Long-term impact of protected colonial birds on a
jeopardized cork oak population: conservation bias
leads to restoration failure
Jos
e M. Fedriani
1,2
*, Luis V. Garc
ıa
3
, Mar
ıa E. S
anchez
4
, Juan Calder
on
1
and
Cristina Ramo
1
1
Estaci
on Biol
ogica de Do
~
nana (EBD-CSIC), c/Am
erico Vespucio s/n, 41092 Seville, Spain;
2
Centre for Applied
Ecology ‘Prof. Baeta Neves’/InBIO, Institute Superior of Agronomy, University of Lisbon, Tapada da Ajuda, 1349-017
Lisboa, Portugal;
3
Instituto de Recursos Naturales y Agrobiolog
ıa de Sevilla (IRNAS-CSIC), P.O. Box 1052, 41080
Seville, Spain; and
4
Agronomy Department (Agroforest Pathology), University of C
ordoba, Ctra. Madrid-C
adiz Km
396, 14014 C
ordoba, Spain
Summary
1. Conservation bias towards flagship species sometimes threatens other species of chief con-
cern. Long-term studies of potential harm by favoured species on other sensitive species,
though seldom adopted, are required to fairly evaluate the suitability of management and
restoration efforts.
2. We illustrate the potential detrimental outcomes of conservation biased towards birds by
investigating the long-term (1963–2009) impact of a large waterbird colony (up to 13 000
pairs) on a remnant cork oak Quercus suber population at a World Biosphere Reserve in
south-western Spain (the Do
~
nana National Park). To this end, we compared changes in per-
formance (growth, crown vigour and survival) of oaks occup ied and unoccupied by the water-
bird colony.
3. After 46 years of occupation, the risk of death to centenarian oaks in the area occupied
by the colony was over twofold higher than for trees outside the area. Non-centenarian
planted and naturally regenerated oaks showed similar trends, leading to restoration failure.
4. Synthesis and applications. Our long-term study reveals that waterbirds and centenarian
oaks cannot coexist, at the most local scale, but they can at a regional scale including within the
Do
~
nana area. We propose immediate plantin g efforts in suitable colony-free areas, while
managers evaluate the feasibility of relocating colonial waterbirds to an alternative location. To
preserve the Do
~
nana oak genetic pool, such reforestation should be accomplished using local
seeds and seedlings. New trees should not be planted in close proximity of co lony-occupied trees
since it significantly reduces their survival. Do
~
nana stakeholders should both overcome current
conservation bias in favour of birds and enter into a process of settlement to best preserve the
overall biodiversity of the system.
Key-words: conservation bias towards waterbirds, conservation conflicts, Do
~
nana National
Park, ecological restoration, genetic pool, long-term monitoring, Medit erranean ecosystem,
plant–animal interactions, Quercus suber, tree mortality
Introduction
Conservation action is prone to bias due to habitats or
ecosystems (e.g. Pressey et al. 2002; Mokhatla et al.
2012), species’ economic importance, threat status, size or
emotional appeal (Ceballos and Brown 1995; Simberloff
1998; Root-Bernstein et al. 2013). Conservation biased
towards flagship species (Simberloff 1998) may threaten
other species of concern and, thus, generate divergences
among conservationists with contrasting sensitivity or pri-
orities (Oro et al. 2009; Redpath et al. 2013; McAlpine
et al. 2016). Importantly, even moderate yet temporally
persistent habitat alterations by favoured species could
eventually harm other coexisting species (Tilman et al.
1994; Gustavsson, Lennartsson & Emanuelsson 2007).
*Correspondence author. E-mail: fedriani@ebd.csic.es
© 2016 The Authors. Journal of Applied Ecology © 2016 British Ecological Society
Journal of Applied Ecology 2017, 54, 450–458 doi: 10.1111/1365-2664.12672
Especially paradoxical are those situations in which flag-
ship species compromise their own local persistence by
reducing or even decimating the long-term survival of
other populations upon which they rely (Roemer &
Wayne 2003; Redpath et al. 2013). Long-term studies of
potential harm by favoured species on other sensitive spe-
cies, though seldom adopted, are required to fairly evalu-
ate the suitability of conservation efforts.
