Predation of young tortoises by ravens: the effect of habitat structure on tortoise detectability and abundance

Abstract

The predation of young tortoise is considered a major cause of mortality for many tortoise species. The predation by common ravens has been identified as being responsible for significant decreases in tortoise populations. Mediterranean spur-thighed tortoise hatchlings and juveniles in Maamora forest (Morocco) were studied in order to describe the size/age class predation of common ravens on young tortoises and infer the drivers of predation risk and population abundance. The results showed a high level of predation on young tortoises (<75 mm carapace length) attributed to ravens in areas with low vegetation cover, representing 100% of the cases of mortality (n = 147), but it was moderate in covered areas (n = 19), representing 12–27%. The population structure of living juveniles differed significantly between covered and uncovered areas, thus suggesting that raven predation might modify juvenile population structure. Finally, N-mixture models showed a positive relationship between (i) bare cover and tortoise detectability that is only evidenced when the plot is far from a perch and (ii) population abundance and shrub species-richness, being higher in uncovered areas. Our results improve the knowledge on predation and survival on this critical stage in life, which is crucial for the conservation of the Mediterranean spur-thighed tortoise.

Full text

1
SCIENTIFIC REPORTS | (2020) 10:1874 | https://doi.org/10.1038/s41598-020-58851-5
www.nature.com/scientificreports
Predation of young tortoises
by ravens: the eect of habitat
structure on tortoise detectability
and abundance
Amalia Segura
1*, José Jimenez1 & Pelayo Acevedo
1,2
The predation of young tortoise is considered a major cause of mortality for many tortoise species.
The predation by common ravens has been identied as being responsible for signicant decreases in
tortoise populations. Mediterranean spur-thighed tortoise hatchlings and juveniles in Maamora forest
(Morocco) were studied in order to describe the size/age class predation of common ravens on young
tortoises and infer the drivers of predation risk and population abundance. The results showed a high
level of predation on young tortoises (<75 mm carapace length) attributed to ravens in areas with low
vegetation cover, representing 100% of the cases of mortality (n = 147), but it was moderate in covered
areas (n = 19), representing 12–27%. The population structure of living juveniles diered signicantly
between covered and uncovered areas, thus suggesting that raven predation might modify juvenile
population structure. Finally, N-mixture models showed a positive relationship between (i) bare cover
and tortoise detectability that is only evidenced when the plot is far from a perch and (ii) population
abundance and shrub species-richness, being higher in uncovered areas. Our results improve the
knowledge on predation and survival on this critical stage in life, which is crucial for the conservation of
the Mediterranean spur-thighed tortoise.
Tortoise populations are characterised by high adult survival and low recruitment rates, which probably inuence
their demographic dynamics1. e lack of reliable information on recruitment complicates evaluations of the real
importance of young individuals as regards population demography2. In this respect, threats that limit recruit-
ment bouts may have an important regulatory eect on population dynamics3, especially when these threats
continue over long periods of time (e.g.4).
Hatchlings and juveniles are more vulnerable than adults owing to their smaller size and higher susceptibility
to environmental conditions, such as the temperature or rain, which aect their physiology, but also to vegetation
cover, which inuences thermoregulation, the availability of food and the risk of predation3,5,6. Indeed, the low
temperature and high rains of winter2,7 and the high temperature of summer8,9 are considered some of the most
common causes of mortality in hatchlings. e predation of young tortoises –their shell is so or not suciently
solid to protect them from predators– by mammals7,10,11 and birds12–14 is also considered a major cause of mortal-
ity in many tortoise species, which was particularly studied in North America7,10,12,15,16.
