Combining ecological, social and technical criteria to select species for forest restoration

Abstract

QuestionHow to evaluate and integrate relevant ecological, social and technical criteria to select species to be introduced in restoration projects of highly diverse ecosystems such as tropical riparian forests. LocationRiparian forest, Marqués de Comillas municipality, southeast Mexico (16°54′N, 92°05′W). Methods
We proposed a ‘species selection index’ (SSI) using five independent criteria related to ecological, social and technical information. SSI targeted species that (1) are important in the reference forest; (2) are less likely to establish following disturbance; (3) are not specific to a particular habitat; (4) are socially accepted; and (5) their propagation requires a reasonable time and financial investment. SSI may range between zero and 50, with higher values meaning higher potential for restoration purposes. ResultsOut of a local pool of 97 species, we identified 30 target tree species that together represented >60% of total importance value index in the reference riparian forests. SSI averaged 28.3 ± 1.0 over the studied species, suggesting that species with high values are not frequent. For 20 species, reintroduction by means of active forest restoration was deemed necessary. Species that established through natural regeneration, following secondary regrowth, had lower social value among local farmers. Nearly half of the identified species showed technical constraints for easy propagation and seeding. Conclusions
The proposed procedure is useful for selecting species to initiate forest restoration projects and of other woody ecosystems that harbour high biodiversity, and is suitable for several stakeholders interested in restoration.

Full text

Applied Vegetation Science
&&
(2014)
Combining ecological, social and technical criteria to
select species for forest restoration
Paula Meli, Miguel Mart

ınez-Ramos, Jos
eMar

ıa Rey-Benayas & Julia Carabias
Keywords
Indicators; Mexico; Natural regeneration;
Propagation; Re-vegetation; Social value;
Tropical riparian forest
Nomenclature
Mart

ınez, E., Ramos, C.H., & Chiang, F. 1994.
Lista flor

ıstica de la Lacandona, Chiapas.
Bolet

ın de la Sociedad Bot
anica de M
exico
54:99–177
Received 6 August 2013
Accepted 11 December 2013
Co-ordinating Editor: Joerg Ewald
Meli, P. (corresponding autor,
paula@naturamexicana.org.mx): Natura y
Ecosistemas Mexicanos A.C., San Jacinto 23-D,
Col. San

Angel, M
exico D.F., 01000, M
exico
Mart
ınez-Ramos,M.
(mmartinez@cieco.unam.mx): Centro de
Investigaciones en Ecosistemas, Universidad
Nacional Aut
onoma de M
exico, Antigua
Carretera a P
atzcuaro No. 8701, Morelia,
Michoac
an, 58190, M
exico
Rey-Benayas, J.M. (josem.rey@uah.es):
Departamento de Ciencias de la Vida,
Universidad de Alcal
a, Edificio de Ciencias,
Alcal
a de Henares, Madrid, 28871, Spain
Carabias, J. (jcarabias@ecologia.unam.mx):
Departamento de Ecolog