Birds are often prioritized in conservation and restora-
tion programmes because they are considered charismatic
species and also provide critical ecosystem services (Whe-
lan, Wenny & Marquis 2008; Verissimo, MacMillan &
Smith 2011; Green & Elmberg 2014). Nonetheless, tree
damage due to the persistent nesting of colonial water-
birds has been reported world-wide (e.g. Telfair & Bister
2004; Kolb et al. 2012). Bird detrimental effects include
direct damage by breaking branches and foliage for nest
construction or by covering leaves with faeces (Aderman
& Hill 1995; Ishida 1997) as well as indirect harm by
changing soil composition through deposition of large
amounts of guano (e.g. Breuning-Madsen et al. 2010;
Garc
ıa et al. 2011; but see Fedriani et al. 2015a). Large
scattered trees are keystone elements in savannah-like
ecosystems because their large cavities and complex
canopy structure provide vital shelter and other benefits
to numerous animal species (Fischer, Stott & Law 2010;
D
ıaz, Tietje & Barrett 2013; Lindenmayer et al. 2013). In
these ecosystems, birds can limit tree survival and recruit-
ment and thus jeopardize these keystone tree species,
especially when they act in combination with other detri-
mental factors (e.g. logging, overgrazing, droughts).
In a Mediterranean area of southern Spain, Do
~
nana
Biological Reserve (DBR), only a few scattered stands of
formerly abundant cork oaks Quercus suber currently per-
sist mostly due to a history of human-related disturbances
(e.g. timber exploitation, fires; Granados, Mart
ın &
Garc
ıa-Novo 1988). In 1967, Do
~
nana was declared a
National Park, mostly because of its importance as a
European bird reserve (Garc
ıa-Novo & Mar
ın-Cabrera
2005; Mart
ın-L
opez et al. 2009). Since then, several fac-
tors such as the decline of human disturbances and the
availability of novel food resources (e.g. the invasive cray-
fish Procambarus clarkii, creation of fish-farming ponds)
have enabled a remarkable increase in several nationally
protected tree-nesting waterbirds (Tablado et al. 2010; del
Estado 2011; Ramo et al. 2013). This enhanced waterbird
guild rapidly colonized a large fraction of the remaining
centenarian cork oaks in the most productive area of
DBR (Ramo et al. 2013). The large amount of faeces
delivered by the waterbird colony (up to 13 000 pairs)
altered soil composition and led to a decline in the health
of the oak crowns (Garc
ıa et al. 2011). To offset those
and other detrimental effects on centenarian oaks
(droughts, herbivory; Herrera 1995), several costly oak
reforestation campaigns have been undertaken during the
last few decades with limited success. Surprisingly, despite
such a palpable ‘birds vs. oaks’ conservation conflict, and
despite the pressing need for well-informed restoration
programmes, whether and how the waterbird colony has
impinged on the long-term performance and survival of
both naturally regenerated and planted oaks remains
unknown.
In this study, we investigated the potential impacts of a
large waterbird colony on a charismatic remnant popula-
tion of cork oaks at Do
~
nana World Biosphere Reserve
for five decades. We hypothesized that such a large water-
bird colony would have detrimental long-term effects on
critical individual oak performance components such as
growth, crown vigour and survival. Our analysis consid-
ered two distinct tree age classes (centenarian and non-
centenarian oaks; hereafter referred to as CO and NCO,
respectively) and sought to answer the following four
specific questions: (i) Have tree-nesting waterbirds altered
growth, crown vigour and survival of Do
~
nana CO? (ii)
Does the proximity of colony-occupied conspecifics harm
the survival of CO? (iii) Is the impact of nesting water-
birds on NCO similar to their impact on CO? and (iv)
Are colony effects similar on planted and naturally regen-
erated NCO?