e common raven Corvus corax is considered to be one of the most relevant predators of tortoises and
is responsible for 70–91% of the mortality of desert tortoises Gopherus agassizii4,13,17. Both single individuals
and breeding pairs have been identied as responsible for signicant decreases in tortoise populations, also in
addition to modifying population structure by aecting juvenile size/age classes.12,18,19 Indeed, its high numbers
and opportunistic feeding habits have severe impacts on its target prey, whose population size may be reduced
without aecting the condition of the predator, which will switch to another prey when this resource becomes
scarce (e.g.4). But the prey species has mechanisms by which to reduce predation. Vegetation cover has been
documented as a key determinant as regards reducing young tortoises’ detectability, since it facilitates their cam-
ouage20. Crypsis has anti-predatory benets, and refuge areas, such as those areas dominated by vegetation
cover, may reduce the risk of predation. It has also been documented that ravens’ predation on tortoises diers
1Instituto de Investigación en Recursos Cinegéticos, IREC (CSIC-UCLM-JCCM). Ronda de Toledo, 12, 13071, Ciudad
Real, Spain. 2Escuela Técnica Superior de Ingenieros Agrónomos (UCLM), Ronda de Calatrava, 7, 13071, Ciudad Real,
Spain. *email: amaliasegura@gmail.com
OPEN
Content courtesy of Springer Nature, terms of use apply. Rights reserved

2
SCIENTIFIC REPORTS | (2020) 10:1874 | https://doi.org/10.1038/s41598-020-58851-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
according to their spatial distribution of single individuals or breeding pairs. Tortoise predation is greater in
adjacent human developments, which attract large numbers of single ravens, and in adjacent successful nests
of breeding pairs throughout human developed and undeveloped areas4. Even the success of juvenile tortoise
releases is compromised in the head-starting programs included in conservation strategies owing to the fact that
certain ravens are attached to the predation of certain tortoise size classes17,21. It could, therefore, be hypothesized
that the risk of raven predation on young tortoises might be high in areas with high dense tortoise populations
and mediated by the overlaid eect of raven presence and vegetation cover.
Maamora forest, an anthropogenic cork oak forest located in northern Morocco, is considered to be close to
the optimum niche –the core range– of the Mediterranean spur-thighed tortoise’s distribution Testudo graeca22
and one of the areas with the highest density populations documented to date23. However, the common raven
has, over the last 40 years, increased its numbers and distribution worldwide, and this cork oak forest is no
exception24. is increase in predators may be a threat to the Mediterranean spur-thighed tortoise, and especially
to highly susceptible young individuals. Concretely densities of 0.8 breeding pairs km−2 had been observed in
the lustrum (A. Segura unpublished data). In this context, our specic objectives were to: (i) describe predation
and size classes’ preference of common ravens on young tortoises, and (ii) infer the main drivers of predation
risk and abundance. ese goals are relevant as regards improving knowledge on the predation and survival of
the Mediterranean spur-thighed tortoise at this critical stage in its life and, therefore, the conservation of this
threatened species.
Results
Raven predation on hatchlings and juveniles. Twenty-two occurrences of ravens, not including groups
(7, 3, 6 and 6, in A, B, C and D, respectively), were detected on our four study sites (Table1). ree raven nests
with active breeding pairs were found and two breeding pairs had between 1 and 2 chicks (Table1).
One hundred and sixty-six dead young tortoises (<100 mm Carapace Length; hereaer CL) (11, 8, 55 and
92, in A, B, C and D, respectively) were detected during the surveys. Most of the dead juvenile tortoises found
(88%) were located in the uncovered areas (areas C and D), where the main CL size category of dead tortoises
corresponded to between 41 and 70 mm (Fig.1). e cause of mortality in those areas was associated with preda-
tion by the common raven (100%, n = 147), where only < 75 mm CL tortoise carcasses were found with signs of
raven predation under perch and nest trees. e ravens ate the hatchling and juvenile tortoises by pulling o their
head and limbs (6%) or pecking holes through the carapace (60%) or plastron (34%). Indeed, 74 and 15 juvenile
tortoises were predated by two active breeding pairs of ravens in spring 2018 (in areas D and C, respectively).
Nevertheless, unknown causes of mortality dominated in the covered areas, and only 12–27% were related to
raven predation. e threshold size above which young tortoises are safe from raven predation in both covered
and uncovered areas was 75 mm CL. Dead young tortoises associated with raven predation diered signicantly
between covered and uncovered areas (A vs. C: X2 = 4.07, p < 0.05, n = 66; A vs. D: X2 = 4.25, p < 0.05, n = 103; B
vs. C: X2 = 5.05, p < 0.05, n = 63; B vs. D: X2 = 5.19, p < 0.05, n = 100), but did not dier between either covered
(A vs. B: X2 = 0.08, p = 0.77, n = 19) or uncovered areas (C vs. D: X2 = 0, p > 0.99, n = 147).
