Diversity and distribution of the last remnants of endemic juniper woodlands on Tenerife, Canary Islands

Abstract

Identifying ecological requirements, species diversity patterns and distribution ranges of habitats of interest is an important task when developing conservation and restoration programs. The Canarian juniper woodland formed by Juniperus turbinata ssp. canariensis is listed as a priority habitat by the European Union. Although very common in the past, this vegetation type has suffered immense destruction and degradation over the last five centuries on the Canary Islands, especially on the largest most populated island of Tenerife. We evaluated the geographical distribution range of the last remnants of Canarian juniper woodlands on Tenerife and analyzed their ecological status, floristic composition and plant species diversity. Despite the degradation of the original vegetation, we still observed outstanding species diversity. Endemic species richness and number of typical habitat species were best predicted by summer rainfall, which seems to be the limiting factor for this habitat in the lower drier regions. Human disturbance has had a negative effect on endemic species richness but a positive effect on the distribution of alien plants, highlighting the potential threat to this habitat. Ecological characterization and floristic composition were most influenced by climatic factors related to the dichotomy of a humid windward and a drier leeward slope of the island and by altitude. However, vegetation structure and human disturbance also determined species composition. Environmental requirements indicated a circuminsular potential distribution of this habitat. Given the exceptional plant diversity, the scarcity of dense stands and the low protection status, immediate protection of the remaining stands and future restoration programs should be the priority for conservation strategies of this endemic vegetation type.

Full text

ORIGINAL PAPER
Diversity and distribution of the last remnants
of endemic juniper woodlands on Tenerife, Canary
Islands
Ru
¨
diger Otto
•
Rube
´
n Barone
•
Juan-Domingo Delgado
•
Jose
´
-Ramo
´
n Are
´
valo
•
Vı
´
ctor Garzo
´
n-Machado
•
Francisco Cabrera-Rodrı
´
guez
•
Jose
´
-Marı
´
a Ferna
´
ndez-Palacios
Received: 2 December 2011 / Accepted: 22 March 2012 / Published online: 17 April 2012
Ó Springer Science+Business Media B.V. 2012
Abstract Identifying ecological requirements, species diversity patterns and distribution
ranges of habitats of interest is an important task when developing conservation and
restoration programs. The Canarian juniper woodland formed by Juniperus turbinata ssp.
canariensis is listed as a priority habitat by the European Union. Although very common in
the past, this vegetation type has suffered immense destruction and degradation over the
last five centuries on the Canary Islands, especially on the largest most populated island of
Tenerife. We evaluated the geographical distribution range of the last remnants of
Canarian juniper woodlands on Tenerife and analyzed their ecological status, floristic
composition and plant species diversity. Despite the degradation of the original vegetation,
we still observed outstanding species diversity. Endemic species richness and number of
typical habitat species were best predicted by summer rainfall, which seems to be the
limiting factor for this habitat in the lower drier regions. Human disturbance has had a
negative effect on endemic species richness but a positive effect on the distribution of alien
plants, highlighting the potential threat to this habitat. Ecological characterization and
floristic composition were most influenced by climatic factors related to the dichotomy of a
humid windward and a drier leeward slope of the island and by altitude. However, veg-
etation structure and human disturbance also determined species composition. Environ-
mental requirements indicated a circuminsular potential distribution of this habitat. Given
the exceptional plant diversity, the scarcity of dense stands and the low protection status,
immediate protection of the remaining stands and future restoration programs should be the
priority for conservation strategies of this endemic vegetation type.
R. Otto (&)  R. Barone  J.-R. Are
´
valo  F. Cabrera-Rodrı
´
guez  J.-M. Ferna
´
ndez-Palacios
Departamento de Ecologı
´
a, Facultad de Biologı
´
a, Universidad de La Laguna, 38206 La Laguna,
Tenerife, Canary Islands, Spain
e-mail: rudiotto@ull.es
J.-D. Delgado
A
´
rea de Ecologı
´
a, Departamento de Sistemas Fı
´
sicos, Quı
´
micos y Naturales, Universidad Pablo de
Olavide, Ctra. de Utrera, km 1, 41013 Seville, Spain
V. Garzo
´
n-Machado
Departamento de Biologı
´
a Vegetal (Bota
´
nica), Universidad de La Laguna, C/Profesor Wolfredo
Wildpret s/n, 38071 La Laguna, Tenerife, Islas Canarias, Spain
123
Biodivers Conserv (2012) 21:1811–1834
DOI 10.1007/s10531-012-0278-2

Keywords Species richness  Floristic composition  GLM  MRPP 
Thermophilous woodland  Canary Islands
Introduction
The characterization of the ecological status, the analysis of species diversity and the
delimitation of the geographical distribution of habitats and plant populations of interest
are fundamental in conservation biology (Scott et al. 2001; Linares-Palomino et al. 2010;
Bacaro et al. 2011). This also holds true for some juniper woodlands, growing in semiarid
regions (Gardner and Fisher 1996; Gauquelin et al. 1999; Mun
˜
oz-Reinoso 2004; El-Bana
et al. 2010). When restricted to just a few areas, isolated plant populations are more
susceptible to climate change, human pressure and to suffering from stochastic events that
can threaten their resources and habitats. Consequences might be the loss of biodiversity of
associated floras and faunas, including the genetic diversity of the species populations
(Thompson 1999). These negative effects are of special concern when target communities
are rich in endemic species (El-Bana et al. 2010). Therefore, identifying the geographical
range and environmental requirements of rare target species or habitats represents an
important tool in conservation planning and biodiversity monitoring.
Endemic plant species richness is known to be high on oceanic islands and is related to
multiple factors, such as isolation in time and space from the continent, water-energy-
dynamics, environmental gradients resulting in high habitat diversity or environmental
heterogeneity, actual size of islands or habitats as well as their historical commonness
(Whittaker and Ferna
´
ndez-Palacios 2007; Kreft et al. 2008; Jakobs et al. 2010; Zobel et al.
2011). Furthermore, oceanic islands are particularly prone to invasions by alien species
due to their unique ecological and biogeographical conditions (Cronk and Fuller 1995;
Denslow et al. 2009). These biological invasions are considered the main threats to native
biodiversity (Mack et al. 2000).
On the island of Tenerife (Canary Islands), climatic gradients related to elevation and
slope orientation have been identified as the most important factors shaping plant species
richness and communities (Ferna
´
ndez-Palacios 1992; Ferna
´
ndez-Palacios and de Nicola
´
s
1995; Whittaker and Ferna
´
ndez-Palacios 2007). In contrast, human activities, such as road
infrastructures and land use types, play an important role in the distribution of alien plant
species (Are
´
valo et al. 2005; Arteaga et al. 2009; Are
´
valo et al. 2010), which has also been
reported from other oceanic islands (Jakobs et al. 2010; Kueffer et al. 2010).
Although recent information exists on the estimated potential area of the thermophilous
forest, including Canarian juniper woodlands, for each Canary Island based on bioclimatic
and phytocoenotic data (Del Arco et al. 2006a) and the diversity of its species pool
(Domı
´
nguez-Lozano et al. 2010; Zobel et al. 2011), there is a lack of knowledge with
respect to geographical distribution, ecological status, floristic composition and diversity
pattern of existing populations of Juniperus turbinata ssp. canariensis on Tenerife. This is
of great conservation concern since these woodlands have been included in the list of
priority habitats of the European Union (9560: Endemic forests of Juniperus, Montesinos
et al. 2009). On Tenerife, this habitat has almost completely been destroyed over the last
five centuries, and its current extension is 290 ha, whereas thermophilous woodlands, as a
whole, occupy today 437 ha, 1.5 % of their original extension of about 29.700 ha (Del
Arco et al. 2010). As a consequence of the immense loss of juniper woodlands on Tenerife,
the local authorities (Cabildo Insular de Tenerife) launched a project in 2005, financed by
1812 Biodivers Conserv (2012) 21:1811–1834
123

