat themselves. Figure 1A- Percentages found in the 0-3" and 9-12" layers on May 4 & 5. Slender vertical bars indicate dawn or dusk; arrows indicate high or low tide. Figure 1B- Percentage found in the top and bottom layers on May 20 & 21. Lines shown connect data points taken at the same frequency as in Figure 1A. Figure 2 - Percentage of core population in each layer through the course of 24 hours. Shows variable trends in behavior of second and third layer; these will be deleted in sub- sequent graphs. Heavy vertical bar indicates power out- age; data to the right not analyzed. Figure 3 - A comparison of endogenous trends in the lowest layer during night hours for six studies conducted between May 4 and May 25. Note the occurrence of nocturnal high tides in all but F. No high tide data point for B. Figure 4 - An experiment showing anomalous activity at both high and low tides for the 0-3" layer. Note statistically significant difference in height of high tide and low tide peaks. Figure 5 - Percentage per layer found when dawn and high tide coincide (May 16). Unexpected increase in percentage found in top layer when high tide is occurring. Fiq May 485 Fig 14 50 4 -O O 2 18 6 S a + + May 20 Fiq. 1B 8 21 30 O- — — + —— M E + 1 A A 10 M Ahde —: — Key toayerst 2 -—00 0 19 - Adual idata pt-6-2 A + Adual dato et-412 9 — ++ 4 +8am 6 X V R + 4 8am Low Tide . --. +1. ——: — — I 8 0 619626 Nn +6 E . 1. + 4 — 5 od. X —. —-—+-- . .-- ... .. . C 2 O I e- - - C —— — R — . 1. 1: O 4 ——— / C —E Fiq. 3 —— — 55 A. 5/4 45 35 5 — 45 ——— B. 5/9 35 V 35- C. 5/20 25 1 15- 5 L 40 5/24 O 30 20 + T 45- — 35- 5- —— — 15- —. 5- V —.— — 20 F. 5716 10- 4 O- A 4 1. A A 9 M digh Tde. —— Lou tide — . â —: — — V Time (O percent O OO S ) O D O D 2 Co — — — — 1 11 ——— —— .. + L 0 0 â â â ââ —1 S 1 .— — 00 ——— ..... — ... 6. 8 — —— — 1 2 .. 2 —.— X .. —— ... . S — .. X —1 4 ... ... .. ... — — : (- : — 0 O: 7. — 9 O.O 0 t —— — DC — 77 . —--.- — L — — s S T — 3 PER LAYER II 10 HDEE &am DA HIGH COIAC 1 ..... — — 16 . . . .... ... : — : + LER-KE +— E — — + —— — SUNRISE — — 90 — — 49am. —— — — 1 + —- — O + — O — — 6 3 0 2 3 4 ) Bam 4 ..:. ..... . HIGH TIDE —: V 5:57a.. W ... —1 —------ ... . — — — .