ıa y Recursos
Naturales,Facultad de Ciencias, Universidad
Nacional Aut
onoma de M
exico, Av.
Universidad 3000, Circuito Exterior S/N, M
exico
D.F., 04510, M
exico
Abstract
Question: How to evaluate and integrate relevant ecological, social and techni-
cal criteria to select species to be introduced in restoration projects of highly
diverse ecosystems such as tropical riparian forests.
Location: Riparian forest, Marqu
es de Comillas municipality, southeast Mexico
(16°54′N, 92°05′W).
Methods: We proposed a ‘species selection index’ (SSI) using five independent
criteria related to ecological, social and technical information. SSI targeted spe-
cies that (1) are important in the reference forest; (2) are less likely to establish
following disturbance; (3) are not specific to a particular habitat; (4) are socially
accepted; and (5) their propagation requires a reasonable time and financial
investment. SSI may range between zero and 50, with higher values meaning
higher potential for restoration purposes.
Results: Out of a local pool of 97 species, we identified 30 target tree species that
together represented >60% of total importance value index in the reference
riparian forests. SSI averaged 28.3 1.0 over the studied species, suggesting
that species with high values are not frequent. For 20 species, reintroduction by
means of active forest restoration was deemed necessary. Species that estab-
lished through natural regeneration, following secondary regrowth, had lower
social value among local farmers. Nearly half of the identified species showed
technical constraints for easy propagation and seeding.
Conclusions: The proposed procedure is useful for selecting species to initiate
forest restoration projects and of other woody ecosystems that harbour high bio-
diversity, and is suitable for several stakeholders interested in restoration.
Introduction
The re-establishment of native plant species is a wide-
spread tool in ecological restoration, but in many ecosys-
tems, such as forests in the humid tropics, the large
regional species pool makes it difficult to effectively iden-
tify target species for restoration projects. Thus, a system-
atic approach is desirable to screen the widest possible
range of native taxa for possible inclusion in restoration
programmes (Knowles & Parrotta 1995). Species selection
requires extensive background studies, and sometimes
monitoring of hundreds of species over several years
(Knowles & Parrotta 1995; Blakesley et al. 2002a,b; Elliott
et al. 2003). However, restoration projects usually require
short-term results with limited economic resources. There-
fore, once the main objectives of restoration efforts based
on a census of all stakeholders have been defined, the gen-
eration of a list of target species for re-vegetation (Brudvig
& Mabry 2008) should be accomplished.
There is a wide variety of criteria to select target species
for forest restoration. These depend on the ecosystem to be
restored and the particular needs of each project. For
example, in Australia and Thailand, the ‘framework spe-
cies method’ (FSM) selected species with ecological prop-
erties, such as (1) high survival and growth rates in
degraded sites; (2) dense crowns that shade out herbaceous
1
Applied Vegetation Science
Doi: 10.1111/avsc.12096©2014 International Association for Vegetation Science

