FRST90015 Landscape Ecology - Practical 3: Connectivity and Conservation

Connectivity and Conservation: Student Notebook

Exercise 1: Code and plots

#Step 2 code

all_binary <- list.files(path = 'data/raw',pattern = "^Binary.*.asc$", full.names = TRUE)
all_binary <- lapply(all_binary, raster)
binary_stack <- do.call(stack, all_binary)

#Step 2 code

spplot(binary_stack)

Figure 1. Spatial configuration of binary habitats for each of the 9 landscapes

#Alternative approach for adding figure captions and text
#library("grid")
#grid.text("Figure 1 test", x=unit(0.9, "npc"), y=unit(0.1, "npc"))

#Step 3 code

binary_curmap <- list.files(path = "data/landscapes", pattern = "*curmap.asc", full.names = TRUE)
binary_curmap <- lapply(binary_curmap, raster)
curmap_stack <- do.call(stack, binary_curmap)

#Step 3 code

spplot(curmap_stack, col=plasma(30))

Figure 2a. Cumulative current of binary habitats for each of the 9 landscapes

#Step 3 code

install.packages("RColorBrewer")

## Installing package into '/cloud/lib/x86_64-pc-linux-gnu-library/4.1'
## (as 'lib' is unspecified)

library(RColorBrewer)
#display.brewer.all(colorblindFriendly = TRUE)

plot(curmap_stack, col=brewer.pal(n = 11, name = 'RdBu'))

Figure 2b. Cumulative current of binary habitats for each of the 9 landscapes

#Step 4 code

res<-read.csv('data/resistances.csv')
head(res)

##   Node1 Node2 Resistance Landscape
## 1     1     2         -1        10
## 2     1     3         -1        10
## 3     1     4         -1        10
## 4     1     5         -1        10
## 5     2     3         -1        10
## 6     2     4         -1        10

#Step 4 code

res_ex1<-res%>%filter(!Landscape %in% c('sp_a', 'sp_b', 'sp_c'))%>%
  mutate(res_fix=ifelse(Resistance == -1, 50, Resistance), flow = paste0('N',Node1,'-N',Node2))%>%
  droplevels()

head(res_ex1)

##   Node1 Node2 Resistance Landscape res_fix  flow
## 1     1     2         -1        10      50 N1-N2
## 2     1     3         -1        10      50 N1-N3
## 3     1     4         -1        10      50 N1-N4
## 4     1     5         -1        10      50 N1-N5
## 5     2     3         -1        10      50 N2-N3
## 6     2     4         -1        10      50 N2-N4

#Step 4 code

ggplot(res_ex1, aes(Landscape, res_fix)) +
  stat_summary(geom = "bar", fun = mean, position = "dodge")+
  stat_summary(geom = "errorbar", fun.data = mean_se, position = "dodge")+
  labs(x = 'Landscape', y = 'Resistance')

Figure 3. Relationship between proportion of the landscape that is habitat and average resistance

#Step 5 code


ggplot(data=res_ex1, aes(x=Landscape, y=res_fix, group=as.factor(flow), color = as.factor(flow))) +
  theme_set(theme_bw())+
  geom_point()+
  geom_line(lwd = 1.5)+
  labs(x = 'Landscape', y = 'Resistance', color = 'Flow')

Figure 4. Relationship between proportion of the landscape that is habitat and resistance for each patch-to-patch pairwise comparison

#Step 6 code

connect<-res_ex1%>%group_by(Landscape)%>%tally(Resistance>0)

ggplot(connect,(aes(Landscape, n)))+
  stat_summary(geom = "bar", fun = max, position = "dodge")+
  labs(x = 'Landscape', y = 'Number of connections')

Figure 5. Relationship between number of connected habitat patches and proportion of the landscape that is habitat

#Step 6 code

write.csv(res_ex1, 'outputs/landscape_resistances.csv')
write.csv(connect, 'outputs/landscape_connections.csv')

Exercise 1: Questions

1. How is resistance related to the proportion of the landscape that is habitat? Describe the relationship.

As the proportion of the landscape that is habitat increases, the resistance decreases. This is because under this model, habitat is required for movement of organisms and resistance is being defined as a proxy for the restriction of movement of organisms within a landscape and is being used to determine how well core habitat patches are connected.

The resistance across a landscape matrix will determine the changes in suitable habitat availability that an organism may encounter as it attempts to travel between habitat patches. Suitable habitats that are of low availability for an organism (and therefore restrict their movement, represented by restricted flow of current) are represented by high resistance vales. Suitable habitats that are of high availability for an organism (and therefore facilitate their movement, represented by increased flow of current) are represented by low resistance values.

As the proportion of the landscape that is suitable, available habitat increases, the facilitation of movement of an organisms between core habitat patches increases (represented by increased flow of current), and resistance decreases.

Figure 1 demonstrates the binary stacks which shows that increasing proportions of habitat landscape (10 – 90%) results in less habitat fragmentation and isolation of the core habitat patches.

Figure 3 demonstrates the relationship of resistance decreasing with increasing proportion of landscape that is classed as suitable habitat. It also shows that the relationship between resistance and habitat is not linear but follows a truncated normal distribution curve shape. It shows that the resistance in a landscape remains at its maximum (maximum resistance = 50) if the proportion of suitable habitat is between 10 – 30%, demonstrating that the landscape must have at least 40% suitable habitat for facilitation of movement by an organism. The non-linear nature of the relationship also shows that resistance decreases dramatically once the proportion of the landscape that is suitable habitat reaches 50%, indicating that maintaining a proportion of suitable habitat at 50% of the landscape will significantly assist the movement of the organisms.

Figure 4 shows the specific resistance flows with increasing proportions of landscape that is suitable habitat. This shows that the relationship of resistances is not uniform across the landscape, but spatially varied according to the network or patch structure of the areas of suitable habitat.

2. Is there a threshold at which connectivity is lost? If yes, what is this threshold?

There is a threshold at which connectivity is lost.

Figures 2a and 2b show that current does not flow when the proportion of the landscape that is suitable habitat is 30% or less. Figure 5 shows that there are no connections between core habitat patches (represented as nodes) when the landscape has proportions of suitable habitat between 0 – 30%.

This means that the landscape must have at least 40% suitable habitat for movement of organisms to be facilitated (ie. for current to flow). The threshold at which connectivity is lost is at 30% proportion of the landscape as suitable habitat.

3. Do all habitat patches respond the same? Explain why or why not.

The core habitat areas do not all respond the same.

The spatial configuration of the binary habitats (Figure 1) show that the presence of the core habitat patches does not change with different proportions of suitable habitat in the landscape. This is because these core habitat patches have been designated as reserves.