Materials and methods
STUDY AREA AND SPECIES
The study was conducted at the DBR (6794 ha) in the heart of
the Do
~
nana National Park (south-west Spain). The climate is
Mediterranean with some Atlantic influences. Average annual
rainfall is 586 mm, with 84% occurring in autumn and winter,
while average monthly temperatures range from 100 °C (Jan-
uary) to 247 °C (July). The DBR has three main ecosystems:
mobile dunes, stabilized sands and marshes (Fig. 1). Oaks tend
to occupy low-lying locations with relatively moist soils and
form sparse and open savannah-like woodland (Garc
ıa-Novo
1979).
The cork oak is an evergreen species of remarkable ecological
and economic importance, endemic to the Mediterranean region
(Aronson, Pereira & Pausas 2009). It has low tolerance to cold
temperatures (frost) and severe droughts and tends to grow in
acidic soils (Aronson, Pereira & Pausas 2009). Changes in land
use and forest management (Aronson, Pereira & Pausas 2009)
and the presence of invasive soil-borne pathogens (Phytophthora
cinnamomi and Pythium spiculum;G
omez-Aparicio et al. 2012)
led to a significant decline in these forests, which are now consid-
ered endangered (Aronson, Pereira & Pausas 2009) and highly
vulnerable to forecasted climate change (Felic
ısimo et al. 2011).
Since its establishment in 1963, the wading bird colony has
moved along a strip (~365 ha) occupying oaks and other tree spe-
cies ( Populus alba, Salix sp., and Olea europaea) within the DBR.
This highly productive area, locally known as ‘Vera’ (Fig. 1), cor-
responds to the scrubland–marshland ecotone. Today the colony
consists of seven nationally protected waterbird species (del
Estado 2011): black-crowned night heron Nycticorax nycticorax,
squacco heron Ardeola ralloides, cattle egret Bubulcus ibis, little
egret Egretta garzetta, grey heron Ardea cinerea, white stork Cico-
nia ciconia and spoonbill Platalea leucorodia. The colony size var-
ies yearly from 150 to 13 000 breeding pairs (Ramo et al. 2013).
© 2016 The Authors. Journal of Applied Ecology © 2016 British Ecological Society, Journal of Applied Ecology, 54, 450–458
Conservation bias leads to restoration failure 451
WATERBIRD COLONY EFFECTS ON GROWTH, CROWN
CONDITION AND SURVIVAL OF CENTENARIAN OAKS
In 1963, most of the existing CO in the DBR (n = 454) were
marked and numbered. In subsequent years (1981, 1989, 1992,
1996 and 2009), surveys were carried out registering the number
of surviving trees both within and outside the Vera (Sol
ıs 1996).
To evaluate whether the colony had an effect on CO crown con-
dition (e.g. Carnicer et al. 2011), we compared photographs of
136 CO (occupied, n = 72; unoccupied n = 64) taken from
around the same camera positions in 1966 and 2010 (see Fig. 1).
The tree crown condition of photographed oaks was visually esti-
mated using an index of crown vigour (ICV). It consists of an
ordinal index that rises as crown conditions improve, and ranges
between 1 (alive tree in very poor condition) and 5 (very healthy
reference tree; Garc
ıa et al. 2011). For analysis, we rescaled the
index from 20 (alive tree in very poor condition) to 100% (very
healthy tree). Tree occupancy by the colony was based on data
from the annual nest census performed by the DBR Monitoring
Team (period 1985–2010), the presence of recent and old nests,
and valuable guard communications. For occupied trees, we also
evaluated whether their mean ICV values (averaged for the last
3 years of monitoring) were related to intensity of occupation,
measured as: (i) the number of years each tree was occupied dur-
ing a 25-year monitored period (1985–2009) and (ii) the accumu-
lated number of nests during a 17-year monitored period (1993–
2009). Similarly, to evaluate the long-term potential colony effect
on oak growth, the circumference at breast height of a subset of
120 trees was measured both in 1966 and 2010 (occupied, n = 75;
unoccupied, n = 45). For analyses, the standard diameter at
breast height (d.b.h.) was estimated assuming circular trunk
section (Nogueira, Nelson & Fearnside 2006).
