Paper 28: The Influence of Rainfall, Evaporation, and Atmospheric Temperature on Fluctuations in the Size of a Natural Population of Thrips imaginis (Thysanoptera)
by J. Davidson and H.G. Andrewartha (1948)
Thrips imaginis
Introduction
Thrips imaginis are a small insect with adults about 1 mm long that are indigenous to southern Australia. They feed on the flowers of weeds most of the time, but also feed on apple blossoms and rose blossoms. The authors noticed that significant fluctuations in their populations had been taking place for the past 40 years prior to this study. These fluctuations were separated by thousands of miles, and so they speculated that the long term fluctuations in weather lead to a growth or decline in the populations based on temperature and soil moisture. They then decided to focus their study on measuring the degree of association between the number of thrips and the components of the weather they considered to be related to the size of the population. They would then analyze this data with partial regressions.
Materials
Over a 14-year period from 1932 - 1946, daily counts of the number of adult thrips in roses in Adelaide were taken. 1944 was excluded, and from 1932 - 1938, a sample of 20 roses was taken daily over the entire year while from 1939 on, only 10 roses were included in the daily sample and the observation time was only from September to December. Samples were not taken on Sundays or holidays. Table 1 displays the number of samplings taken each month for 14 years.
Method
The total variability of the daily records of thrips may be ascribed to four components:
- The growth of the population with time within each year.
- The variation in the growth of the population from one year to another.
- The variation due to variable “activity” of the insects in seeking out roses, and
- The residual variability which could be random with respect to the three components selected for this study.
The study mainly focused on the second and third components, one at a time. The authors go on to state that from their experience, thrip populations don’t change by a constant quantity, but by a constant proportion. They then describe why they are choosing to display their data on a logarithmic scale, and then use a partial regression to analyze their data.
Daily Fluctuations in the Number of Thrips in the Roses
It’s stated that in any one year during the time when a population is increasing in size, the variability in the daily counts may be ascribed to:
- natural increase with time in the number of thrips in the garden
- variable “activity” of the thrips in seeking out the flowers
- residual or random variation
They then calculated a curve of predetermined degree for the 60 days which preceded the day when the population attained its maximum and turned them into 3-day intervals - giving 20 groups to calculate the mean number from each. Table 2 displays the extent and direction of the daily variation.
They then choose to look at the days preceding the sample taking and the day of as their independent variates. They choose to examine the effects of daily maximum air temperature, daily total rainfall, and changes in the barometric pressure. They continue by excluding the effects of barometric pressure and choose to just look at air temperature and daily total rainfall.
The x variables (X1 - X6) all represent some aspect of the climate for either the previous day or the day before that. They perform regressions for the effects of y (thrips numbers) on all of the X variates for the 8 year data. Ultimately they come to the conclusion that the daily number of thrips per rose increased by 25% for each 10 degree fahrenheit increase in temperature, and decrease about 66% for each 1 inch in total rainfall.
Annual Fluctuations in the Number of Thrips in the Roses
They follow a similar process here, selecting independent and dependent variates and investigating their relationships. The perform many partial regressions to find out what is the main cause of the annual fluctuations in these populations. They mention the germination time of the thrips’ main food source, which is in autumn. Ultimately they find in this section that the maximum density is reached by the population in the early spring. This is largely determined by the weather the preceding autumn, although abnormal early spring weather may modify this.
Annual Fluctuations in the Numbers of Thrips After the Elimination of Variance due to “Activity”
In this section the authors are aiming explain for any variation in their data due to the activity of the insects seeking out the flowers. They calculate a correction factor for each dataset, and then continue to complete partial regressions with their new dependent variates.
Discussion
The authors present their alternative theory for the density of the thrips population in this section. They found through all of their statistical analyses that the most influential factors on the population of thrips is the variable X2, which is total daily rainfall. They then state many of the factors that also may influence the populations, including the pollen count, at what time the “break” of the season happened, etc. Figure 5 is a hypothetical plot of the control of the population density (density independent) components of the environment.
They conclude that there are two major facts that they found through this study:
- There are four “density-independent” components of the physical environment that account for 78% of the variance within thrips populations.
- Rainfall (the variable X2) was the most important component.
They spend a good time at the end discussing the results put forward by Nicholson and Bailey, and how they are completely wrong. Davidson and Andrewartha say competition plays no part in determining the maximum density of a population. The “balance” that keeps a population from getting too large is the race against time to reproduce in what they call the “favorable period”, or a season. They say weather is the most important influence on the density, and that it never decreases to zero because they can’t all die out by the time it’s spring and the population starts to increase again.