the European Union (Project LIFE04/NAT/ES/000064), to restore Canarian juniper
woodlands in the northwestern region of the island.
The present study is part of this project and focuses on the following objectives: (1) to
evaluate the distribution of the last remnants of juniper woodlands on Tenerife in order to
reconstruct the potential geographical range of Canarian juniper, (2) to obtain insights
into the ecological and floristic characterization of the J. turbinata stands, (3) to identify
patterns of species richness, with special interest in endemic species richness and alien
species distribution, and (4) to provide information for conservation strategies as well as
for future monitoring and restoration projects of this habitat of priority. In general, we
hypothesized that richness and composition of native and endemic species is mainly related
to climatic factors, i.e. to the differences between the more humid and colder northern
slope and the drier and warmer southern slope of the island, whereas the distribution of
alien species is more influenced by human activities on the landscape scale and certain land
use types or infrastructures such as roads.
Materials and methods
Study area
The study was carried out on Tenerife, the largest and highest island of the Canarian
Archipelago. The island’s surface area is about 2,034 km
2
and the highest point is reached
at 3,718 m asl. at the peak of the Teide volcano. Owing to the steep elevation gradient,
the following five zonal ecosystems can be found from the coast to the peak of the Teide
(Del Arco et al. 2006b): (1) coastal sub-desert scrub, an open shrub vegetation adapted to
the subtropical, semi-arid climate and dominated by stem succulents of the genus
Euphorbia and leaf succulents or sclerophyllous shrubs; (2) thermophilous forest,
including juniper woodlands, the object of this study; (3) evergreen laurel forest growing
on the north and northeastern sides of the island and formed by evergreen tree species
belonging to genera such as Laurus, Apollonias, Persea, Ilex, Prunus; (4) Canary pine
forest, exclusively made up of Pinus canariensis, and distributed above the laurel forest on
the windward slope and above the thermophilous woodlands on the leeward slope; and (5)
summit broom scrub restricted to areas above the timber line with common shrub species,
such as Spartocytisus supranubius, Descurainia bourgeauana and Pterocephalus lasio-
spermus, adapted to low temperature in winter and warm, dry conditions in summer.
The Canarian juniper, J. turbinata ssp. canariensis (Guyot) Rivas-Mart et al., is cur-
rently considered an endemic tree of the Canarian and Madeiran archipelagos (Acebes
et al. 2010), despite the ongoing debate about its taxonomical status (Adams et al. 2002).
Adams et al. (2006) and Adams (2008) classified the Canarian juniper populations as
Juniperus phoenicea var. turbinata, whereas Farjon (2005) grouped them within Juniperus
phoenicea var. phoenicea. Adams et al. (2009) found differentiation in the leaf volatile oils
of populations from Madeira and the Canary Islands compared to populations in Spain and
Morocco, but they concluded that these differences did not justify the recognition of
J. phoenicea subsp. canariensis.
Nowadays, this species is absent from the more arid eastern islands of Fuerteventura and
Lanzarote (Acebes et al. 2010), where it has probably been eliminated by human activity
over the last few centuries (Reyes-Betancort et al. 2001). Thermophilous forests would
potentially grow at intermediate altitudes between 0–200 and 500 m a.s.l. on the windward
slope and between 300–500 and 700–1100 m a.s.l. on the leeward slope of the islands
Biodivers Conserv (2012) 21:1811–1834 1813
123

(Del Arco et al. 2006a; Ferna
´
ndez-Palacios et al. 2008). Canarian juniper woodlands are
considered to constitute the most extended thermophilous woodland on the western Canary
Islands (Del Arco et al. 2010). However, specific models predicting the potential distri-
bution of this species do not yet exist.
Other communities belonging to this vegetation type are dominated by species such as
Pistacia atlantica, P. lentiscus, Olea cerasiformis, Phoenix canariensis, Retama rhodo-
rhizoides or Heberdenia excelsa (Ferna
´
ndez-Palacios et al. 2008; Nezadal and Welss
2009). The climate is Mediterranean with annual precipitation between 250 and 450 mm,
mostly occurring in winter, and with average temperature between 15 and 19 °C,
depending on aspect and elevation. The Canarian juniper is not very demanding with
respect to soil conditions, since it is able to grow on poorly developed, stony soils.
However, soils of juniper woodlands can substantially vary according to altitude and
exposure to the dominant northeastern trade winds (von Gaisberg 2005).
Local studies have mainly focused on the floristic composition and geographical aspects
of some thermophilous communities, including the juniper woodlands on Tenerife (Criado
1982; Santos and Ferna
´
ndez 1983; Rodrı
´
guez et al. 1990; Marrero et al. 1992; Luis et al.
2005). Detailed distribution maps of this species exist for the islands of El Hierro (von
Gaisberg 2005) and Gran Canaria (Gonza
´
lez-Artiles 2006), but no attempt has ever been
made to identify the geographical range and the ecological status of this species on
Tenerife. In general, phytosociological characterizations and ecological differentiations
were presented by Rivas-Martı
´
nez et al. (1993) and by Del Arco et al. (2006b), whereas
Domı
´
nguez-Lozano et al. (2010) highlighted the richness of endemic species of this
vegetation type.
On Tenerife, juniper forests have almost completely been destroyed over the last five
centuries, since the potential zone of this vegetation type was the most favorable place for
human settlements and agriculture (Del Arco et al. 2010). Furthermore, timber was used
for many kinds of tools and for constructing houses (Rodrı
´
guez and Marrero 1991;
Ferna
´
ndez-Palacios et al. 2008).
Data collection
We searched for juniper patches within the potential area of the thermophilous woodlands
on the island of Tenerife between 2006 and 2009 as part of the activities carried out within
the LIFE project. In total, we sampled 108 sites where J. turbinata ssp. canariensis was
present. We installed 10 m radius circular plots (area: 314 m
2
) around selected adult
juniper individuals, identified all perennial understory vascular plant species within these
plots and visually estimated their cover. Since remnants of Canarian juniper woodlands on
this island are usually represented by very small groups of individuals, we selected at least
one adult individual to study its associated flora. Annual species were not recorded because
it was not possible to sample sites twice a year in order to obtain a complete list of this
species group. Furthermore, we were mainly interested in characterizing the understory
shrub community for regeneration purposes. The percentage of annual species in this
vegetation type varies considerably with the degree of human disturbance, structure and
local climatic conditions. The highest participation of annuals, up to 50 % of the total
species richness per plot, was found in substitution shrub communities within the potential
area of humid juniper woodlands on the island of El Hierro (von Gaisberg 2005).
For each plot, we obtained UTM coordinates with a Global Positioning System device
(model GPS, Garmin, Olathe, Kansas, USA), and several environmental and biotic vari-
ables were recorded in the field, such as altitude, slope inclination, slope aspect, cover of
1814 Biodivers Conserv (2012) 21:1811–1834
123

rocks and superficial soil, as well as total cover of trees, shrubs and perennial herbs. Cover
values were visually estimated.
Climatic variables were interpolated applying spatial interpolation tools implemented in
a Geographical Information System (GIS). After testing various interpolation techniques,
we selected ordinary co-kriging (OCK) incorporating elevation from a Digital Elevation
Model (DEM), since it provided the most accurate results after testing for prediction errors
of cross-validation. The application of this geo-statistical approach is particularly justified
in areas where landform is very complex (Diodato 2005). All the calculations were carried
out using the Geostatistical Analyst module implemented in ArcGIS-ESRI software.
Interpolation maps were elaborated with grid cell size of 50 9 50 m for several climatic
variables provided by the Botany Department of La Laguna University.
Additionally, we extracted spatial information at a landscape scale by analyzing the
surroundings of the plots and using spatial analysis tools incorporated in ArcGis. Areas of
main land use types were calculated within a buffer of 500 m around the plots (area:
0.785 km
2
) and nearest distances from plot center to land use or infrastructure types were
analyzed (Table 1). Information on thematic layers was obtained from GRAFCAN S.L.,
Tenerife.
Plants were grouped, according to their status, into single island endemics (SIE),
restricted to the island of Tenerife, Canarian endemics, Macaronesian endemics (endemic
to the Canary Islands and Madeira), non-endemic natives and alien species. Recent
checklists were used for classification (von Gaisberg 2005; Stierstorfer and von Gaisberg
2006; Acebes et al. 2010).
Statistical analysis
We applied multivariate statistical techniques to analyze the influence of selected
explanatory variables on species composition and richness of remnants of juniper wood-
lands. To avoid multi-colinearity effects in multiple regression analysis, a correlation
matrix was constructed using non-parametric Spearman rank correlation coefficient to
explore relationships among explanatory variables. We then selected a set of explanatory
variables to enter in the regression analysis, by eliminating those variables that were highly
correlated with each other (r [ 0.70) and exhibited low tolerance statistics (\0.3) in
ordinary least square regression analysis (OLS). In a second step, we used generalized
additive models (GAM; Zuur et al. 2007) to evaluate the effect of each selected explan-
atory variable on species richness groups. This non-parametric model is especially rec-
ommended to detect non-linear relationships among variables. Since dependent variables
and their error distributions were not normally distributed, species richness was finally
analyzed applying generalized linear models (GLMs) with Poisson error distribution, using
a log-link function, as recommended for count data in ecological analysis (Crawley 1993).
In a GLM, the probability distributions of the dependent variable also include distributions
of the exponential family such as a Poisson or binomial distribution (McCullagh and
Nelder 1989; Dobson 1990). A link function provides the relationship between the linear
predictor that incorporates the information of the independent variables and the mean of
the distribution function. Predictor variables may be either numerical or categorical.
Explanatory variables that showed uni-modal relationships in GAMs were included in
GLMs with an additional quadratic term. To obtain the optimal set of predictor variables,
we used AIC (Akaike Information Criteria) with forward stepwise selection, choosing
the lowest AIC value for every possible combination of explanatory variables. Regression
analyses were run using the software STATISTICA 8.
Biodivers Conserv (2012) 21:1811–1834 1815
123