Live hatchlings and juveniles. One hundred and sixty-three live young tortoises (39, 30, 65, 39 in A, B, C
and D, respectively) were found during the surveys. Figure1 shows the size-class distribution on the four study
sites. e juvenile structure of live tortoises did not dier between the covered areas (A vs. B; X2 = 7.68, p = 0.26,
n = 69) but diered signicantly between the uncovered areas (C vs. D; X2 = 21.19, p < 0.05, n = 104) and also
A B C D
Vegetation
Bare ground cover 0 0 ± 0.2 17.9 ± 17.4 21.4 ± 13.8
Shrub cover 59.7 ± 21.1 52.0 ± 14.8 45.0 ± 23.0 28.2 ± 26.1
Shrub height 62.1 ± 17.7 71.1 ± 30.6 31.9 ± 27.3 19.8 ± 17.6
Shrub richness 2.0 ± 0.8 1.5 ± 0.7 0.8 ± 0.4 0.7 ± 0.7
Predation risk
Perch distance 203.6 ± 35.4 510.8 ± 120.8 72.5 ± 36.3 89.2 ± 35.3
Nest distance 999.0 ± 84.5 1076.7 ± 131.1 184.9 ± 89.2 144.7 ± 37.7
Raven occurrence
Single individuals 5 2 4 5
Breeding pairs 2 1 2 1
Groups 1(28) 1(5) 0 1(17)
Number of live tortoises 39 30 65 39
Number of dead tortoises 11 8 55 92
Raven predated tortoises 3 1 55 92
Table 1. Bare ground cover (%), shrub cover (%), shrub height (cm) and shrub species richness; raven perch
and nest distance (m) in the four study sites, A, B, C and D (mean ± SD); maximum raven occurrence in the
Mediterranean spur-thighed tortoise survey distributed by single individuals, breeding pairs and groups of
>2 individuals (number of individuals); live (<100 mm carapace length, CL), dead (<100 mm CL) and raven
predated tortoises (<75 mm CL) (in all of them, the number of individuals).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

3
SCIENTIFIC REPORTS | (2020) 10:1874 | https://doi.org/10.1038/s41598-020-58851-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
between the covered and the uncovered areas (C vs. A: X2 = 16.06, p < 0.05, n = 104; C vs. B: X2 = 16.06, p < 0.05,
n = 95; D vs. A: X2 = 36.89, p < 0.05, n = 78; D vs. B: X2 = 36.77, p < 0.05, n = 69). e main dierences concerned
the CL size category of 81–100 mm, which represented 41% versus 16% in the covered and uncovered areas,
respectively.
Tortoise detectability and abundance: risk of predation by ravens. According to previous results
regarding dead animals found on our study sites, only those animals < 75 mm CL were susceptible to predation
in Maamora forest. Predation risk analysis was, therefore, restricted to these size classes: 140 detections (A: 27
tortoises in 11 occupied grids; B: 21 tortoises in 10 occupied grids; C: 57 tortoises in 13 occupied grids, and D: 35
tortoises in 7 occupied grids). It varied from a maximum of 3 to 5–11 tortoises per grid in covered and uncovered
areas, respectively (Table1).
In the N-mixture model, we used a negative binomial model (Table2). e stepwise procedure carried out
to select predictors explaining detectability and abundance processes is summarized in Table2. e bootstrap
p-values for the nal model based on the SSE, Freeman-Tukey, and Chi-square statistics were 0.03, 0.00 and 0.01,
respectively. e value of ĉ (ratio of observed/expected) was 1.59.
e nal model included the interaction between bare ground cover and the distance to the raven perch in
the detection process (Fig.2), and the site and the number of shrub species in the abundance process (Fig.3,
Table3). e detectability increased with the percentage of bare ground, but this eect was apparent only when
the distance to a raven perch was greater. e abundance was higher in uncovered areas and in grids with high
shrub species richness.
Discussion
Evidence of common raven predation on hatchlings and juveniles. is study reveals the predation
of common ravens on juveniles of Mediterranean spur-thighed tortoises in certain areas of Maamora forest. Both,
single individuals and breeding ravens were observed killing, carrying away and consuming juvenile tortoises,
their preference being for class sizes 40–70 mm CL. e selection by ravens of certain size classes of tortoises over
others has been found in other Testudinidae populations too18,19. Despite some tortoises being paint-marked to
identify them as recaptures, no evidence of any increased risk of predation was observed for the recognisable
fraction of the population (but see25); only in C uncovered area was one individual found dead, representing 2%
of the marked individuals.
Bearing in mind that raven predation might vary between years and among individuals, and the fact that our
study comprised only one year, our results still suggest that the predation of juvenile tortoise in the study area
was higher when compared with that of populations of Mediterranean spur-thighed tortoises in southern Spain8.