What do you think of their conclusion about population density? Would you tend to agree more with Nicholson and Bailey (density dependent), or Davidson and Andrewartha (density independent)?
What do you think of their data collection methods? Although the study was over a long period of time, do you think their methods are sound?
Overall, it was a great read. I really think this paper is given us a valuable insight to understand modern studies, which are aiming to understand the effects of climate change upon the distribution and population biology of invasive species. I was reading a paper published by leading group of ecologist in Chile, where they basically were looking at the impact of precipitation and temperature upon the distribution of two stored grain insects using model of infraspecific competition and climate change scenarios. With regard to the question posted, I think both approaches (density dependent and density independent) are valid in order to understand questions which are aiming to understand how populations are fluctuating through time. This paper came just right on time in order to push the progress that ecology was having further back in those days. I think, it was one of the first times where people doing ecology are connecting cause and effect relationships.
ReplyDeleteA great effort to really understand and pinpoint the forces driving population fluctuations. This was a very thorough and detailed study. I found it most interesting that they were able to confidently state that predator or competition interactions were not responsible for population fluctuations. I wonder if this was really true and if there are systems where certain organisms are not influenced by predation or competition. I would think there would be at least for a period of time maybe not “forever.” It’s also interesting how they mention climate as being a density-dependent influence because if the population is larger there will be more die off than if the population was smaller for the same climatic event (explained on page 637). Is this an acceptable definition of “density-dependence?”
ReplyDeleteI feel like this paper was included for two reasons: its major use of partial regressions to analyze causal relationships, and its support of density independent population growth. The paper doesn’t make the density independent argument outright, but does so, quite effectively I might add, in the conclusion leading up to it with strong stats and data analysis. The rationale for the study was well laid out with the goal of helping mitigate the pest impact on apple yields. I did think more could have been said about the biology of the Thrip in the background, but enough was said to understand the study. The authors do a nice job of walking you through their regressions and carefully explain all their assumptions and conclusions. It was really fascinating to see so much of the variation in the population being explained by direct and indirect effects of climate. I don’t feel they try to bash the density-dependent crowd, but rather they take a stance that both are likely plausible depending on the system and organism examined.
ReplyDeleteI feel like this paper was included for two reasons: its major use of partial regressions to analyze causal relationships, and its support of density independent population growth. The paper doesn’t make the density independent argument outright, but does so, quite effectively I might add, in the conclusion leading up to it with strong stats and data analysis. The rationale for the study was well laid out with the goal of helping mitigate the pest impact on apple yields. I did think more could have been said about the biology of the Thrip in the background, but enough was said to understand the study. The authors do a nice job of walking you through their regressions and carefully explain all their assumptions and conclusions. It was really fascinating to see so much of the variation in the population being explained by direct and indirect effects of climate. I don’t feel they try to bash the density-dependent crowd, but rather they take a stance that both are likely plausible depending on the system and organism examined.
ReplyDeleteQuite the interesting read, although I'm not entirely in agreement on their assertions about predation not playing a role in population. There's something about all the papers we've read in the past few weeks... we've spoken about the "n-dimensional hyper-space" and how much of a key concept that is to understand. Yet the vast majority of the papers we've read seem to be hyper focused on a single dimension. Granted, I completely understand that you can't make an experiment to test all of the n-dimensions at once, but there's something about how focused the authors get on their own conclusions about whatever dimension they happen to be focusing on that seems to ignore the fact that there is more than one dimension to be examined. Key to this paper- the assertion that predation is not controlling population. Just because the thrip itself is not being directly predated doesn't mean predation isn't influential. What should happen to the thrip population if, for example, and invasive species of deer were to kill all the apple trees and flowers that the thrip eats?
ReplyDeleteI agree with predation must play a role. As Kat mentioned it obviously does, even if not directly. However the effects of the weather on the populations were, shown to be significant. I look at these experiments people have been conducting and I find it amazing, spending 14 years counting thrips on roses, or barnacles on rocks, that takes dedication!
ReplyDeleteDavidson and Andrewartha make good points about density independence, but I am not completely sold on this idea that the population does have competition or predation. In my opinion, the role of density dependence and density independence vary in both time and space (n-dimensional hyperspace). I think both regulate populations in different capacities and probably fluctuate between the two depending on the situation. Secondly, I liked that this study sought to make predictions about the thrips populations. It's cool to see a study that looks beyond the past and present like most of the papers we have read.
ReplyDeleteIn the Davidson and Andrewartha study, at first I couldn’t see why the months September through December were considered spring, until I realized that the University of Adelaide is in Australia, so Sept-Dec is spring in Australia!