Ordination techniques represent useful tools to explain variation in species composition
of communities (Gauch 1982) and to evaluate major floristic gradients in time as well as in
space (Ter Braak and Smilauer 1998). We chose the indirect gradient analysis based on
Detrended Correspondence Analysis (DCA; Hill and Gauch 1980) of the software package
CANOCO (Ter Braak and Smilauer 1998) to evaluate major floristic gradients and to
examine how species composition was related to explanatory variables. We therefore
extracted the coordinates of the first three DCA axes and correlated them against the
explanatory variables using Spearman correlation coefficients. Additionally, we con-
structed Canonical Correspondence Analyses (CCAs) to confirm and visualize the results
obtained from the correlation analysis. The classification of typical species for each eco-
system follows Zobel et al. (2011).
In order to floristically classify the juniper stands, we performed cluster analysis using a
hierarchical, agglomerative cluster analysis on the samples with a relative Sørensen
Table 1 Results of basic statistics of all explanatory and dependent variables for all 108 plots studied, and
differentiated between windward (n = 46) and leeward slope (n = 62) of the island of Tenerife
Explanatory variables Mean Min. Max. Std. Mean W Mean L p value
Altitude (m) 505.3 10.0 1108.0 255.5 273.1 677.5 \0.001*
Slope (8) 35.5 0 75.0 14.60 32.5 39.2 0.016
MAT (mm) 17.2 14.3 19.5 1.4 18.2 16.5 \0.001*
MAP (mm) 383.4 194.3 610.1 131.7 476.1 313.7 \0.001*
MSP (mm) 14.1 3.9 37.3 7.5 19.0 10.4 \0.001*
Soil cover (%) 29.9 0 95.0 23.9 36.7 21.6 \0.001*
Tree cover (%) 27.6 0 75.0 19.4 33.1 21.0 \0.001*
Shrub cover (%) 38.6 5.0 90.0 21.3 39.0 38.1 0.824
Herb cover (%) 12.9 0 85.0 17.5 16.0 9.2 0.046
Urbanized areas (ha) 1.1 0 7.3 1.7 0.9 1.2 0.377
Cultivated areas (ha) 10.0 0 59.7 15.7 12.8 6.7 0.045
Abandoned areas (ha) 10.8 0 70.9 17.3 9.2 12.8 0.287
Forests (ha) 3.2 0 66.5 12.0 0.3 6.7 0.005
Shrubland (ha) 51.1 2.3 78.5 22.2 52.3 49.5 0.521
Dist main road (km) 0.7 0 2.3 0.58 738.6 543.4 0.080
Dist cultivated areas (km) 0.5 0 2.3 0.5 484.0 555.9 0.449
Dist urbanized areas (km) 1.0 0 3.0 0.8 1062.4 992.3 0.634
Dist forests (km) 1.4 0 3.5 1.0 1325.3 1571.1 0.188
Dependent variables
Total species richness 21.3 10.0 42.0 6.4 20.8 21.8 0.401
Endemic species 15.6 5.0 34.0 5.3 15.5 15.8 0.832
SIE 1.9 0 6.0 1.3 2.1 1.7 0.138
Native species 4.4 0 10.0 2.0 3.9 4.9 0.009
Alien species 1.2 0 5.0 1.2 1.3 1.1 0.434
Thermophilous species 3.8 1.0 10.0 2.1 4.3 3.3 0.017
Differences of means between both slope types were tested applying non-parametric Kolmogorov–Smirnov
test (W windward, L leeward, p value of t-test comparing windward and leeward slope, *Significant after
Bonferroni corrections). MAT mean annual temperature, MAP mean annual precipitation, MSP mean
summer precipitation (J, A, S), land use types area occupied (ha) within a buffer of 500 m around the juniper
plot, Dist distance from plot centre to nearest land use type, SIE single island endemic species
1816 Biodivers Conserv (2012) 21:1811–1834
123

distance measure and a flexible beta of -0.25 (McCune and Grace 2002). Then, the
optimal number of clusters was chosen using a MRPP (Multiple Response Permutation
Procedures). This is a non-parametric multivariate test similar to a multivariate ANOVA,
which can be used to compare results of different groups (McCune and Grace 2002). It was
performed on data separated into at least two clusters and up to 17 clusters. We used
Sørensen distances and PC-ORD default group weightings for all MRPP analyses (McCune
and Grace 2002). Results from the MRPPs that showed high separation between groups
(T-statistic) and high homogeneity within groups (A statistic) were used to select the best
number of plot clusters (Dolan and Parker 2005). The more negative T is, the stronger the
separation is between groups. A statistic ranges from -1 to 1, where 1 signifies that all
objects are identical within groups (Cha
´
vez and Macdonald 2005). Even with significant
separation of groups, A statistic values of less than 0.1 are common with community data
(McCune and Grace 2002).
After the optimum number of clusters was determined, an indicator species analysis
(ISA) was performed to identify which species were important to each cluster group.
Indicator species analysis provides a method of combining the relative abundance and
relative frequency of each species into an indicator value (Dufre
ˆ
ne and Legendre 1997).
Indicator values were then tested for statistical significance using a randomization
technique (Monte Carlo test) with 4,999 iterations. The randomizations were used to test
the statistical significance of each species. The statistical software PC-Ord Version 6.0
(McCune and Mefford 2011) was applied for the vegetation classification.
Results
Habitat characterization
Patches of juniper woodlands were found within a circuminsular distribution with two
major gaps, one between the Gu
¨
ı
´
mar Valley and the Anaga Mountains in the Northeast and
another one between Anaga Mountains and the Orotava Valley in the North (Fig. 1). We
found a significant difference in altitudinal distribution between windward (mean: 273 m,
range: 10–580 m) and leeward slopes (mean: 678 m, range: 312–1108 m; Table 1; Fig. 2).
Mean annual precipitation ranged from 200 to 600 mm, with a mean of 383 mm, while
mean annual temperature ranged from 14 to 19.5 °C with an average of 17.2 °C. The
juniper habitat was confirmed to be significantly drier and colder on the leeward compared
to the windward slope. Furthermore, stands in the North of the island showed significantly
higher soil and tree cover than those in the South. For the rest of habitat characteristics, we
did not observe significant differences between slope types. In general, stands grow at
present at sites with a considerable inclination, showing low tree and high shrub cover due
to the degradation of the original vegetation. Consequently, shrubland dominated around
the juniper stands (within an area of 0.785 km
2
) followed by abandoned and cultivated
areas. Main roads, urbanized and cultivated areas were, on average, not further than 1 km
and never further than 3 km away from the stands studied, reflecting considerable land-
scape and habitat fragmentation.
Richness pattern
Despite the clear signs of habitat fragmentation and degradation mentioned above, we still
found a very high species richness, especially in endemic species, in the juniper stands
Biodivers Conserv (2012) 21:1811–1834 1817
123

studied on Tenerife. In all plots, we recorded 214 perennial vascular plant species, out of
these, 132 were Macaronesian endemics (62 %), 57 non-endemic native (26 %) and 25
alien species (12 %), while the entire perennial flora of the Canary Islands only exhibits a
Fig. 1 Location of the studied patches of Juniperus turbinata ssp. canariensis woodlands on Tenerife,
Canary Islands
100 300 500 700 900 1100
0 200 400 600 800 1000 1200
Altitude (m)
2
4
6
8
10
12
14
16
18
20
Number of plots
Windward
Leeward
Fig. 2 Altitudinal distribution of Canarian Juniper woodlands at the windward and leeward slope of
Tenerife, Canary Islands (with Gaussian distribution model shown)
1818 Biodivers Conserv (2012) 21:1811–1834
123

corresponding value of 43 % endemics (Acebes et al. 2010). With respect to the perennial
flora of Tenerife, we detected 47 % of all endemic plants present on this island within the
108 plots, covering a total surface of only 34 ha.
The understory layer of this habitat harbored, on average, 15.6 endemic plants per plot
(314 m
2
) (73 % of the total average plot richness of 21.3) showing a maximum of 34
endemics (Table 1). Within this group, we detected 1.9 single island endemics per plot
(max. 6 SIE). Thermophilous species showed a mean value of 3.8 and a maximum value of
ten species per plot. The number of perennial alien species was quite low (1.2 on average,
maximum 5), but one of them, Opuntia maxima, considered invasive in the Canary Islands,
was found in 56 % of the stands with maximum cover values of 25 %. The most diverse
juniper woodland patch, located on the southern slope of the Anaga mountains, was
composed of five single island endemics, 24 Canarian endemics, five Macaronesian en-
demics and eight non-endemic native species.
The results of the correlation matrix showed some important correlations between
explanatory variables, such as the correlation between altitude and mean annual temper-
ature, altitude and slope type or the correlations between most of the land-use distance
measurements with the corresponding surface measurements around the stands. Results of
GAM showed an important uni-modal relationship between mean summer precipitation
and most of the species groups studied. Therefore, a quadratic term for this variable in the
regression analysis was also included.
Modeling species richness by means of GLMs revealed interesting insights into the
existing richness pattern (Table 2). Herb and soil cover, as well as urbanized areas,
around the plots had an overall negative effect on total species richness. Mean summer
precipitation was observed to have a clear uni-modal relationship, showing higher
richness values at intermediate precipitation levels. Richness of endemic species followed
the same patterns. Regarding native species richness, we only obtained a weak negative
linear relationship with mean annual precipitation. Richness of alien species was best
predicted by herb cover at the plot scale and by the abundance of disturbed areas (areas
of cultivated and urbanized land) at the landscape scale. The best predictor for the
number of thermophilous species per plot was mean summer precipitation, revealing a
uni-modal relationship. Additionally, tree cover had a positive effect on this species
group. On the whole, deviance explained by the models was relatively low, indicating
that there might be other important factors influencing richness pattern not included in
this study.
Species composition
Detrended Correspondence Analysis indicated that slope orientation was by far the
strongest factor influencing plant species composition of juniper stands on Tenerife.
Species with highest abundance on the windward slope are located on the right side of the
DCA diagram, species common on the leeward slope on the opposite side (Fig. 3). Slope
orientation was strongly correlated with the sample scores of the first DCA axis (Table 3).
Lower, but still significant correlation coefficients were shown by altitude and all the
climatic variables. Overall, the first DCA axis was related to a climatic gradient separating
more humid and warmer windward sites from drier and colder leeward sites. The length of
the gradient of the first DCA axis reached 3.6 SD, indicating high b-diversity and an almost
full species turnover, which occurs at four SD units (Gauch 1982).
The highest correlation coefficients with the coordinates of the second DCA axis were
shown by the tree and herb cover variables, separating closed from rather open juniper
Biodivers Conserv (2012) 21:1811–1834 1819
123