weeds; (3) provision of resources that attract seed dispersal
vertebrates at early restoration age; and (4) germination
traits enabling easy propagation in nurseries (Blakesley
et al. 2002a,b; Elliott et al. 2003). In India (Sharma &
Sunderraj 2005) and Brazil (dos Santos et al. 2008), spe-
cies were selected based on their natural regeneration
capacity. However, besides ecological criteria, other criteria
related to social acceptance and technical feasibility for
propagation are required to optimize identification of suit-
able native species for restoration.
We distinguished tree species that were passively
restored by natural regeneration from those requiring
active restoration in a previous study based on ecological
criteria, namely dominance and regeneration potential
(Meli et al. 2013a). However, given that biodiversity con-
servation and ecological restoration must embody societal
values to improve their success (Garibaldi & Turner 2004),
it is critical to recognize and take into account the cultural
perceptions and acceptance of the species used in restora-
tion projects. Successful restoration actions need the par-
ticipation of local stakeholders, and the potential of species
to be used in such actions should be evaluated not only on
the basis of their ecological traits, but also on criteria that
consider both social benefits and technical limitations,
such as germination and propagation requirements under
nursery conditions. In this study, we propose a procedure
to select target species for forest restoration projects, which
is illustrated by a case study related to restoration of Neo-
tropical riparian forest. This work does not constitute a
framework for implementing restoration activities (SER
2004). Rather, it pursues (1) the identification of the spe-
cies pool at a reference ecosystem; (2) the selection of spe-
cies from this pool based on ecological, social and technical
criteria that are considered relevant for restoration; and
(3) the integration of such criteria into a single and opera-
tional species selection index (SSI). It aims to link the ecol-
ogy and management of degraded forests and to be
suitable for implementation by various stakeholders in for-
est restoration efforts. We also discuss the potential imple-
mentation of the proposed procedure in other ecosystem
types and in scenarios with uneven information availabil-
ity related to social values and technical requirements. We
finally provide some suggestions that could be addressed in
future studies of species selection for restoration of tropical
riparian forests and other species-rich ecosystem types.
Study site
We conducted this study at the Marqu
es de Comillas
municipality (16°54′N, 92°05′W), Selva Lacandona region,
southeastern Mexico. The climate is typically hot (25 °C
annual mean), with a mean annual precipitation of ca.
3000 mm and a short dry season (<100 mmmo
1
)
between January and April. Due to its diversity of soil
types, heterogeneous topography (Siebe et al. 1995) and
complex fluvial network, several tropical ecosystems are
present in this municipality but rain forest is the dominant
one. Although the Maya and other human groups inhab-
ited and abandoned this municipality more than 500 years
ago, human colonization restarted in the early 1970s,
when governmental programmes encouraged immigra-
tion, and this settlement has been portrayed as spontane-
ous and unorganized (De Vos 2002). Former old-growth
forest has been extensively converted to agricultural fields.
Deforestation also includes riparian vegetation, which
impacts both terrestrial and aquatic ecosystems. Marqu
es
de Comillas adjoins Montes Azules Biosphere Reserve
across the Lacant
un River, and contains a complex net-
work of permanent and temporal streams. Therefore, the
conservation of remnant old-growth forest in the region
has been recognized as of high priority, both in Mexico
and Guatemala (Mendoza & Dirzo 1999).
Methods
Procedure and criteria
To obtain a list of target species for the re-vegetation of
riparian degraded zones, we considered five criteria that
are based on ecological, social and technical information
(Table 1).
1 Natural species dominance (D). This criterion evaluates
dominance of individual species in the reference forest,
which in our case was represented by six sites with pristine
old-growth riparian forest. Sites were identified through
prospective routes along stream-sides. We estimated rela-
tive density, relative frequency and relative basal area of
all woody species with DBH ≥0.5cmalonga50910 m
transect parallel to the stream in each site. Basal area was
estimated using the DBH and the formula p*(DBH*0.5)
2
assuming a circular shape of the stem cross plane. For each
transect and species, we calculated an importance value
index (IVI) as the sum of relative density, relative fre-
quency and relative basal area of a species divided by three
(Curtis & McIntosh 1951). The measured IVI
i
was used as
an indicator of D, and adopted values between 0 and 100.
2 Natural regeneration potential (NRP). This criterion
evaluates the potential of the species to re-establish after
disturbance, and was first elaborated in Meli et al.
(2013a). To quantify NRP, we used five sites representing
the typical secondary riparian forest. This secondary forest
grew on sites formerly covered with old-growth forest sim-
ilar to the studied reference forest that was totally defor-
ested and abandoned later. Age of the secondary forest
sites varied between 3 and 10 yrs. In equal transects
(50 910 m each) to reference forest sites, we obtained for
every species abundance (N
i
, number of stems of species i
Applied Vegetation Science
2Doi: 10.1111/avsc.12096 ©2014 International Association for Vegetation Science
Selecting species for forest restoration P. Meli et al.

