Biodiversity in National Parks

Data Science Notebook using U.S. National Park Service data.
Author: E. J. Kluge, 2020, ekluge@ikp.uni-koeln.de

Intro

Due to deforestation, mono agricultures, overhunting or other human disturbances, the natural habitats of flora and fauna are ever shrinking. Thus, multiple species got endangered over the years, finding a habitat only in the protected and confined national parks of their countries. Some species are so severely endangered, they can only to be found in the zoo [1], [2].
The purpose of this project is to investigate the state of the remaining flora and fauna, its diversity and to find specie categories or types that became endangered.

Methodology

For that matter, the bio diversity

• overall,
• per species category (i.e. mammals, reptiles, fish etc.) and
• per national park

is investigated.
Also, the conservation state per species category is investigated.
Since the available data originates form the U.S. National Park Service [3], the investigation is confined to the U.S. only.

Summary & Conclusions

In the investigated habitats, flora live is four times more abundant than fauna. Within the flora category, nine out of ten plants are vascular. Within the animal realm, birds make out half the population, mammals a fifth and the remaining one third is almost evenly divided between the fish, amphibians and reptiles.
Independent of their geographical size, the Bryce national park contains the fewest individuals, while the Yellowstone national park inhabits the most individual beings. In dependence of their U.S. location and provided environment, naturally, the inner-species abundance vary. However, for the investigated category excerpt, the variations in magnitude are smaller than initially expected and no specific fluctuation stands out.
The investigation shows, that mammals are the most threatened, followed by the fish and then birds. Plant species are mostly out of concern.
Finally, the investigated data set reveals a bimodal correlation between plant species observation frequencies and animal species observation frequencies, showing a functional of the food chain and suggesting a symbiosis between plants and animals.

Suggested Proceeding

To fully reflect the actual state of biodiversity on a global scale, data from different countries and more national parks per country could be investigated. By comparing the biodiversity and conservation states by countries, possible patterns or correlations might be found that remain hidden otherwise.

Setup

Library imports

# - Data manipulation -
import numpy as np                                     # numerical python
import pandas as pd                                    # python based analysis for data scientists

import missingno as missing                            # missing data explorer
from sklearn.preprocessing import MinMaxScaler         # data scaler
from scipy.interpolate import interp1d                 # interpolator

# - ML models -
from sklearn.linear_model import LinearRegression
from scipy.stats import pearsonr                       # lin. corr. strength

# - Visualization -
import matplotlib.pyplot as plt                        # plotter
import seaborn as sb                                   # advanced plotter

# - Interactive plots (offline mode) -
import plotly                                          
import plotly.graph_objects as go
import plotly.offline as ply
import cufflinks as cf

# - Local libraries -
import sys
import copy                                           # for deepcopy

# - Misc extension -
from IPython.display import HTML, display
import re

Settings and Parameters

# --Global parameters and settings--

# pandas:
pd.options.display.max_columns = 50
pd.options.display.max_rows = 30

# matplotlib and seaborn:
%matplotlib inline
sb.set_theme(color_codes= True)
sb.set_palette("pastel")

# plotly and cufflinks
plotly.offline.init_notebook_mode(connected=True)
cf.go_offline(connected=True)
cf.set_config_file(theme='white')

# misc:
if 'autoreload' not in get_ipython().extension_manager.loaded:  
    %load_ext autoreload
%autoreload 2;

# parameters:
table_count = 1
figure_count = 1

Data info and processing

Observation data from NPS

# OBVERSATION DATA:
# load data
observ_data = pd.read_csv('observations.csv')

############################################################################

# OBSERVATION data processing:

# check for missing or duplicates
# missing.matrix(observ_data, figsize=(30,5))
# result: no missing or NAN labeled data

# check labels
# result: some labels are to long
# action: drop 'National' from park names
drop_NP = lambda entry : re.sub(' National Park', '', entry)
observ_data.park_name = observ_data.park_name.apply(drop_NP)