stands. Mean annual temperature was a less important factor correlated with this axis that,
overall, revealed structural differences of the vegetation. The third DCA axis was clearly
related to land use types and the degree of landscape transformation since the cultivated
and urbanized area variables, as well as most of the land-use distance measurements, were
correlated with the scores of this axis.
The species scatter plot of the two first DCA axes (Fig. 3) clearly displays the vege-
tation belts of Tenerife in contact with the juniper woodlands, with the typical species of
each habitat represented by their centroids. On the northern slope, we can find remnants of
thermophilous woodlands from the coast up to 200–500 m, where the transition to laurel
forest occurs (right side of the figure). In the South of the island (left in the figure), juniper
stands are mixed with succulent scrub formed mainly by Euphorbia species at low altitudes
and with pine forest species at higher altitudes, indicating the transition from juniper
woodlands to Canary Island pine forest. In one location, in the Gu
¨
ı
´
mar Valley, in the
Table 2 Results of generalized linear models (GLMs) for all 108 plots studied, showing the best set of
explanatory variables explaining richness of different species groups as response variable, using AIC
(Akaike Information Criteria) best set selection and Poisson distribution with log-link function (MAP mean
annual precipitation, MSP mean summer precipitation, Dev. expl. deviance explained)
Species group Parameter estimates Model building results
Estimate Wald’s v
2
p value AIC p value Dev. expl.
Total richness
Intercept 2.9061 1211.14 \0.001 671.8 \0.001 36.2
Herb cover -0.0087 31.29 \0.001
MSP 0.0458 16.97 \0.001
MSP 9 MSP -0.0012 12.26 \0.001
Soil cover -0.0027 6.51 0.011
Urbanized areas -0.0001 5.56 0.018
Endemic species
Intercept 2.5804 714.16 \0.001 608.3 \0.001 39.8
Herb cover -0.0082 21.93 \0.001
MSP 0.0529 17.02 \0.001
MSP 9 MSP -0.0013 11.84 0.001
Soil cover -0.0030 6.05 0.014
Urbanized areas -0.0004 5.90 0.015
Native species
Intercept 1.6422 119.74 \0.001 14.6
MAP -0.0008 4.21 0.040
Alien species
Intercept -2.5481 4.01 0.045 318.2 \0.001 15.6
Herb cover -0.0240 7.67 0.006
Disturbed areas 0.0001 6.64 0.010
Thermophilous species
Intercept 0.4096 3.78 0.049 421.7 \0.001 26.4
MSP 0.1078 12.41 0.000
MSP 9 MSP -0.0031 10.75 0.001
Tree cover 0.0074 7.51 0.006
1820 Biodivers Conserv (2012) 21:1811–1834
123

Southeast of the island, laurel forest, pine forest and thermophilous woodland species
coexist. CCA ordination confirmed the relationship between climatic variables with the
first axis, structural variables (tree and herb cover) with the second axis and human dis-
turbance with the third axis (Fig. 4).
Fig. 3 DCA ordination diagram of the first two axes displaying centroids of typical species found in 108
juniper woodland patches on the island of Tenerife. The eigenvalues of the axes were 0.436 and 0.310, the
cumulative percentage variance of species data of the first two axes reached 14.5 %. Square root
transformation of species cover values and down weighting of rare species were selected as options of
analysis (circles thermophilous species, triangles succulent scrub species, squares pine forest species, open
hexagons laurel forest species, abbreviation of species: first four letters of genus name and first four letters of
species name, see ‘‘Appendix’’)
Biodivers Conserv (2012) 21:1811–1834 1821
123

Table 3 Spearman coefficients of correlations between explanatory variables and coordinates of the first
three DCA axes (abbreviations see table 1)
Explanatory variables DCA_axis 1 DCA_axis 2 DCA_axis 3
Altitude -0.699 0.237 0.217
Windward/Leeward 0.851 -0.105 -0.124
UTM X 0.361 0.236 0.201
UTM Y 0.786 -0.035 0.115
Slope inclination -0.218 0.325 0.103
MAT 0.518 -0.344 -0.267
MAP 0.561 0.176 0.205
MSP 0.561 0.176 0.205
Soil cover 0.366 0.083 -0.063
Tree cover 0.296 0.406 -0.044
Shrub cover 0.045 -0.042 -0.283
Herb cover 0.164 -0.411 0.371
Urbanized areas 0.084 -0.136 -0.394
Cultivated areas 0.288 0.115 -0.568
Abandoned areas -0.206 0.034 -0.183
Forests -0.121 0.052 0.322
Shrubland -0.010 -0.196 0.459
Dist main road 0.047 -0.004 0.457
Dist cultivated areas -0.203 -0.129 0.547
Dist urbanized areas 0.012 0.115 0.495
Dist forests -0.289 -0.180 -0.378
-1.5 1.5
-1.5
2.0
Altitude
Precipitation
Temperature
Slope inclination
Rock cover
Soil cover
Tree cover
Herb cover
DCA axis 1
DCA axis 2
A
-1.0
1.0
-1.0
1.0
Agriculture
Abandoned fields
Urbanization
Dist road
Dist agriculture
Dist urbanization
Tree cover
Herb cover
DCA axis 2
DCA axis 3
B
Fig. 4 CCA ordination diagrams showing biplots of significant explanatory variables and 108 sites of
juniper woodland patches on the island of Tenerife. a Ordination representing the first and second axis of the
CCA and b Second and third CCA axis. The eigenvalues of the axes were a 0.242 and 0.203, and b 0.230
and 0.197. Square root transformation of species cover values and down weighting of rare species were
selected as CCA options (dist distance from plot centre to nearest land use type)
1822 Biodivers Conserv (2012) 21:1811–1834
123

Vegetation classification
The MRPP indicated that the optimal number of clusters was between six and eight sample
groups. We chose a classification with eight clusters, since it provided the optimal com-
bination of low T-statistic and high A-value (T =-38.4; A = 0.185). The indicator
species analysis identified 36 species, only considering species with p values \0.1, out of
214 as significant indicators of one of the eight juniper woodland types (Table 4). Two
groups (G1 and G8) were characterized by the combination of locally abundant endemic
shrubs. The first type showed high cover values of Euphorbia atropurpurea, R. rhodo-
rhizoides, Phagnalon purpurascens and Echium aculeatum and was located in
the Southwest of the island at relatively high altitudes compared to the mean values of
the other groups (Table 5). The first two species were not selected by ISA, but showed the
highest cover values in this rather species poor type, also characterized by the highest herb
cover and a low shrub cover. Type G8 represents a variant in the Anaga Mountains in the
northeastern part of Tenerife, where local endemics such as Aeonium lindleyi frequently
grow at lower altitudes. The juniper patches of this type exhibited high herb and shrub
cover and low species richness. The only indicator species of group G2 was Euphorbia
lamarckii ssp. lamarckii, an endemic spurge growing mainly in the South of the island.
Analyzing the whole floristic composition of this type, we also detected high abundances
of shrubs, such as Cistus monspeliensis and Artemisia thuscula, although these species did
not reach significant indicator species values, since they are also present in lower abun-
dances in other groups.
Two single island endemic shrubs, Pericallis lanata and Echium virescens, were
selected as indicator species of type G3, which was characterized by the highest values of
total species richness and pine forest species richness, as well as by low tree, herb and soil
cover. Type G4 included P. canariensis as an indicator species and clearly represented a
transition from juniper woodlands to pine forest, including in some cases Erica arborea.
Laurel forest tree species, such as Ilex canariensis and Visnea mocanera, and the ther-
mophilous tree O. cerasiformis were indicators of type G5, growing mainly on the northern
slope of the island and showing the highest participation of laurel forest and thermophilous
species as well as the highest tree cover among all the different types. Endemic shrub
species present in the northern part of the island, such as Sonchus congestus, Echium
giganteum and Atalanthus pinnatus, were typical of group G6 that exhibited highest shrub
and soil cover. Juniper patches of group G7 had ten indicator species, all of them typical of
succulent scrub growing at low altitudes in the dry South of the island. Herb and tree cover
was low in this type.
Discussion
Species richness
The Canarian Archipelago, biogeographically included in the Mediterranean Basin
(Blondel and Aronson 1999), is considered one of the most important biodiversity hotspots
in the world (Me
´
dail and Que
´
zel 1997) due to the high level of endemism of its biota
(Whittaker and Ferna
´
ndez-Palacios 2007). Within the Canarian Archipelago, Tenerife is
the most diverse island with respect to the number of habitats and endemic plants because
of its altitude, age and size (Ferna
´
ndez-Palacios and de Nicola
´
s 1995; Zobel et al. 2011).
Recent studies analyzing species pools of vascular plants at the habitat level on the Canary
Biodivers Conserv (2012) 21:1811–1834 1823
123