WATERBIRD COLONY EFFECTS ON CROWN CONDITION
AND SURVIVAL OF NON-CENTENARIAN OAKS
To assess the potential colony effect on oak reforestation pro-
grammes within the Vera, we initiated in 2008 a detailed monitor-
ing of NCO that were either planted before 2005 (n = 256) or
naturally recruited (n = 56). Because the smallest size (d.b.h.) of
occupied NCO (i.e. trees that could potentially be occupied by
the colony) was 13 cm, only trees above that size were considered.
During the autumns of seven consecutive years (2008–2014), we
assessed in the field the crown vigour (ICV) and the status (alive/
dead) of each focal tree.
STATISTICAL ANALYSIS
To evaluate potential colony effects on CO crown vigour (ICV)
and growth (d.b.h.), we fit generalized linear mixed models using
Proc Glimmix in SAS (Littell et al. 2006). Centenarian oaks were
categorized within three types according to location and colony
occupancy: trees occupied and unoccupied in Vera and trees in the
scrubland (all unoccupied). Tree type, year (1966 and 2009) and
their second-order interaction were specified in the models as fixed
factors. Individual tree (i.e. block) was included as a random fac-
tor. Arcsine-transformed ICV data, normal error and identity link
function were used in the models (Bolker et al. 2009). When the
interaction between tree type and year was significant, we per-
formed tests for the effect of a given factor at the different levels of
the other factor using the SLICE option in the LSMEANS state-
ment of the MIXED procedure (Littell et al. 2006). Also, for col-
ony-occupied trees, we evaluated a potential long-term relationship
between the intensity of tree occupation (as defined above) and the
average ICV (last 3 years) by means of simple regression.
Fig. 1. Main habitat types in the Do
~
nana Biological Reserve (south-west Spain) and location of the centenarian cork oaks (CO). Solid
dots represent CO occupied by the colony at any time during the period 1963–2010, while white dots represent unoccupied CO. Dashed
lines indicate the scrubland–marshland ecotone locally known as ‘Vera’. (a) Stabilized sands (elevated ground); (b) stabilized sands (low
ground); (c) marshes; (d) mobile dunes; (e) freshwater ponds. On the right are examples of photographs taken to assess changes in oak
crown condition.
© 2016 The Authors. Journal of Applied Ecology © 2016 British Ecological Society, Journal of Applied Ecology, 54, 450–458
452 J. M. Fedriani et al.
Waterbird colony effect on CO survivorship was first evaluated
by means of failure-time analyses, fitting Cox proportional hazard
regression models (e.g. Fedriani, Zywiec & Delibes 2012). The
response variable was the number of years between the start of
monitoring in 1963 and oak death (or end of monitoring),
whereas tree type (as defined above) was the explicative variable.
Data were modelled as right-censored given the uncertainty about
when CO alive at the study end (2009) will eventually die. The
significance of tree type was evaluated following Therneau &
Grambsch (2000). We also hypothesize that dead CO in the Vera
would be surrounded by a higher number of occupied conspecifics
than expected given the mean number neighbouring occupied
oaks. To address this question, we used recent techniques of
marked spatial point pattern analysis (e.g. Fedriani et al. 2015b;
see Appendix S1, Supporting information for details).
Potential colony effects on non-centenarian planted and natu-
rally regenerated oaks within the Vera were evaluated as above.
Generalized linear mixed models (Littell et al. 2006) were used to
evaluate the effect of tree type (either occupied or not by the col-
ony; all within Vera), tree origin (planted or naturally regener-
ated) as well as their second-order interaction on mean ICV
(averaged across 2008–2014). For colony-occupied NCO, we also
evaluated the potential relationship between the intensity of col-
ony occupation and their mean ICV using simple regression.