Indeed in those areas raven predation did not aect tortoise populations at all. Nevertheless, some similarities
Figure 1. Distribution of carapace length (mm; CL) in the four study areas: covered (A,B) and uncovered
(C,D). Data represent the frequencies of young tortoises: live shown in black, dead by raven predation in light
grey and dead by any other reason in dark grey.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

4
SCIENTIFIC REPORTS | (2020) 10:1874 | https://doi.org/10.1038/s41598-020-58851-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
with other Testudinidae populations aected by raven predation were found12,16,26, although the ravens involved
in predation in our study appeared to have a slightly lower size threshold above which juveniles are safe from
predation (75 mm CL) than reported for other Testudinidae populations (85 and 100 mm CL; 27, 21 respectively).
is might be associated with the length of time shell hardening takes in Mediterranean spur-thighed tortoises,
which has been documented to limit the probability of predation by ravens (e.g.27), and this merits further studies.
Nevertheless, when comparisons of raven predation on tortoises were restricted to covered versus uncovered
areas, in the former such occurrences were all anecdotal while high mortality rates were rigorously recorded in
the latter. Certainly, we were surprised by the high amount of predation by one pair of breeding ravens, which
predated 74 tortoises of <75 mm CL in a single breeding season. Indeed, raven predation might be modifying
juvenile tortoise population structure through altering recruitment (e.g.4). Further studies are required to dis-
entangle the role played by predation in tortoise population structure within the Maamora forest.
Model specication nPars AIC ΔAICc < 2 AICwt cltvWt
(1) Latent abundance distribution
Negative binomial 3 541.51 0.00 1.00 1.00
Zero Inated 3 584.01 42.50 0.00 1.00
Poisson 2 620.03 78.52 0.00 1.00
(2) Covariates of abundance (γ) and detection (p)
γ (Shrub richness + Site) p (Bare ground cover*Perch distance) 10 527.82 0.00 0.98 0.98
γ (Site) p (Bare ground cover*Perch distance) 9 536.24 8.42 0.01 0.99
γ (.) p (Bare ground cover*Perch distance) 6 538.68 10.86 0.00 1.00
γ (.) p (.) 3 541.51 13.69 0.00 1.00
Table 2. Model selection of Mediterranean spur-thighed tortoise hatchlings and juveniles (<75 mm carapace
length; see text for details): (1) Latent abundance distribution and (2) covariates of abundance and detection.
Covariates considered: shrub richness; site, bare ground cover (%) and raven perch distance (m). Model
selection based on Akaike’s Information Criterion (AIC), number of parameters (nPars), the dierence in AICc
from the best tted models (ΔAICc < 2), model weights (AICwt), and cumulative model weights (cltvWt).
D=350 D=400 D=450
D=150 D=200 D=250
D=20 D=50 D=100
0.00.2 0.40.6 0.00.2 0.40.6 0.00.2 0.
40.6
0.00
0.25
0.50
0.75
1.00
0.00
0.25
0.50
0.75
1.00
0.00
0.25
0.50
0.75
1.00
Bare ground rate
Detection probability
Figure 2. Interaction between two continuous covariates: bare ground cover rate and distance to the nearest
raven perch (D, in meters) over the probability of detection.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

5
SCIENTIFIC REPORTS | (2020) 10:1874 | https://doi.org/10.1038/s41598-020-58851-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
Predation risk by ravens, population size and structure of young tortoises. In this study, tortoise
detectability –associated with predation risk– was mediated by the interaction between predator presence and
bare or low cover areas; with tortoise detectability increasing with greater areas of bare ground, mainly in loca-
tions far from perch trees (e.g.18,28). is might suggest that ravens could be modulating the behavioral response
of young tortoises, e.g. they will reduce their activity in areas near perch trees in order to be less detectable by the
ravens (e.g.29). In this respect, it is reasonable to assume that the survivorship of juveniles in areas with higher
predation risk might be lower and, therefore, a lower juvenile population size will characterise populations in
such areas. Conversely, we found a higher abundance of young tortoises in uncovered areas, where they suered
higher predation, but also in areas where there was high diversity of shrub species in both covered and uncovered
areas (e.g.30). It is thus plausible to assume that many other factors involving juvenile physiological costs (e.g.29)
or even female reproduction traits –the number of clutches, clutch size and recruitment success–3, might explain
part of the variation found in the size of juvenile populations in covered and uncovered areas3. However, in higher
risk predation areas, it might also be expected that juvenile population structure would be modied and show
dierences in size/age classes12,18,19. Indeed, the higher percentage of longer/older juveniles (76–100 mm) –which
are not considered susceptible to predation– found in covered areas compared to in uncovered ones evidenced a
threat to juvenile survivorship in the lower size/age classes in those areas where hatchlings are more detectable,
e.g. uncovered areas, and whose eects, among others, might be mediated by vegetation cover (e.g.30).