However, Figure 2a shows that the cumulative current of the core habitat areas changes with proportions of suitable habitat. Excluding landscapes with no connectivity (10 – 30% suitable habitat), the amount of current (or potential for connectivity) is lowest when the proportion of the landscape as suitable habitat is 40%. Not only is the amount of current flowing from these patches reduced at 40% suitable habitat, but only 3 out of 5 patches allow current to flow. This is supported by Figure 4, which shows that resistance decreases between three of the core habitats at 40% suitable habitat, whereas connections/flow (demonstrated by resistance reduced from its maximum of 50) between all the core habitats occurs at 50% suitable habitat. This implies that at 40% suitable habitat and lower, two of the core habitat patches are completely isolated. There are a number of potential drivers of isolationism:

1. Amount of available suitable habitat within a specified radius of a patch

2. Arrangement and location of habitat areas in relation to edges and other habitat areas

3. Quality of the matrix habitat or corridors (Spiesman et al. 2018)

4. Size of the patch

As this is a binary model, the quality of the matrix habitat or corridors is not considered a driver of patch isolation in this case. However, the amount of available habitat within the radius of the isolated patches is likely to be a driver for their isolationism. It is also possible that the relatively small size of the isolated patches at 40% suitable habitat plays a role, as Hill et al (1996) states that patches are more likely to be colonised if they are larger in size.

Figure 2b shows that at 50% suitable habitat, all of the core habitat areas are connected. This is supported by Figure 5 which shows that from 50 – 90% suitable habitat, the number of connections between the core habitat areas is at its maximum. However, the amount of current flowing from the core habitat areas is different depending on the core habitats at 50% suitable habitat, demonstrating that although the connection corridor may exist, the quantity and arrangement of the suitable habitat is still the driving factor of movement of organisms. Figure 2b shows that at 50% suitable habitat the edge core habitat areas have less current flowing through them than the central core habitat area. Figures 2a and 2b shows that from 60 – 90% suitable habitat, all core habitat patches have similar levels of current flowing through them.

Figure 2b also shows that when only three core habitat patches are connected (40% suitable habitat), the levels of current flowing between them is stronger than when all five core habitat patches are connected (50 – 90% suitable habitat). Figure 2b shows that once connections are established between patches (at 40% suitable habitat), the number of corridors increases with increasing suitable habitat, but the strength of current flowing through these corridors decreases with increasing suitable habitat. Figure 2a also shows that higher amount of current flow between the core habitat areas through the corridors at 40% and 50% suitable habitat, than at 70 – 90% suitable habitat. A reason for this could be that when there are fewer connections/corridors, they experience increased movement (current flow) as the same amount of current/organism movement has a limited number of connections/corridors to move through or flow. However, with increasing connections/corridors and suitable habitat, the same amount of current/organism movement has more suitable habitat to move through or flow. At 90% suitable habitat, the movement of current is displayed to demonstrate that the landscape has almost become one connected patch of suitable habitat.

References

Spiesman B J, Stapper A P & Inouye B D. 2018. Patch size, isolation, and matrix effects on biodiversity and ecosystem functioning in a landscape microcosm. Ecosphere 9(3): e02173. 10.1002/ecs2.2173

Hill J K, Thomas C D & Lewis O T. 1996. Effects of habitat patch size and isolation on dispersal by Hesperia comma butterflies: Implications for metapopulation structure. Journal of Animal Ecology 65(6): 725 – 735.

4. What is a key factor underpinning the response of organisms when applying a binary landscape model?

The response of the organism under a binary landscape model is underpinned by the quantity of the habitat.

Other factors to consider are:

1. Survival and reproduction rates

2. Dispersal and recruitment: Species will undertake fundamental processes of dispersal between patches, if there is enough area of corridor habitat

5. How do species perceive a binary landscape?

Species perceive a binary landscape mosaic comprised of focal habitats (ie. suitable habitat) in an inhospitable background matrix (ie. non-habitat) (Lecture 11).

6. Which meta-population model is synonymous with the binary landscape model?

In ecology, a metapopulation model is defined as a model of the occupancy of habitat patches across time linked by dispersal (Berkeley, CA 2022).

The metapopulation model synonymous with the binary landscape model is the patch dynamics model. An example of the binary patch dynamics model as a metapopulation model is described by the Levins model (1969), which is a metapopulation model based on the concept that local populations will reproduce and die within local patches and their offspring disperse into different, unoccupied patches (vlab.amrita.edu (2011).

References

University of California, Berkeley. Metapopulation ecology. 2022. https://nature.berkeley.edu/beislab/rail/html/rail_metapopulation_ecology.html#:~:text=RESEARCH%20APPROACH,of%20habitat%20patches%20through%20time.

vlab.amrita.edu,. (2011). Metapopulation Dynamics -Levins Model. Retrieved 23 April 2022, from vlab.amrita.edu/?sub=3&brch=65&sim=772&cnt=1

7. In this model what determines community structure and how is it measured?

Under this binary metapopulation model, community structure is determined by colonisation and extinction (Lecture 11).

The process of colonisation and extinction under this metapopulation model of community structure is described by Kareiva & Marvier (2001): “Metapopulation theory describes a species as a collection of populations (or patches) in which there is turnover due to local extinction and subsequent recolonization by dispersing individuals, and in which the fate of a species can only be understood by tracking the fate of a collection of populations or patches, as opposed to a single local population or patch” (Kareiva & Marvier 2001). Therefore:

• Colonisation is defined as the successful movement or immigration of organisms to a patch

• The colonisation rate is the number of successful colonisations of organisms to unoccupied patches

• Extinction is defined as the movement of organisms from occupied patches or absence of organisms in previously occupied patches

• The extinction rate is the number of patches where local communities are completely absent from their previously occupied patches

Colonisation and extinction is measured through connectivity (Lecture 11). For example, in this practical, connectivity between focal nodes (ie. suitable habitat) is being measured through conductance and resistance across a matrix (ie. corridors).

This analysis of connectivity will allow the rates of colonisation and extinction to be estimated, as the flowing current to core habitat patches represents immigration and movement of organisms. In this model, the analysis of colonisation and extinction (measured through connectivity) shows that:

• The landscape with 30% or less proportion of suitable habitat results in reserve patches (core habitat areas) that are completely isolated from each other. There is no colonisation from dispersing individuals. This risks local extinction in the core habitat areas.

• The landscape with 40% proportion of suitable habitat results in potential for colonisation of three reserve patches (core habitat areas). Two core habitat areas remain isolated, do not experience colonisation and risk local extinction.

• The landscape with 50% proportion of suitable habitat areas results in potential for colonisation of all five reserve patches (core habitat areas).

References

Peter Kareiva, Michelle Marvier, in Encyclopedia of Biodiversity (Second Edition), 2001.