Finally, Cox proportional hazard regression models were used to
assess the effect of tree type, tree origin and their second-order
interaction on NCO survivorship. All other procedures and con-
ventions were as detailed above.
Results
LONG-TERM WATERBIRD COLONY EFFECTS ON
CENTENARIAN OAKS
Size of CO (d.b.h.) was variable both within Vera
[7314 208 (mean 1SE); range, 2419–15113, n = 178]
and within scrubland (7885 458; range, 1910–15342;
n = 62). When we controlled for the effect of the random
factor (i.e. individual tree), our mixed model showed no
effect of tree type on oak d.b.h. (F
1,117
= 149, P = 0231).
However, there was a marked effect of year (1966 and 2010)
on d.b.h. (F
1,117
= 58685, P < 00001; Fig. 2); as would be
expected since they were 44 years older, CO in 2010 were
on average wider (~15-fold) than in 1966. Interestingly, we
found a significant interaction between tree type and year
(F
2,117
= 679, P < 0002), indicating temporal change in
d.b.h. among tree types. Specifically, tests of slices showed
that although there were no differences among tree types at
the start of the monitoring in 1966 (F
2,117
= 038,
P = 0682), by 2010 the diameter of occupied oaks was on
average 14% smaller than unoccupied trees both within and
outside Vera (F
2,117
= 322, P = 0044; Fig. 2).
Index of crown vigour (ICV) for CO was most variable
between years [8689 0013 (1966) vs. 9045 0013
(2010)]. Indeed, our mixed model showed that average ICV
was significantly higher in 2010 than in 1966 (F
1,133
= 441,
P = 0037; Fig. 3a). Though tree type did not have an effect
as main factor (F
2,133
= 129, P=0278), it showed a sig-
nificant interaction with year (F
2,133
= 359, P < 0030)
indicating that the effect of tree type was temporally incon-
sistent. Specifically, tests of slices indicated that whereas in
1966 there were no differences in ICV between occupied
and unoccupied trees (F
2,133
= 018, P=0837), during
2010 unoccupied trees (both outside and inside Vera)
showed on average ICV 7–9% higher than occupied trees
(F
2,133
= 424, P=0016; Fig. 3a). Furthermore, for col-
ony-occupied trees, there was a weak though significant
negative linear trend between the number of years occupied
by the colony and ICV (r = 0223, P = 0033; R
2
= 0050;
Fig. 4a). No relationship across CO between average ICV
and the accumulated number of nests during the monitored
period was detected (P = 0344).
Finally, Cox regression analysis indicated that survival
differed among CO inside and outside the Vera
(v
2
= 1922, d.f. = 1, P < 00001). Specifically, risk of oak
death during the 46-year monitored period was 21 times
higher in the Vera than in the scrubland (Fig. 5a). Also,
marked spatial point pattern analyses revealed that the
number of colony-occupied conspecific neighbours around
dead oaks was always much higher than expected by the
random marking model (see Appendix S1).
WATERBIRD COLONY EFFECT ON NON-CENTENARIAN
OAKS
During the 7-year period examined (2008–2014), a much
larger fraction of planted oaks (297%, n = 256) was
occupied by the colony as compared to naturally regener-
ated ones (89%, n = 56). Mean ICV values for NCO
were variable for both planted (058 0 01; range, 027–
091; n = 256) and naturally regenerated individuals
(057 001; range, 036–080; n = 56). Average ICV did
not vary between planted and naturally regenerated trees
(F
1,308
= 068, P=0410; Fig. 3b). However, there was a
significant colony effect (F
1,308
= 1406, P<0001) with
Scrubland
d.b.h. (cm)
0
25
50
75
100
125
1966
2010
Vera
unoccupied
Vera
occupied
a
a
b
a
b
c
Fig. 2. Model-adjusted mean diameter (d.b.h.; 1SE) during
1966 and 2010 of CO occupied and unoccupied by nesting water-
birds in the area used (Vera) and not used (Scrubland) by the col-
ony. Different letters denote significant differences.