Table 4 Results of the Indicator
Species Analysis for the (8 or
eight) juniper woodland types (or
groups) identified by means of a
Multi-response Permutation Pro-
cedure (MPPP) after a cluster
analysis
Juniper woodland type/species Indicator value p value
G1 Euphorbia atropurpurea-Type
Umbilicus horizontalis 75 0.001
Phagnalon purpurascens 74 0.001
Echium aculeatum 68 0.001
Gonospermum fruticosum 62 0.001
Lobularia canariensis 56 0.001
G2 Cistus-Artemisia-Type
Euphorbia lamarckii 11 0.082
G3 Pericallis lanata-Type
Pericallis lanata 32 0.013
Echium virescens 24 0.028
Monanthes brachycaulos 23 0.028
Tolpis laciniata 18 0.072
G4 Pinus-Erica-Type
Pinus canariensis 23 0.051
Bituminaria bituminosa 42 0.012
Paronychia canariensis 25 0.021
Descurainia millefolia 22 0.043
G5 Ilex canariensis-Type
Ilex canariensis 15 0.100
Olea cerasiformis 27 0.012
Visnea mocanera 20 0.050
Bystropogon canariensis 16 0.050
G6 Sonchus congestus-Type
Sonchus congestus 49 0.001
Atalanthus pinnatus 41 0.004
Echium giganteum 40 0.001
Asparagus umbellatus 37 0.008
Aeonium canariense 36 0.006
Bystropogon origanifolius 36 0.002
G7 Euphorbia balsamifera-Type
Euphorbia balsamifera 40 0.001
Euphorbia canariensis 36 0.005
Lavandula buchii 56 0.001
Cenchrus ciliaris 50 0.001
Drimia maritima 46 0.001
Kleinia neriifolia 43 0.001
Kickxia scoparia 42 0.001
Neochamaelea pulverulenta 33 0.008
Ceballosia fruticosa
25 0.026
Campylanthus salsoloides 20 0.033
G8 Aeonium lindleyi-Type
Aeonium lindleyi 85 0.001
Plantago arborescens 59 0.001
1824 Biodivers Conserv (2012) 21:1811–1834
123

Islands revealed that thermophilous woodlands, including juniper woodlands, together with
the summit scrub showed the highest levels of diversity of endemic species (Domı
´
nguez-
Lozano et al. 2010; Zobel et al. 2011; Steinbauer et al. 2011).
The present study carried out at the level of plot or a-richness confirmed the outstanding
diversity of perennial vascular plants, especially of endemics, of the last remnants of
juniper woodlands on Tenerife. Although there is a lack of comparative studies of richness
patterns using the same plot size for all main ecosystems of the island, we can show that
the remaining juniper woodland patches represent high local biodiversity spots within the
recognized regional biodiversity hotspot of the Canarian Archipelago. With respect to
perennial vascular plants, and depending on plot size (100–400 m
2
), mean richness values
per plot of 12–19 species were recorded for the succulent scrub, 21.3 for juniper woodlands
(present study), 10–15 for the laurel forest, 4–8 for the pine forest and 3–6 for the summit
scrub (Ferna
´
ndez-Palacios 1987; Otto et al. 2001; Otto 2003; Otto et al. 2010). This would
indicate a hump-shaped distribution of habitat richness along the elevation gradient on
Tenerife with maximum richness at mid-altitudes, i.e. within the potential area of ther-
mophilous woodlands and lower laurel forests. Similar patterns have been reported in other
regions of the world and on islands (McCain 2007; Jakobs et al. 2010). Several expla-
nations have been put forward for this pattern such as the mid-domain effect caused by
overlapping altitudinal species ranges (Rahbek 1995), decreasing area effect with
increasing elevation (Ko
¨
rner 2007) or water-energy-availability (O’Brien et al. 2000;
Currie et al. 2004; McCain 2007).
Modeling within habitat richness patterns by applying GLMs, we found that overall
richness, richness of endemic species and the number of thermophilous species recorded in
juniper patches were best predicted by mean summer rainfall showing a uni-modal rela-
tionship. This observation might be explained by the water-energy-hypothesis (Rosen-
zweig and Abramsky 1993), since the drier, lower part of the island within the habitat of
succulent scrub perennial plant richness was found to positively correlate with mean
annual precipitation (Otto et al. 2001). The positive correlation of richness with precipi-
tation at juniper sites with low and intermediate water availability would represent the
continuation of this trend. On the other hand, structural vegetation changes could be
responsible for the slight decrease of richness in juniper patches at sites with higher water
availability in the transition zone to laurel and pine forest, where increasing tree cover
Table 5 Mean richness values of typical habitat species and some structural characteristics for the clas-
sified juniper woodland types
Characteristics G1 G2 G3 G4 G5 G6 G7 G8
Thermophilous sp. 4.8 3.8 3.1 4.5 5.4 4.7 3.6 2.4
Succulent scrub species 8.6 9.5 12.3 6.2 8.5 10.3 14.6 6.5
Laurel forest species 1.2 1.1 1.4 3.5 5.0 1.7 1.0 1.0
Pine forest species 1.2 2.6 5.5 4 1.9 1.2 1.5 0.9
Total richness 15.8 17.0 22.3 18.2 20.8 17.9 21.7 10.8
Altitude (m) 660.0 708.0 569.0 676 403.0 398.0 390.0 282.0
Tree cover (%) 30.0 21.1 14.2 22.8 38.8 23.5 11.3 14.5
Shrub cover (%) 25.0 45.3 26.6 21.7 38.7 51.7 38.2 44.5
Herb cover (%) 49.0 10.7 2.8 14.1 4.1 18.5 6.5 46.0
Soil cover (%) 20.4 18.1 15.7 21.5 40.7 48.3 28.6 41.0
Biodivers Conserv (2012) 21:1811–1834 1825
123

possibly limits understory plant richness due to competition for light. A similar hump-
shaped richness pattern has been reported for roadside plant communities along the
principal elevation gradient on Tenerife (Are
´
valo et al. 2005). The fact that thermophilous
woodland species richness was not predicted by temperature seems to indicate that pre-
cipitation, i.e. mean summer precipitation, is limiting the distribution of this habitat at
lower altitudes. This is consistent with the findings that compared to the windward slope,
juniper stands grow at higher and colder sites in the South of the island where water
availability is sufficient.
The increase of richness of native, non-endemic species with decreasing mean annual
precipitation in the studied juniper woodlands can be explained by the increasing partic-
ipation of succulent scrub species in the drier South of the island and their floristic rela-
tionship with shrub communities in northern Africa (Otto et al. 2001). Some typical species
of the succulent scrub (Euphorbia balsamifera, Launaea arborescens, Lycium intricatum,
etc.) are not endemic but shared with similar communities in Northwest Africa.
The strong negative effect of plot herb cover on richness of both endemic and alien perennial
plants in juniper patches is probably related to human disturbance. Herb cover here is mainly
formed by perennial grasses, which clearly indicate the influence of grazing and agricultural
activities in the past, in the form of abandoned fields. Sites with high grass cover ([30 %) are
strongly degraded and support lower number of shrub species independent of origin.
The degree of human activity within the landscape, here represented by the area of
urbanized and agricultural land in the surroundings of the juniper patches, had a weak
negative effect on richness of endemic species but a positive effect on the number of
perennial alien species. In contrast to this negative relationship, a positive correlation
between alien and endemic species was found in roadside communities along an elevation
gradient on Tenerife (Are
´
valo et al. 2005). However, annuals were also included in the
latter study, a species group that comprised the highest proportion of the alien flora of the
Canary Islands. Therefore, the interpretation of our findings is limited in this context.
Our results show that perennial alien plants are rather scarce in the juniper woodlands
but one invasive species, Opuntia maxima, frequently grows in this habitat with inter-
mediate cover values, where it clearly competes with many endemic species, including the
Canarian juniper. This noncolumnar cactus was introduced from Mexico to the Canary
Islands in the sixteenth century for cultivation of fruits, fencing and the production of a red
dye that was elaborated from the infesting cochineal insect Dactylopius coccus. As in other
regions with Mediterranean climate (Vila
`
et al. 2003; Erre et al. 2009), O. maxima has
rapidly spread into not only human disturbed areas, such as abandoned fields, but also
semi-natural shrublands in the lower parts of all the Canary Islands due to the very
successful recruitment by seedlings and cladodes and the positive interaction with native
dispersers (Gimeno and Vila
`
2002; Padro
´
n et al. 2011). Cover of O. maxima in Canarian
juniper woodlands was weakly negatively correlated (Pearson coefficient: 0.33, p = 0.017)
with the distance to urban nuclei, which highlights the importance of landscape transfor-
mation in understanding the distribution and spread of this species (Vila
`
et al. 2003).
Overall, our results fit with the general findings that human disturbance is a strong driver
of alien species richness and determines the invasion process on oceanic islands, which has
been observed not only at island level (Denslow et al. 2009; Jakobs et al. 2010; Kueffer et al.
2010), but also at landscape and habitat level (Pretto et al. 2010). The importance of distance
to nearest urban nuclei for alien plant richness in roadside communities has already been
reported on the Canary Islands (Are
´
valo et al. 2005; Arteaga et al. 2009).
The most detailed study on J. turbinata ssp. canariensis has so far been carried out on
the island of El Hierro (von Gaisberg 2005), where a negative correlation between canopy
1826 Biodivers Conserv (2012) 21:1811–1834
123