Exercise 2: Code and plots

#Step 1 code

all_species <- list.files(path = "data/raw", pattern = "^Species.*.asc$",  full.names = TRUE)
all_species <- lapply(all_species, raster)

species_stack <- do.call(stack, all_species)

#Step 2 code

spplot(species_stack)

Figure 6. Raw resistances of binary habitats for Species A, B and C

#Step 3 code

#get curmaps
species_curmap <- list.files(path = "data/species", pattern = "*curmap.asc", full.names = TRUE)
species_curmap <- lapply(species_curmap, raster)

#Step 3 code

sp_A <- species_curmap[[1]]
plot(sp_A, col=brewer.pal(n = 11, name = 'RdBu'))

Figure 7a. Conductance map for Species A

#Step 3 code

sp_B <- species_curmap[[2]]
plot(sp_B, col=brewer.pal(n = 11, name = 'RdBu'))

Figure 7b. Conductance map for Species B

#Step 3 code

sp_C <- species_curmap[[3]]
plot(sp_C, col=brewer.pal(n = 11, name = 'RdBu'))

Figure 7c. Conductance map for Species C

#Step 4 code

reclass_raster<-function(x, threshold){
  z<-reclassify(x, c(-Inf,threshold,0, threshold,+Inf,1))
  return(z)
}

#Step 4 code

sp_A_re<-reclass_raster(sp_A, 0.3)
plot(sp_A_re, col = c('white', 'black'))

sp_B_re<-reclass_raster(sp_B, 0.2)
plot(sp_B_re, col = c('white', 'black'))

sp_C_re<-reclass_raster(sp_C, 0.4)
plot(sp_C_re, col = c('white', 'black'))

#Step 5 code

reserves<-raster('data/raw/Reserves.asc')
plot(reserves)

#Step 5 code

reserves_poly<-rasterToPolygons(reserves, dissolve = T)

## Loading required namespace: rgeos

#Step 5 code

plot(sp_A_re, col = c('white', 'black'), legend = FALSE, main = 'Species A')

plot(reserves_poly, col = 'yellow', add = T)
legend("topright", 
       legend = c("Matrix", "Suitable habitat corridors", "Habitat reserves"),
       title = "Species A landscape", 
       cex = 0.5,
       fill = c('white', 'black', 'yellow'))

Figure 8a. Reserve and habitat map for Species A

#Step 5 code

plot(sp_B_re, col = c('white', 'black'), legend = FALSE, main = 'Species B')

plot(reserves_poly, col = 'blue', add = T)
legend("topright", 
       legend = c("Matrix", "Suitable habitat corridors", "Habitat reserves"),
       title = "Species B landscape", 
       cex = 0.5,
       fill = c('white', 'black', 'blue'))

Figure 8b. Reserve and habitat map for Species B

#Step 5 code

plot(sp_C_re, col = c('white', 'black'), legend = FALSE, main = 'Species C')

plot(reserves_poly, col = 'green', add = T)
legend("topright", 
       legend = c("Matrix", "Suitable habitat corridors", "Habitat reserves"),
       title = "Species C landscape", 
       cex = 0.5,
       fill = c('white', 'black', 'green'))

Figure 8c. Reserve and habitat map for Species C

Exercise 2: Questions

8. Compare and contrast the pairwise resistances for the three species. Which species will have a harder time moving around the landscape?

#Question 8 code

res_sp<-res%>%filter(Landscape %in% c('sp_a', 'sp_b', 'sp_c'))%>%
  mutate(res_fix=ifelse(Resistance == -1, 50, Resistance), flow = paste0('N',Node1,'-N',Node2))%>%
  droplevels()

pairwise<-pivot_wider(res_sp, names_from = c('Landscape'), values_from = c('Resistance'))%>%
  group_by(flow) %>%
  summarize(sp_a=max(sp_a,na.rm=TRUE),sp_b=max(sp_b,na.rm=TRUE),sp_c=max(sp_c,na.rm=TRUE))

pairwise

## # A tibble: 10 × 4
##    flow   sp_a  sp_b  sp_c
##    <chr> <dbl> <dbl> <dbl>
##  1 N1-N2  36.1  4.66 177. 
##  2 N1-N3  39.4  4.89 203. 
##  3 N1-N4  68.6  8.24 337. 
##  4 N1-N5  18.3  2.32  96.5
##  5 N2-N3  44.5  5.52 224. 
##  6 N2-N4  69.5  8.20 336. 
##  7 N2-N5  20.4  2.61 103. 
##  8 N3-N4  48.6  5.87 248. 
##  9 N3-N5  27.1  3.34 141. 
## 10 N4-N5  54.3  6.38 264.

#Question 8 code

write.csv(pairwise, 'outputs/species_pairwise.csv')

#Question 8 code

ex2_q8 <- read.csv('outputs/species_pairwise.csv')

ex2_q8_tidy <- ex2_q8 %>% 
  select(1, flow = 2, sp_a = 3, sp_b = 4, sp_c = 5) 

ex2_q8_summary <- ex2_q8_tidy %>%
  group_by(flow) %>%
  summarise(sp_a_resistance = sp_a,
            sp_b_resistance = sp_b,
            sp_c_resistance = sp_c)

knitr::kable(ex2_q8_summary,
             caption = "Table 1. Pairwise resistances for habitats of Species A, B and C across nodes",
             digits = 2,
             col.names = c('Flow between nodes', 
                           'Resistance (Species A habitat)', 
                           'Resistance (Species B habitat)',
                           'Resistance (Species C habitat)'))

Table 1. Pairwise resistances for habitats of Species A, B and C across nodes

Flow between nodes	Resistance (Species A habitat)	Resistance (Species B habitat)	Resistance (Species C habitat)
N1-N2	36.06	4.66	176.58
N1-N3	39.37	4.89	202.91
N1-N4	68.57	8.24	336.73
N1-N5	18.34	2.32	96.45
N2-N3	44.49	5.52	224.50
N2-N4	69.54	8.20	336.21
N2-N5	20.41	2.61	102.69
N3-N4	48.59	5.87	248.34
N3-N5	27.06	3.34	141.30
N4-N5	54.26	6.38	264.19

#Question 8 code

sum(ex2_q8_summary$sp_a_resistance)

## [1] 426.6807

sum(ex2_q8_summary$sp_b_resistance)

## [1] 52.03917

sum(ex2_q8_summary$sp_c_resistance)

## [1] 2129.896

#Question 9 code

df_sp_a <-ex2_q8_summary[order(ex2_q8_summary$sp_a_resistance),]
df_sp_a

## # A tibble: 10 × 4
##    flow  sp_a_resistance sp_b_resistance sp_c_resistance
##    <chr>           <dbl>           <dbl>           <dbl>
##  1 N1-N5            18.3            2.32            96.5
##  2 N2-N5            20.4            2.61           103. 
##  3 N3-N5            27.1            3.34           141. 
##  4 N1-N2            36.1            4.66           177. 
##  5 N1-N3            39.4            4.89           203. 
##  6 N2-N3            44.5            5.52           224. 
##  7 N3-N4            48.6            5.87           248. 
##  8 N4-N5            54.3            6.38           264. 
##  9 N1-N4            68.6            8.24           337. 
## 10 N2-N4            69.5            8.20           336.