Finally, hatchlings are challenging to study since they are rarely encountered in the eld31,32 and, as such, very
low capture rates and practically nonexistent recapture rates of Mediterranean spur-thighed tortoise hatchlings
have been reported in southern Spain33,34. Further studies are required to discover the role played by predation in
ABCD
020406
08
0
Sites
Abundance
0123
050 100 150 200 250 300
Number of species
Figure 3. Relationship between sampling sites (A, B, C and D; le graph) and the richness of shrub species
(number of species; right graph) with the abundance of Mediterranean spur-thighed tortoise hatchlings and
juveniles (<75 mm carapace length).
Process Var iable Estimate Standard
Error ZP (>|Z|)
Abundance
Intercept 1.025 0.49 2.094 ***
Shrub richness 0.709 0.222 3.196 **
Site B −0.419 0.619 −0.677 ns
Site C 2.697 0.564 4.78 ***
Site D 2.102 0.586 3.588 ***
Detection
Intercept −2.416 0.454 −5.32 ***
Bare ground cover 0.636 0.516 1.23 ns
Perch distance 1.9 0.588 3.23 **
Bare ground cover* Perch
distance 1.961 0.78 2.52 **
Table 3. Summary and statistical parameters of the nal N-mixture model parameterized to estimate
Mediterranean spur-thighed tortoise hatchlings and juveniles (<75 mm carapace length) as regards both
detectability and abundance (signicance codes: ns no signicative; ** < 0.01 and *** < 0.001).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

6
SCIENTIFIC REPORTS | (2020) 10:1874 | https://doi.org/10.1038/s41598-020-58851-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
the recruitment process, which will have later impacts on the size and structure of tortoise populations. is will
also help to disentangling the possible physiological costs associated with predation risk.
Material and Methods
Study area. e study was conducted in areas of low elevation (72–185 m a.s.l.) and sandy soil in Maamora
forest (northwest Morocco; 34° 02′ 54.19″ N, 6° 27′ 19.24″ W). e climate is Mediterranean, with hot and dry
summers, and the annual range of average rainfall is from 300 to 500 mm. Maamora forest is dominated by cork
oak trees, Quercus suber, with scattered endemic wild pear, Pyrus mamorensis, wild olive Olea europaea, green
olive Phyllirea latifolia and mastic Pistacia lentiscus, and a sparse understory represented by bush and shrub
species, such as Mediterranean broom Genista linifolia, Cytisus arboreus, Stauracanthus genistoides, dwarf palm
Chamaerops humilis, French lavender Lavandula stoechas, sage-leaved rockrose Cistus salviifolius, Halimium hal-
imifolium and ymelaea lythroides35.
e study specically took place on four sites that were close together (separated by 3 km), but always ensuring
that the tortoise populations’ territories were separated by a sucient distance for them to be considered as inde-
pendent populations34,36; Fig.4. Despite the fact that all the sampling sites were located on private land on which
there has been no pet trade (>10 years protected) and that the undergrowth is well represented when compared
with other sites in unprotected Maamora, the study sites diered as regards their vegetation cover. Two of them
were characterized by high cover and the diversity of their shrub and herbs (hereaer denominated as covered
areas, A and B), while the other two were characterized by a lower cover and diversity of shrubs and a high per-
centage of bare ground (hereaer denominated as uncovered areas, C and D; see Table1 for further details). is
experimental design allowed us to test for dierences in tortoise detectability, and, therefore, infer predation risk,
in relation to vegetation cover.
Mediterranean spur-thighed tortoise and common ravens. e study sites were surveyed during
the 2018-breeding period, comprising the end of February until the end of May, covering an area of 15 ha in
each of the covered areas and of 18 ha in each of the uncovered ones. Each of the four sites was intensively sur-
veyed in order to detect young tortoises. is was done for four (uncovered areas) or ve (covered areas) days
by four trained people. e survey consisted of searching the entire territory for the occurrence of individual
tortoises, but focusing on the detection of hatchlings and juveniles (<100 mm CL). e tortoises were recorded
from 12 h until 16 h on foot and in adequate weather conditions (sunny days with temperatures of between 20
and 24 °C). Each recorded individual was georeferenced using a GPS and the CL was measured using a vernier
Figure 4. Location of the study area in Morocco and tortoise populations studied (sites A, B, C and D) is
shown. Grey circles size is proportional to the maximum abundance of live young tortoises (<75 mm CL)
observed per 1ha-grid (1, 2–3, 4–5, 6–11 ranges). Small grey squares mark grids where no tortoises were found.