per transect) in each of ten DBH classes (range:
0.5–>50 cm, class intervals: 5 cm). For each transect and
species, we calculated the correlation (Spearman rank cor-
relation, r
s
)betweenabundance[log(N
i
+1)] and the mid-
point of the DBH classes (hereafter called abundance–size
correlation). A high NRP is represented by a diminishing
number of individuals as diameter sizes increase; this
change will result in a significant negative correlation and
therefore an acceptable potential for passive establishment
of the species (Meli et al. 2013b). A null or a positive corre-
lation for a particular species indicates that it does not
establish naturally (i.e. lack of regeneration) and, there-
fore, it needs to be actively restored or reintroduced. We
focused on the last kind of species considering that in our
study site the establishment of some species could be
impeded or slowed by physical, chemical or biological bar-
riers (Holl 2007). The NRP is a continuous variable that
varied between 1and1.
3 Habitat breadth (H). This criterion is a surrogate of the
ability of the species to develop in habitats of different geo-
morphology, which differ in soil and topographical proper-
ties. We assumed that species found in more habitats have
higher ability to establish after disturbances. Selecting
those species with higher habitat breadth implies selecting
generalist species, which may be detrimental for riparian
specialist species. However, we envisage the selection of
generalist species as an initial restoration step that will lead
to the rapid establishment of an initial canopy, thus creat-
ing environmental conditions for the re-establishment of
specialist species in a later step. This criterion selects wide-
spread, but not necessarily abundant, species. We used
data from 14 permanent 20 9250 m plots that were pre-
viously established within five geomorphological units
that differed in soil and topography in pristine rain forest:
floodplain, karst, alluvial, savanna and low-hill rain forest
(Siebe et al. 1995). We then counted the units where each
species occurred. As H is an ordinal criterion, it ranged
between zero and 5.
4 Social value (SV). This criterion identifies locally salient
species that shape the perceptions of local people with
respect to (1) the natural abundance of the species in the
riparian forest (in a rank of zero to 5); and (2) the local val-
ues of species for provision of food, materials, medicine
and/or cultural practices (Garibaldi & Turner 2004). These
two components of the SV in our study are comparable
because the number of different use types never exceeded
four (see below). The information related to these two
aspects was confirmed from participatory interviews with
farmers in four local communities. In groups of four or five
persons each, they shared photos of the 30 species with
highest IVI
i
at reference forest sites (App. S1). Farmers
were also consulted about other suitable species for ripar-
ian restoration that were not included in the previous list.
The SV was calculated as the rank of abundance plus the
number of local use types; as SV was an ordinal variable, it
took values >0.
5 Technical constraints (Tc). We collected seeds in the
field, and germinated and propagated them in a nursery
for all available species of those selected 30 species with
highest IVI
i
at reference forest sites, and then scored these
species. This criterion identifies cost-effective techniques
for successful species propagation. We used our own data
in an adapted scoring system from Knowles & Parrotta
(1995) that included three aspects with three categories
each: (1) ease of seed collection (combining seed size and
dispersal syndrome: large and zoochorous, small and
zoochorous, and small and anemochorous/hydrochorous;
note that seed availability is included in this component of
Tc); (2) seed germination treatment requirements (none,
mechanical and chemical treatment); and (3) alternatives
for introduction in the field (direct seeding, wildlings/
stumps, seedlings produced in nurseries; App. S2). The
categories received numerical values (1–3), with higher
values for the easiest/lowest cost option and lower values
for the most difficult/expensive options. These three values
were added; as Tc was an ordinal variable, it ranged
between 3 and 9.
For all abbreviations used see App. S3.
Assembling the index
Considering that some criteria were continuous and others
were ordinal, and that they varied at different scales, to
make them comparable we calculated the Zscore for each
criterion by obtaining the difference between a datum
value and the mean of the variable and dividing this differ-
ence by the SD. Finally, we divided these individual Z
scores into ten classes, from <2and>2, with 0.5 class
Table 1. Species selection criteria included in the proposed procedure.
Criteria Indicator Information type
Natural dominance (D) Importancevalue index (IVI
i
)Ecological
Natural regeneration potential (NRP) Spearman rank correlation ofabundance across size classes (r
s
)Ecological
Habitat breadth (H) Occurrence in five geomorphological units Ecological
Social value (SV) Natural abundance in riparian systems and local use according to social perception Social
Technical constraints (Tc) Ease of propagation (seed collection +germination +introduction alternatives) Technical
3
Applied Vegetation Science
Doi: 10.1111/avsc.12096©2014 International Association for Vegetation Science
P. Meli et al. Selecting species for forest restoration