#Question 9 code

df_sp_b <-ex2_q8_summary[order(ex2_q8_summary$sp_b_resistance),]
df_sp_b

## # A tibble: 10 × 4
##    flow  sp_a_resistance sp_b_resistance sp_c_resistance
##    <chr>           <dbl>           <dbl>           <dbl>
##  1 N1-N5            18.3            2.32            96.5
##  2 N2-N5            20.4            2.61           103. 
##  3 N3-N5            27.1            3.34           141. 
##  4 N1-N2            36.1            4.66           177. 
##  5 N1-N3            39.4            4.89           203. 
##  6 N2-N3            44.5            5.52           224. 
##  7 N3-N4            48.6            5.87           248. 
##  8 N4-N5            54.3            6.38           264. 
##  9 N2-N4            69.5            8.20           336. 
## 10 N1-N4            68.6            8.24           337.

#Question 9 code

df_sp_c <-ex2_q8_summary[order(ex2_q8_summary$sp_c_resistance),]
df_sp_c

## # A tibble: 10 × 4
##    flow  sp_a_resistance sp_b_resistance sp_c_resistance
##    <chr>           <dbl>           <dbl>           <dbl>
##  1 N1-N5            18.3            2.32            96.5
##  2 N2-N5            20.4            2.61           103. 
##  3 N3-N5            27.1            3.34           141. 
##  4 N1-N2            36.1            4.66           177. 
##  5 N1-N3            39.4            4.89           203. 
##  6 N2-N3            44.5            5.52           224. 
##  7 N3-N4            48.6            5.87           248. 
##  8 N4-N5            54.3            6.38           264. 
##  9 N2-N4            69.5            8.20           336. 
## 10 N1-N4            68.6            8.24           337.

The total resistance across each landscape for each of the three species has been calculated. The species which experience the greater resistances will have the harder time moving around the landscape. The following lists the total resistance for the species in descending order:

1. Species C (Total resistance = 2129.90)

2. Species A (Total resistance = 426.68)

3. Species B (Total resistance = 52.04)

Therefore, Species C will have the hardest time moving around the landscape, followed by Species A. Species B will have a relatively easy time moving around the landscape.

Table 1 shows the comparison of pairwise resistances for the three species across the landscape. The same trend (Species C < Species A < Species B) is observed across all node connections, meaning that Species B has consistently better potential for dispersal than Species C and Species A, irrespective of which core habitat patch emigration is occurring from.

9. Which habitat patches are better connected for each species? How does the connectivity between nodes differ between species?

Table 1 also shows which node connections facilitate the most dispersal. The node connections which facilitate the most or least dispersal is the same across each species, indicating that potential connectivity and movement patterns are similar, irrespective of the species.

The lowest resistance for Species A, B and C occurs at node N1-N5, indicating the best connected patches. This is likely due to the fact that the N1 and N5 patches are within a close spatial proximity. This is also interesting as under the binary model, patch N1 was isolated below a minimum suitable habitat proportion of 50%. This highlights the changing roles of habitats once habitat quality is considered in tandem with habitat quantity. The node connections of high resistance are similarly consistent across species types, with the highest resistance for Species A occurring at node N2-N4, and the highest resistances for Species B and C occurring at node N1-N4, indicating the least connected patches.

The quality of the N4 patch for all species may be low due to its small size and location at a distance from other patches in the landscape. The smaller core habitat patches may be of reduced habitat quality due to potential resource pressures and edge effects. There is difference in that in combination with N4, Species A has highest resistance between reserve N2 and Species B and C have highest resistance between N1. This is an illustration of the difference in perceived habitat quality between the species, as patch N2 was a lower quality for Species A and patch N4 was lower quality for Species B and C.

Sorting the resistance values in ascending order for all three species demonstrates that the potential for dispersal between specific nodes, although the resistances are different, have the same relationship across species (with the exception of the highest resistances, described above). The following list describes which habitat patches are the best connected to which habitat patches are the worst connected, in order from best to worst:

1. N1-N5

2. N2-N5

3. N3-N5

4. N1-N2

5. N1-N3

6. N2-N3

7. N3-N4

8. N4-N5

9. N1-N4/ N2-N4

10. N2-N4/ N1-N4

The demonstrates a couple of trends. Patch number 4 is consistently at the bottom of the list, meaning that patch number 4 is the worst connected patch for each species in the landscape. Similarly, patch number 5 is consistently at the top of the list (except for when it is connected to patch number 4), meaning that patch number 5 is one of the best connected patches for each species in the landscape. This is unsurprising, as patch number 5 is spatially located centrally relative to the other patches. Patch number 1 is also well connected, as it occupies the top half of the list (with the exception of its connection to patch number 4).

10. Do the species have the same network of habitat through the matrix? Compare and contrast.

The species do not have the same network of habitat through the matrix. Although network patterns are similar across all three species, the area of quality habitat networks between species is very different, which results in different networks of quality habitat through the matrix across species.

This pattern is demonstrated in Figures 8a, 8b and 8c. The increased resistances translate to reduced number and area of corridors of quality habitat. The limited corridors for Species C can be clearly seen, particularly in contrast to the number and area of corridors for Species B. Species A has a habitat network which is more established than Species C, but not as extensive as Species B.

Exercise 3: Code and plots

#Step 1 code

sum<-sp_A+sp_B+sp_C
plot(sum)

#Step 2 code

corridors<-reclassify(sum, c(-Inf, 0.4,0, #Matrix becomes 0
                             0.4,10,1, # Corridors become 1
                             10,Inf,2)) #Reserves become 2

plot(corridors, col = plasma(3), legend = FALSE)
legend("topright", legend = c("Matrix", "Corridors", 'Reserves'), fill = plasma(3))

Figure 9. Corridor network for Species A, B and C

#Step 3 code

corridors_df<-corridors%>%as.data.frame(xy = T, na.rm = T)

corridors_df$ID<-1

reso<-res(corridors)[1]

areas<-corridors_df%>%group_by(layer)%>%
  summarise(area = sum(ID), area_ha = (area*reso^2)/10000)%>%
  mutate(sumA = sum(area), per = 100*area/sumA)%>%rename(class = layer)

areas

## # A tibble: 3 × 5
##   class  area area_ha  sumA   per
##   <dbl> <dbl>   <dbl> <dbl> <dbl>
## 1     0  6610    6610 10000  66.1
## 2     1  2780    2780 10000  27.8
## 3     2   610     610 10000   6.1

write.csv(areas, 'outputs/species_areas.csv')

Exercise 3: Questions

11. How many hectares is your corridor network (do not include the area occupied by the habitat patches)?

The corridor network is 2780 ha.