# check data types
# display(observ_data.dtypes)
# result: some categorical data has type object
# action: categorical data gets proper data type
observ_data.park_name = observ_data.park_name.astype('category')
observ_data.scientific_name = observ_data.scientific_name.astype('category')

############################################################################

# OBSERVATION DATA INFO:

# OBSERVATION data categories
display(HTML('<h3>Observation Data Categories: %s</h3>' % ', '.join(observ_data.columns).upper()))

# OBSERVATION final data types
observ_types = pd.DataFrame(observ_data.dtypes)
observ_types.reset_index()
observ_types.columns = ['data type']
display(HTML('<h3>Table %d: Observation Data Types</h3>' % table_count))
table_count += 1
display(observ_types)

# OBSERVATION data content:
display(HTML('<h3>Table %d: Observation Data Excerpt</h3>' % table_count))
table_count += 1
display(observ_data)

# OBSERVATION data length:
display(HTML('<h3>Observation Data Length: %d species</h3>' % len(observ_data)))

Observation Data Categories: SCIENTIFIC_NAME, PARK_NAME, OBSERVATIONS

Table 1: Observation Data Types

	data type
scientific_name	category
park_name	category
observations	int64

Table 2: Observation Data Excerpt

	scientific_name	park_name	observations
0	Vicia benghalensis	Great Smoky Mountains	68
1	Neovison vison	Great Smoky Mountains	77
2	Prunus subcordata	Yosemite	138
3	Abutilon theophrasti	Bryce	84
4	Githopsis specularioides	Great Smoky Mountains	85
...	...	...	...
23291	Croton monanthogynus	Yosemite	173
23292	Otospermophilus beecheyi	Bryce	130
23293	Heterotheca sessiliflora ssp. echioides	Bryce	140
23294	Dicranella rufescens	Yosemite	171
23295	Cucurbita pepo	Yosemite	164

23296 rows × 3 columns

Observation Data Length: 23296 species

Species Info Data from NPS

# SPECIES DATA:
# load data
species_data = pd.read_csv('species_info.csv')

############################################################################

# SPECIES DATA PROCESSING:

# check for missing data
# missing.matrix(species_data, figsize=(30,5))      
# result:'conservation_status' has NAN labels.
# action: rename NAN to 'Unknown'
species_data.conservation_status = species_data.conservation_status.fillna('Unknown')

# check for duplicated data
num_species_duplicates = species_data.scientific_name.duplicated().sum()
# result: duplicated 'scientific_name' due to multiple 'common_name' entries
# action: drop of duplicated 'scientific_name' entris
species_data = species_data.drop_duplicates(subset='scientific_name', keep="first")
#display(HTML('<p>Note: Due to multiple "common name" entries, the data had %s duplicate "scientific name" entries. The table was cleaned. Furthermore, the "type" column was added.</p>' % num_species_duplicates))

# check data types:
# species_data.dtypes
# result: some categorical data has type object
# action: categorical data gets proper data type
category_cat = pd.Categorical(species_data.category, categories=['Reptile', 'Amphibian', 'Fish', 'Mammal', 'Bird', 'Nonvascular Plant', 'Vascular Plant'], ordered=False)
species_data['category'] = category_cat
species_data.scientific_name = species_data.scientific_name.astype('category')
conserv_cat = pd.Categorical(species_data.conservation_status, categories=['Unknown', 'Species of Concern', 'In Recovery', 'Threatened', 'Endangered'], ordered=True)
species_data['conservation'] = conserv_cat

# introduction of species types (animal, plant): will come in handy later
PLANTS = ['Vascular Plant', 'Nonvascular Plant']
ANIMALS = ['Amphibian', 'Bird', 'Fish', 'Mammal', 'Reptile']
type_sort = lambda row : 'Animal' if row['category'] in ANIMALS else 'Plant'
species_data['type'] = species_data.apply(type_sort, axis=1)
species_data.type = species_data.type.astype('category')
species_data = species_data[['type', 'category', 'scientific_name', 'common_names', 'conservation']]