intervals. We assigned a value of 0 to the lowest class and
10 to the highest class. We considered all criteria equiva-
lent and calculated SSI using the following formula:
SSI =D+NRP +H+SV +Tc. This SSI is an ordinal vari-
able that ranges between 0 and 50.
To explore possible relationships among the five criteria,
we performed non-parametric correlations (Spearman r
s
)
across the normalized data (Zscores) of all criteria.
Results
Criteria values
A total of 97 species were found in the reference forests, of
which Ficus sp. had the maximum IVI
i
(11%) and only ten
species had an IVI
i
>2% (Table S1). We found 92 species in
the disturbed forests, of which Dialium guianense had the
maximum IVI
i
(5%) and only 14 species had an IVI
i
>2%
(Table S2). The first 15 species accumulated 50% of total
IVI in the reference sites (Fig. 1a) and 48% in the dis-
turbed sites (Fig. 1b). We restricted all our analysis to those
30 species that showed the highest IVI
i
in the reference
sites, which together covered >60% of the total commu-
nity IVI.
Eight out of these 30 dominant species showed negative
abundance–size correlation coefficients (r
s
<0.6, P<
0.05), which suggested that passive restoration could be
sufficient for their successful establishment (Table S3).
Twelve species did not occur at disturbed sites and ten spe-
cies showed a non-significant abundance–size correlation,
thus hinting at the necessity of introducing them by means
of active restoration.
More than half of the species occurred in three or four
geomorphological units (54%), whereas nine species
occurred in one or two (30%), and only three species
(Brosimum alicastrum,D. guianense,Protium copal) occurred
in all geomorphological units (10%; Fig. 2, Table S3). Two
sampled species (6%) were totally absent in the five geo-
morphological units (Miconia glaberrima and Nectandra sle-
neri).
Farmers recognized most of the species (80%; App. S1).
Ten species (33%) were recognized in all cases, while
seven species (23%) were mostly unknown. In general,
farmers notably distinguished Lacant
un river valley and
stream banks (our reference ecosystem) as environments
with different hydrologic dynamics, soil types and species
composition. According to their perception, only Inga vera,
D.guianense and Albizia leucocalyx (4% of the species) were
abundant in riparian ecosystems (Fig. 3). Most species
(70%) were considered of low to medium abundance, and
only two species (Blepharidium mexicanum,Eugenia mexica-
na) were considered absent. There was no agreement
about the abundance of five species (8%), namely E.nigri-
ta,Jacataria dolichaula,Licania platypus,M.glaberrima and
N. reticulata. The relative species abundance denoted by
farmers was not correlated (r
s
=0.0414, P=0.8475)
with the species abundance recorded in the reference site
surveys (App. S1).
Most species (41%) were used only for timber (i.e. fuel-
wood, fence posts, handles, boards and shelves) and five
species (17%) had two use types besides timber (i.e. medi-
cine and fodder). Only B. alicastrum had four use types:
timber, food, medicine and fodder. Eleven species (38%)
were reported as not used by local people.
Species producing seeds that were considered easy to
collect represented 40% of the 30 species. A total of 53%
of the species were deemed easy to propagate, with no pre-
sowing treatment or only a simple mechanical scarification
required (App. S2). However, we did not have suitable
information about the appropriate introduction method
for 33% of the species. Finally, 43% of the species attained
aTcvalue>5, which could be a limitation when attempt-
ing to reintroduce native vegetation on disturbed sites.
Selection index and species selected
We calculated the SSI for the list of the 30 target woody
species to restore disturbed riparian zones (Table 2). SSI
was normally distributed, with a mean (SE) of
28.3 1.0, and ranged between 18 and 43. Less than half
of the species (43%) scored an SSI higher than the mean.
The species with the lowest SSI values were those with
null SV (i.e. not used or accepted by the local farmers).
We found a significant negative correlation only
between the natural regeneration potential (NRP) and the
social value (SV; r
s
=0.7036, P=0.0008), suggesting
that those species that naturally established following sec-
ondary regrowth have lower social value among local
farmers than those species that need to be actively
restored.
Discussion
Criteria for species selection
Natural dominance was the first criterion that we used for
species selection. We targeted selection of woody species to
initiate forest restoration projects. Although tropical ripar-
ian ecosystems contain other than woody species, these
species can: facilitate the establishment of other plants
(Parrotta et al. 1997) when their architecture (e.g. leaf and
canopy area) buffers harsh abiotic conditions (Meli & Dirzo
2013); attract seed dispersers when having fresh fruits
(Slocum 2001); and outcompete (typically) shade-intoler-
ant grasses through reducing their cover (Zimmerman
et al. 2000). They also provide organic matter to the ripar-
ian soil and promote shore stabilization in the medium-
term through their dense roots (Meli et al. 2013b). All
Applied Vegetation Science
4Doi: 10.1111/avsc.12096 ©2014 International Association for Vegetation Science
Selecting species for forest restoration P. Meli et al.