12. What proportion of the landscape is within your corridor network (do not include the area in the habitat patches)?

The landscape is assumed to comprise of only the matrix, corridors and reserves. Therefore, the total area of the landscape is 10,000 ha.

The corridor network is 27.8% of the total landscape.

13. What is a key factor underpinning the response of organisms when applying a mosaic landscape model? Hint: Look in Landscape Models Lecture.

The response of the organism under a mosaic landscape model is underpinned by the quality of the habitat.

14. How do species perceive a mosaic landscape?

Under a mosaic landscape model, the species perceive the entire landscape as a mosaic of different quality habitats (Lecture 11). The species perceive the quality of a habitat depending on what resources are available for them in the habitat. Under this model, species dispersal is not limited.

15. Which meta-population model is synonymous with the mosaic landscape model?

The metapopulation model synonymous with the mosaic landscape model is the Species Sorting model. According to Leibold & Loeuille (2015), “the species-sorting view [is] in which community assembly in local communities always goes to an endpoint that depends only on local environmental conditions”.

References

Leibold M A & Loeuille N. 2015. Species sorting and patch dynamics in harlequin metacommunities affect the relative importance of environment and space. Ecology 96 (12): 3227 – 3233.

16. In this model, what determines community structure and how is it measured?

Community structure is determined by the local environmental conditions, which “filter the regional pool of species along different resources gradients” (Lecture 11). In other words, “species self-assemble into clusters based on their niche requirements” (Li & Ma 2020).

Species sorting is measured by determining the number of species and individuals in a specified habitat area or community and calculating how the species are distributed or clustered within that community.

References

Li L & Ma Z. 2020. Species Sorting and Neutral Theory Analyses Reveal Archaeal and Bacterial Communities Are Assembled Differently in Hot Springs. Front. Bioeng. Biotechnol., 28 May 2020, https://doi.org/10.3389/fbioe.2020.00464

17. In the landscape, 50% of the area outside the reserve network is habitat. As a conservation planner you are tasked with determining the minimum amount of habitat that needs to be maintained to conserve the three species of interest. How much of the remaining habitat on the landscape must be maintained?

The minimum amount of habitat outside of the reserve network has already been defined, based on the species perceptions of habitat quality (ie. their resistance thresholds).

As such, 27.8% of the habitat area outside of the reserve network must be maintained as a minimum (2780 ha). This area must be maintained in the spatial configuration of the corridors already defined.

18. Are any of the habitat patches more important than the others to the network?

As previously described in Question 9, patch numbers 1 and 5 are the most connected patches in the network. As they are the most connected, they are more critical in providing dispersal potential for organisms. Although dispersal is not limited under this model, connected patches represent habitats of higher quality. Patch number 5 is also centrally located, which reduces any potential edge effects. Similarly, patch numbers 2 and 3 are larger in size, which is beneficial as larger patches are typically more resilient to disturbance. This implies that patch number 4 is the least important in the network.

19. If you lost a habitat patch in the network would it still function as a meta-community? Why or why not?

The discussion is based on the following definition: “A metacommunity is defined as a set of local communities that are linked by dispersal” (Altermatt 2016).

Under the mosaic model

As the model is based on habitat quality, it is not clear whether the loss of a patch would impact the local species demography. This is because under the mosaic model, patch types cause differences in the outcomes of local species’ interactions, dispersal and community composition (Leibold et al 2004). As the species-sorting/mosaic model emphasises spatial niches, it is not clear whether the removal of a habitat patch would remove a niche upon which a species is dependent, and therefore impact dispersal, survival or extinction.

It can be assumed that all patches represent an ecological niche for each species. As the same reserves are used to model the connectivity of each species, it may be that removal of a habitat patch may cause a local extinction in that patch but would not necessarily mean the extinction of all patch communities and that meta-community functions could still occur.

Under the binary model

The binary model is based on habitat quantity, where it is assumed that all patches are identical and capable of containing populations (Leibold 2004). As there are five patches in the landscape, losing one habitat patch would impact the dispersal network, but the network would still function as a meta-community. Removing one patch would increase the resistance across the matrix, reduce the number of corridors and reduce habitat quantity, but species would still be able to disperse between the remaining habitat patches.

This is assuming that the loss of one habitat patch does not reduce the proportion of total suitable habitat in the landscape to less than 50%. Under this scenario, the remaining patches may become isolated. In this case, the network would no longer function as a meta-community, as dispersal would not be possible.

References

Altermatt F. 2016. Metacommunity dynamics. Oxford Bibliographies Online. https://www.oxfordbibliographies.com/view/document/obo-9780199830060/obo-9780199830060-0021.xml

Leibold M A, Holyoak M, Mouquet N, Amarasekare P, Chase J M, Hoopes M F, Holt R D, Shurin J B, Law R, Tilman D, Loreau M & Gonzalez A. 2004. The metacommunity concept: a framework for multi-scale community ecology. Ecology Letters 7: 601 – 613.

20. Compare and contrast the connectivity results when considering the landscape using a binary model versus mosaic model. What are the implications of these approaches for conservation planning of the three species in the landscape?

Connectivity results - binary model	Connectivity results - mosaic model
Minimum quantity of suitable habitat required for connectivity to occur (greater than 30% of landscape) Minimum quantity of suitable habitat required for connectivity to occur between all patches (greater than 40% of landscape) Best connected patches: Patch 2, Patch 3, Patch 5 Worst connected patches: Patch 1, Patch 4	Minimum quantity of corridor habitat required for connectivity to occur for all three species (27.8%) Best connected patches: Patch 1, Patch 5 Worst connected patches: Patch 4

Implications of approaches for conservation planning of three species in the landscape

Using the binary approach assumes that dispersal only occurs between connected corridors. The mosaic approach assumes that dispersal occurs based on resource gradients and patch types.

From a conservation planning perspective, using a binary model assumes that all the species have the same requirements from the habitat. As such, using this model results in a conservation planning manager results in conserving greater areas of habitat within corridor networks (minimum 30%). Additionally, using this approach would results in focusing conservation and connectivity efforts on Patches 2, 3 and 5.

However, using the mosaic approach results in conserving a lower minimum area of habitat within corridor networks, but requiring the consideration of habitat quality in determining which habitat corridor networks to conserve. This decision must be based on the resources available in the corridor networks for the species in question. As such, conservation and connectivity efforts would be focused on Patches 1 and 5.