cover and understory plant richness of juniper stands was reported. Although canopy cover
of some of our plots reached 85 %, we did not detect any correlations between this
structural variable and plant richness. However, we cannot reject the hypothesis that the
very high diversity of endemic species in the understory vegetation is partly related to the
degradation of the tree layer and the subsequent colonization of endemic shrubs typical of
substitution communities. Nevertheless, except when considering the transitional zones to
forests, the habitat of the Canarian juniper is expected to be a rather open woodland with
participation of many shrub species (von Gaisberg 2005). Decrease of understory plant
diversity with increasing tree cover has been observed for other juniper woodlands (Miller
et al. 2000).
Species composition
In contrast to the richness patterns, DCA and cluster analyses revealed that the climatic
differences between windward and leeward slope of the island had the strongest influence
on plant species composition within the habitat of juniper woodlands. The effect of the
exposure to the humid northeastern trade winds has already been highlighted for the whole
island of Tenerife (Ferna
´
ndez-Palacios and de Nicola
´
s 1995), as well as for a single habitat,
the pine forest (Rivas-Martı
´
nez et al. 1993). Here, we can also confirm this pattern for the
juniper woodlands on Tenerife, since the whole distribution range of this species that
potentially rings the island was covered (Del Arco et al. 2006a). Furthermore, the contact
of the studied habitat with three zonal ecosystems present on the Canary Islands, laurel
forest, pine forest and succulent scrub, can be confirmed.
On the windward slope and at altitudes of 300–500 m a.s.l., we observed that laurel
forest species usually participate in the formation of humid juniper woodland, which is
here represented by cluster groups G5 and G6. The second one can be considered a
degraded variant of the more conserved first type with higher species richness. The selected
indicator species of type G5, as well as its high tree and soil cover, confirm the transition
character between juniper woodland and laurel forest. A similar formation has been
reported from the island of El Hierro (von Gaisberg 2005). Juniper patches at altitudes of
500–600 m on the windward slope of Tenerife are usually found on steep rocky slopes,
since this zone, where more developed soils are available, would potentially already belong
to the laurel forest. On the other hand, J. turbinata ssp. canariensis can grow close to the
coast at favorable sites in the North of the Western Canary Islands (von Gaisberg 2005; Del
Arco et al. 2006a; Ferna
´
ndez-Palacios et al. 2008).
In the South of Tenerife, the Canarian juniper has a wide altitudinal distribution range
(300–1100 m), and some isolated individuals have even been found in the Teide crater of
Las Can
˜
adas at more than 2,000 m (Sventenius 1946). In lower regions, juniper stands are
mixed with the succulent scrub: a vegetation type that is highly adapted to hydric stress
almost over the whole year (Otto et al. 2001; Otto 2003). This habitat transition is rep-
resented by cluster group G7 and many indicator species. At the upper limit of the southern
distribution range, juniper patches with participation of pine forest species were found. The
exact location of this transitional zone from juniper woodlands to pine forest depends not
only on local climatic conditions but also on the substrate type, since pine forest has been
found to descend to the coast on salic lava flows in the SW and SE sector of the island (Del
Arco et al. 2006b). Cluster groups G3 and G4 represented this influence of pine forest.
Finally, cluster type G2 included strongly degraded juniper stands at intermediate and
higher elevations on both slopes, characterized by high abundances of substitution shrub
species such as Euphorbia lamarckii, Cistus monspeliensis and Artemisia thuscula.
Biodivers Conserv (2012) 21:1811–1834 1827
123

In general, the last remnants of juniper woodlands on Tenerife have been found to
exhibit both an extraordinary plant diversity at the plot level, i.e. high a-diversity, espe-
cially of endemic plants and a high floristic variation within the island, i.e. high species
turnover between sites or high b-diversity. Both of these findings are related firstly to
climatic conditions, the result of the steep environmental gradients typical for most of the
Canary Islands, and secondly to human disturbance and corresponding structural changes
of the vegetation. The effect of landscape transformation by humans on species compo-
sition has also been reported for other Juniperus species (Milios et al. 2007).
Although two major distribution gaps of the Canarian juniper on Tenerife currently
exist, coinciding with the most populated areas around the capital Santa Cruz and between
the cities of La Laguna and Puerto de la Cruz, ecological characterization supported the
idea that the habitat would potentially be circuminsular.
Consequences for conservation
The closeness of all juniper patches to intensive human activities (agriculture, urbaniza-
tions, road constructions), the scattered distribution of the last remnants over the island and
the very low number of juniper patches with more than 100 individuals confirmed not only
the immense destruction and degradation of the original vegetation and the heavy land-
scape transformation at mid-altitudes of Tenerife, the so called ‘‘medianı
´
as’’ (Ferna
´
ndez-
Palacios et al. 2008; Del Arco et al. 2010), but also demonstrated that this habitat is
obviously threatened on Tenerife. Given the exceptional plant diversity of this habitat,
a priority at European level, and the fact that 39 % of the studied juniper patches are not
included in protected natural areas (Martı
´
n-Esquivel et al. 1995), the priority for conser-
vation should be the immediate protection of all the remnants of juniper woodlands on the
island of Tenerife. This is also justified by the findings that J. turbinata ssp. canariensis has
a low regeneration capacity on this island due to low growth rates, dispersal difficulties and
regeneration niches that depend on favorable environmental conditions and structural
characteristics of the vegetation (Ferna
´
ndez-Palacios et al. 2008; Otto et al. 2010). In most
of the juniper patches studied in the drier South of the island, no regeneration of the
Canarian juniper and rarely fruit production have been observed (Otto and Barone, unpubl.
data.). Considering that global climate change will also affect Tenerife by increasing
temperatures (Martı
´
n-Esquivel et al. 2012), many of the juniper stands at lower altitudes in
the South of the island will probably disappear in the future due to increasing environ-
mental stress and lack of regeneration causing a local loss of biodiversity, including the
loss of genetic diversity of these populations (Terrab et al. 2008).
After protecting the remaining juniper patches, eradication of the aggressive invader
Opuntia maxima should be considered. Since this habitat revealed the highest degree of
destruction and alteration of all major zonal ecosystems of the Canary Islands (Del Arco
et al. 2010), restoration activities are urgently needed and should have priority in con-
servation plans. Considering that the highest diversity of juniper patches in endemic and
thermophilous species are expected where water availability is higher, restoration projects
would have best success on the more humid windward slope and probably also in the upper
parts of the Gu
¨
ı
´
mar Valley in the Southeast of the island.
Acknowledgments We thank the local authorities (Cabildo Insular de Tenerife, A
´
rea de Medio Ambiente
y Paisaje) and the European Commission of Environment for funding the LIFE Project (LIFE04/NAT/ES/
000064) including this study. We are grateful to Carlos Gonza
´
lez Escudero, M
a
Candelaria Rodrı
´
guez
Rodrı
´
guez and Silvia Ferna
´
ndez Lugo for field data collection.
1828 Biodivers Conserv (2012) 21:1811–1834
123

Appendix
Table 6.
Table 6 List of species included
in Fig. 3
Species Status
Aeonium Lindley SIE
Aeonium smithii SIE
Ageratina adenophora ALI
Aichryson laxum CAN
Allagopappus canariensis CAN
Andryala pinnatifida CAN
Apollonias barbujana MAC
Arbutus canariensis CAN
Argyranthemum gracile SIE
Aristida adscensionis NAT
Artemisia thuscula CAN
Asparagus arborescens CAN
Asparagus umbellatus MAC
Asplenium onopteris NAT
Atalanthus pinnatus CAN
Bosea yervamora CAN
Bupleurum salicifolium MAC
Bystropogon origanifolius CAN
Campylanthus salsoloides CAN
Canarina canariensis CAN
Carlina salicifolia MAC
Ceballosia fruticosa CAN
Ceropegia fusca CAN
Chamaecytisus proliferus CAN
Cheilanthes pulchella NAT
Cistus monspeliensis NAT
Cistus symphytifolius CAN
Convolvulus floridus CAN
Crambe strigosa CAN
Daphne gnidium NAT
Descurainia millefolia CAN
Echium aculeatum CAN
Echium strictum CAN
Echium virescens SIE
Erica arborea NAT
Erysimum bicolor MAC
Euphorbia atropurpurea SIE
Euphorbia balsamifera NAT
Euphorbia canariensis CAN
Euphorbia lamarckii ssp. lamarckii SIE
Biodivers Conserv (2012) 21:1811–1834 1829
123

Table 6 continued
Species Status
Euphorbia lamarckii ssp. wildpretii CAN
Globularia salicina MAC
Hyparrhenia sinaica NAT
Hypericum glandulosum MAC
Hypericum grandifolium MAC
Ilex canariensis MAC
Isoplexis canariensis CAN
Jasminum odoratissimum MAC
Kickxia scoparia CAN
Kleinia neriifolia CAN
Laurus novocanariensis NAT
Lavandula buchii SIE
Lavandula canariensis CAN
Lotus sessilifolius CAN
Maytenus canariensis CAN
Myrica faya NAT
Neochamaelea pulverulenta CAN
Olea cerasiformis CAN
Opuntia dillenii ALI
Opuntia maxima ALI
Origanum vulgare NAT
Pericallis lanata SIE
Pericallis tussilaginis CAN
Periploca laevigata MAC
Phagnalon saxatile MAC
Phoenix canariensis CAN
Phyllis viscosa CAN
Picconia excelsa MAC
Pinus canariensis CAN
Pistacia atlantica NAT
Polycarpaea aristata CAN
Polypodium macaronesicum NAT
Retama rhodorhizoides CAN
Rhamnus crenulata CAN
Rubia fruticosa MAC
Rumex lunaria CAN
Ruta pinnata CAN
Sideritis oroteneriffae SIE
Sonchus acaulis CAN
Sonchus congestus CAN
Tamus edulis MAC
Teline canariensis CAN
Teucrium heterophyllum MAC
1830 Biodivers Conserv (2012) 21:1811–1834
123