these characteristics may be also considered as species
selection criteria in forest restoration projects, but their
inclusion will depend mainly on the ecological condition
of the degraded ecosystem, and should be complemented
with other criteria, as we have shown in this work.
Once the restoration project has been established, it is
necessary to consider a wider range of species to fill under-
represented niches with other life forms (e.g. herbs, palms
and ferns) and with rare, endangered, endemic and/or
riparian specialist species, and thus to improve the struc-
ture and function of the riparian forest (Meli et al. 2013a)
and promote higher diversity and functional redundancy
(Brudvig & Mabry 2008). This will ensure the effectiveness
of critical ecological processes that sustain ecosystems
(Society for Ecological Restoration International Science &
Policy Working Group (SER) 2004).
6%
13%
17%
27%
27%
10% Number of
geomorphological
units
0
1
2
3
4
5
Fig. 2. Proportion of species out of the 30 studied native tree species
occurring in different numbers of geomorphological units found in
Marqu
es de Comillas.
0
1
2
3
4
5
6
Importance value index (%)
0
2
4
6
8
10
12
Importance value index (%)
(a)
(b)
Fig. 1. Importance value index (IVI) of species accounting for >60% of total IVI in the six riparian reference forests (a) and in the five disturbed or secondary
growth riparian forests (b).
5
Applied Vegetation Science
Doi: 10.1111/avsc.12096©2014 International Association for Vegetation Science
P. Meli et al. Selecting species for forest restoration

We used natural regeneration potential as the second
criterion. The predictive potential of the abundance–size
correlations for selecting target species from disturbed sites
could be limited by the small sample size, and hence
decrease as their age increases and its species composition
starts to resemble that of the reference sites (Meli et al.
2013a). However, the typically low species abundance in
highly diverse humid tropics makes it difficult to perform
accurate correlations without higher statistical power.
Assessing some preferred ecological characteristics of
target species is a different way to estimate the potential for
establishment. For example, longevity, resistance to herbi-
vores or physical damage, and tolerance to flooding in the
case of riparian systems, could also be important features
for assessing the potential for establishment. These features
focus on the species responses to particular abiotic or biotic
factors. Some of these ecological features are indirectly
included in our habitat breadth score, since generalist spe-
cies may have life-history and functional attributes to cope
with biotic and abiotic environmental filters better than
specialist species (Young et al. 2005).
Young fallows such as those we surveyed to estimate
the NRP are not always present in areas where restoration
is being planned, but they are good sites to identify poten-
tial species for passive restoration purposes at the initial
stages of restoration efforts (Meli et al. 2013a). In subse-
quent stages of the restoration project, other sites such as
older regeneration patches and other ecological species
characteristics could be used.
Our target species list is useful to restore typical dis-
turbed riparian forests in the studied region, including
those human-disturbed sites that were abandoned recently
(with minimal natural regeneration) or long ago (with
substantial natural regeneration). Unlike Brudvig & Mabry
(2008), we did not consider the species of the regional pool
that were already established at the disturbed sites because
they may not be the most suitable species in social or eco-
nomic terms when degradation is not very severe, as was
the case in our study. The ability of such species to establish
naturally in degraded areas is high, and therefore it may be
more appropriate to use these species for restoration of
severely degraded lands, such as mined sites (Parrotta &
Knowles 2001; Sharma & Sunderraj 2005) or sites highly
susceptible to erosion on steep slopes (dos Santos et al.
2008). Seed size and dispersal mechanism syndromes have
also been used to understand which species might require
active re-establishment and which might passively recolo-
nize degraded sites (Pausas & Lavorel 2003). For example,
regenerating species in disturbed sites are frequently those
with small seeds, which are widely dispersed (Chazdon
et al. 2007). We believe that regeneration indices (cf. dos
8%
17%
46%
17%
8%
4%
Rank
abundance
0
1
2
3
4
5
Fig. 3. Proportion of species out of the 30 studied native tree species
occurring at six rank abundance categories according to local people’s
perceptions found in Marqu
es de Comillas. See main text for details on
rank abundance calculation.
Table 2. Species selection index (SSI) values for 30 woody species tar-
geted for re-vegetation of riparian forest in Marqu
es de Comillas. The SSI
integrates standardized values (categories of Z-values, see text for details)
of Natural dominance (D), Natural regeneration potential (NRP), Habitat
breadth (H), Social value (SV) and Technical constraints (Tc).
Species D NRP H SV Tc SSI
Dialium guianense 810 99 743
Brosimum alicastrum 66 99 838
Brosimum costarricanum 610*76 837
Ficus sp.10 9 2 8 7 36
Cojoba arborea 10 6 4 5 7 32
Vochysia guatemalensis 57 78532
Trophis racemosa 410*66 632
Albizia leucocalyx 58 38731
Ampelocera hottlei 63 76931
Calophyllum brasiliense 56 76731
Licania platypus 510*66 431
Posoqueria latifolia 510*65 531
Guarea glabra 53 76829
Protium copal 73 96328
Castilla elastica 53 66727
Hirtella americana 44 75727
Pouteria durlandii 55 76 427
Swartzia simplex 510*35 427
Blepharidium mexicanum 45 64726
Inga vera 45 39526
Eugenia negrita 410*70526
Quararibea yunckerii 410*36 326
Nectandra reticulata 510*60 425
Miconia argentea 45 46524
Jacaratia dolichaula 410*60 424
Croton schiedeanus 52 65523
Eugenia mexicana 56 40 520
Licaria capitata 410*40 220
Nectandra sanguinea 510*10420
Miconia glaberrima 410*10 318
*Species absent in disturbed forest and therefore considered to need
active reintroduction (high NRP values).
Applied Vegetation Science
6Doi: 10.1111/avsc.12096 ©2014 International Association for Vegetation Science
Selecting species for forest restoration P. Meli et al.