Black stars represent the raven perch or nest locations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

7
SCIENTIFIC REPORTS | (2020) 10:1874 | https://doi.org/10.1038/s41598-020-58851-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
calliper (accuracy ± 1 mm). Size classes were used to characterize the tortoises’ population structure in covered
and uncovered areas24. e Chi-square test was used to assess any dierences between covered and uncovered
areas in terms of hatchling and juvenile size-classes structure, considering measurements of 30–40, 41–50, 51–60,
61–70, 71–80, 81–90 and 91–100 mm. e tortoises’ carapaces were lightly marked with non-toxic paint so as
to be able to identify any recaptured individuals. In addition, two additional days per area at the beginning and
the end of the raven breeding season were carried out looking for any dead animals, which were collected and
measured and, when possible, their cause of death was determined on the basis of a visual examination of the
remains of the animal and the area in which it was found. ese data were then used to determine the relevance of
predation by ravens and to characterize the size-classes of the tortoises that are predated by this species.
During the tortoise survey, the location of common raven perches and nests were registered with a GPS. For
this purpose, we also surveyed a buer area of 1.5 km around the sampling areas, which is slightly greater than the
average distance between the ravens’ territories4. e number of single individuals and breeding pairs of ravens
was also recorded, along with their breeding status and their recruitment success (chicks or edglings).
Modelling tortoise detectability and abundance. e recorded data were referred to 1-ha grids, which
were our territorial unit for analytical purposes. e number of young tortoises susceptible to predation (accord-
ing to our data < 70 mm CL, see below; but see21,37) was, therefore, quantied (our response variable) for each
grid and survey. In addition, each grid was characterized during the survey in relation to vegetation cover, and
specically to the shrub cover (%), shrub height (cm) and species richness, and the bare ground cover (%). In
order to incorporate the eect of the common raven into species detectability, the mean distance to the nearest
raven perch and nest (m; two variables) in each grid was also calculated.
We used N-mixture models to model the detectability and abundance of the hatchlings and juveniles that
were, according to the count data, susceptible to predation, while we accounted for imperfect detection38 using
the unmarked package39 in R40. Our assumption is that the detectability of young tortoises during the surveys
can be considered a proxy of young individuals’ detectability by the ravens. Hatchlings and juveniles are well
camouaged and dicult to follow in their environment. eir detectability is very low8,10,16,41, since it is linked
to the central hours of the day when animals are active. We, therefore, assume that detectability by observers is
a proxy of detectability by ravens in order to be able to explore the eect of dierent factors on predation risk.
is does not mean that the observer is able to detect the same number of young tortoises, but that the higher the
detectability for the observer, the higher detectability for the ravens.
We used Akaike’s Information Criterion corrected for small sample sizes (AICc)41 to select the most appro-
priate error distribution by comparing the performance of Poisson, zero-inated Poisson and Negative Binomial
models. e detection and abundance processes were then modelled sequentially. A forward stepwise procedure
was used to identify the most relevant predictors explaining the detection process. e same procedure was sub-
sequently followed in order to identify the predictors explaining the abundance process. AICc was also used to
compare models in the stepwise procedures by following the ΔAICc < 2 rules42. N-mixture models require an
integer value that species the upper bound used in the integration (K). In our study, this upper bound was set at
K = 120, which is suciently large for it not to have an eect on the model results. We used the parametric boot-
strap approach (1000 samples) in unmarked to attain p-values from the sums of squares, along with Chi-square
and Freeman-Tukey t statistics as a measure of the goodness of t of the nal model.
Ethic statements. Sampling of Mediterranean spur-thighed tortoises in Maamora forest was conducted
under the authorization and following the protocols approved by Le Haut-Commissariat aux Eaux et Forêts et
à la Lutte Contre la Désertication of Morocco (High Commission for Waters and Forests and the Fight against
Desertification). The field surveys were done in accordance with the guidelines and regulations. There was
non-invasive sampling.
Data availability
e datasets generated during and/or analyzed during the current study are available from the corresponding
author on reasonable request.
Received: 22 October 2019; Accepted: 22 January 2020;
Published: xx xx xxxx
References
1. Epperson, D. M. & Colleen, D. H. Nestling and hatchling ecology of Gopher tortoises (Gopherus polyphemus) in southern
Mississippi. J. Herpetol. 37, 315–324 (2003).
2. Campbell, S. P., Steidl, . J. & Zylstra, E. . ecruitment of desert tortoises (Gopherus agassizii and G. morafai): a synthesis of
reproduction and rst-year survival. Herpetol. Conserv. Biol. 10, 583–591 (2015).