############################################################################

# SPECIES DATA INFO:

# SPECIES data categories
display(HTML('<h3>Species Data Categories: %s</h3>' % ', '.join(species_data.columns).upper()))

# SPECIES final data types:
species_datatypes = pd.DataFrame(species_data.dtypes)
species_datatypes.reset_index()
species_datatypes.columns = ['data type']
display(HTML('<h3>Table %d: Species Data Types</h3>' % table_count))
table_count += 1
display(species_datatypes)

# SPECIES data content
display(HTML('<h3>Table %d: Species Data Excerpt</h3>' % table_count))
table_count += 1
display(species_data)

# SPECIES data length:
display(HTML('<h3>Species Data Length: %d species</h3>' % len(observ_data)))

Species Data Categories: TYPE, CATEGORY, SCIENTIFIC_NAME, COMMON_NAMES, CONSERVATION

Table 3: Species Data Types

	data type
type	category
category	category
scientific_name	category
common_names	object
conservation	category

Table 4: Species Data Excerpt

	type	category	scientific_name	common_names	conservation
0	Animal	Mammal	Clethrionomys gapperi gapperi	Gapper's Red-Backed Vole	Unknown
1	Animal	Mammal	Bos bison	American Bison, Bison	Unknown
2	Animal	Mammal	Bos taurus	Aurochs, Aurochs, Domestic Cattle (Feral), Dom...	Unknown
3	Animal	Mammal	Ovis aries	Domestic Sheep, Mouflon, Red Sheep, Sheep (Feral)	Unknown
4	Animal	Mammal	Cervus elaphus	Wapiti Or Elk	Unknown
...	...	...	...	...	...
5819	Plant	Vascular Plant	Solanum parishii	Parish's Nightshade	Unknown
5820	Plant	Vascular Plant	Solanum xanti	Chaparral Nightshade, Purple Nightshade	Unknown
5821	Plant	Vascular Plant	Parthenocissus vitacea	Thicket Creeper, Virginia Creeper, Woodbine	Unknown
5822	Plant	Vascular Plant	Vitis californica	California Grape, California Wild Grape	Unknown
5823	Plant	Vascular Plant	Tribulus terrestris	Bullhead, Caltrop, Goathead, Mexican Sandbur, ...	Unknown

5541 rows × 5 columns

Species Data Length: 23296 species

Merged data

# MERGED DATA:
# group by observations
observ_data_total = observ_data.groupby(observ_data.scientific_name).sum().reset_index()
# merge
merged_data = pd.merge(species_data.drop(columns=['common_names']), observ_data_total, how='left', on=['scientific_name'])
display(HTML('<h3>Table %d: Merged tables with summed observations</h3>' % table_count))
table_count += 1
display(merged_data)

# Merged data encoded:
data_catcode = copy.deepcopy(merged_data)
# create encoding map:
encode_map = list()
encode_map.append(dict(zip(data_catcode.type.values, data_catcode.type.cat.codes)))
encode_map.append(dict(zip(data_catcode.category.values, data_catcode.category.cat.codes)))
encode_map.append(dict(zip(data_catcode.conservation.values, data_catcode.conservation.cat.codes)))
# encode data:
data_catcode.type = data_catcode.type.cat.codes
data_catcode.category = data_catcode.category.cat.codes
data_catcode.conservation = data_catcode.conservation.cat.codes
# index and column sort:
data_catcode = data_catcode.set_index('scientific_name')
data_catcode.columns = ['type', 'category', 'conservation', 'observations']
encode_map = dict(zip(data_catcode.columns[:-1], encode_map))

display(HTML('<h3>Table %d: Merged tables with summed observations (encoded)</h3>' % table_count))
table_count += 1
display(data_catcode)

display(HTML('<h3>Categorical Encoding Map:</h3><p>%s</p>' %encode_map))

