
give and use peer critique, stress-test the colony plan and its genetic sampling with statistics, and revise the plan and its explanation into version 2.
Phase 4: Build
Do Now. Read two pieces of feedback: (1) “Looks fine.” (2) “I like your 50 m viewing line. I wonder what happens when a storm pushes the harems up the beach. What if the line could move?” Which helps more, and why?
Which feedback follows the “I like, I wonder, what if” protocol best?
Gallery critique. Teams post version 1: the beach map, the matrix and the genetic model. Visitors leave sticky notes in three colors: I like…, I wonder…, What if…. Every note must point to a number, a map feature or a criterion.
The stress test. Each plan is tested against the criteria under a harder scenario (sample scenario): during peak pupping, a winter storm with high tides pushes the harems 30 m up the beach, toward the viewing area, on a holiday weekend when visitor numbers triple. Does the plan still keep disturbance below 2 per hour? Does monitoring still work?
Worked example 1: a failed test. Team Jamal’s boardwalk sits 50 m from the usual harems. After the storm push, the seals are only 20 m away: from the volunteer data, about 6 disturbance responses per hour, three times the limit. Version 2: a movable rope line that docents shift to keep 50 m, and a rule that the boardwalk’s end section closes when harems move up.
Worked example 2: is the genetic sample big enough? Team Jamal planned 20 molt samples a year to estimate the proportion of seals heterozygous at a marker. If the true proportion is about 0.20, the 95% margin of error is about 2 × √(0.20 × 0.80 ÷ n). With n = 20: 2 × √0.008 ≈ 2 × 0.089 ≈ 0.18, so the estimate could be anywhere from about 0.02 to 0.38: useless for spotting a slow decline. With n = 80: 2 × √0.002 ≈ 0.09. Version 2 collects 80 samples, spread across the beach so one family group is not counted many times. Their explanation changed too: “We used to think any sample would show the colony’s genes. The calculation showed that small samples are dominated by chance, just like small populations.”
Iterate. Engineers rarely get it right the first time. They test, learn, change one thing, and test again, keeping a record of what changed and why.
After the gallery walk, test and revise.
Version 2.
Write your change log: two pieces of feedback you used, what you changed, and the new stress-test result.
Use the panther case to explain how a change in allele frequencies changed the distribution of traits in a population, and what that suggests about the elephant seals' risk. Use the words heterozygosity and survival.
By the early 1990s, the Florida panther, a population of pumas isolated in southern Florida, had fallen to only about 20 to 30 adults. Biologists found signs of inbreeding: kinked tails, heart defects, poor sperm quality and low genetic variation. In 1995, wildlife agencies released eight female pumas from Texas, a population that had once exchanged genes with Florida’s. Several of them bred with Florida males. In the following years, kittens with mixed ancestry had higher heterozygosity, inbreeding defects such as kinked tails became less common, and survival improved. The population has since grown to well over 100 adults, although habitat loss and road deaths remain serious threats.
Think about it: How is the panther’s history like the elephant seals’, and why would genetic rescue be much harder for a species with no other population to draw from?
Version 2 is tested and explained. Next: the Prove It showcase for the sanctuary.
Exit ticket: what is the single most important change between your version 1 and version 2, and what evidence caused it?
How confident are you that you can revise a plan and its explanation using statistics, test results and feedback?
Materials: posted version 1 work, sticky notes in three colors, the stress-test card, calculators with square roots.
Case study check: the 1995 release of eight female Texas pumas and the later fall in inbreeding defects are documented by the U.S. Fish and Wildlife Service and in a 2010 Science paper by Johnson and colleagues. Population figures are approximate; current estimates are published by state wildlife agencies.
Listen for: “any sample is good enough”; revisions that change looks but not evidence; teams hiding a failed stress test.
Support / stretch: give the margin-of-error formula with √ values pre-computed for n = 20, 40, 80. Stretch: decide the smallest sample that keeps the margin under 0.10.