ANALYSIS: FINDINGS

On this page, we aim to tell a whole story of how the research has been conducted, including thoughts, visualisations/findings and analysis. If you wish to see the methodology behind each visualisation, you can check the red boxes. If don't, just skip and it will not affect the story!

Starting Point

After reviewing the data, there are two aspects on which we can further explore:

Cross-Sectional Analysis: Firstly, there are 4533 grids in the UK assigned with 7 different urban types, with all observations occurring in the 5-year period. Differences in bird patterns between urban and rural areas are evident.

Time Series Analysis: Secondly, each observation has a record of time. We also aim to present how bird patterns change from 2020 to 2025.

Defining Metrices

So far, we've been using the term “patterns.” We define it in two ways:

1. Richness – The number of unique bird species observed in a location.
2. Observation Count – The number of bird observations recorded within a period of time.

Pre-handling of Data

We aggregate the 7 types into two general ones—urban/suburban and rural—based on their “smod” values. Urban and suburban grids are then combined, as the urban heat island effect often extends into surrounding suburban areas. (Voogt & Oke, 2003)

With this sorting:

number of blocks in "Rural" :3696
number of blocks in "Suburban" :420
number of blocks in "Urban" :416

1. Cross-sectional Analysis

We first plot a boxplot on the adjusted richness in 3 urban types:

( Rationale for choosing Boxplot: The boxplot was ideal for highlighting differences in central tendency and variability between rural and urban/suburban blocks. It allowed us to directly compare the variation and median adjusted richness between groups.)

Figure 1 Boxplot of adjusted richness in different urban types

Based on the similar adjusted richness data, a histogram is plotted:

( Rationale for choosing Histogram: The histogram was used to show the shape and frequency of adjusted richness values across the two landscape types. It lets us visualise how many blocks fell into each richness bracket.)

Our boxplot showed a distinct shift: Rural blocks had higher median adjusted richness, with relatively tighter variability. Urban/suburban blocks not only had lower averages, but they also showed greater inconsistency.

Figure 2 Histogram of adjusted richness in different urban types

The histogram revealed a striking frequency difference. This is a clear right-skew for rural areas, with a large number of rural blocks achieving high adjusted richness values (e.g. >0.4), while very few urban/suburban blocks did.

2. Time Series Analysis

Given the growing interest in how urbanisation affects bird distribution, we began by examining the overall spatiotemporal patterns of bird observations. Specifically, we wish to see whether urbanisation might be associated with seasonal or long-term shifts in bird activity.

We then plotted the total number of bird observations in urban/suburban and rural areas for each month from 2020 to 2024:

Figure 3 Total number of bird observations in urban/suburban and rual areas over time, 2020 to 2024

Interestingly, the observation trends in urban/suburban and rural areas were seasonal and inversely related. Bird observations in rural areas peaked during summer and dropped in winter, while observations in urban/suburban areas increased in winter and declined in summer.

WHY?

Assuming that these patterns arise solely from bird behavior—excluding human observational bias—we propose the following hypotheses:

Hypothesis 1: Some bird species migrate seasonally between urban/suburban and rural areas.
Hypothesis 2: Urban-dwelling birds tend to be more visible in winter and less active in summer, while rural birds exhibit the opposite pattern.

H1/H2 are used in the following paragraphs.

To evaluate these hypotheses, we adopted a classification framework from existing literature, which categorizes bird species into three groups based on their typical habitat preferences:

• Urban Avoiders – Species predominantly found in rural areas.
• Urban Exploiters – Species predominantly found in urban and suburban areas.
• Urban Adapters – Species commonly observed in both urban/suburban and rural areas.

While these category labels are drawn from prior research, we applied them to our dataset based on the observed distribution patterns of bird species in our own analysis.

If Hypothesis 1 is correct, Urban Adapters—being capable of moving across both environments—should exhibit a seasonal observation pattern similar to the overall trend.

If Hypothesis 2 is correct, Urban Avoiders and Urban Exploiters should show periodic fluctuations aligned with the general rural and urban/suburban trends, respectively.

Note: H1 and H2 are not mutually exclusive.

We plotted observation counts over time for each group and area type:

Figure 4 Numbers of observations of three types of birds over time, 2020-2024

Our results show that Urban Adapters follow a trend closely matching the overall pattern, supporting Hypothesis 1. This suggests that seasonal migration between urban/suburban and rural areas is primarily driven by this group.

We also observed periodic fluctuations in Urban Exploiters, consistent with the urban/suburban trend. This may support Hypothesis 2 or indicate that some migratory species were misclassified as Exploiters.

In contrast, Urban Avoiders did not show a clear seasonal pattern. Although their observation counts fluctuated, the absence of a consistent cycle weakens support for Hypothesis 2 in their case.