A senior capstone in field biology

Pollinators,
priced out.

Presented by Maya Ortiz, B.S. Biology '26 Advised by Dr. Elena Cho · Dept. of Ecology & Evolutionary Biology
14 · V · 2026
For: the defense committee.
Six sites. Seven visits. One clear signal.
Capstone Defense Bio 4900Spring 2026

What's
inside.

01 The open question Context · 3 min
02 Survey design Method · 4 min
03 What the data show Evidence · 5 min
04 Contribution & what's next Close · 3 min
Principles Ortiz CapstonePhase II

Three commitments we held to.

Pinned above the field notebook. Every transect walked and every bee logged got checked against these three before it went in the dataset.
Rule · 01

Standardize
the transect.

Same 200m walk, same 10am–2pm window, same six urban sites — from a fenced community garden to a highway median planting.

— same route, every time.
Rule · 02

Log every
visit, not just hits.

Zero-count observations are recorded with the same rigor as sightings, so absence carries real weight in the model.

seven visits per site, logged.
Rule · 03

Verify the
ID twice.

Every specimen photo gets a second call from Dr. Cho's lab before a species label is locked into the dataset.

412 specimens, double-checked.
Section II
Why urban
bees are thinning

Show the
decline
clearly.

— turn the page —
Findings · detail Ortiz CapstonePhase III
Notice · 05
Why this matters

Since 2019, three citywide surveys have shown pollinator counts falling — but none isolated which urban feature drives the drop.

What's known

Regional counts are falling — three consecutive citywide surveys point the same direction, but at a resolution too coarse to explain why.

  • Citywide surveys (2019, 2021, 2023) show a steady decline in wild bee sightings.
  • Native plant cover is shrinking as infill development replaces yards with turf and pavement.
  • Warmer, drier summers are shifting bloom timing across the whole city.
Three prior surveys, one open question

Where the evidence falls short

Coarse · citywide · one visit a year.

Existing surveys sample once a year at the city scale, so they can't tell a fenced pocket garden from a highway median — the two ends of the urban gradient get averaged together.

That means no one can say whether it's plant diversity, paved cover, or something else driving the decline.

No site-level resolution, no repeated visits

What this capstone contributes

  • Six sites spanning the full urban gradient, walked seven times each over ten weeks.
  • A model that separates impervious-surface cover from bloom diversity as distinct predictors.
  • A replicable protocol the next cohort can pick up and repeat next spring.
Six sites · seven visits · one growing season
Visitation rate, by site Phase IIIEvidence

Visits drop
63 percent
by midsummer.

Across the six sites, pollinator visits per ten-minute observation fall from an average of 14 in early May to about 5 by mid-July — and the paved, high-traffic sites lead the decline.

Community garden (least paved)
Mixed residential yards
Highway median & lot (most paved)
Pollinator visits per 10-min count
1410.573.50
V1V2V3V4V5V6V7
Field protocol Ortiz CapstonePhase IV

From transect
to dataset,
in five steps.

The same protocol repeated at all six sites, seven times each, from early May to mid-July.
1

Scout

Walk the 200m transect once to flag flowering patches and set fixed observation points.

2

Observe

Ten-minute timed counts at each point, recording every visit and species where identifiable in the field.

3

Collect

Net and photograph unclear specimens for lab confirmation; release healthy insects on site.

4

Verify

Dr. Cho's lab cross-checks every photographed specimen before a species label is locked in.

5

Model

Counts feed a generalized linear model with impervious cover and bloom diversity as predictors.

Step 1 · Scout Step 2 · Observe Step 3 · Collect Step 4 · Verify Step 5 · Model
Three study designs, side by side Ortiz CapstonePhase IV

Where this
design earns
its keep.

Scored against the four levers that decide whether a survey can actually isolate cause from correlation.
Lever
Citywide surveys
Single-site study
This capstone
Site-level resolution
city-averaged
one site only
six-site gradient
Repeated visits per site
once a year
weekly
7 visits, 10 weeks
Blind ID verification
field call only
single reviewer
double-checked in lab
Controls for pavement cover
not modeled
not modeled
GLM covariate
In numbers Phase IIIEvidence

The case,
by the numbers.

Three figures from ten weeks of fieldwork across six urban sites — every count logged the same day it was taken.
6

Urban sites
surveyed

From a fenced community garden to a highway median planting, spanning the full paved-to-planted gradient.

412

Specimens
logged & ID'd

Every photographed specimen cross-checked in Dr. Cho's lab before its species label was locked in.

63%

Drop in visits,
May to July

Pollinator visits per ten-minute count fell from 14 to about 5 at the most paved sites.

Committee feedback Phase IIIEvidence
"
This is the first dataset from our program to actually separate pavement cover from bloom diversity — that's a real methodological contribution, not just a nice survey.
Dr. Elena Cho Faculty advisor · Dept. of Ecology & Evolutionary Biology
What's next Ortiz CapstonePhase V
Six sites · Ten weeks · One clear signal

The answer:
pavement,
not plants.

Impervious surface cover — not bloom diversity — is the strongest predictor of pollinator decline at these six sites. More paved cover means fewer visits, independent of how many flowering species were present.

Where this goes next

1

Submit to the Undergraduate Research Journal

Manuscript draft due to Dr. Cho by June 1; targeting the campus journal's fall issue.

2

Present at the spring symposium

Poster session, May 22 — same six-site figure and GLM, condensed to a three-foot board.

3

Hand the protocol to next year's cohort

Limitation: one growing season, six sites, a single city — the next cohort repeats the transects to test if the pattern holds.