Exercise 4: Code and plots

#Step 1 code

lbp_res<-read.csv('data/lbp_resistances.csv')

lbp_res<-lbp_res%>%mutate(threshold = 20,
                          connected = ifelse(resistance<=threshold, 1, 0))

lpb_res_summary<-lbp_res%>%
  group_by(replicate, scenario, timestep)%>%
  summarise(connected_nodes = sum(connected))

## `summarise()` has grouped output by 'replicate', 'scenario'. You can override
## using the `.groups` argument.

lpb_res_summary

## # A tibble: 14 × 4
## # Groups:   replicate, scenario [6]
##    replicate scenario timestep connected_nodes
##        <int> <chr>       <int>           <dbl>
##  1         1 BAU             0             364
##  2         1 BAU            20             405
##  3         1 BAU           100              13
##  4         1 NM              0             405
##  5         1 NM             20             405
##  6         1 NM            100             124
##  7         2 BAU            20               2
##  8         2 BAU           100             109
##  9         2 NM             20              54
## 10         2 NM            100              19
## 11         3 BAU            20             405
## 12         3 BAU           100               0
## 13         3 NM             20             405
## 14         3 NM            100               0

#Step 1 code

knitr::kable(lpb_res_summary,
             caption = "Table 2. Summary of total connected nodes per replicate, scenario and timestep for modelling of Leadbeaters Possum habitat",
             digits = 2,
             col.names = c('Replicate', 
                           'Scenario', 
                           'Timestep',
                           'Number of connected nodes'))

Table 2. Summary of total connected nodes per replicate, scenario and timestep for modelling of Leadbeaters Possum habitat

Replicate	Scenario	Timestep	Number of connected nodes
1	BAU	0	364
1	BAU	20	405
1	BAU	100	13
1	NM	0	405
1	NM	20	405
1	NM	100	124
2	BAU	20	2
2	BAU	100	109
2	NM	20	54
2	NM	100	19
3	BAU	20	405
3	BAU	100	0
3	NM	20	405
3	NM	100	0

#Step 2 code

lpb_res_summary%>%filter(scenario == 'NM')%>%
ggplot(aes(as.factor(timestep), connected_nodes))+
  stat_summary(geom = 'bar', fill = 'dodgerblue', width = 0.6)+
  stat_summary(geom = 'errorbar', width = 0.1)+
  labs(y = 'Connected nodes', x = 'Time (years)', title = 'Connected Nodes - No Management')+
  scale_x_discrete(labels = c("0","20", '100'))+
  theme_bw()

## No summary function supplied, defaulting to `mean_se()`
## No summary function supplied, defaulting to `mean_se()`

Figure 10a. Connected nodes for Leadbeaters Possum habitat under No Management disturbance scenario

#Step 2 code

lpb_res_summary%>%filter(scenario == 'BAU')%>%
ggplot(aes(as.factor(timestep), connected_nodes))+
  stat_summary(geom = 'bar', fill = 'green', width = 0.6)+
  stat_summary(geom = 'errorbar', width = 0.1)+
  labs(y = 'Connected nodes', x = 'Time (years)', title = 'Connected Nodes - Business As Usual')+
  scale_x_discrete(labels = c("0","20", '100'))+
  theme_bw()

## No summary function supplied, defaulting to `mean_se()`
## No summary function supplied, defaulting to `mean_se()`

Figure 10b. Connected nodes for Leadbeaters Possum habitat under Business As Usual disturbance scenario

#Step 2 code

#Alternative solution using facet_grid function
#Alternative solution provided to check accuracy of above solutions and provide comparative view


ConnectedNodes <- lpb_res_summary%>%
  select(opening = 1, scenario = 2, timestep = 3, connected_nodes = 4)%>%
  ggplot(aes(as.factor(scenario), connected_nodes, fill = as.factor(scenario)))+
#  geom_bar(stat = 'identity', position = "dodge", width = 0.6)+
  stat_summary(geom = 'bar', width = 0.6)+
  stat_summary(geom = 'errorbar', width = 0.1)+
  labs(x = '', 
       y = 'Connected nodes', 
       title = 'Relationship between number of connected nodes, management scenario 
       and time for Leadbeaters Possum habitat',
       fill = '',
       caption = '')+
  scale_y_continuous(expand = c(0,0), limits = c(0, 500))+
  theme_bw()+
  theme(axis.text.x = element_blank())

ConnectedNodes + 
  facet_grid(. ~ timestep)

## No summary function supplied, defaulting to `mean_se()`
## No summary function supplied, defaulting to `mean_se()`
## No summary function supplied, defaulting to `mean_se()`
## No summary function supplied, defaulting to `mean_se()`
## No summary function supplied, defaulting to `mean_se()`
## No summary function supplied, defaulting to `mean_se()`

Figure 10c. Number of connected nodes for Leadbeaters Possum habitat under Business As Usual and No Management disturbance scenarios, grouped by time (0, 20, 100)

#Step 3 code

NM_curmap <- list.files(path = "data/LBP/NM", pattern = "*curmap.asc", full.names = TRUE)
NM_curmap <- lapply(NM_curmap, raster)

NM1_T0 <- NM_curmap[[1]]
NM1_T20 <- NM_curmap[[3]]
NM1_T100 <- NM_curmap[[2]]
NM2_T0 <- NM_curmap[[4]]
NM2_T20 <- NM_curmap[[6]]
NM2_T100 <- NM_curmap[[5]]
NM3_T0 <- NM_curmap[[7]]
NM3_T20 <- NM_curmap[[9]]
NM3_T100 <- NM_curmap[[8]]


NM_stack <- stack(NM1_T0,NM1_T20,NM1_T100, NM2_T0,NM2_T20, NM2_T100,NM3_T0, NM3_T20, NM3_T100)

plot(NM_stack)

Figure 11. Conductance graphs for Leadbeaters Possum under the No Management scenario

#Step 4 code

BAU_curmap <- list.files(path = "data/LBP/BAU", pattern = "*curmap.asc", full.names = TRUE)
BAU_curmap <- lapply(BAU_curmap, raster)

BAU1_T0 <- BAU_curmap[[1]]
BAU1_T20 <- BAU_curmap[[3]]
BAU1_T100 <- BAU_curmap[[2]]
BAU2_T0 <- BAU_curmap[[4]]
BAU2_T20 <- BAU_curmap[[6]]
BAU2_T100 <- BAU_curmap[[5]]
BAU3_T0 <- BAU_curmap[[7]]
BAU3_T20 <- BAU_curmap[[9]]
BAU3_T100 <- BAU_curmap[[8]]


BAU_stack <- stack(BAU1_T0,BAU1_T20,BAU1_T100, BAU2_T0,BAU2_T20, BAU2_T100,BAU3_T0, BAU3_T20, BAU3_T100)

plot(BAU_stack)

Figure 12. Conductance graphs for Leadbeaters Possum under the Business As Usual scenario