Table 5: Merged tables with summed observations

	type	category	scientific_name	conservation	observations
0	Animal	Mammal	Clethrionomys gapperi gapperi	Unknown	615
1	Animal	Mammal	Bos bison	Unknown	542
2	Animal	Mammal	Bos taurus	Unknown	514
3	Animal	Mammal	Ovis aries	Unknown	542
4	Animal	Mammal	Cervus elaphus	Unknown	1218
...	...	...	...	...	...
5536	Plant	Vascular Plant	Solanum parishii	Unknown	574
5537	Plant	Vascular Plant	Solanum xanti	Unknown	575
5538	Plant	Vascular Plant	Parthenocissus vitacea	Unknown	583
5539	Plant	Vascular Plant	Vitis californica	Unknown	562
5540	Plant	Vascular Plant	Tribulus terrestris	Unknown	556

5541 rows × 5 columns

Table 6: Merged tables with summed observations (encoded)

	type	category	conservation	observations
scientific_name
Clethrionomys gapperi gapperi	0	3	0	615
Bos bison	0	3	0	542
Bos taurus	0	3	0	514
Ovis aries	0	3	0	542
Cervus elaphus	0	3	0	1218
...	...	...	...	...
Solanum parishii	1	6	0	574
Solanum xanti	1	6	0	575
Parthenocissus vitacea	1	6	0	583
Vitis californica	1	6	0	562
Tribulus terrestris	1	6	0	556

5541 rows × 4 columns

Categorical Encoding Map:

{'type': {'Animal': 0, 'Plant': 1}, 'category': {'Mammal': 3, 'Bird': 4, 'Reptile': 0, 'Amphibian': 1, 'Fish': 2, 'Vascular Plant': 6, 'Nonvascular Plant': 5}, 'conservation': {'Unknown': 0, 'Species of Concern': 1, 'Endangered': 4, 'Threatened': 3, 'In Recovery': 2}}

Data Analysis (EDA)

Biodiversity Overall

# BIODIVERSITY
species_types = merged_data.type.unique()
display(HTML('<h3>Species Types: %s</h3>' % ', '.join(species_types)))
species_categories = merged_data.category.unique()
display(HTML('<h3>Species Categories: %s</h3>' % ', '.join(species_categories)))

# species per category:
species_count = merged_data.groupby('category', axis=0).count()
species_count = species_count.type.sort_values(ascending=False).reset_index()
species_count.columns = ['category', 'species_count']
display(HTML('<h3>Table %d: Number of Species per Category</h3>' % table_count))
table_count += 1
display(species_count)

Species Types: Animal, Plant

Species Categories: Mammal, Bird, Reptile, Amphibian, Fish, Vascular Plant, Nonvascular Plant

Table 7: Number of Species per Category

	category	species_count
0	Vascular Plant	4262
1	Bird	488
2	Nonvascular Plant	333
3	Mammal	176
4	Fish	125
5	Amphibian	79
6	Reptile	78

# Sum and display species types and categories:
# -- TYPES --
animal_plant_data = merged_data.groupby('type').sum().reset_index()
species_colors = ["#ca684d", "#7ba24f"]
#display(animal_plant_data)

# -- ANIMALS --
animal_data = merged_data[merged_data.type == 'Animal'].groupby('category').sum().reset_index()
animal_data = animal_data.drop(animal_data.index[-2:])
animal_colors = ["#e6b8b3", "#9bcde5", "#d6d8b4", "#cdbedd", "#aad2bf"]
#display(animal_data)

