(Foto dari Google)
ENTOMOLOGY
DEVELOPMENT OF TIME-LAPSE CINEMATOGRAPHIC
TECHNIQUE
Nur Tjahjadi
Abstract
Thirty mealworm pupae (Tenebrio mollitor) were placed in the soil in front of one the rhizotron windows. This was done at three different depths (10 at the soil surface, 10 at 3 cm and 10 at 6 cm), so that they could be seen throught the rhizotron window. An important parameter in this study was the legth of time interval between photographs. Too many frames per hour is costly in film and analysis time. Too few frames means that important events are missed. The result from 4.000 exposed frames shot at the rate of one frame every 4 minutes, were first analysed one frame at a time. Subsequently, only the results from every second fifth, tenth, fiftieth and hundredth frames were used, to simulate the results of more economical time lapse filming regimes. Minitab macros were created to automate the procedures and macro statements stored in a file which could be executed. Macros enabled data to be looked at every second, fifth, tenth, fiftieth and hundredth frame using the commands.The cost of an observation consists of the time taken for analysis and the The rhizotron and time-lapse cinematography can only be effectively used for studying predator-prey interactions if baits are provided. The predators are highly mobile and the area of window which the camera can adequately observe is limited to 15 cm x 10 cm. In any study where the prey is also mobile, particulary if reliance is placed on naturally occuring prey alone, the number of useful observations will be extremely small. cost of the film used. In order to estimate the total cost of predator observation in the soil using various time lapse intervals, the total cost of analysis was calculated by taking 11 frames from a single film as samples.The benefits, in contrast, are the amount of detailed information about predation which can be obtained. In this case, it was decided to see how many significant biological effects could be detected with the various numbers of time lapse pictures used. Optimising the length of the time interval between photographs, involves achieving the lowest possible cost per significant effect detected.The results showed that the rhizotron and time-lapse cinematography can only be effectively used for studying predator-prey interactions if baits are provided. The predators are highly mobile and the area of window which the camera can adequately observe is limited to 15 cm x 10 cm.
INTRODUCTION
Time-lapse cinematography has been used to monitor the growth of plants and to study the behaviour of animals. Smith et al. (1993) monitored mobile megafaunal activity in the deep-sea with a time-lapse camera using a one hour time interval between frames, Perbal & Drissecole (1994) monitored the growth of plant roots using one second and above.
Time-lapse photography allows events occurring over a long period in nature to be examined over a short period when viewing the film. For continuous, slowly changing situations such as plant growth, speeding them up allows the human viewer to see the changing shapes and relationships in a more comprehensible form. In the present case however, some of the activities observed can occur very quickly, although invertebrates acting within the confines of the soil structure act relatively slowly compared with vertebrates feeding on the surface where prey can be seized very rapidly and moved elsewhere. In this situation, time-lapse is being used as a time sampling device, rather than as away of speeding up slow changes to make them more understandable.
We are concerned with picking out relatively rare events from long periods of inactivity, so that the 4 minutes it takes to view a single film represents a storing of events which actually took place over 10 days. Changes of interest such as the disappearance of a pupa ar damage to it, can then be timed and examined. The problem here is to obtain an optimal time interval. Short time intervals involve much viewing and little action, long time intervals give plenty of evidence of action having occurred, but the key event is incresingly likely to have taken place between frames which have missed capturing the participants on camera.
To determine the most suitable interval between frames for time-lapse recording of predation, it was decided to film with a short time interval and to simulate the effects of longer time-lapse intervals by omitting frames in subsequent analyses. The choice of intervals ultimately depends upon which events are of most interest. The interactions between some predators and their prey may be adequately recorded with a 400 minute time interval, but for other predators, a 4 minute time interval may be insufficient. It was dicided to use standardised types and amounts of prey placed in front of the windows, rather than rely on the relatively rare occurrence in the field of natural prey.
MATERIALS AND METHODS
Placing the pupae at different depths
Thirty mealworm pupae (Tenebrio mollitor) were placed in the soil in front of one the rhizotron windows. This was done at three different depths (10 at the soil surface, 10 at 3 cm and 10 at 6 cm), so that they could be seen throught the rhizotron window. To do this, a 15 cm x 6 cm x 2 cm block of soil was removed from in the front of the glass of the rhizotron, the glass was cleaned with tissue paper and 10 pupae were put into the bottom of the pit. A 3 cm layer of soil was then replaced and again 10 pupae were set out. A futher 3 cm layer of soil was replaced and the final 10 pupae were placed on the soil surface. A time-lapse camera was set up to record predator-prey interactions.
