Stopover Population Estimate and Migration Ecology of Red Knots C. c. rufa at Delaware Bay, USA, 2025

Using a Bayesian Jolly-Seber mark-resight model, this study estimates that the 2025 Red Knot stopover population at Delaware Bay reached approximately 54,000 individuals with stable migration timing, providing critical data to support the Adaptive Resource Management framework for balancing horseshoe crab harvest and shorebird conservation.

The Gist

Imagine a massive, high-stakes pit stop at a Formula 1 race. The drivers are Red Knots, a type of tiny shorebird, and the race is their epic journey from South America to the Arctic. The Delaware Bay is the only gas station on the entire route where they can refuel.

But here's the catch: the "fuel" isn't gasoline; it's horseshoe crab eggs. These eggs are like high-octane energy bars, packed with calories that allow the birds to make the final, grueling leg of their trip.

This paper is a report card on how well the birds made it to this gas station in 2025, and whether the "gas station" is still open for business.

The Story So Far: A Race Against Time

Back in the 1990s, humans started harvesting horseshoe crabs for bait (to catch eels and whelks). It was like someone draining the gas station's tanks. The birds arrived, found empty plates, and couldn't refuel. As a result, the population of Red Knots crashed from about 50,000 to just 13,000.

To fix this, scientists and managers created a "Adaptive Resource Management" (ARM) plan. Think of this as a smart thermostat for the ecosystem. The goal is to find the perfect balance: harvest enough crabs for the fishing industry, but leave enough eggs so the birds don't starve. To make sure the thermostat is working, scientists need to count the birds every year.

How They Counted the Birds (The "Flag" System)

Counting 50,000 tiny, fast-moving birds is impossible if you just try to guess. So, the scientists use a clever trick: The Flag System.

1. The Tags: Over the years, scientists have caught thousands of these birds and put unique, colorful plastic flags on their legs (like a license plate).
2. The Hunt: In 2025, volunteers and staff walked along 20 different beaches in Delaware Bay. They didn't count every single bird; instead, they looked for the ones with flags.
3. The Math: They used a sophisticated computer model (called a Jolly-Seber model) that acts like a detective. It looks at:
   • How many flagged birds they saw.
   • How many times they saw the same bird.
   • How many unflagged birds were in the crowd (estimated by taking quick "scan" photos of flocks).

By combining these clues, the model can estimate the total number of birds, even the ones without flags that the humans never saw.

The 2025 Results: Good News, But Not a Miracle

Here is what the report found for the 2025 race season:

• The Crowd Size: The estimated number of Red Knots stopping in Delaware Bay was 54,044.

  • The Analogy: This is like a stadium that was nearly empty a few years ago finally filling up again. It's about 17% more than last year and is only the second time since 2011 that the crowd has exceeded 50,000.
  • The Caveat: While the number went up, the "confidence interval" (the margin of error in the math) overlaps with previous years. This means the population is likely stable, not necessarily exploding. It's a good sign, but the birds are still living on the edge.

• The Timing: The birds arrived right on schedule.

  • The Analogy: The "gas station" opens its doors when the crabs lay their eggs. The birds are like commuters who know exactly when to show up. In 2025, half the population arrived by May 18th, just like they have for the last 15 years. They haven't gotten confused by climate change or arrived too early/late.

• The "Flag" Problem: There is a slight worry in the data. The percentage of birds wearing flags is dropping.

  • The Analogy: Imagine if you went to a concert and realized fewer people were wearing the special VIP wristbands you gave out years ago. This doesn't mean there are fewer birds; it means fewer of the new birds are getting tagged (perhaps because tagging efforts in South America have slowed down). The scientists had to adjust their math to account for this.

What Does This Mean for the Future?

This report is the "thermostat reading" for the ARM plan.

• The Verdict: The system is working. The horseshoe crab harvest is currently being managed in a way that allows the Red Knots to refuel successfully. The population is holding steady around 50,000, which is a massive recovery from the low point of 13,000.
• The Takeaway: Nature is resilient. When we give the birds a fighting chance (by protecting their food source), they bounce back. However, the job isn't done. The population is stable, but not booming yet. The "smart thermostat" needs to stay tuned to ensure the crab eggs remain plentiful enough for the birds to keep thriving.

In short: The birds showed up, they ate well, and the numbers look healthy. The plan to save them is working, but we can't stop watching the clock.

Technical Summary

1. Problem Statement

The subspecies of Red Knot (Calidris canutus rufa) relies heavily on the eggs of the Atlantic horseshoe crab (Limulus polyphemus) during its northward migration through the Delaware Bay to refuel. Historical overharvesting of horseshoe crabs in the 1990s led to a dramatic decline in Red Knot populations (from ~50,000 to ~13,000). To address this, the Atlantic States Marine Fisheries Commission (ASMFC) implemented an Adaptive Resource Management (ARM) framework. This framework requires annual estimates of both the horseshoe crab spawning population and the Red Knot stopover population to balance sustainable crab harvest with shorebird recovery.

The specific objectives of this study were:

1. To estimate the 2025 passage (stopover) population size of Red Knots at Delaware Bay.
2. To assess potential shifts in migration timing (arrival dates) since 2011.
3. To provide data-driven inputs for the ARM framework to guide future harvest regulations.