# -- PLANTS --
plant_data = merged_data[merged_data.type == 'Plant'].groupby('category').sum().reset_index()
plant_data = plant_data.drop(plant_data.index[0:5])
plant_colors = ["#d7da5b", "#81e799"]
#display(plant_data)


display(HTML('<h3>Figure %d: Biodiversity per Category</h3>' % figure_count))
figure_count += 1
fig = plt.figure(figsize=(15,5))
ax1 = plt.subplot(1,3,1)
plt.pie(x=animal_plant_data.observations, autopct="%.0f%%", pctdistance=0.7, explode=[0.05]*2, startangle=0, colors=species_colors, labels=animal_plant_data.type)
ax1.axis('equal')
ax1.set_title('Species Categories')
ax2 = plt.subplot(1,3,2)
plt.pie(x=plant_data.observations, autopct="%.0f%%", pctdistance=0.7, explode=[0.05]*2 , startangle=100, colors=plant_colors, labels=plant_data.category)
ax2.axis('equal')
ax2.set_title('Plant Categories')
ax3 = plt.subplot(1,3,3)
plt.pie(x=animal_data.observations, autopct="%.0f%%", pctdistance=0.7, explode=[0.05]*5 , startangle=-30, colors=animal_colors, labels=animal_data.category)
ax3.axis('equal')
ax3.set_title('Animal Categories')
plt.show()
plt.clf()

Figure 1: Biodiversity per Category

<Figure size 432x288 with 0 Axes>

Figure Analysis

The pie charts show, that flora live is four times more abundant than fauna live (right pie).
Within the flora realm, almost nine out of ten are vascular plants (center pie).
Within the animal realm, birds make out 50% of the population, mammals about 20% and the remaining 30% is almost evenly divided between the fish, amphibians and reptiles.

Biodiversity per National Park

# BIODIVERSITY PER NATIONAL PARK
NATIONAL_PARKS = observ_data.park_name.unique()
display(HTML('<h3>National Parks: %s</h3>' % ', '.join(NATIONAL_PARKS)))

# Counts per park pivoted:
observ_park = observ_data.pivot_table(index='scientific_name', columns='park_name', values='observations', aggfunc=[sum], margins=True, margins_name='Total Count')
display(HTML('<h3>Table %d: Observed Spiecies per National Park</h3>' % table_count))
table_count += 1
display(observ_park)

display(HTML('<h3>Figure 1: Observed Individuals per National Park</h3>'))
temp_frame = pd.DataFrame([NATIONAL_PARKS,[observ_park[('sum', park)][-1] for park in NATIONAL_PARKS]], columns=['key', 'value'])
#display(temp_frame)
sb.barplot(x=temp_frame.value, y=temp_frame.key, data=temp_frame)
plt.xlabel('Individual Counts (millions)')
plt.ylabel('National Park')
plt.title('Flora and Founa Count per National Park')
plt.show()
plt.clf()

National Parks: Great Smoky Mountains, Yosemite, Bryce, Yellowstone

Table 8: Observed Spiecies per National Park

	sum
park_name	Bryce	Great Smoky Mountains	Yellowstone	Yosemite	Total Count
scientific_name
Abies bifolia	109	72	215	136	532
Abies concolor	83	101	241	205	630
Abies fraseri	109	81	218	110	518
Abietinella abietina	101	65	243	183	592
Abronia ammophila	92	72	222	137	523
...	...	...	...	...	...
Zonotrichia leucophrys oriantha	73	123	227	135	558
Zonotrichia querula	105	83	268	160	616
Zygodon viridissimus	100	71	270	159	600
Zygodon viridissimus var. rupestris	102	102	237	210	651
Total Count	576025	431820	1443562	863332	3314739

5542 rows × 5 columns

Figure 1: Observed Individuals per National Park

<Figure size 432x288 with 0 Axes>

Figure Analysis

Independent of their provided land size, the Bryce national park contains the fewest individuals, while the Yellowstone national park inhabits the most beings.