3. Díaz-Paniagua, C., eller, C. & Andreu, A. C. Long-term demographic uctuations of the spur-thighed tortoise, Testudo graeca in
SW Spain. Ecography 24, 707–721 (2001).
4. ristan, W. B. & Boarman, W. I. Spatial pattern of ris of common raven predation on desert tortoises. Ecology 84, 2432–2443
(2003).
5. Ben addour, ., Slimani, T., El Mouden, E. H., Lagarde, F. & Bonnet, X. Population structure, population density and individual
catchability of Testudo graeca in the central Jbilets (Morocco). Vie et Milieu 56, 49–54 (2006).
6. aymond, . M. N. Determining habitat preferences of the juvenile Gopher polyphemus using spatially model vegetation on central
Florida sandhill. Graduate theses and dissertations (2007).
7. Butler, J. A. & Sowell, S. Survivorship and predation of hatchling and yearling gopher tortoises. Gopherus poliphemus. Chelon.
Conserv. Biol. 1, 173–180 (1996).
8. eller, C., Díaz-Paniagua, C. & Andreu, A. C. Survival rates and mortality causes of Testudo graeca hatchlings in southwestern Spain.
J. Her pet. 32, 238–243 (1998).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

8
SCIENTIFIC REPORTS | (2020) 10:1874 | https://doi.org/10.1038/s41598-020-58851-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
9. Nafus, M. G., Esque, T. C., Averill-Murray, . C., Nussear, . E. & Swaisgood, . . Habitat drives dispersal and translocated juvenile
desert tortoises. J. Applied Ecol. 54, 430–438 (2017).
10. Tom, J. Microhabitats and use of burrows of Bolson tortoise hatchlings in Biology of North American tortoises (ed. Bury, . B. &
Germano, D. J.) 137–146 (National Biological Survey, 1994).
11. Smith, L. L., Steen, D. A., Conner, L. M. & utledge, J. C. Eects of predator exclusion on nest and hatchling survival in the gopher
tortoise. J. Wildl. Manage. 77, 352–358 (2013).
12. Wilson, D. S. Estimates for survival gopher tortoises. Gopherus poliphemus. J. Herpetol. 25, 376–379 (1991).
13. Camp, . J., night, . L., night, H. A. L., Sherman, M. W. & awashima, J. Y. Food habits of nesting common ravens in the eastern
Mojave Desert. Southwest. Nat. 38, 163–165 (1993).
14. Barje, F., Slimani, T., El Mouden, E. H., Lagarde, F. & Bonnet, X. Shrewd shries and spiny shrubs: a calamity for hatchling Moorish
tortoise. Amphibia-eptilia 26, 113–115 (2005).
15. Germano, D. J. Comparative life histories of North American tortoises in Biology of North American tortoises (ed. Bury, . B. &
Germano, D. J.) 175–185 (National Biological Survey, 1994).
16. Pie, A. Movement patterns, habitat use and growth hatchling tortoises. Gopherus polyphemus. Copeia 1, 68–76 (2006).
17. Nagy, . A., Hillard, L. S., Tuma, M. W. & Morafa, D. J. Head-started desert tortoises (Gopherus agassizii): movements, survivorship
and mortality causes following their release. Herpetol. Conserv. Biol. 10, 203–215 (2015).
18. Berry, . H., Yee, J. L., Coble, A. A., Perry, W. M. & Shields, T. A. Multiple factors aect a population of Agassiz´s desert tortoise
(Gopherus agassizii) in the northwestern Mojave desert. Herpetolological Monographs 27, 87–109 (2013).
19. Boarman, W.I. Problems with management of a native predator on a threatened species: raven predation on desert tortoises in
Proc.15th Vertebrate Pest Conf. (ed. Borrecco, J.E., & Marsh .E.). University of California, Davis (1992).
20. Nafus, M. G. et al. Hiding in plain sight: a study on camouage and habitat selection in slow-moving desert herbivore. Behav. Ecol.
26, 1389–1394 (2015).
21. Hazard, L. C. Post-release dispersal and predation of head-started juvenile desert tortoises (Gopherus agassizii): Eect of release site
distance on homing behaviour. Herpetol. Conserv. Biol. 10, 504–515 (2015).
22. Anadón, J. D. et al. Distribution of Testudo graeca in the western Mediterranean according to climatic factors. Amphibia-eptilia 33,
285–296 (2012).
23. Segura, A. & Acevedo, P. The importance of protected and unprotected areas for the Mediterranean spur-thighed tortoise
demography in northwest Morocco. Amphibia-eptilia 40, 369–371 (2019).