References
Acebes JR, Leo
´
n MC, Rodrı
´
guez ML, Del Arco MJ, Garcı
´
a-Gallo A, Pe
´
rez-de-Paz PL, Rodrı
´
guez-Delgado
O, Martı
´
n VE, Wildpret W (2010) Pteridophyta, spermatophyta. In: Arechavaleta M, Rodrı
´
guez S,
Zurita N, Garcı
´
a A (eds) Lista de especies silvestres de Canarias. Hongos plantas y animales terrestres.
Gobierno de Canarias, Tenerife, pp 119–172
Adams RP (2008) Junipers of the world: the genus Juniperus, 2nd edn. Trafford Publication, Vancouver
Adams RP, Pandey RN, Rezzi S, Casanova J (2002) Geographic variation in the Random Amplified
Polymorphic DNAs (RAPDs) of Juniperus phoenicea, J. p. var. canariensis, J. p. subsp. eumediter-
ranea, and J. p. var. turbinata. Biochem Syst Ecol 30:223–229
Adams RP, Nguyen S, Achak N (2006) Geographic variation in Juniperus phoenicea (Cupressaceae) from
the Canary Islands, Morocco and Spain based on RAPDs analysis. Phytologia 88(3):270–278
Adams RP, Rumeu B, Nogales M, Fontinha S (2009) Geographic variation and systematics of Juniperus
phoenicea l. from Madeira and the Canary islands: analyses of leaf volatile oils. Phytologia
91(1):39–52
Are
´
valo JR, Delgado JD, Otto R, Naranjo A, Salas M, Fernandez-Palacios JM (2005) Distribution of alien
vs. native plant species in roadside communities along an altitudinal gradient in Tenerife and Gran
Canaria (Canary Islands). Perspect Plant Ecol Evol Syst 7:185–202
Are
´
valo JR, Otto R, Escudero C, Ferna
´
ndez-Lugo S, Arteaga M, Delgado J, Ferna
´
ndez-Palacios JM (2010)
Do anthropogenic corridors homogenize plant communities at a local scale? A case studied in Tenerife
(Canary Islands). Plant Ecol 209:23–25
Arteaga MA, Delgado JD, Otto R, Fernandez-Palacios JM, Are
´
valo JR (2009) How do alien plants distribute
along roads on oceanic islands? A case study in Tenerife, Canary Islands. Biol Invasions 11:1071–1086
Bacaro G, Santi E, Rocchini D, Pezzo F, Puglisi L, Chiarucci A (2011) Geostatistical modelling of regional
bird species richness: exploring environmental proxies for conservation purpose. Biodivers Conserv
20:1677–1694
Blondel J, Aronson J (1999) Biology and Wildlife of the Mediterranean region. Oxford University Press,
Oxford
Cha
´
vez V, Macdonald SE (2005) The influence of canopy patch mosaics on understory plant community
composition in boreal mixedwood forest. Forest Ecol Manag 259:1067–1075
Crawley MJ (1993) GLIM for ecologists. Blackwell Scientific Publications, Oxford
Criado C (1982) Nota geogra
´
fica sobre los sabinares de Anaga. VVAA Homenaje a Alfonso Trujillo. Aula
de Cultura de Tenerife, Santa Cruz de Tenerife, pp 453–469
Cronk QCB, Fuller JL (1995) Plant invaders. The threat to natural ecosystems. Chapman and Hall, London
Currie DJ, Mittelbach GG, Cornell HW, Field R, Gue
´
gan JF, Hawkins BA, Kaufman DM, Kerr JT,
Oberdorff T, O’Brien EM, Turner JRG (2004) Predictions and tests of climate-based hypotheses of
broad-scale variation in taxonomic richness. Ecol Lett 7:1121–1134
Del Arco M, Wildpret W, Pe
´
rez-de-Paz PL, Rodrı
´
guez-Delgado O, Acebes JR, Garcı
´
a-Gallo A, Martı
´
n VE,
Reyes-Betancort JA, Salas M, Bermejo JA, Gonza
´
lez R, Cabrera MV, Garcı
´
a S (2006a) Mapa de
vegetacio
´
n de Canarias. GRAFCAN SL, Santa Cruz de Tenerife
Del Arco M, Pe
´
rez-de-Paz PL, Acebes JR, Gonza
´
lez-Mancebo JM, Reyes-Betancort JA, Bermejo JA, De
Armas S, Gonza
´
lez-Gonza
´
lez R (2006b) Bioclimatology and climatophilous vegetation of Tenerife
(Canary Islands). Ann Bot Fenn 43(3):167–192
Del Arco M, Gonza
´
lez-Gonza
´
lez R, Garzo
´
n-Machado V, Pizarro-Herna
´
ndez B (2010) Actual and potential
natural vegetation on the Canary Islands and its conservation status. Biodivers Conserv 19:3089–3140
Denslow JS, Space JC, Thomas PA (2009) Invasive exotic plants in the tropical pacific islands: patterns of
diversity. Biotropica 41:162–170
Diodato N (2005) The influence of topographic co-variables on the spatial variability of precipitation over
small regions of complex terrain. Int J Climatol 25:351–363
Dobson AJ (1990) An introduction to generalized linear models. Chapman and Hall, London
Table 6 continued
SIE Single island endemic, CAN
Canary endemic, MAC
Macaronesian endemic, NAT
native non-endemic, ALI alien
Species Status
Tolpis laciniata CAN
Viburnum rigidum CAN
Visnea mocanera MAC
Biodivers Conserv (2012) 21:1811–1834 1831
123

Dolan BJ, Parker JR (2005) Ecosystem classification in a flat, highly fragmented region of Indiana, USA.
Forest Ecol Manag 219:109–131
Domı
´
nguez-Lozano F, Price J, Otto R, Ferna
´
ndez-Palacios JM (2010) Using taxonomic and phylogenetic
evenness to compare diversification in two Island Floras. Perspect Plant Ecol Evol Syst 12:93–106
Dufre
ˆ
ne M, Legendre P (1997) Species assemblages and indicator species: the need for a flexible asym-
metrical approach. Ecol Monogr 67:345–366
El-Bana M, Shaltout K, Khalafallah A, Mosallam H (2010) Ecological status of the Mediterranean Juni-
perus phoenicea L. Relicts in the desert mountains of North Sinai. Egypt Flora 205:71–178
Erre P, Chessa I, Nieddu G, Jones PG (2009) Diversity and spatial distribution of Opuntia spp. in the
Mediterranean Basin. J Arid Environ 73:058–1066
Farjon A (2005) Monograph of Cupressaceae and Sciadopitys. Royal Botanic Gardens, Kew, London
Ferna
´
ndez-Palacios JM (1987) Ana
´
lisis de la variacio
´
n espacial de la vegetacio
´
n de Tenerife e interpretacio
´
n
en te
´
rminos ambientales. Doctoral thesis, Universidad de La Laguna (unpublished)
Ferna
´
ndez-Palacios JM (1992) Climatic response of plant species on Tenerife (Canary Island). J Veg Sci
3:595–602
Ferna
´
ndez-Palacios JM, de Nicola
´
s JP (1995) Altitudinal pattern of vegetation variation on Tenerife. J Veg
Sci 6:183–190
Ferna
´
ndez-Palacios JM, Otto R, Delgado JD, Are
´
valo JR, Naranjo A, Gonza
´
lez-Artiles F, Morici C, Barone
R (2008) Los Bosques Termo
´
filos de Canarias. Proyecto LIFE04/NAT/ES/000064. Excmo. Cabildo
Insular de Tenerife, Santa Cruz de Tenerife
Gardner AS, Fisher M (1996) The distribution and status of the montane juniper woodlands of Oman.
J Biogeogr 23:791–803
Gauch HG Jr (1982) Multivariate analysis in community ecology. Cambridge University Press, Cambridge
Gauquelin T, Bertaudiere V, Montes N, Badri W, Asmode JF (1999) Endangered stands of thuriferous
juniper in the western Mediterranean basin: ecological status, conservation and Management. Biodi-
vers Conserv 8:1479–1498
Gimeno I, Vila
`
M (2002) Recruitment of two Opuntia species invading abandoned olive groves. Acta Oecol
23:239–246
Gonza
´
lez-Artiles FJ (2006) El Bosque Termo
´
filo de Gran Canaria. Doctoral thesis, Departamento de Bio-
logı
´
a, Universidad de Las Palmas de Gran Canaria
Hill MO, Gauch HG Jr (1980) Detrended correspondence analysis, an improved ordination technique.
Vegetatio 42:47–58
Jakobs G, Kueffer C, Daehler CC (2010) Introduced weed richness across altitudinal gradients in Hawai’i:
humps, humans and water-energy dynamics. Biol Invasions. doi:10.1007/s10530-010-9816-6
Ko
¨
rner C (2007) The use of ‘‘altitude’’ in ecological research. Trends Ecol Evol 22:569–574
Kreft H, Jetz W, Mutke J, Kier G, Barthlott W (2008) Global diversity of island floras from a macroeco-
logical perspective. Ecol Lett 11:116–127
Kueffer C, Daehler CC, Torres-Santana C, Lavergne C, Meyer JY, Otto R, Silva L (2010) A global
comparison of invasive plant species on oceanic islands. Perspect Plant Ecol Evol Syst 12(2):145–161
Linares-Palomino R, Kvist LP, Aguirre-Mendoza Z, Gonza
´
les-Inca C (2010) Diversity and endemism of
woody plant species in the Equatorial Pacific seasonally dry forests. Biodivers Conserv 19:169–185
Luis M, Ferna
´
ndez-Pello L, Quirantes F (2005) El papel de las transiciones laterales en el escalonamiento
vegetal: el ejemplo del sabinar de Afur (Tenerife, Islas Canarias). In: VVAA Espacios pu
´
blicos,
espacios privados. Un debate sobre el territorio, pp 276–279
Mack RN, Simberloff D, Lonsdale WM, Evans HC, Clout M, Bazzaz FA (2000) Biotic invasions: causes,
epidemiology, global consequences and control. Ecol Appl 10:689–710
Marrero MV, Rodrı
´
guez O, Wildpret W (1992) Contribucio
´
n al estudio fitocorolo
´
gico de los restos de
sabinares y otras comunidades termo
´
filas del Sur de Tenerife (Islas Canarias). Rev Acad Canar Cienc
3(4):25–44
Martı
´
n-Esquivel JL, Garcı
´
a Court H, Redondo-Rojas CE, Garcı
´
a-Ferna
´
ndez I, Carralero Jaime I (1995) La
Red Canaria de Espacios Naturales Protegidos. Consejerı
´
a de Polı
´
tica Territorial, Viceconsejerı
´
ade
Medio Ambiente, Gobierno de Canarias
Martı
´
n-Esquivel JL, Bethencourt J, Cuevas-Agullo
´
E (2012) Assessment of global warming on the island of
Tenerife, Canary Islands (Spain). Trends in minimum, maximum and mean temperatures since 1944.
Clim Chang. doi:10.1007/s10584-012-0407-7
McCain CM (2007) Could temperature and water availability drive elevational species richness patterns? A
global case study for bats. Global Ecol Biogeogr 16:1–13
McCullagh P, Nelder JA (1989) Generalized linear models. Chapman and Hall, London
McCune B, Grace JB (2002) Analysis of ecological communities. MjM Software Design. Gleneden Beach,
El Orego
´
n
1832 Biodivers Conserv (2012) 21:1811–1834
123