© 2016 The Authors. Journal of Applied Ecology © 2016 British Ecological Society, Journal of Applied Ecology, 54, 450–458
Conservation bias leads to restoration failure 453
unoccupied trees having on average a 22% higher ICV
than occupied ones. There was no significant interaction
between tree origin and colony occupancy (F
1,308
= 036,
P = 0550). When considering only NCO occupied by the
colony (n = 81), average ICV showed a strong negative
relationship with both the number of years trees were
occupied by the colony (r = 0490, P < 00001,
R
2
= 02402; Fig. 4b) and the accumulated number of
nests (r = 0310, P < 0005, R
2
= 0096; Fig. 4c).
Survivorship of NCO clearly differed between occupied
and unoccupied ones (Cox regression, v
2
= 1116, d.f. 1,
P < 0001). Risk of death across the 7-year monitored
period was 67 times higher for occupied than for unoccu-
pied oaks (Fig. 5b). This trend was consistent for both
planted and naturally regenerated oaks, as indicated by
the lack of significant interaction between colony
occupancy and tree origin (v
2
= 094, d.f. = 3, P = 0816;
Fig. 5b). Cox regression did not detect differences in mor-
tality rate between planted and naturally regenerated
NCO (v
2
= 164, d.f. = 1, P = 0 200), though sample size
for occupied naturally regenerated oaks was small (n = 5).
Natural
Index of crown vigour
0·6
0·7
0·8
0·9
1·0
Vera occupied
Vera unoccupied
Planted
0·57
0·60
0·63
0·66
0·69
0·72
Scrubland
Vera unoccupied
Vera occupied
P < 0·050
1966
2010
Index of crown vigour
(b) Non-centenarian oaks
(a) Centenarian oaks
Fig. 3. (a) Model-adjusted mean (1SE) index of crown vigour
(ICV) during 1966 and 2010 of CO occupied and unoccupied by
the nesting waterbirds in the area used (Vera) and not used
(Scrubland) by the colony. (b) Model-adjusted mean ICV [average
across a 7-year period (2008–2014)] of NCO occupied and unoccu-
pied by the nesting waterbirds in the Vera. Differences were signif-
icant (P < 0001) for both planted and naturally regenerated trees.
(a)
(b)
(c)
0 10203040506070
Accumulated number of nests
0·5
0·6
0·7
0·8
0·9
1·0
1·1
Index of crown vigour
(2008–2014)
012345678
Occupied years
0·5
0·6
0·7
0·8
0·9
1·0
1·1
Index of growth vigour
(2008–2014)
0 2 4 6 8 10121416182022242628
Occupied years
0·5
0·6
0·7
0·8
0·9
Index of growth vigour
(last three years)
Fig. 4. Significant negative linear relationships and confidence
intervals between (a) index of crown vigour (ICV) and total num-
ber of years each CO was occupied by the colony (r = 0242,
P = 0020; R
2
= 0059), (b) ICV and total number of years each
NCO was occupied by the colony during the period 2008–2014
(r = 0490, P < 00001, R
2
= 02402), and (c) ICV and accumu-
lated number of nests held by each NCO during the period 2008–
2014 (r = 0310, P < 0005, R
2
= 0096).
© 2016 The Authors. Journal of Applied Ecology © 2016 British Ecological Society, Journal of Applied Ecology, 54, 450–458
454 J. M. Fedriani et al.
![Fig. 3. (a) Model-adjusted mean ( 1SE) index of crown vigour (ICV) during 1966 and 2010 of CO occupied and unoccupied by the nesting waterbirds in the area used (Vera) and not used (Scrubland) by the colony. (b) Model-adjusted mean ICV [average across a 7-year period (2008–2014)] of NCO occupied and unoccupied by the nesting waterbirds in the Vera. Differences were significant (P < 0 001) for both planted and naturally regenerated trees.](/figures/fig-3-a-model-adjusted-mean-1se-index-of-crown-vigour-icv-pqb4d59y.webp)