2. Methodology

Field Data Collection

• Survey Design: Surveys were conducted on 20 beaches along the Delaware and New Jersey shores of Delaware Bay from May 7 to June 6, 2025.
• Sampling Periods: The migration season was divided into 10 sampling occasions, each lasting three days.
• Mark-Resight Data: Field observers recorded individually marked birds (identified by unique alphanumeric leg flags) and performed "scan sampling" to estimate the ratio of marked to unmarked birds.
  • Total Resightings: 2,035 flag resightings were recorded.
  • Valid Data: After filtering for misread errors (5.8% error rate), 1,276 unique individuals were analyzed.
  • Scan Samples: 590 marked-ratio scan samples were collected (327 in DE, 263 in NJ).

Statistical Modeling

The study utilized a Bayesian state-space formulation of the Jolly–Seber (JS) mark-resight model for open populations. This approach accounts for:

• Population Turnover: Birds arriving and departing continuously during the season.
• Imperfect Detection: Not all birds are seen during every survey.
• Superpopulation Estimation: Estimating the total number of birds that pass through the site (the "superpopulation"), rather than just the number present at a single point in time.

Key Model Components:

1. Marked Subpopulation: The JS model estimated the abundance of flagged birds ($M^*$) using encounter histories.
   • Detection Probability ($p$): Modeled as a function of time and flag color. A specific parameter was included for dark green flags, which have historically shown lower detection rates than light green flags.
   • Arrival ($\beta$) and Departure ($\phi$): Time-specific probabilities for birds entering and leaving the study area.
2. Unmarked Subpopulation Adjustment: The total population ($N^*$) was derived by adjusting the marked estimate ($M^*$) using the proportion of marked birds ($\pi$) estimated from the scan sample data via a binomial generalized linear mixed model (GLMM).
   • Formula: $\hat{N}^* = \hat{M}^* / \hat{\pi}$
3. Stopover Duration: Calculated using the latent parameter approach from Lok et al. (2019), which includes individuals estimated to be present even if never resighted, reducing bias toward longer stays.
4. Migration Timing: The date by which 50% of the cumulative population had arrived was calculated for each year from 2011 to 2025 to test for temporal trends using linear regression.

3. Key Results

Population Estimates

• 2025 Passage Population: Estimated at 54,044 individuals (95% Credible Interval [CI]: 47,926 – 61,928).
• Trend: This represents a ~17% increase from 2024 (46,127) and is only the second year since 2011 to exceed 50,000.
• Stability: Despite the numerical increase, the 95% credible intervals for 2025 overlap significantly with recent years (2020–2024), indicating a stable stopover population rather than a definitive upward trend.
• Marked Proportion: The proportion of flagged birds in the population dropped to 4.5% (95% CI: 3.4%–5.5%), the lowest since monitoring began in 2011. This decline is attributed partly to reduced banding efforts in South America (Argentina/Uruguay), where orange flags are used.

Migration Ecology

• Arrival Timing: 84% of the population arrived by May 21. The median date for 50% arrival was May 18.
• Temporal Trends: Linear regression of 50% arrival dates (2011–2025) showed no significant trend (slope = -0.03, $p = 0.235$). Migration timing has remained consistent, with no evidence of advancement or delay.
• Stopover Duration: The mean stopover duration in 2025 was 8.6 days (95% CI: 7.9–9.5). This is slightly longer than 2023–2024 but shorter than the 10–12 day durations observed in 2013–2016.

Model Performance

• Detection Probability: Resighting probability ranged from 14% to 25% between May 9 and 27. Dark green flags did not show a statistically significant difference in detection probability compared to other colors in 2025, though the model retained separate parameters for robustness.
• Data Quality: The study successfully utilized all 10 sampling periods, unlike some previous years where the final period (early June) was excluded due to sparse data.

4. Key Contributions

1. Robust Population Estimation: Provided the first 2025 estimate using a rigorous Bayesian JS model that explicitly corrects for imperfect detection and population turnover, offering a more accurate metric for the ARM framework than simple counts.
2. Long-term Trend Analysis: Confirmed that while the 2025 count was high, the population remains within the historical variability range observed since 2011, with no statistically significant directional trend.
3. Migration Phenology: Demonstrated that Red Knot migration timing at Delaware Bay has remained stable over 15 years, suggesting that current food resource availability (horseshoe crab eggs) is sufficient to maintain established arrival schedules.
4. Methodological Refinement: Highlighted the declining proportion of marked birds and the need to account for changing flag colors and international banding effort variations in population models.

5. Significance

• Management Implications: The findings support the current Adaptive Resource Management (ARM) framework. The stable population size and consistent arrival timing suggest that current horseshoe crab harvest limits are likely maintaining adequate foraging opportunities for Red Knots.
• Conservation Status: The population estimate of ~54,000 is a positive indicator for the recovery of the rufa subspecies, which was listed as threatened under the U.S. Endangered Species Act.
• Data Utility: The study provides critical, decision-relevant metrics (arrival schedules, departure rates, and total abundance) that allow managers to adjust crab harvest quotas dynamically to ensure ecosystem integrity and species recovery.
• Future Monitoring: The decline in the proportion of flagged birds necessitates continued vigilance in data collection and potentially adjustments to modeling to account for the changing composition of the marked population.