# merging of species info and counts per park
temp_dict = {'scientific_name':observ_park.index[:-1].values}
for park in NATIONAL_PARKS:
    temp_dict[park] = observ_park[('sum', park)][:-1].values
counts_per_park = pd.DataFrame(data=temp_dict)
merged_counts_parks = pd.merge(species_data.drop(columns=['type', 'common_names', 'conservation']), counts_per_park, how='left', on=['scientific_name'])
#display(merged_counts_parks)

# separation in categories for park diversity
park_palette = sb.color_palette("pastel", 4)
park_colors = dict(zip(NATIONAL_PARKS, park_palette))
park_dict = {'1':'Bryce', '0':'Great Smoky Mountains', '2':'Yosemite', '3':'Yellowstone'}

def return_zero(x):
    return 0

# Species < 500 tables:
selected_types = ['Reptile', 'Amphibian', 'Mammal', 'Bird']
type_counts_park = dict()
type_counts_interpol = dict()
for species_type in selected_types:
    display(HTML('<h3>Figure {figure_number}: {type_string} Species Distribution per Park:</h3>'.format(figure_number=figure_count, type_string=species_type)))
    figure_count += 1
    # Tables:
    type_counts_park[species_type] = merged_counts_parks[merged_counts_parks.category == species_type].reset_index(drop=True)
    type_counts_park[species_type].scientific_name = type_counts_park[species_type].scientific_name.astype('category')
    type_counts_park[species_type].insert(loc=1, column='name_id', value=type_counts_park[species_type].scientific_name.cat.codes)
    type_counts_park[species_type] = type_counts_park[species_type].drop(columns=['scientific_name']) 
    # Figures:
    x_length = len(type_counts_park[species_type].name_id)
    x_new = np.linspace(0, x_length-1, num=x_length*3, endpoint=True)
    type_counts_interpol[species_type] = [return_zero]
    plt.figure(figsize=(20,4))
    plt.title('%s Species Distribution per National Park' % species_type)
    plt.xlabel('Species (Encoded)')
    plt.ylabel('Species Count')
    for i in range(4):
        type_counts_interpol[species_type].append(interp1d(type_counts_park[species_type].name_id, type_counts_park[species_type][park_dict[str(i)]], kind='cubic'))
        plt.plot(x_new, type_counts_interpol[species_type][i+1](x_new), '-', color=park_colors[park_dict[str(i)]], label=park_dict[str(i)])
        plt.fill_between(x_new, type_counts_interpol[species_type][i](x_new), type_counts_interpol[species_type][i+1](x_new), color=park_colors[park_dict[str(i)]])
        plt.legend()
    plt.show()
    plt.clf()

Figure 2: Reptile Species Distribution per Park:

Figure 3: Amphibian Species Distribution per Park:

<Figure size 432x288 with 0 Axes>

Figure 4: Mammal Species Distribution per Park:

<Figure size 432x288 with 0 Axes>

Figure 5: Bird Species Distribution per Park:

<Figure size 432x288 with 0 Axes>

<Figure size 432x288 with 0 Axes>

Figure Analysis

In dependence of the provided environment, naturally, when divided into categories, the inner-species abundance vary. However, there is no magnitude fluctuation, that stands out.

Species Conservation states

# CONSERVATION STATES:
species_states = copy.deepcopy(species_data[['category', 'scientific_name', 'conservation']])
species_states['threat_level'] = data_catcode.conservation.values

conservation_labels = species_states.conservation.unique()
display(HTML('<h3>Species Conservation Categories: %s</h3>' % ', '.join(conservation_labels)))

# conservation status counts:
conservation_counts = species_states.groupby(['conservation', 'threat_level']).count().sort_values(by='conservation', ascending=False).reset_index().drop(columns=['scientific_name']).dropna()
conservation_counts.columns = ['conservation', 'threat_level', 'species count']
display(HTML('<h3>Table %d: Species Conservation State</h3>' % table_count))
table_count += 1
display(conservation_counts)
display(HTML('<p>Note: Here, the threat level" respectively, is the orderly encode of "conservation". The order is based on the "IUCN Red List of Threatened Species".</p>'))


