Skip to content

Latest commit

 

History

History
705 lines (471 loc) · 22.9 KB

ialeuk25_analysis.md

File metadata and controls

705 lines (471 loc) · 22.9 KB
Raw
title author output linkcolor citecolor urlcolor
analysis of ialeUK conference proceedings
James Millington
html_document pdf_document word_document
number_sections toc code_folding df_print keep_md
true
true
hide
paged
true
highlight number_sections toc
tango
true
true
toc
true
red
cyan
blue
 
#Load Data
#(After slightly cleaning column titles - in future include code to do that here)

rm(list=ls())
library(tidyverse)
library(ggplot2)
path <- "C:/Users/k1076631/Google Drive/Research/Papers/InProgress/ialeUK_25years/QuantAnalysis/Rproject"
setwd(path)
filename <- "abstract_review_export_2018-06-11.csv"
cpdata <- read_csv(filename)

This document contains analysis by year. Future analysis could examine contribution attributes by:

  • author affiliation (e.g. do NGOs conduct studies at particular scales?)
  • landscape type (e.g. what species do studies in Urban landscapes focus on?)
  • species (e.g. are birds studied more using empirical studies or GIS?)

etc.

#spec(cpdata)

yrdata <- cpdata %>%
  select_if(is.numeric) %>%
  group_by(`Conference Year`) %>%
  summarise_all(sum, na.rm=T)

Total Conference Contributions

Quick observations:

  • general increase through time to early 2000s then drop but steady through 2010s
authorCounts <- yrdata %>%
  select(`Conference Year`,Academic, Government,NGO,Business,Private) %>%
  mutate(yrsum = rowSums(.[2:6])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum)  #calculate proportion

ggplot(authorCounts, aes(x=`Conference Year`, y=count)) + geom_bar(stat="identity")

Analysis by Conference Year

Stacked bar plots of contributions (by types and year)

Author Affiliation

Quick observations:

  • Academic contributors generally dominate
  • Government contributors have decreased through time
  • NGO attendance has replaced declines in Government? (could check sum of Gov + NGO through time)
authorCounts <- yrdata %>%
  select(`Conference Year`,Academic, Government,NGO,Business,Private) %>%
  mutate(yrsum = rowSums(.[2:6])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum)  #calculate proportion


ggplot(authorCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(authorCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(authorCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Landscape Type

Quick observations:

  • Lowland rural generaly dominates (but lesser contribution in later years)
  • Spikes in some years for types (corresponding to special themes)
  • Urban and Seascape both appear for first time in 1998; urban then constant presence, but seascape more variable until recent years
lspCounts <- yrdata %>%
  select(`Conference Year`,`Upland rural`, `Lowland rural`, Urban,	Riverscape, Seascape, `Undefined LspType`,Other) %>%
  mutate(yrsum = rowSums(.[2:8])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 


ggplot(lspCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(lspCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(lspCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Species

Quick observations:

  • no clear patterns
  • some years contain no Generic Habitat - is this real or a data entry issue?
sppCounts <- yrdata %>%
  select(`Conference Year`,Mammals, Humans, Birds, Reptiles, Inverts, Plants, Amphibians, Fish, `Generic Habitat`,`Woodland Forests`) %>%
  mutate(yrsum = rowSums(.[2:11])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(sppCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(sppCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(sppCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Methods

Quick observations:

  • empirical studies have decreased through time
  • GIS and qualitative have increased through time
  • Quantitative and theoretical quite steady through time (although theoretical does seem to have reduced after initial years)
methodsCounts <- yrdata %>%
  select(`Conference Year`, Empirical, Theoretical, Qualitative, Quantitative, GIS, `Remote sensing`) %>%
  mutate(yrsum = rowSums(.[2:7])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(methodsCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(methodsCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(methodsCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Spatial Extent

Quick observations:

  • no clear trends?
  • Global studies only appear from 2014 onwards
extentCounts <- yrdata %>%
  select(`Conference Year`, Micro, Mini, Local, Regional, National, Continental, Global,`Undefined Extent`) %>%
  mutate(yrsum = rowSums(.[2:9])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(extentCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(extentCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(extentCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Temporal Extent

Quick observations:

  • most studies have undefined temporal duration
  • those that do are dominated by studies over decades and years

Concepts

Quick observations:

