Tropical bees and nectar choice

These data are from a PhD project regarding amino acids and floral nectar. Tropical stingless bees (Trigona hockingsii) were presented with a choice of nectar:

Sugar – plain sugar water (a mix of carbohydrates).
Amino – added amino acids. A sugar solution with additional amino acids to mimic “natural” nectar.

The number of bees feeding at each type was recorded at intervals, over a period of 9 days. The bee hive was located at James Cook University, Townsville, Queensland, Australia.

Table 1. Bee visits to different feed types. Values are daily totals.

Date	Amino	Sugar
20/11/2000	96	135
21/11/2000	97	136
22/11/2000	115	78
23/11/2000	129	100
24/11/2000	137	121
25/11/2000	79	120
27/11/2000	93	117
28/11/2000	183	109
29/11/2000	134	201

The table shows the daily totals. The full dataset is also available, which gives the counts at each time interval (400 observations for each feed type).

Download

You can download the dataset as a CSV file using this link: <Trigona-pairs.csv>. Alternatively, you might copy the table to the clipboard and paste into a spreadsheet. The full dataset, which includes the counts at each time interval is available using this link: <Trigona-indiv.csv>.

Usage

You can use these data to practice/illustrate various topics:

Using a Pivot Table.
Simple summary statistics.
Graphical summary.
Paired differences test.

Keywords

Invertebrate, pollinator, bee, paired test, differences, graphics, Pivot Table, scatter plot, isocline.

Examples

The following examples will give you a few ideas about how you might explore these data.

Pivot Table

The dataset Trigona-indiv.csv contains the complete set of observations. This dataset could be summarized and explored using a Pivot Table. The column headings are:

Day – a simple integer giving the day of observation.
Obs – an integer value as an index, the observation number.
Count – the number of bees observed.
Feed – a categorical variable giving the feed type (A = amino, S = sugar).

You might also use the Pivot Table to make a Pivot Chart. A bar chart showing the sum of counts compared by day might be useful for example.

Summary statistics

These datasets can be summarized in various ways, including:

Averages – mean, median.
Dispersion – standard deviation, inter-quartile ranges.
Shape – parametric or not (e.g. Histogram or Shapiro-Wilk test).

The “shape” is important, as it will inform you which kind of differences test is most appropriate.

Paired test

Since the observations for the two samples are matched, we can use a matched-pair version of the t-test or U-test to compare differences between feed types. The choice of test will depend on the distribution of the data.

The Trigona-pairs.csv data are normally distributed so you could use a paired t-test. A paired U-test can also be conducted:

T-test: t = -0.38589, df = 8, p-value = 0.7096.
U-test: V = 17, p-value = 0.5533.

The individual data, Trigona-indiv.csv are definitely not normally distributed so a paired U-test would be the best option:

U-test: V = 7492.5, p-value = 0.2915

Graphics

Paired tests can be problematic to display visually. A “traditional” bar chart or box-whisker plot would compare the overall mean or median values. However, in a paired test we are more interested in the pair by pair comparisons.

Daily visits to different feed types. The line is an isocline (where Amino = Sugar).

In the scatter plot, each datum is a paired observation with the daily sum of visits to the Amino feed plotted against the Sugar feed. The line is an isocline, drawn with a slope of 1 and an intercept of 0.

If all the points lay to one side of the line it would be a strong indication that one feed-type was more visited than the other. In this case we see more or less equal spread.

References

Gardener, M. C., Rowe, R. J. & Gillman, M. P. 2003. Tropical bees (Trigona hockingsi) show no preference for nectar with amino acids. Biotropica, 35, 119-125. [PDF].