# pivoted table:
species_pivot_table = species_states.pivot_table(index=['category', 'scientific_name'], columns='conservation', values='threat_level', aggfunc=[max], margins=True, margins_name='Threat Level')
display(HTML('<h3>Table %d: Spiecies Conservation states overview</h3>' % table_count))
table_count += 1
display(species_pivot_table)

Species Conservation Categories: Unknown, Species of Concern, Endangered, Threatened, In Recovery

Table 9: Species Conservation State

	conservation	threat_level	species count
0	Endangered	4	15.0
6	Threatened	3	9.0
12	In Recovery	2	3.0
18	Species of Concern	1	151.0
24	Unknown	0	5363.0

Note: Here, the threat level" respectively, is the orderly encode of "conservation". The order is based on the "IUCN Red List of Threatened Species".

Table 10: Spiecies Conservation states overview

		max
	conservation	Unknown	Species of Concern	In Recovery	Threatened	Endangered	Threat Level
category	scientific_name
Reptile	Agkistrodon contortrix mokasen	0.0	NaN	NaN	NaN	NaN	0
	Anolis carolinensis carolinensis	0.0	NaN	NaN	NaN	NaN	0
	Apalone spinifera spinifera	0.0	NaN	NaN	NaN	NaN	0
	Aspidoscelis tigris munda	0.0	NaN	NaN	NaN	NaN	0
	Carphophis	0.0	NaN	NaN	NaN	NaN	0
...	...	...	...	...	...	...	...
Vascular Plant	Zigadenus venenosus var. venenosus	0.0	NaN	NaN	NaN	NaN	0
	Zizia aptera	0.0	NaN	NaN	NaN	NaN	0
	Zizia aurea	0.0	NaN	NaN	NaN	NaN	0
	Zizia trifoliata	NaN	1.0	NaN	NaN	NaN	1
Threat Level		0.0	1.0	2.0	3.0	4.0	4

5542 rows × 6 columns

# distill observ over conserv data
observ_conserv_data = copy.deepcopy(merged_data[['category', 'observations']])
observ_conserv_data['conservation'] = data_catcode.conservation.values

display(HTML('<h3>Figure %d: Species Observations per Category over Conservation Status</h3>' % figure_count))
figure_count += 1
ax = sb.lmplot(data=observ_conserv_data, x='conservation', y='observations', col='category', col_wrap=3, x_estimator=np.mean)
plt.xticks([x for x in range(5)])
plt.xlim(-0.2,4.2)
ax.set_xticklabels(['Unknown', 'Concern', 'Recov.', 'Threat.', 'Endang.'])
plt.show()
plt.clf()

Figure 6: Species Observations per Category over Conservation Status

<Figure size 432x288 with 0 Axes>

Figure Analysis

The above figures show, that the conservation state for reptiles and non-vascular plants is healthy at the moment. However, to discearn the status by numbers and not visuals, a correlation investigation has been done and statistics were calculated:

Correlation investigation

# Category to conservation correlation:
display(HTML('<h3>Table %d: Correlations</h3>' % table_count))
table_count += 1
correlation_frame = data_catcode.groupby('observations').mean().reset_index()
correlation_frame = correlation_frame.corr()
display(correlation_frame)

Table 11: Correlations

	observations	type	category	conservation
observations	1.000000	0.081908	0.156756	-0.415458
type	0.081908	1.000000	0.922266	-0.507811
category	0.156756	0.922266	1.000000	-0.550609
conservation	-0.415458	-0.507811	-0.550609	1.000000

Table Analysis

While "species type to species category" and "observation to conservation" correlate by design, possible correlations exist for:

• species category and conservation and/or
• species type and conservation.