#step 5 code

T0<-stack(NM1_T0, NM2_T0, NM3_T0)
names(T0)<-c('T0_NM1', 'T0_NM2', 'T0_NM3')

re_T0 <- reclass_raster(T0, 5) #function from above, set threshold to 5

cellStats(re_T0, 'sum') #in ha per rep

## T0_NM1 T0_NM2 T0_NM3 
##   1867   1867   1867

#step 5 code

T20<-stack(NM1_T20, NM2_T20, NM3_T20)
names(T20)<-c('T20_NM1', 'T20_NM2', 'T20_NM3')

re_T20 <- reclass_raster(T20, 5)

cellStats(re_T20, 'sum')

## T20_NM1 T20_NM2 T20_NM3 
##    1925    1660    1925

#step 5 code

T100<-stack(NM1_T100, NM2_T100, NM3_T100)
names(T100)<-c('T100_NM1', 'T100_NM2', 'T100_NM3')

re_T100 <- reclass_raster(T100, 5)

cellStats(re_T100, 'sum')

## T100_NM1 T100_NM2 T100_NM3 
##     1697     1631     1655

#step 5 code

T0_BAU<-stack(BAU1_T0, BAU2_T0, BAU3_T0)
names(T0_BAU)<-c('T0_BAU1', 'T0_BAU2', 'T0_BAU3')

re_T0_BAU <- reclass_raster(T0_BAU, 5) #function from above, set threshold to 5

cellStats(re_T0_BAU, 'sum') #in ha per rep

## T0_BAU1 T0_BAU2 T0_BAU3 
##    1867    1867    1867

#step 5 code

T20_BAU<-stack(BAU1_T20, BAU2_T20, BAU3_T20)
names(T20_BAU)<-c('T20_BAU1', 'T20_BAU2', 'T20_BAU3')

re_T20_BAU <- reclass_raster(T20_BAU, 5)

cellStats(re_T20_BAU, 'sum')

## T20_BAU1 T20_BAU2 T20_BAU3 
##     1915     1994     1914

#step 5 code

T100_BAU <- stack(BAU1_T100, BAU2_T100, BAU3_T100)
names(T100_BAU)<-c('T100_BAU1', 'T100_BAU2', 'T100_BAU3')

re_T100_BAU <- reclass_raster(T100_BAU, 5)

cellStats(re_T100_BAU, 'sum')

## T100_BAU1 T100_BAU2 T100_BAU3 
##      1589      1718      1535

#step 5 code

NM_Net_stack <- stack(re_T0[[1]],re_T20[[1]],re_T100[[1]],
                       re_T0[[2]],re_T20[[2]], re_T100[[2]],
                       re_T0[[3]],re_T20[[3]], re_T100[[3]])

plot(NM_Net_stack)

Figure 13. Habitat corridor network for each replicate and time period under No Management scenario

#Step 5 code

BAU_Net_stack <- stack(re_T0_BAU[[1]],re_T20_BAU[[1]],re_T100_BAU[[1]],
                       re_T0_BAU[[2]],re_T20_BAU[[2]], re_T100_BAU[[2]],
                       re_T0_BAU[[3]],re_T20_BAU[[3]], re_T100_BAU[[3]])

plot(BAU_Net_stack)

Figure 14. Habitat corridor network for each replicate and time period under Business As Usual scenario

#Step 6 code

NM_LBP<-sum(NM_Net_stack)

plot(NM_LBP,col = plasma(9))

Figure 15. Aggregated habitat corridor network under No Management scenario

#Step 6 code

BAU_LBP<-sum(BAU_Net_stack)

plot(BAU_LBP,col = plasma(9))

Figure 16. Aggregated habitat corridor network under Business As Usual scenario

#Step 7 code

NM_LBP_nozero<-NM_LBP #duplicate layer to not overwrite it

NM_LBP_nozero[NM_LBP_nozero == 0] <- NA #replace no data values with NA 

corridors_LBP_NM<-NM_LBP_nozero%>%as.data.frame(xy = T, na.rm = T) #make raster into table

corridors_LBP_NM$ID<-1 #assign unit (1 = 1ha)

areas_LBP_NM<-corridors_LBP_NM%>%group_by(layer)%>%
  summarise(area = sum(ID))%>%
  mutate(sumA = 6784, per = 100*area/sumA)%>%
  rename(class = layer)

areas_LBP_NM

## # A tibble: 9 × 4
##   class  area  sumA   per
##   <dbl> <dbl> <dbl> <dbl>
## 1     1   657  6784  9.68
## 2     2   247  6784  3.64
## 3     3   420  6784  6.19
## 4     4   188  6784  2.77
## 5     5   217  6784  3.20
## 6     6   308  6784  4.54
## 7     7   159  6784  2.34
## 8     8   475  6784  7.00
## 9     9   565  6784  8.33

write.csv(areas_LBP_NM, 'outputs/LBP_areas_NM.csv')

#Step 7 code

BAU_LBP_nozero<-BAU_LBP

BAU_LBP_nozero[NM_LBP_nozero == 0] <- NA

corridors_LBP_BAU<-BAU_LBP_nozero%>%as.data.frame(xy = T, na.rm = T)

corridors_LBP_BAU$ID<-1

areas_LBP_BAU<-corridors_LBP_BAU%>%group_by(layer)%>%
  summarise(area = sum(ID))%>%
  mutate(sumA = 6784, per = 100*area/sumA)%>%
  rename(class = layer)

areas_LBP_BAU

## # A tibble: 10 × 4
##    class  area  sumA    per
##    <dbl> <dbl> <dbl>  <dbl>
##  1     0  8140  6784 120.  
##  2     1   682  6784  10.1 
##  3     2   491  6784   7.24
##  4     3   319  6784   4.70
##  5     4   296  6784   4.36
##  6     5   372  6784   5.48
##  7     6   268  6784   3.95
##  8     7   353  6784   5.20
##  9     8   156  6784   2.30
## 10     9   586  6784   8.64

write.csv(areas_LBP_BAU, 'outputs/LBP_areas_BAU.csv')

Exercise 4: Questions

21. Compare and contrast the change in the meta-population network within and between scenarios. Does the meta-population network remain static over time? What do you think is influencing the connectivity between LBP colonies overtime? Integrate the maps and graphs created above into this answer. Note: this question is worth ~20% of the entire assessment for this Prac!

In regards to this analysis, the metapopulation in question is a group of 30 spatially separated populations of Leadbeater’s Possum that interact.

No Management Scenario

The metapopulation network does not remain static over time under the NM scenario, as demonstrated by the following data:

• Figure 10a shows that the number of connected nodes steadily decreases over time

• Figure 11 shows that without thresholds, conductance across the habitats changes with time, demonstrated by reduced patchiness and increased connected conductive surface in conductance networks with time. This is potentially due to the impacts of recent disturbances.

