Stocking Bass in Two Lakes: A Cascade That Worked Once
A lake district stocks largemouth bass into two small seepage lakes that had no piscivorous fish and had been turbid for a decade, to test whether top predators control algal blooms. North Lake is 8 ha; South Lake is 12 ha. Technicians sampled zooplankton (vertical hauls, >200 micrometers), chlorophyll a at 1 m, and Secchi depth at three stations per lake, monthly May-September, for two years before and three years after stocking. North: zooplankton dry mass rose from 12 to 45 micrograms per liter, chlorophyll a fell from 28 to 9 micrograms per liter, and Secchi depth improved from 0.8 to 2.1 m. South: zooplankton rose from 10 to 14 micrograms per liter, chlorophyll a stayed near 30 micrograms per liter, and Secchi depth did not change. South has dense submerged macrophytes covering about 40% of its littoral zone.
1. State the trophic cascade the managers are hypothesizing, predict the direction of each link, and explain why the North Lake data support the mechanism only indirectly.
Read the worked answer
The hypothesized chain (four levels)
- Bass (added) increase in abundance.
- Planktivorous fish (bluegill, minnows) decline through predation.
- Large-bodied grazing zooplankton (Daphnia) increase because predation is relaxed.
- Phytoplankton decline through grazing, so chlorophyll a falls and water clears.
Direction: each link alternates - predator up, prey down, next level up, next level down. A cascade is diagnosed by the alternating sign across three or more levels, not by the first and last alone.
Why North is only indirect evidence
- The observed changes are at the ends of the chain: zooplankton up, chlorophyll down, Secchi up. The middle link - planktivore abundance - was never measured.
- A two-level explanation fits the same numbers: bass could have reduced something else, or the lakes could have changed for unrelated reasons (nutrient loading, weather, water level) that affect zooplankton and phytoplankton independently.
- North also has no replication: one lake, one stocking event. The three stations are subsamples of one lake, not three independent lakes.
What would strengthen it: planktivore abundance and gut contents before and after stocking, zooplankton body-size distributions (large grazers appearing), and grazer exclusion experiments in situ.
2. Explain why South Lake did not respond, giving at least two competing hypotheses and a specific test that would separate them.
Read the worked answer
Hypothesis 1: macrophyte refuge for planktivores.
- With 40% of the littoral zone in weed beds, small planktivores can shelter where bass cannot forage effectively. The middle of the cascade never weakens: planktivores stay abundant, Daphnia stay suppressed, phytoplankton stay high.
- Test: sample planktivore abundance and gut contents inside and outside weed beds; run paired enclosure/exclosure experiments in vegetated and open water; compare zooplankton size structure between habitats.
Hypothesis 2: the phytoplankton is largely inedible.
- If South is dominated by colonial or filamentous cyanobacteria (large colonies, toxins), grazing cannot convert zooplankton biomass into reduced chlorophyll. Zooplankton may rise slightly (10 to 14) without any algal response.
- Test: identify phytoplankton to genus and measure size distribution; run a grazer bioassay (South Lake water with and without added Daphnia) and check whether chlorophyll and cell counts decline.
Hypothesis 3: bottom-up control dominates.
- If nutrient loading is high enough, or light limits growth, grazing cannot overcome the supply. Test with nutrient-addition and light-shading bioassays on lake water, plus a nutrient budget.
Cheap discriminating step first: the phytoplankton community composition and the planktivore data cost little and would already rule out one or more hypotheses. Note also that South is larger and physically different, so lake identity remains an alternative explanation for the whole comparison.
3. Critique the study design and propose an analysis plan that the data can legitimately support.
Read the worked answer
Design problems
- No replication at the level of treatment. There are two lakes and two "treatments," and each treatment is applied to exactly one lake. Treatment is completely confounded with lake identity, area, depth, watershed, and macrophyte cover.
- Pseudoreplication. Three stations per lake are subsamples of a single experimental unit; treating them as n = 3 would inflate significance.
- Uncontrolled trends. A before-after comparison cannot separate stocking effects from a regional trend (climate, nutrient decline, shoreline development) or from the natural multi-year variability of shallow lakes.
- Short post-period. Three years may capture a transient, and Secchi depth responds to factors other than algal biomass (suspended sediment, carp).
What the current data can support
- Descriptive statements: in North, zooplankton rose and chlorophyll fell concurrently with stocking; in South, they did not.
- The North pattern is consistent with a cascade - it is not a demonstration.
Better plan
- BACI design: multiple stocked and unstocked reference lakes matched on area, depth, nutrients, and macrophyte cover; sample all lakes before and after.
- Analyze lake-by-period interactions with lake as the unit of analysis and date as a repeated measure; report effect sizes with confidence intervals.
- Add the missing mechanistic variables (planktivore abundance, zooplankton body size, phytoplankton composition) and a grazing bioassay.
- Confirm the stocking actually established bass populations (recruitment surveys) - an unmeasured failure at step one would mimic "no cascade."