Vertical Variation of the Caterpillar Community in Oak (Quercus robur) Canopies

This study reveals that vertical stratification of caterpillar communities in oak canopies is inconsistent across years and driven more by inter-annual and tree-level variation than by canopy height or phenology, highlighting the need to combine targeted canopy sampling with ground-based proxies for effective monitoring.

The Gist

Imagine a giant, living skyscraper made of oak trees. Inside this skyscraper, millions of tiny tenants (caterpillars) live, eat, and grow. For years, scientists have been trying to count these tenants, but they've mostly been looking at the building from the ground up, peering through the windows on the first floor. They've been guessing what's happening on the 20th floor based on what they see on the 1st.

This paper is like sending a team of researchers up a cherry picker (a mobile crane) to actually walk the floors of the oak tree "skyscraper" and see exactly who lives where, how many there are, and how they change from year to year.

Here is the story of what they found, broken down simply:

1. The Big Question: Is the Top Floor Different?

Scientists have long wondered if the caterpillars at the very top of the tree (the sunny, windy penthouse) are different from those in the middle (the cozy apartments) or the bottom (the shaded basement).

• The Theory: Maybe the top floor is a luxury suite with better food, or maybe it's a dangerous place with too much sun. Maybe the "landlord" (the tree) opens the windows (buds) at different times on different floors, causing the tenants to move around.

2. The Experiment: Climbing the Tree

The researchers went to a forest in the UK and picked 34 mature oak trees. Over three years (2023, 2024, and 2025), they used a cherry picker to reach the bottom, middle, and top of the tree crowns.

• They didn't just look; they cut off branches, bagged them, and brought them down to count every single caterpillar, measure how much leaf they had eaten, and identify the species.
• They also set up "traps" on the ground to catch falling caterpillars or their droppings (frass) to see if the ground methods matched the real data from the top.

3. The Surprising Results: It Depends on the Year!

If you asked, "Are there more caterpillars at the top?" the answer was: "It depends on which year you ask."

• 2023 (The "Top Floor" Year): The top of the tree was packed with caterpillars. It was like a crowded party at the penthouse. The density and variety of species were highest up high.
• 2024 (The "Basement" Year): The pattern flipped! The bottom and middle of the tree had the most caterpillars, while the top was surprisingly empty.
• 2025 (The "Quiet Year"): Everyone was gone. The caterpillar population crashed across the whole tree, from top to bottom. It didn't matter which floor you were on; it was just quiet everywhere.

The Analogy: Imagine a hotel where the guests usually prefer the top floor. But one year, a storm hits the roof, so everyone moves to the lobby. The next year, the roof is fine, but the lobby is hosting a free buffet, so everyone moves down. The "preference" isn't fixed; it changes with the weather and the year.

4. The "Tree Personality" vs. The "Floor Plan"

The researchers also checked if the timing of the tree mattered. They found that the top branches of a single tree do indeed "wake up" (bloom) about two days earlier than the bottom branches.

• The Expectation: You'd think the caterpillars would rush to the top to eat the fresh, early leaves.
• The Reality: It didn't matter. The caterpillars didn't seem to care about that two-day difference. They were more influenced by the overall weather of the year and the specific personality of the tree (some trees just had more bugs than others, regardless of height).

5. The Ground Trap Problem: Are We Looking at the Right Data?

For decades, scientists have used two main methods to count caterpillars from the ground:

1. Water Traps: Buckets of water placed under trees to catch falling bugs.
2. Frass Traps: Cloths to catch bug poop (droppings).

What they found:

• The Poop Traps (Frass): These were actually pretty good! The amount of poop on the ground correlated well with how many bugs were actually in the tree. It's like knowing how many people are in a house by counting the empty pizza boxes on the doorstep.
• The Water Buckets: These were unreliable. They only caught bugs that fell out (either because they were knocked off by rain/wind or because they were leaving to pupate). They missed the ones that stayed put or climbed up. It's like trying to guess how many people are in a building by only counting the ones falling out the window.

The Big Takeaway

This study teaches us that nature is messy and changes fast.

• Don't assume the top is always different: The "vertical layers" of a forest aren't static. They flip-flop based on the weather and the year.
• Multi-year is key: If you only study one year, you might think the top is always the best place to live, or that the bottom is. You need to watch for several years to see the real picture.
• Ground methods have limits: While counting poop on the ground is a decent shortcut, relying on water buckets to count bugs is like trying to count fish in a lake by only looking at the ones jumping out.

In short: To understand the forest, we need to stop just looking at the ground and start climbing the trees, and we need to keep climbing them for many years to see the full story.

Technical Summary

1. Problem Statement

Understanding the spatial distribution of caterpillar communities within tree canopies is critical for interpreting forest trophic dynamics and predicting responses to environmental change. However, this variation remains poorly quantified due to the logistical challenges of sampling in three dimensions.

• Knowledge Gap: Most existing monitoring relies on ground-based methods (e.g., water traps, frass traps) or accessible lower branches. It is unclear how accurately these methods reflect the diversity, density, and phenology of the full canopy community.
• Hypothesized Drivers: Vertical stratification may be driven by species-specific life history traits (e.g., flightless females of the winter moth laying eggs higher up), microclimatic gradients (light, temperature), leaf trait variation (nutritional quality, toughness), and within-crown phenological differences (budburst timing).
• Objective: The study aimed to quantify fine-scale vertical heterogeneity in caterpillar density, diversity, and herbivory across lower, middle, and upper canopy strata of mature oak trees, assess the influence of host tree phenology, and evaluate the reliability of ground-based monitoring proxies.