  • Ecosystem services appear from 1998 and have grown recently
  • climate change interactions have only become common recently (since 2008)
  • 'Scale and scaling' and 'connectivity and fragmentation seem to have decreased oin recent years
  • LUCC and Spatial Analysis are mainstays throughout
conceptCounts <- yrdata %>%
  select(`Conference Year`, `PPS of landscapes`,
`Connectivity and fragmentation`, `Scale and scaling`,`Spatial analysis and modeling`,LUCC,`History and legacy`,`Climate change interactions`,`Ecosystem services`,`Landscape sustainability`,`Accuracy and uncertainty`
) %>%
  mutate(yrsum = rowSums(.[2:11])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(conceptCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(conceptCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(conceptCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Other Concepts

Quick observations:

  • socio-economic studies have increased through time
  • biodiversity has decreased through time
  • Landscape management and Biodiversity peak in early 2000s
othCCounts <- yrdata %>%
  select(`Conference Year`, `Green Infrastructure`,`Planning and Architecture`,`Management and Conservation`,`Cultural Landscapes`,`Socio-economic Dimensions`,Biodiversity,`Landscape Assessment`,`Catchment Based Approach`,`Invasives Pests Diseases`
) %>%
  mutate(yrsum = rowSums(.[2:10])) %>%
  gather(key = Type, value = count, -`Conference Year`, -yrsum) %>%
  mutate(prop = count / yrsum) 

ggplot(othCCounts, aes(x=`Conference Year`, y=count)) + geom_line(aes(colour=Type))

ggplot(othCCounts, aes(x=`Conference Year`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(othCCounts, aes(x=`Conference Year`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Analysis by Author Affiliation

#spec(cpdata)

affildata <- cpdata %>%
  select_if(is.numeric) %>%
  gather(key = Affiliation, value = count, Academic:Private) %>%
  filter(count > 0) %>%
  group_by(`Affiliation`) %>%
  summarise_all(sum, na.rm=T)

Total Conference Contributions

Quick observations:

  • Academic contributors dominate, followed by Government (but as shown above, Government contributions have decreased recently, replaced by NGOs)
lspACounts <- affildata %>%
  select(`Affiliation`,`Upland rural`, `Lowland rural`, Urban,	Riverscape, Seascape, `Undefined LspType`,Other) %>%
  mutate(Asum = rowSums(.[2:8])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(lspACounts, aes(x=`Affiliation`, y=count)) + geom_bar(stat="identity")

Landscape Type

Stacked bar plots of contributions (by types and author affiliation)

Using all landscape types

Quick observations:

  • Business not good at reporting landscape type!
  • Private have greatest proportions of Seascape and Other
ggplot(lspACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(lspACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Without 'Undefined LspType' and 'Other' landscape types

Quick observations:

  • Government has greatest proportion of Upland Rural
  • Business has greatest Urban proportion and smallest Lowland Rural proportion
  • Academic dominates total number of all landscape types (with possible exception of Upland Rural)
lspACounts <- affildata %>%
  select(`Affiliation`,`Upland rural`, `Lowland rural`, Urban,	Riverscape, Seascape) %>%
  mutate(Asum = rowSums(.[2:6])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(lspACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(lspACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Species

Quick observations

  • Academic seem to be majority by absolute number for all species
  • Business and Private have greatest proportions of Generic Habitat
  • NGOs have greatest proportion of Birds (RSPB?)
speciesACounts <- affildata %>%
  select(`Affiliation`,Mammals, Humans, Birds, Reptiles, Inverts, Plants, Amphibians, Fish, `Generic Habitat`,`Woodland Forests`) %>%
  mutate(Asum = rowSums(.[2:11])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(speciesACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(speciesACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Methods

Quick observations

  • Academic seem to be majority by absolute number for all methods
  • Business obviously lower proportion of empirical studies (expensive?), substituted by GIS and qualitative
  • Government has smallest proportion of qualitative
  • Private has greatest proprtion of theoretical, no RS and relatively little GIS (technical training?)
methodsACounts <- affildata %>%
  select(`Affiliation`,Empirical, Theoretical, Qualitative, Quantitative, GIS, `Remote sensing`) %>%
  mutate(Asum = rowSums(.[2:7])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(methodsACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(methodsACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Spatial Extent

Quick observations

  • Academic seem to be majority by absolute number for all extents
  • Business have largest proportion of Global and National studies, with smallest proprtion of Local studies
  • Private has larest proportion of Local and Mini studies (cost-related and given no RS and few GIS studies?)
  • Academic: decreasing proportion Local -> Regional -> National -> Global
  • Government: greater proportion of National than Regional
spatialACounts <- affildata %>%
  select(`Affiliation`,Micro, Mini, Local, Regional, National, Continental, Global,`Undefined Extent`) %>%
  mutate(Asum = rowSums(.[2:9])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(spatialACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(spatialACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Temporal Extent