Santos et al. 2008) are more accurate indicators of these
two types of species. Although not all second-growth for-
ests have recolonized degraded sites, and some species may
be adapted to several forms of degradation (e.g. degraded
soils, fires and weed infestations), the regeneration poten-
tial is a good indicator of the potential use of the species for
restoration purposes.
Habitat breadth was the third criterion. We found that
half of the species were present in at least three geomor-
phological units, suggesting that these species could estab-
lish in the riparian forest as in other ecosystem types. Few
species showed high habitat breadth for a particular unit,
and only A. leucocalyx was present in the floodplain and
should be re-established in riparian restoration sites in our
case study. The occurrence of species at particular habitats
is implicitly related to their recruitment niche and should
be strongly linked to ecological restoration projects. Many
species can persist as adults in a far broader niche than that
into which they can successfully recruit (Young et al.
2005) because habitat associations of adults do not neces-
sarily emerge at early life stages (Comita et al. 2007). Res-
toration activities may broaden the dispersal or
recruitment niche through translocation of propagules and
assisted establishment, and create non-regenerating popu-
lations by planting saplings where adults can develop but
seeds fail to germinate or seedlings have limitations to
establish themselves (Young et al. 2005).
Social value was the fourth criterion and a salient con-
tribution of our proposed procedure for restoration. Our
selected species were socially accepted or, at least, had
some appraisal or utility for local people, mostly for timber.
However, selecting only socially valuable species may put
at risk their establishment in the harsh conditions of a
degraded site. Non-pioneer species are a typical case of this
situation, but in the humid tropics they show high plastic-
ity in their growth rates and often establish successfully
when they are directly transplanted to open sites, even
when these sites have not been previously colonized by
pioneer species (Mart