2. Methodology

Study Site and Design:

• Location: Wytham Woods, UK (temperate deciduous woodland).
• Subjects: 34 mature pedunculate oak trees (Quercus robur).
• Duration: Three consecutive springs (2023–2025).
• Sampling Strategy: Direct canopy access was achieved using Mobile Elevated Work Platforms (MEWPs).
  • Strata: Branches were sampled from three vertical levels: lower, middle, and upper thirds of the crown.
  • Timing: Sampling occurred during the spring abundance peak (mid-May) for three years. In 2024, additional early-season sampling (mid-April) was conducted to capture early instar emergence.
  • Sample Size: Approximately 285 branches total (96 in 2023, 98 in 2024, 91 in 2025).

Data Collection & Processing:

• Caterpillar Metrics: Branches were bagged, cut, and processed in the lab. Caterpillars were shaken out, identified to species/morphospecies, and counted.
  • Metrics Calculated: Total density, density of key species (Operophtera brumata - winter moth; Tortrix viridana - green oak tortrix), species richness, Shannon diversity index, and herbivory (percentage leaf area consumed).
  • Standardization: All density metrics were scaled per 100g of fresh leaf mass.
• Phenology Monitoring:
  • Intra-annual (2024): Manual bud scoring (scale 1–7) on 12 sections per tree to determine stratified budburst dates.
  • Inter-annual (2023–2025): Drone-based multispectral imagery (RedEdge-P) was used to calculate NDVI (Normalized Difference Vegetation Index). The "half-NDVI date" (50% of seasonal max) served as a proxy for whole-crown budburst timing.
• Ground-Based Proxies:
  • Frass Traps: Muslin cloths placed 1m above ground to collect fecal pellets (biomass proxy).
  • Water Traps: Trays filled with water to catch falling larvae (abundance/diversity proxy).
• Statistical Analysis:
  • Bayesian Generalized Linear Mixed Models (GLMMs) using the brms package in R.
  • Model Structure: Hurdle lognormal models were used to handle zero-inflated data (distinguishing between probability of absence and conditional density).
  • Fixed Effects: Canopy level, year, and phenology (budburst/NDVI).
  • Random Effects: Tree identity.

3. Key Results

A. Vertical Stratification is Inconsistent and Year-Dependent

• 2023: Densities, species richness, and herbivory increased significantly with canopy height (Upper > Mid > Lower).
• 2024: The pattern reversed; densities were highest in the lower canopy and lowest in the upper canopy.
• 2025: Densities were uniformly low across all strata with no significant vertical differentiation.
• Early Instars (April 2024): No significant vertical stratification was detected at the time of larval emergence, suggesting that vertical patterns emerge later in the season or are driven by post-emergence dynamics.

B. Drivers of Variation

• Phenology: While upper branches burst approximately 2 days earlier than lower branches, this fine-scale phenological heterogeneity did not explain variation in caterpillar communities. Stratified budburst dates did not improve model fit over whole-crown dates.
• Inter-annual Variation: Year-to-year differences were the dominant driver. 2024 had the highest overall densities, while 2025 saw a marked decline.
• Tree Identity: Significant variation existed between individual trees, suggesting tree-level factors (genetics, micro-site) outweigh within-crown vertical gradients.

C. Evaluation of Ground-Based Methods

• Frass Traps: Showed moderately strong, consistent positive correlations ($r_s = 0.49–0.68$) with total caterpillar density and herbivory across canopy levels and years. They serve as a reliable proxy for total community biomass.
• Water Traps: Showed weak and inconsistent relationships with canopy measures.
  • Correlations were only significant in specific years (2025) and for specific strata (mid/upper).
  • They failed to capture species with different life histories (e.g., T. viridana, which pupates in the canopy, was underrepresented).
  • Seasonal cumulative totals improved correlations for winter moths but not for diversity metrics, indicating water traps are biased toward species that descend to the ground.

4. Key Contributions

1. Methodological Advancement: Demonstrated the feasibility and necessity of using MEWPs to access the full vertical profile of tree canopies, moving beyond the limitations of ground-based sampling.
2. Temporal Dynamics: Revealed that vertical stratification patterns are not static; they can reverse or disappear entirely between years, challenging the assumption of consistent vertical gradients in ecological models.
3. Phenological Decoupling: Provided empirical evidence that small-scale within-crown phenological differences (approx. 2 days) are insufficient to drive vertical structuring of caterpillar communities, likely due to larval dispersal or starvation tolerance.
4. Monitoring Validation: Quantified the biases of common monitoring tools, confirming that while frass traps are robust proxies for biomass, water traps are poor proxies for community composition and diversity due to behavioral biases.

5. Significance and Implications

• Trophic Synchrony: The findings suggest that climate-driven shifts in phenology will likely impact caterpillar populations at the between-tree and between-year scales rather than creating fine-scale vertical mismatches within a single tree.
• Forest Resilience: Understanding that inter-annual variability (weather, host quality) outweighs vertical structure is crucial for modeling forest resilience and trophic cascades (e.g., impacts on insectivorous birds like the Great Tit).
• Future Monitoring: The study advocates for integrating targeted canopy sampling with ground-based proxies (specifically frass traps) to improve the accuracy of arboreal Lepidoptera monitoring. Relying solely on water traps or single-year data may lead to erroneous conclusions about community dynamics and vertical stratification.

Conclusion: Caterpillar communities in oak canopies are shaped more strongly by inter-annual environmental variation and tree-level heterogeneity than by consistent vertical stratification. Ground-based methods have utility but require careful interpretation regarding species-specific behaviors.