• However Figure 13 shows that with thresholds, the spatial configuration and area of corridor habitat changes and fluxes with time. Generally, the area of habitat corridor area decreases with time:

o Total average habitat corridor area T0 = 1867 ha

o Total average habitat corridor area T20 = 1837 ha

o Total average habitat corridor area T100 = 1661 ha

• Figure 15 demonstrates the aggregated habitat corridor network and provides a basis for calculation of habitat quality according to classes. The following summary table of habitat quality for the NM scenario demonstrates the impact of non-static spatial and temporal habitat configurations on habitat quality:

o Total area of low quality habitat (classes 1 – 3) = 657 + 247 + 420 = 1,324 ha

o Total area of moderate quality habitat (classes 4 – 6) = 188 + 217 + 308 = 713 ha

o Total area of high quality habitat (classes 7 – 9) = 159 + 475 + 565 = 1,199 ha

• This shows that although average habitat corridor area decreases with time, and habitat quality declines across the entire time period, habitat quality changes non-linearly with time.

Business As Usual Scenario

The metapopulation network does not remain static over time under the BAU scenario, as demonstrated by the following data:

• Figure 10b shows that the number of connected nodes steadily decreases over time

• Figure 12 shows that without thresholds, conductance across the habitats changes with time, demonstrates by reduced patchiness in conductance networks with time

• However Figure 14 shows that with thresholds, the spatial configuration and area of corridor habitat changes non-linearly and fluxes with time. This may be due to the impact of disturbances and timber harvesting activities.

o Total average habitat corridor area T0 = 1867 ha

o Total average habitat corridor area T20 = 1941 ha

o Total average habitat corridor area T100 = 1614 ha

• Figure 16 demonstrates the aggregated habitat corridor network and provides a basis for calculation of habitat quality according to classes. The following summary table of habitat quality for the BAU scenario demonstrates the impact of non-static spatial and temporal habitat configurations on habitat quality:

o Total area of low quality habitat (classes 1 – 3) = 682 + 491 + 319 = 1,492 ha

o Total area of moderate quality habitat (classes 4 – 6) = 296 + 372 + 268 = 936 ha

o Total area of high quality habitat (classes 7 – 9) = 353 + 156 + 586 = 1,095 ha

Between the NM and BAU scenarios

Table 2 shows the number of connected nodes for both scenarios. Across time, the No Management scenario achieves a greater number of connections over time than the Business As Usual scenario (Figures 10a and 10b).

Interestingly, Figures 11 and 12 demonstrate that, when not restricted by thresholds, the conductance of the landscapes across replicates and timesteps varies very little between NM and BAU scenarios. This implies that if the LBP populations did not have specific habitat quality requirements, that undertaking timber harvesting would have very little impact on the conductance of the landscape.

However, with the application of conductance thresholds, it can be seen that there is a significant difference in the spatial arrangement and area of habitat corridors between NM and BAU scenarios, across replicates and timesteps (Figures 13 and 14). This difference is made the most obvious by the aggregated habitat corridor networks displayed in Figures 15 and 16.

This spatial and temporal differences of corridor habitat is summarised by the area of each corridor network in each numeric category. The higher the number in the category, the more consistent the quality of habitat has been over time. Based on this approach, the following table summarises the quality of the habitat corridor networks over time across management scenarios:

		NM sce nario	NM sce nario	BAU sce nario	BAU sce nario
	C ategory n umber	*Area**	% of ha bitat	*Area**	% of ha bitat
Lowest quality habitat corridor network	1	657	9.68	682	10.05
	2	247	3.64	491	7.24
	3	420	6.19	319	4.70
	4	188	2.77	296	4.36
	5	217	3.20	372	5.48
	6	308	4.54	268	3.95
	7	159	2.34	353	5.20
	8	475	7.00	156	2.30
Highest quality habitat corridor network	9	565	8.33	586	8.64

This shows that the Business As Usual scenario has the higher amount of lowest quality habitat corridor area (682 ha). Interestingly, the Business As Usual Scenario also has the higher amount of highest quality corridor area (586 ha). Due to this paradoxical result, the data was generalised in an attempt to establish an understanding of the habitat quality under both scenarios.

		NM sce nario	NM sce nario	BAU sce nario	BAU sce nario
	C ategory n umber	*Area**	% of ha bitat	*Area**	% of ha bitat
Low quality habitat corridor network	1 - 4	1,512	22.28	1,788	26.34
Moderate quality habitat corridor network	5	217	3.20	372	5.48
High quality habitat corridor network	6 - 9	1,507	22.22	1,363	20.09

Based on this generalised data, it can be seen that the No Management scenario results is less Low-Moderate quality habitat corridor area (total area = 1,512 + 217 = 1,729 ha) than the Business As Usual management scenario (total area = 1,788 + 372 = 2,160 ha). Similarly, it can be seen that the No Management scenario results is greater areas of High quality habitat corridor (total area = 1,507 ha) than the Business As Usual management scenario (total area = 1,363 ha).

Influences on connectivity between LBP colonies over time

The connectivity between LBP colonies is driven by the quality and suitability of the habitat between the patches. As the quality of the habitat reduces, it can be assumed that the connectivity between LBP colonies also reduces.

Disturbance is a cause of reduction in habitat quality. Spatially-correlated disturbance events have the potential to affect proportions of the patch areas and corridor networks.

A type of disturbance that would influence the changing spatial connectivity of LBP colonies over time under both NM and BAU scenarios is fire. This is due to the fact that there are three vegetation species required to make up a habitat of sufficient quality for the successful colonisation of LBP populations: Living mountain ash trees, dead mountain ash trees and Acacia. These three vegetation type respond to fire differently. Additionally, LBP requires hollows to be present in mountain ash trees for use as habitat. The different responses of the three types of quality LBP habitat variables to fire, as defined by McCarthy & Lindenmayer (2000):

• “Living mountain ash trees: hollows only develop after 100 – 200 years

• Dead mountain ash trees after fire: good habitat but become unsuitable after 10 – 30 years

• Acacia increases quickly following fire, die after about 50 – 100 years and regenerate again from soil seed back following another disturbance” (McCarthy & Lindenmayer (2000).

As it can be seen, the required vegetation for LBP habitat achieves sufficient habitat quality across different time cycles in response to fire. These spatially and temporally-dependent responses of vegetation types to fire impacts the habitat quality and suitability for LBP populations over time and impacts the connectivity between colonies. It means that habitat quality will be a changing gradient depending on location of fire disturbance in relation to the patches and time since the fire disturbance.

A type of disturbance that would influence the connectivity between LBP colonies under the BAU scenario would be planned logging activities. Timber harvesting is time-dependent, and occurs on rotations of 50 – 80 years, impacting the development of suitable habitat for LBP colonies and connectivity between them (McCarthy & Lindenmayer 2000).

References

McCarthy M A & Lindenmayer D B. 2000. Spatially-correlated extinction in a metapopulation model of Leadbeater’s Possum. Biodiversity and Conservation 9: 47 – 63.