ReplyDeleteI would like to have been able to ask Davidson and Andrewartha why they chose to collect data based on the numbers of thrips found in the roses, rather than on the apple blossoms or the group of annual weeds deemed most important as hosts for T. imaginis. since they write that “it was not possible to select independent variables which would serve to discriminate between the influence of the weather on the host plants and its more direct influence on the insects themselves (p. 627).” They partially account for the potential difference between numbers on the thrips’ major hosts and numbers on the roses with their analysis of daily rose migration activity, but why not just measure them on the plants considered their main hosts in the first place?
Still, their overall result that 78% of the variance in thrip population could be associated with rainfall and temperature, was impressive. Would this number have been higher if “activity” had not been a factor?
I can’t help wondering if the apparent divide between the density-independent and density dependent camps is just a reflection of the type of organisms and environments they chose to study. Davidson, Andrewartha, and Birch were studying populations of insects that had been observed to have wide fluctuations over large geographic areas apparently related to the weather (grasshoppers, thrips); the density-dependent researchers possibly were studying types of organisms where the animals could saturate the environment (eg. run out of food) before the season changed. As they say on p. 636, “on the average the physical environment remained favourable for far too short a period to permit the insects to multiply to a point where competition became important.”
Great point in your last paragraph there Julie! I never considered that but it totally makes sense that studying different organisms might lead to coming to different conclusions. It would even be interesting to do a meta-study and see if similar conclusions are always drawn from the same model organism. Also, finally nice to see some numbers that help us evaluate how good their results are. Though I assume all of our papers for our paper at the end of the semester will have some kinds of numbers like this.
DeleteIn the Davidson and Andrewartha study, at first I couldn’t see why the months September through December were considered spring, until I realized that the University of Adelaide is in Australia, so Sept-Dec is spring in Australia!
ReplyDeleteI would like to have been able to ask Davidson and Andrewartha why they chose to collect data based on the numbers of thrips found in the roses, rather than on the apple blossoms or the group of annual weeds deemed most important as hosts for T. imaginis. since they write that “it was not possible to select independent variables which would serve to discriminate between the influence of the weather on the host plants and its more direct influence on the insects themselves (p. 627).” They partially account for the potential difference between numbers on the thrips’ major hosts and numbers on the roses with their analysis of daily rose migration activity, but why not just measure them on the plants considered their main hosts in the first place?
Still, their overall result that 78% of the variance in thrip population could be associated with rainfall and temperature, was impressive. Would this number have been higher if “activity” had not been a factor?
I can’t help wondering if the apparent divide between the density-independent and density dependent camps is just a reflection of the type of organisms and environments they chose to study. Davidson, Andrewartha, and Birch were studying populations of insects that had been observed to have wide fluctuations over large geographic areas apparently related to the weather (grasshoppers, thrips); the density-dependent researchers possibly were studying types of organisms where the animals could saturate the environment (eg. run out of food) before the season changed. As they say on p. 636, “on the average the physical environment remained favourable for far too short a period to permit the insects to multiply to a point where competition became important.”
Like others, I found their findings that rainfall and temperature account for the variance in population from year to year to be an important find. However, I also wonder about the impacts of competition and predation on the population. This concern is especially in relation to the Thrips during their developmental stages. The authors state that while the roses are attractive to adults, it is unfavorable for development. While the environment may have the same impact on all stages of life, would competition and predation be different?
ReplyDeleteIn the introduction, Peet suggests that Andrewartha and Birch's conclusion that the thrips example showed no evidence of density-dependant control may have been wrong. He suggests that the researchers were unaware of how many roses were available from year to year. In all, I believe there could be an interplay of density-dependent and independent factors that regulate population.
I had a comment written up that somehow got lost!
ReplyDeleteTo summarize, I think D & A make a very important point in arguing for the importance of density-independence and abiotic factors in population dynamics. Unfortunately they don't offer much support for two important assumptions: 1) no predation/parasitism, and 2) resources not limiting (therefore no competition). Too bad they don't offer better support for these, because I'm convinced by their larger point and discussion of Nicholson & Bailey's shortcomings. It seems that something like climate could operate in either a density-independent manner (e.g. mortality/limited reproduction from heat-wave or frost) or in and indirect density-dependent way via limiting primary production and therefore the food resource of consumers. I wonder how common density independent regulation, abiotic control, and lack of predation/parasitism is in nature. These may be more common in variable or harsh environments (e.g. boreal, temperate continental areas), than in stable, productive environment (e.g. tropics).