McCune B, Mefford MJ (2011) PC-ORD. Multivariate analysis of ecological data. Version 6.0 MjM
Software Design, Gleneden Beach, Oregon, USA
Me
´
dail F, Que
´
zel P (1997) Hot-spots analysis for conservation of plant biodiversity in the Mediterranean
Basin. Ann Mo Bot Gard 84:112–127
Milios E, Pipinis E, Petrou P, Akritidou S, Smiris P, Aslanidou M (2007) Structure and regeneration patterns
of the Juniperus excelsa Bieb. stands in the central part of the Nestos valley in the northeast of Greece,
in the context of anthropogenic disturbances and nurse plant facilitation. Ecol Res 22:713–723
Miller RF, Svejcar TJ, Rose JA (2000) Impacts of western juniper on plant community composition and
structure. J Range Manag 53(6):574–585
Montesinos D, Otto R, Ferna
´
ndez Palacios JM (2009) 9560 Bosques ende
´
micos de Juniperus spp. (*). In:
VVAA Bases ecolo
´
gicas preliminares para la conservacio
´
n de los tipos de ha
´
bitat de intere
´
s comun-
itario en Espan
˜
a. Madrid: Ministerio de Medio Ambiente, y Medio Rural y Marino. Madrid, pp 84
Mun
˜
oz-Reinoso JC (2004) Diversity of maritime juniper woodlands. Forest Ecol Manag 192:267–276
Nezadal W, Welss W (2009) Aportaciones al conocimiento del bosque termo
´
filo en el noroeste de Tenerife
(Islas Canarias). In: Beltran-Tejera E, Afonso-Carillo J, Garcı
´
a-Gallo A, Rodrı
´
guez-Delgado O (eds),
Homenaje al profesor Dr. Wolfredo Wildpret de la Torre, Instituto de Estudios Canarios, La Laguna
(Tenerife, Islas Canarias), Monografı
´
a, vol 78, pp 229–244
O’Brien EM, Field R, Whittaker RJ (2000) Climatic gradients in woody plant (tree and shrub) diversity:
water-energy dynamics, residual variation, and topography. Oikos 89:588–600
Otto R (2003) Spatial and temporal variability of succulent scrub on Tenerife and its ecological interpre-
tation. Doctoral Thesis (English summary), Institute of Geography. University of Zu
¨
rich
Otto R, Ferna
´
ndez-Palacios JM, Kru
¨
si BO (2001) Variation in species composition and vegetation structure
of succulent scrub on Tenerife in relation to environmental variation. J Veg Sci 12:237–248
Otto R, Kru
¨
si BO, Delgado JD, Ferna
´
ndez-Palacios JM, Garcı
´
a-del-Rey E, Are
´
valo JR (2010) Regeneration
niche of the Canarian juniper: the role of adults, shrubs and environmental conditions. Ann For Sci.
doi:10.1051/forest/2010034
Padro
´
n B, Nogales M, Traveset A, Vila
`
M, Martı
´
nez-Abraı
´
n A, Padilla DP (2011) Integration of invasive
Opuntia spp. by native and alien seed dispersers in the Mediterranean area and the Canary Islands. Biol
Invasions 13:831–844
Pretto F, Celesti-Grapow L, Carli E, Blasi C (2010) Influence of past land use and current human distur-
bance on non-native plant species on small Italian islands. Plant Ecol 210:225–239
Rahbek C (1995) The elevational gradient of species richness: a uniform pattern? Ecography 18:200–205
Reyes-Betancort JA, Wildpret W, del Arco M, Pe
´
rez-de-Paz PL (2001) The vegetation of Lanzarote (Canary
Islands). Phytocoenologia 31:185–247
Rivas-Martı
´
nez S, Wildpret M, del Arco M, Rodrı
´
guez O, Pe
´
rez-de-Paz PL, Garcı
´
a-Gallo A, Acebes JR,
Dı
´
az-Gonza
´
lez TE, Ferna
´
ndez-Gonza
´
lez F (1993) Las comunidades vegetales de la Isla de Tenerife
(Islas Canarias). Itinera Geobot 7:169–374
Rodrı
´
guez O, Marrero MV (1991) Evolucio
´
n y aprovechamiento de los bosques termo
´
filos (los ‘‘montes
bajos’’) de la Isla de Tenerife. Anuario de Estudios Atla
´
nticos 36, pp 595–630
Rodrı
´
guez O, Wildpret W, del Arco M, Pe
´
rez-de-Paz PL (1990) Contribucio
´
n al estudio fitosociolo
´
gico de
los restos de sabinares y otras comunidades termo
´
filas de la Isla de Tenerfie (Canarias). Rev Acad
Canar Cienc 2:121–142
Rosenzweig ML, Abramsky Z (1993) How are Diversity and Productivity related? In: Ricklefs RE, Schluter
D (eds) Species diversity in ecological communities. University of Chicago Press, Chicago, pp 52–65
Santos A, Ferna
´
ndez M (1983) Vegetacio
´
n del macizo de Teno. Datos para su conservacio
´
n. Comunic II
Congr Pro-Flora Macarone
´
sica 1977:385–424
Scott JM, Murray M, Wright RG, Csuti B, Morgan P, Pressey RL (2001) Representation of natural vege-
tation in protected areas: capturing the geographic range. Biodivers Conserv 10:1297–1301
Steinbauer MJ, Otto R, Naranjo-Cigala A, Beierkuhnlein C, Ferna
´
ndez-Palacios JM (2011) Increase of
island endemism with altitude-speciation processes on oceanic islands. Ecography. doi:
10.1111/j.1600-0587.2011.07064.x
Stierstorfer C, von Gaisberg M (2006) Annotated checklist and distribution of the vascular plants of El
Hierro, Canary Islands, Spain. Englera 27:1–221
Sventenius ERS (1946) Notas sobre la flora de las Can
˜
adas de Tenerife. Boletı
´
n Instituto Nacional de
Investigaciones Agrono
´
micas 15(78):149–171
Ter Braak CJF, Smilauer P (1998) CANOCO reference manual and user’s guide to Canoco for windows,
software for canonical community ordination (version 4). Microcomputer Power, Ithaca
Terrab A, Scho
¨
nswetter P, Talavera S, Vela E, Stuessy TF (2008) Range-wide phylogeography of Juniperus
thurifera L., a presumptive keystone species of western Mediterranean. Mol Phylogenet Evol
48(1):94–102
Biodivers Conserv (2012) 21:1811–1834 1833
123

Thompson JD (1999) Population differentiation in Mediterranean plants: insights into colonization history
and the evolution and conservation of endemic species. Heredity 82:229–236
Vila
`
M, Burriel JA, Pino J, Chamizo J, Llach E, Porterias M, Vives M (2003) Association between Opuntia
species invasion and changes in land-cover in the Mediterranean region. Glob Chang Biol
9:1234–1239
von Gaisberg M (2005) Die Vegetation der Fußstufe von El Hierro (Kanarische Inseln). Diss Bot 395:1–364
Whittaker RJ, Ferna
´
ndez-Palacios JM (2007) Island biogeography. Ecology, evolution and conservation.
Oxford University Press, New York
Zobel M, Otto R, Laanisto L, Naranjo-Cigala A, Pa
¨
rtel M, Ferna
´
ndez-Palacios JM (2011) The formation of
species pools: historical habitat abundance affects current local diversity. Global Ecol Biogeogr
20(2):251–259
Zuur AF, Ieno EN, Smith GM (2007) Analysing ecological data. Springer, USA
1834 Biodivers Conserv (2012) 21:1811–1834
123