Field Data & modeling

The data, the methods, and the line between them.

All observations in this dataset come from the following sources: 48 entries of a manual fishing logbook, 75 entries of manually recorded pH values, and NOAA's daily temperature and rainfall record for Washington Dulles International Airport. All measured values, such as catch counts and pH, are shown in solid teal. Modeled water temperature is shown in dashed amber.

Methods

What was measured, what was modeled

Measured

Catch log — species names and counts tallied manually at each session's conclusion, May 2023 – August 2026. 48 sessions total, 525 fish.

Session effort — timestamps marking the beginning and end times, totaling 269 hours.

pH — 75 total recordings of three separate points throughout each session over 25 sessions.

Air temperature and rainfall — NOAA/National Centers for Environmental Information Daily Summaries from Washington Dulles Airport Weather Station (USW00093738). Actual values were gathered roughly seven miles north of the study site.

Modeled

Water temperature — no instrument has ever been installed in this body of water to measure temperature. Estimated from the daily air temperatures via a first-order lag equation:

Twater(d) = Twater(d−1) + [ Tair(d) − Twater(d−1) ] / τ

τ represents the time constant of thermal response, which dictates how quickly the water follows the air temperatures. τ was fixed at five days, a literature value for shallow ponds. It is assumed, not fitted, because there is nothing to fit it against.

This model recreates the physical characteristics of a pond, which include lagging the air and oscillating less than the air does. Over 48 sessions, the modeled water temperature range proved to be narrower than the air temperature range, as it must be.

Finding one — the strongest result

The pond's chemistry changes over a single afternoon

Three readings per session, late morning to late afternoon. pH rose every single time.

pH across the day, all 25 sessions

Measured

Each faint line is one session; the heavy line is the mean of all 25. Not a single session ran the other way.

Mean pH 7.55 → 7.71 → 8.07. Rose in 25 of 25 sessions. Wilcoxon signed-rank, first versus last: p = 1.2e-05. Mean within-day swing 0.52 units; largest 1.00.

What causes it

During the day, algae and aquatic plants absorb dissolved carbon dioxide from pond water through photosynthesis, lowering carbonic acid levels and raising the pH. At night, photosynthesis stops, and respiration starts, causing the pH to drop as algae and aquatic plants release carbon dioxide into the water.

The extent of that daily fluctuation is a well-recognized indicator of how biologically productive a body of water is and, thus, how nutrient-rich it is. A swing of half a unit over six daylight hours is a sign of an active and fertile pond; it is exactly what you would expect from a pond collecting lawn fertilizer from the surrounding neighborhood.

Why it matters beyond chemistry

If the pH changes that way, dissolved oxygen must change. Throughout the day, algae and aquatic plants produce oxygen, and dissolved oxygen levels climb, peaking in the late afternoon. When the sun sets, they consume dissolved oxygen through respiration, which causes its level to hit the lowest point just before dawn. Algae and aquatic plants grow quickly with high nutrients, creating enormous oxygen demand at night. During the pre-dawn hours, low oxygen stresses fish and kills them in extreme cases.

Since I have not measured dissolved oxygen, the parallel cycle is an inference based on established limnological knowledge. The next reasonable step to determine whether the nighttime drop in oxygen levels threatens fish populations is to measure dissolved oxygen at dawn.

Finding two — a null result

Total catch has no relationship with temperature

Fish per session against modeled water temperature

x-axis modeled

Each point is one Saturday. The horizontal band line is the mean within each temperature band. There is no trend to describe.

Pearson r = -0.001, p = 0.996 across 48 sessions. This is a null result, and it is reported as one.
Withdrawn
I reported on an earlier version of this site that catch peaked within the 80-85 °F range When I verified the numbers against handwritten records, I found that the catch log contained incorrect species counts. I withdrew that claim and referred to it here rather than simply deleting it. The chart above uses the corrected log.

Finding three

The total is flat because the species move in opposite directions

Bluegill hold the catch steady while bass fall away and bullhead climb. The aggregate hides all of it.

Mean fish per session by temperature band

Counts measured

Same sessions, split by species. Bass and bullhead run in opposite directions across the temperature range; bluegill are indifferent to it.

Largemouth bass: 2.25 per session below 62 °F, 0.25 above 80 °F. Brown bullhead: 1.50 rising to 4.62.

Every species against modeled water temperature

SpeciesSpearman ρpAt p < 0.05
Bluegill-0.0810.5855
Brown bullhead+0.3970.0052significant
Largemouth bass-0.5980.0000significant
Channel catfish+0.0680.6454
Common carp-0.6060.0000significant
Crappie+0.3380.0187significant
Golden shiner-0.0380.7999
There are two confounding factors. First, the day of the year and water temperature are correlated (ρ = +0.659) since sampling only takes place from May to August; for bass the seasonal correlation (−0.703) is greater than the thermal one (−0.598). Largemouth bass build their nests in the spring and are found in shallow water during this season. They are aggressive and easy to catch when guarding their nests; by August they have moved to deeper water and are much less likely to take shore-based tackle. This is an explanation based on the breeding calendar, not one based on temperature, and this study is unable to distinguish between the two. Second, the catch shows only which fish bite. A hook does not sample small fish, and nothing in this study constitutes a full population census.

Finding four — a trend that turned out not to exist

How an apparent bass recovery disappeared

If the figures are added together from year to year they give 7, 6, 11, and 14; this shows that the population has doubled, and it would have been a pleasant figure to have reported.

It is precisely because the sessions occurred at that time that this situation arose. At this pond, bass are basically only seen in May — there were 2.55 per session in May as against 0 in August, and in fact no bass were caught in twelve August sessions over a period of four years. Furthermore, 2026 had four May sessions while 2024 had only one: there was therefore greater activity in May and consequently more bass, even though no alterations had been made to the pond.

If we confine ourselves to the months from June through August — the same periods that were sampled in a similar way each year — then the trend vanishes.

Bass per year, all months vs. June–August only

Measured

The same fish, counted two ways.

June–August: 2, 3, 2, 3. Flat.

Further results

Rain against dry

Measured

Mean fish per session on days NOAA recorded rainfall at Dulles, against days it did not.

11.27 on 37 dry days, 9.82 on 11 wet. Mann–Whitney p = 0.28. Null.

Catch per hour by season

Measured

Fish per hour of recorded effort, averaged within each year. Effort-corrected, so a short season does not read as a poor one.

1.97, 1.92, 1.91, 2.15 fish/hour across the four seasons. Overall mean 1.98.
One more null
The daily pH swing does not indicate the number of fish that were caught on that day (r = +0.049, p = 0.81). The connection between productivity and catch was tested, not taken for granted, and it did not seem to exist.

Species composition

525 fish, seven species

The length of the bar indicates the total number caught; following each bar is the count, together with the number of sessions during which that species appeared at all.

Bluegillforage base
257 · 48/48
Brown bullhead
159 · 42/48
Largemouth basspredator
38 · 19/48
Channel catfish
31 · 27/48
Common carpincl. ornamental
18 · 15/48
Crappie
11 · 11/48
Golden shiner
11 · 11/48

forage   predator   management concern

Open data

Download everything

All the figures on this site can be obtained from these files; the measured and the modeled values are in separate columns and the modeled value includes the method used on each row.