24. Birdlife International. Corvus corax (amended version published in 2016) The IUCN ed List of threatened Species 2017: e.
T22706068A113271893 (2017).
25. Calderon-Chalco, . A. & Putman, B. J. e eect of paint maring on predation ris in western fence lizards: a test using clay
models. Herpetol. Conserv. Biol. 14, 80–90 (2019).
26. Berr y, . H., et al. Changes in desert tortoise populations at the desert tortoise research natural area between 1979 and 1985. Desert
tortoise council proceedings of the 1986 symposia: 100–123 (1986).
27. Nagy, . A., Tuma, M. W. & Hillard, L. S. Shell hardness measurement in juvenile desert tortoises Gopherus agassizii. Herpetol.
eview 42, 191–195 (2011).
28. Hansen, N. A., Sato, C. F., Michael, D. ., Lindenmayer, D. B. & Driscoll, D. A. Predation ris for reptiles is highest at remnant edges
in agricultural landscapes. J. Applied Ecol. 56, 31–43 (2019).
29. Martín, J. When hiding from predators is costly: optimization of refuge use in lizards. Etologia 9, 9–13 (2001).
30. Schneider, M. F. Habitat loss, fragmentation and predator impact: spatial implications for prey conservation. J. Applied Ecol. 38,
720–735 (2001).
31. Dunham, A. E., Morin, P. J. & Wilbur, H. . Methods for the study of reptile populations in Biology of the reptilia vol 16. Defense
and life history (ed. Gans C., & Huey .B.) 331–385 (Alan Liss, 1988).
32. Morafa, D. J. Neonates: missing lins in the life histories of North American tortoises in Biology of North American tortoises (ed.
Bury, . B. & Germano, D. J.) 161–173 (National Biological Survey, 1994).
33. Braza, F., Delibes, M. & Castroviejo, J. Estudio biométrico y biológico de la Tortuga mora (Testudo graeca) en la eserva Biológica
de Doñana, Huelva. Doñana ActaVertebr. 3, 15–41 (1981).
34. Andreu, A. C. Ecología y dinámica poblacional de la Tortuga mora, Testudo graeca, en Doñana. Tesis doctoral. Univ. Sevilla (1987).
35. Aa, A. Etude de la diversité oristique de l´écosystème de chene-liège du foret de la Maamora. ese, Institut. Agron. et Vétér.
Hassan II, abat (2007).
3 6. Anadón, J. D. et al. Factors determining the distribution of the spur thighed tortoise Testudo graeca in south east Spain: a hierarchical
approach. Ecography 29, 339–346 (2006).
37. Boarman, W. I. Managing a subsidized predator population: reducing common raven predation on desert tortoises. Environ.
Manage. 32, 205–217 (2003).
38. oyle, J. A. N-mixture models for estimating population size from spatially replicated counts. Biometrics 60, 108–115 (2004).
3 9. Fise, I. & Chandler, . Unmared: An  pacage for tting hierarchical models of wildlife occurrence and abundance. J. Stat. Sow.
43, 1–23 (2011).
40.  Core Team. : A language and environment for statistical computing. . Foundation for Statistical Computing, Vienna, Austria,
www.-project.org/ (2019).
41. Hellgrem, E. C., azmaier, . T., uthven, D. C. & Synatzse, D. . Variation in tortoise life history: demography of Gopherus
berlandieri. Ecology 81, 1297–1310 (2000).
42. Burnham, . P. & Anderson, D. . Model Selection and Multimodel Inference: A Practical Information-eoretic Approach
(Springer Verlag, 2002).
Acknowledgements
We are very grateful to Oscar Rodríguez, Hassan Belhajjamia, Absallam Belhajjamia and Bouhali Kaddouri for
their eld assistance. We truly appreciate their commitment to the surveys and their ability to detect hatchlings.
We would like to thank HCEFLCD services for their guidance. We would also like to thank Greg Trollip and
Jacob Mwanzia for their support and interest in wild species conservation. PA is supported by MINECO-UCLM
through a “Ramón y Cajal” contract (RYC-2012–11970).
Author contributions
A.S. and P.A. conceived the initial ideas, A.S. did the eld surveys, A.S., J.J. and P.A. analysed the data and led the
writing. All the authors discussed the ndings.
Competing interests
e authors declare no competing interests.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

9
SCIENTIFIC REPORTS | (2020) 10:1874 | https://doi.org/10.1038/s41598-020-58851-5
www.nature.com/scientificreports
www.nature.com/scientificreports/
Additional information
Correspondence and requests for materials should be addressed to A.S.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
© e Author(s) 2020
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com