#conservation medians
median_frame = data_catcode.groupby(data_catcode.conservation).median().reset_index()
median_frame = median_frame[['type', 'category', 'conservation', 'observations']]
#display(median_frame)


def return_key(key_code, key_chain):
    for key in key_chain.keys():
        if key_chain[key] == key_code:
            return key
median_frame_unencode = copy.deepcopy(median_frame)
for entity in encode_map:
    entity_lambda = lambda row : return_key(row[entity], encode_map[entity])
    median_frame_unencode[entity] = median_frame_unencode.apply(entity_lambda, axis=1)

display(HTML('<h3>Table %d: Medians (unencoded)</h3>' % table_count))
table_count += 1
display(median_frame_unencode)

display(HTML('<h3>Figure %d: Spiecies Category over Threat Level</h3>' % figure_count))
figure_count += 1
sb.set_palette('coolwarm')
ax = sb.barplot(x=median_frame.conservation, y=median_frame_unencode.category, ci=None)
plt.xticks([x for x in range(5)])
plt.xlim(-0.2,4.2)
ax.set_xticklabels(['Unknown', 'Concern', 'Recov.', 'Threat.', 'Endang.'])
ax.set_title('Spiecies Category Threat Level')
plt.show()
plt.clf()
sb.set_palette('pastel')

Table 12: Medians (unencoded)

	type	category	conservation	observations
0	Plant	Vascular Plant	Unknown	575
1	Animal	Bird	Species of Concern	515
2	Animal	Bird	In Recovery	465
3	Animal	Fish	Threatened	278
4	Animal	Mammal	Endangered	146

Figure 7: Spiecies Category over Threat Level

<Figure size 432x288 with 0 Axes>

Figure Analysis

The above figure shows, that mammals are most threatened, then fish and finally the birds. Plant species are mostly out of concern.

ratio_dict = {'category':[], 'threat_percent':[]}
for species_cat in species_categories:
    temp_data = merged_data[merged_data.category == species_cat]
    total_count = temp_data.scientific_name.count()
    endangered_count = temp_data[temp_data.conservation.isin(['Endangered', 'Threatened'])].scientific_name.count()
    ratio_dict['category'].append(species_cat)
    ratio_dict['threat_percent'].append(round(endangered_count * 100.0 / total_count,1))
ratio_data = pd.DataFrame(ratio_dict)
ratio_data = ratio_data.sort_values(by='threat_percent', ascending=False)
display(ratio_data)

	category	threat_percent
4	Fish	4.8
0	Mammal	4.5
3	Amphibian	3.8
1	Bird	0.8
5	Vascular Plant	0.1
2	Reptile	0.0
6	Nonvascular Plant	0.0

Additional investigation

# Spieces count frequencies:
animal_freq = merged_data[merged_data.type == 'Animal'][['scientific_name', 'observations']].reset_index().drop(columns='index')
plant_freq = merged_data[merged_data.type == 'Plant'][['scientific_name', 'observations']].reset_index().drop(columns='index')

display(HTML('<h3>Figure %d: Species Count Frequencies</h3>' % figure_count))
figure_count += 1
plt.figure(figsize=(20,8))
sb.histplot(data=plant_freq, x='observations', kde=True, label='Plants', color=species_colors[1])
sb.histplot(data=animal_freq, x='observations', kde=True, label='Animals', color=species_colors[0])
plt.xlim(0,1400)
plt.xlabel('Species Observations')
plt.ylabel('Frequency')
plt.title('Species Ovservations Frequency')
plt.legend()
plt.show()
plt.clf()

Figure 8: Species Count Frequencies

<Figure size 432x288 with 0 Axes>

Figure Analysis

The congruence of plant and animal species observations frequency is striking. Both are bimodal with matching peaks at about 580 and 1150 observations. This correlation of plant and animal spieces occurences is probably highlighting the symbiosis of plants and animals and the food chain.

References

1. Wikipedia, "Conservation status", 2021, Website
2. IUCN, "Red List of Threatened Species", 2021, Website

Data Sources

• U.S. National Park Service (NPS): Tools and Data