Analysis outline:

  1. Detectability correction by distance
  2. Biodiversity differences of far vs. near
    1. Alpha
    2. Beta
    3. Temporal patterns
  3. Community compositional differences, far vs. near
    1. overall
    2. temporal pattern
  4. Biogeographic origin frequencies and the connection to proximity to settlements, precipitation, temperature, NDVI
    1. overall
    2. temporal pattern
  5. Specific species for the previous item.
  6. frequency of shrubland specialists
    1. far vs. near
    2. temporal pattern
  7. frequency of nesting migrants
    1. far vs. near
    2. temporal pattern
  8. frequency of vulnerable species
    1. near vs. far
    2. temporal pattern

Data filtering: use only breeding species (both resident and migratory), only spring data, exclude data from pilot.

Note: Har Amasa from T1 (2016) contains only 2 “near” plots instead of 3. As the other two “near” plots contain >10 observations each (9 and 15 species for plots 1 and 2 respectively), it is unlikely that no birds were observed. Hence this should be treated as missing data that cannot be completed, rather that complete absence of birds.

## Loading required package: readxl

Calculate richness, abundance and geometric mean abundance per plot. Geometric mean abundance calculated following recommendation of Santini et al., Biological Conservation 2017. In calculation of biodiversity metrics only data from <250 m was used. “Rare species” definition (see calculations below): species that were observed in 5 plots AT LEAST.

## [1] "ABUNDANCE WITH RARE SPECIES"

## [1] "RICHNESS WITH RARE SPECIES"

## [1] "RICHNESS WITHOUT RARE SPECIES"

## [1] "GEOMETRIC MEAN ABUNDANCE WITH RARE SPECIES"

## [1] "GEOMETRIC MEAN ABUNDANCE WITHOUT RARE SPECIES"

Run GLM and GLMM for geometric mean abundance

## [1] "Summary of GLM:"
## 
## Call:
## glm(formula = gma ~ year_ct + settlements, family = Gamma(link = identity), 
##     data = P_byplot)
## 
## Deviance Residuals: 
##      Min        1Q    Median        3Q       Max  
## -0.63462  -0.30378  -0.04815   0.13503   1.43661  
## 
## Coefficients:
##                 Estimate Std. Error t value Pr(>|t|)    
## (Intercept)      2.75418    0.16450  16.742  < 2e-16 ***
## year_ct         -0.16989    0.02601  -6.532 1.01e-09 ***
## settlementsNear  0.54699    0.14596   3.748 0.000256 ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Gamma family taken to be 0.1613521)
## 
##     Null deviance: 27.445  on 148  degrees of freedom
## Residual deviance: 18.506  on 146  degrees of freedom
## AIC: 349.92
## 
## Number of Fisher Scoring iterations: 5
## [1] "Summary of GLM with interaction:"
## 
## Call:
## glm(formula = gma ~ year_ct * settlements, family = Gamma(link = identity), 
##     data = P_byplot)
## 
## Deviance Residuals: 
##      Min        1Q    Median        3Q       Max  
## -0.64671  -0.31341  -0.06301   0.13520   1.35757  
## 
## Coefficients:
##                         Estimate Std. Error t value Pr(>|t|)    
## (Intercept)              2.90211    0.20017  14.498  < 2e-16 ***
## year_ct                 -0.19732    0.03268  -6.038 1.25e-08 ***
## settlementsNear          0.18362    0.30334   0.605    0.546    
## year_ct:settlementsNear  0.06914    0.05240   1.319    0.189    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for Gamma family taken to be 0.1553126)
## 
##     Null deviance: 27.445  on 148  degrees of freedom
## Residual deviance: 18.199  on 145  degrees of freedom
## AIC: 349.38
## 
## Number of Fisher Scoring iterations: 6
## [1] "Summary of GLMM:"
## Generalized linear mixed model fit by maximum likelihood (Laplace
##   Approximation) [glmerMod]
##  Family: Gamma  ( identity )
## Formula: gma ~ year_ct + settlements + (1 | point_name)
##    Data: P_byplot
## Control: glmerControl(nAGQ0initStep = FALSE)
## 
##      AIC      BIC   logLik deviance df.resid 
##    342.5    357.5   -166.2    332.5      144 
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -1.4530 -0.6858 -0.1255  0.3321  5.0835 
## 
## Random effects:
##  Groups     Name        Variance Std.Dev.
##  point_name (Intercept) 0.1040   0.3224  
##  Residual               0.1302   0.3609  
## Number of obs: 149, groups:  point_name, 54
## 
## Fixed effects:
##                 Estimate Std. Error t value Pr(>|z|)    
## (Intercept)       2.6943     0.1540  17.491  < 2e-16 ***
## year_ct          -0.1558     0.0209  -7.457 8.86e-14 ***
## settlementsNear   0.5475     0.1757   3.115  0.00184 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) yer_ct
## year_ct     -0.592       
## settlmntsNr -0.533 -0.017

Beta diversity

Ordination of community composition

Distance-based ordination (nMDS)

Model-based ordination (ecoCopula)

Distribution of conservation statuses across Far and Near

LC -> 1

NT -> 2

VU -> 3

EN -> 4

CR -> 5

no cutoff applied for minimum number of individuals

no cutoff applied for minimum number of individuals

cutoff applied for minimum number of individuals: 4

total number of species: 55

number of synanthrope/ invasive species: 26

number of endangered species: 8

number of batha species (marked with *): 18

median of synanthrope / invasive in black; endangered in red; batha in blue; overall in grey

no cutoff applied for minimum number of individuals

total number of species: 79

number of synanthrope/ invasive species: 29

number of endangered species: 16

number of batha species (marked with *): 31

median of synanthrope / invasive in black; endangered in red; batha in blue; overall in grey

Results of Mann-Whitney test for difference of medians between distributions of conservation ranks between near and far:

Wilcoxon rank sum test with continuity correction

data: ConservationCodeIL2018_ordinal by settlements W = 300161, p-value < 2.2e-16 alternative hypothesis: true location shift is not equal to 0

Temporal trend is not caused by reduction in number of endangered species. This is the number of species in the above graph, after cutoff of observations in those years in which there were less than 4 individuals observed from that species.

year V1
2012 5
2014 4
2016 5
2018 4
2020 5

And this is the total number of endangered species observed yearly in the Sfar unit:

year V1
2012 6
2014 5
2016 7
2018 8
2020 8

Unique rarely observed species appearing Far vs. Near

Rarely observed species vs. endangered species

Common species: - Lanius collurio - Cercotrichas galactotes - Oenanthe isabellina - Oenanthe finschii - Coturnix coturnix - Hippolais olivetorum - Motacilla alba - Phylloscopus collybita - Emberiza caesia - Hippolais languida

Endangered but not rarely observed: - Ciconia ciconia - Merops apiaster - Sylvia conspicillata - Oenanthe hispanica - Anthus similis - Emberiza hortulana

Rarely observed but not endangered: - Anas platyrhynchos - Cuculus canorus - Sylvia crassirostris - Sylvia communis - Onychognathus tristramii - Luscinia megarhynchos - Phoenicurus ochruros - Rhodospiza obsoleta - Pterocles orientalis - Buteo rufinus - Ptyonoprogne fuligula - Monticola solitarius - Ammoperdix heyi - Clamator glandarius - Milvus migrans - Dendrocopos syriacus - Psittacula krameri - Corvus ruficollis - Ammomanes deserti - Iduna pallida - Carpospiza brachydactyla - Athene noctua - Cisticola juncidis - Passer hispaniolensis - Petronia petronia - Serinus serinus