Quick observations

  • Vast majority of all affiliations did not list temporal extent of the study Academic seem to be majority by absolute number for all methods
  • Not much more of interest here...
temporalACounts <- affildata %>%
  select(`Affiliation`,Hours, Days, Weeks, Months, Years, Decades, Centuries, Longer, `Undefined Temporal`) %>%
  mutate(Asum = rowSums(.[2:10])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(temporalACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(temporalACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Concepts

Quick observations

  • Academic seem to be majority by absolute number for all extents
  • Business have greatest proportions of climate change and ecosystem services, less interested in history and legacy
  • All other affiliations reasonably similar in terms of proportions
conceptACounts <- affildata %>%
  select(`Affiliation`,`PPS of landscapes`,
`Connectivity and fragmentation`, `Scale and scaling`,`Spatial analysis and modeling`,LUCC,`History and legacy`,`Climate change interactions`,`Ecosystem services`,`Landscape sustainability`,`Accuracy and uncertainty`) %>%
  mutate(Asum = rowSums(.[2:11])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(conceptACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(conceptACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Other Concepts

Quick observations

  • Academic seem to be majority by absolute number for all extents
  • Government and NGOs have greater proportion of Management and Conservation than Academic
  • Private low on biodiversity but higher on cultural landscapes, landscape assessment and planning
oconceptACounts <- affildata %>%
  select(`Affiliation`,`Green Infrastructure`,`Planning and Architecture`,`Management and Conservation`,`Cultural Landscapes`,`Socio-economic Dimensions`,Biodiversity,`Landscape Assessment`,`Catchment Based Approach`,`Invasives Pests Diseases`) %>%
  mutate(Asum = rowSums(.[2:10])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -`Affiliation`, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(oconceptACounts, aes(x=`Affiliation`, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(oconceptACounts, aes(x=`Affiliation`, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Analysis by Landscape Type

#spec(cpdata)

lspdata <- cpdata %>%
  select_if(is.numeric) %>%
  gather(key = LspType, value = count, `Upland rural`:Other) %>%
  filter(count > 0) %>%
  group_by(`LspType`) %>%
  summarise_all(sum, na.rm=T)

Quick observations:

  • Lowland rural dominate, followed by 'undefined' and Upland rural
AlspCounts <- lspdata %>%
  select(LspType,Academic, Government,NGO,Business,Private) %>%
  mutate(Asum = rowSums(.[2:6])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -LspType, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(AlspCounts, aes(x=LspType, y=count)) + geom_bar(stat="identity")

Author Affiliation

Quick observations:

  • Academic are majority of all landscape types, with possible exception of Upland rural (Government?)
ggplot(AlspCounts, aes(x=LspType, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(AlspCounts, aes(x=LspType, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Species

Quick observations

  • Animal types quite evenly distributed across Lowland rural
  • Humans are large contributor to seascape studies (possibly by absolute number as well as relative)
  • Generic habitat is large contributor across all landscape types
specieslspCounts <- lspdata %>%
  select(LspType,Mammals, Humans, Birds, Reptiles, Inverts, Plants, Amphibians, Fish, `Generic Habitat`,`Woodland Forests`) %>%
  mutate(Asum = rowSums(.[2:11])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -LspType, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(specieslspCounts, aes(x=LspType, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(specieslspCounts, aes(x=LspType, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Methods

Quick observations

  • 'Undefined landscape' studies are largely theoretical
  • Lowland rural largely studies using empirical and quantitative methods
  • Seascape studies have largest proportion of qualitative methods
methodslspCounts <- lspdata %>%
  select(LspType,Empirical, Theoretical, Qualitative, Quantitative, GIS, `Remote sensing`) %>%
  mutate(Asum = rowSums(.[2:7])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -LspType, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(methodslspCounts, aes(x=LspType, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(methodslspCounts, aes(x=LspType, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

Spatial Extent

Quick observations

  • Urban landscape studies are dominated by Local scale analysis
  • Upland rural have larger proportion of national studies than Lowland rural
spatiallspCounts <- lspdata %>%
  select(LspType,Micro, Mini, Local, Regional, National, Continental, Global,`Undefined Extent`) %>%
  mutate(Asum = rowSums(.[2:9])) %>%   #calculate total for subsquent calcultation of proportion
  gather(key = Type, value = count, -LspType, -Asum) %>%
  mutate(prop = count / Asum)  #calculate proportion

ggplot(spatiallspCounts, aes(x=LspType, y=count, fill=Type)) + geom_bar(stat="identity", colour="white")

ggplot(spatiallspCounts, aes(x=LspType, y=prop, fill=Type)) + geom_bar(stat="identity", colour="white")

More here