Time-lapse cinematography
An important parameter in this study was the legth of time interval between photographs. Too many frames per hour is costly in film and analysis time. Too few frames means that important events are missed. The result from 4.000 exposed frames shot at the rate of one frame every 4 minutes, were first analysed one frame at a time. Subsequently, only the results from every second fifth, tenth, fiftieth and hundredth frames were used, to simulate the results of more economical time lapse filming regimes.
Data Selection and Storage
The number of each frame (1 - 4.000) was entered in C1 (column one) of the minitab spread sheet. The number of each pupa (1 - 30) was entered in C2 (column two) and the various events which befell each pupa were recorded in C3 (column three). Codes were used for each type of activity as outlined in Table 3.1. The minitab routine allows the production of data identified by code with varying time-lapse intervals and the producers, illustrated by the carabid data, are shown in appendix 1a.
Minitab macros were created to automate the procedures and macro statements stored in a file which could be executed. Macros enabled data to be looked at every second, fifth, tenth, fiftieth and hundredth frame using the commands in Appendix 1b.
Technique used for estimating the costs and benefits of different numbers of observations
The cost of an observation consists of the time taken for analysis and the cost of the film used. In order to estimate the total cost of predator observation in the soil using various time lapse intervals, the total cost of analysis was calculated by taking 11 frames from a single film as samples. Frame numbers 1, 100, 200, 300, 400, 500, 600, 700, 800, 900 and 1.000 were chosen to be reanalysed. To look at the eleven frames, identity the predators and note down all the events required 9 minutes. To enter the data into minitab needed 23 minutes and to analyse the data required 8 minutes (a total of 40 minutes for 11 frames). Each frame therefore required 3.64 minutes and the breakdown of costs is shown in Table 2.
The benefits, in contrast, are the amount of detailed information about predation which can be obtained. In this case, it was decided to see how many significant biological effects could be detected with the various numbers of time lapse pictures used. Optimising the length of the time interval between photographs, involves achieving the lowest possible cost per significant effect detected.
Definition of predator contacts with pupae
The definition of a predator contact in this study is when a predator or any part of its body, either touches the pupa, or is within 0.5 mm of it. Predator contacts are important in predator – prey interactions, because they may lead to prey death and consumption.
Determining the predator responsible for each prey death
Each type of predator has different ways of attacking and consuming the pupae. Carabids eat the whole pupae, its integument as well as its body content. In the laboratory, where the search arena was limited, carabids normally killed and ate a whole pupae in less than half an hour. Sometimes they took away the pupa to eat it elsewhere. In the field, where the search arena is much greater, carabids sometimes made contacts with pupae, but did not eat or kill them. This information is important for considering how long a time lapse interval should be employed, to avoid missing crucial events such as a picture of the predator responsible for killing and eating the pupa.
Unlike carabids, staphylinids, centipedes and campodeans feed on the contents of the pupa without completely destroying it. Centipedes and campodeans need several days to complete their feeding on the pupa and it is not difficult to record their feeding activity in the rhizotron. Staphylinids need a few hours to feed on pupae in the laboratory. When they find the pupa, staphylinids attack and kill it and then feed continously for 2-3 hours. However in the field, staphylinids are not observed feeding for these long periods.
Using time lapse photography, it is extremely difficultto determine which predator is responsible for the death of a given prey. The animal that was first seen consuming any part of the pupa (integument or body content) was assumed to be responsible for thet prey’s death. This assumption becomes less valid as the time interval between frames is increased, because the animal first seen feeding is increasingly likely to be a secondary feeder, taking over after the original predator has gone.
Determining an optimal time interval
As pointed out above, reducing the number of frames observed saves time and money, but also reduces the amount of information that emerges from the data set. A measure of this loss of information was therefore required.
The more information that is available, the more likely it is that true but weak biological hypotheses will be verified. Various hypotheses were tested, each employing different numbers of frames, and the number of significant effects detected in each case was recorded. This provides a measure of the number of biological conclusions which could be reached using different amounts of sampling effort (cost).
RESULTS
The first hypothesis tested was that the number of contacts between predators and their prey was unaffected by prey depth (Table 3.3). This showed that the maximum time-lapse interval required to demonstrate an effect of depth varied with the soil predator studied. The second null hypothesis tested was that the predator type did not consume pupae. The data are shoen in Table 3.4. For each animal, this hypothesis can be rejected by a single observation of consumption and no statistical analysis was required. The used of a 4 or 8 minute time interval was usually adequate to record consumption taking place, but the use of a 20 or 40 minute time interval failed to do so.
The third hypothesis tested was that the fate of none of the pupae was observed (Table 3.5). For each pupa, this hypothesis can be rejected by a single observation of dying, being killed, disappering or surviving, and again no statistical analysis was required. Table 3.5. illustrates that the use of a 4 or 8 minute time interval was usually adequate to record at least one example of each fate, but the use of 20 minute or longer time interval failed to do so accurately.
DISCUSSION
The rhizotron and time-lapse cinematography can only be effectively used for studying predator-prey interactions if baits are provided. The predators are highly mobile and the area of window which the camera can adequately observe is limited to 15 cm x 10 cm. In any study where the prey is also mobile, particulary if reliance is placed on naturally occuring prey alone, the number of useful observations will be extremely small. Gunn and Cherrett (1993) together with the data given in Chapter II of this study, indicate that the occurrence of predators in the windows is normally quite a rare event. Best and Beegle (1977) showed that immobile prey (dead black cutworm) are normally as attractive as, or more attractive than, live larvae as a food for carabids, so using living but generally immobile pupae seems justified.
The choice of the most suitable interval between frames for time-lapse recording is not a simple one. Figure 3.1. is derived from Tables in this Chapter and shows that the reduction in cost is a expected, roughly linearly related to the reduction in frame frequency. The loss in significant information however has a less step gradient initially, as frame frequency reduced. The gap between the two plots is greatest at 7.5 per hour (8 minute intervals), suggesting that this is the most cost effective regime in which to work. However, the choice obviously depends upon which events are of most interest. Very frequent events such as contacts can be monitored with no loss of significance using 20 or 40 minute intervals (Table 3.3), whereas if it is important to observe a rare event such as staphylinid feeding (Table 3.4), no reduction from the four minute intervals would be acceptable. The other factor in the cost equation is the number of days for which continuous filming is carried out.
Clearly, the way to economise on time and money is not to observe every nth frame, but to set the camera to expose frames only at the required frequency. The conclusion from the data reported in this Chapter was that 8 minute intervals would meet the requirements of the rest of this study.
Although 2000 frames were to be exposed (8 minute x 11 days), time could also be saved by selectively analysing the film produced. More than 75 per cent of the analytical time was devoted to entering data into Minitab. Where attention could be concentrated on shorter periods (for example contacts before a significant proportion of the prey had disappeared) futher savings in time were possible.
Figure 1. The estimation of observation costs related to the number of significant observation obtained.
Table 1. Codes of soil fauna activity
Various taxa eventsCodes in column three
Carabid contact1
Staphylinid contact2
Slug contact3
Pupa disappearence4
Spider contact5
Pupa death6
Replacement7
Worm contact8
Campodean contact9
Centipede contact10
Ant contact11
Pupa survival12
Table 2. The estimation of observation costs using different time intervals (4, 8, 20, 40, 200 and 400 minutes)
FrequencyNo. framesTime requiredCost of Cost of Total
of filmingfilmedfor analysisanalysisfilm cost
(minutes)(hours)(£)1(£)2(£)
44000242.67728.0030.00758.00
82000121.33364.0015.00379.00
2080048.53145.606.00151.00
4040024.2772.803.0075.80
200804.8514.550.6015.15
400402.437.290.307.59
1) Analysis time is obtained by multiplying the number of frames by 3.64 minutes. The cost of labour is estimated by multiplying the number of hours by the cost of labour, assumed to be £ 3.00 per hour.
2) The cost of film (4000 frames) is £ 30.00 including processing.
Table 3. Hypothesis 1 : Depth of prey has no effect on the number of predator contacts. Chisquare values represent the departure from an even distribution of contacts over the three depth categories (0 cm, 3 cm and 6 cm) with 6 time lapse interval (4, 8, 20, 40, 200 and 400 minutes)
Soil animalsChisquare values at different time intervals
482040200400
Cara386***215***104***50*** NS+NS+
Stap32***NS+NS+ NS+ NS+NS+
Camp7026***3506***1220***703***141***73***
+) Numbers inadequate to perform a chisquare test
*) Stars indicate a statistically effect : *P<0.05, **P<0.01 and ***P<0.0001
Table 4. Hypothesis 2 : The various predators do not consume pupae
AnimalEvidence of consumption from different time interfals
taxon482040200400
CaraRRCCCC
StapRCCCCC
CampRRRCCC
CentRRCCCC
R = Hypothesis rejected (consumption observed)
C = Hypothesis confirmed (no consumption recorded)
Table 5. Hypothesis 3 : The fate of pupae cannot be determined by time-lapse photography