ınez-Garza et al. 2005). Monitoring
field performance of these socially valuable species will be
crucial in restoration projects.
Although it is not the case in our study, the number of
use types could be much larger than abundance classes,
making these two components not comparable. In such
cases, averaging the normalized score in a single SV could
be a way to obtain a single SV value. Another option could
be using rank abundance and use types as separated val-
ues.
Interestingly, the species abundance denoted by local
farmers (social information) was not correlated with the
actual species abundance recorded in the reference sites
(ecological information; App. S1). At the same time, we
found that those species that are naturally established
following secondary regrowth had the lower social value
among local people. This is an unusual outcome, consider-
ing that in other tropical regions the young, second-
growth forests have high utilitarian as well as conservation
value and will likely become important sources of timber
and non-timber forest products (Chazdon & Coe 1999;
Vœk 2004; Gavin 2009). This emphasizes the need for fur-
ther research on flora uses among local people, both in
pristine and secondary riparian forest. The fact that people
did not recognize the species by their abundance or ecolog-
ical dominance does not mean that they do no actually use
these species. Other criteria such as utility should be analy-
sed to evaluate the accuracy of our correlation to reflect
real local uses in the region.
Local knowledge collected by interviews is important
and useful to make local people pro-active participants at
all stages of restoration practice (Blakesley et al. 2002b).
Snapshot questionnaires may not reveal the species prefer-
ences of the local communities, but we believe they do
reflect the farmer’s perception, as we infer from other pre-
vious participatory interviews that were conducted since
our conservation project started several years ago.
Supply of ecosystem services (i.e. supporting, regulating,
provisioning and cultural services) is directly related to
human well-being (MEA 2005). Any woody species can
supply more than one supporting and regulating service
(e.g. habitat provision, carbon fixation, soil retention and
many others). Thereby, the differences among these spe-
cies are mostly related to their supply of provisioning or
cultural services, and thus the use of species by local people
could be a surrogate for such services.
Technical constraints for propagation and introduction
of target species were the fifth criterion. This criterion con-
siders ease of seed collection, germination and alternatives
for introduction. Seed availability is indirectly included
when valuing the ease to collect seeds of different sizes
from fruits showing variable dehiscence. However, species
phenology and dioecism (seeds produced only by female
trees) also affect seed availability, especially of mast-fruit-
ing species. Further research about these characteristics of
the 30 selected species would provide important informa-
tion to estimate and value the entire spectrum of efforts to
obtain enough seeds and will be considered as surrogate
variables to score technical constraints in our riparian res-
toration project in the future.
While local people may be interested in propagating
native species for their reintroduction in many restoration
projects, this propagation may be time-consuming and
expensive. Consequently, it is important to select species
that are easily propagated, since local communities cannot
implement techniques that are costly or hazardous (e.g.
use of acid for seed scarification). Research is needed to
better understand the technical constraints to propagate
7
Applied Vegetation Science
Doi: 10.1111/avsc.12096©2014 International Association for Vegetation Science
P. Meli et al. Selecting species for forest restoration

and reintroduce native species, including species identifi-
cation and studies of fruiting phenology, seed germination
and nursery practice (Knowles & Parrotta 1995). Re-vege-
tation projects should emphasize the importance of this
information. Lack of information underestimates the
rating of some species but also guides future research on
species propagation for restoration purposes. This high-
lights the ‘adaptability’ of our procedure. Species could be
selected on the basis of one or two criteria and, at the same
time, they could generate useful information about the
other criteria.
Seeds from species classified as difficult to propagate
should not be collected in the first stages of the restoration
project, as it would be more efficient and less costly to
locate and transplant saplings from the forest (Knowles &
Parrotta 1995). However, the conservation status of some
target species may restrict this technique, because a threa-
tened or endangered species may not bear additional
reduction in its population through harvesting (Garibaldi
& Turner 2004). Also, reintroduction may be a successful
strategy for overcoming dispersal limitations but may not
reflect adult establishment (Turnbull et al. 2000); thus, the
performance of transplanted species in the field should be
included in our Tc index in future stages of the restoration
project (Knowles & Parrotta 1995; Elliott et al. 2003).
Species selection index
The criteria used to constitute the SSI appear to be inde-
pendent and complementary, as we found hardly any sig-
nificant correlation among them. Thus, ideally, they
should be used simultaneously or at least in groups of two
or three. We considered all five criteria to be equivalent
when assembling the SSI. However, as discussed above,
when species establishment faces hard ecological limita-
tions, ecological criteria could be more important than the
technical or social ones (Sharma & Sunderraj 2005; dos
Santos et al. 2008). Technical criteria could be considered
most important when there are monetary or time con-
straints, whereas social criteria are essential and should be
prioritized when there is no consensus among ecological
and social interests. Thus, priority ranking of species in
Table 2 could be re-ordered following these criteria (e.g.
ecological priority, social priority and technical feasibility
priority) in different restoration scenarios. The SSI average
was near the median value, suggesting that species with
high SSI were not frequent. At the same time, some species
showed very low SSI due to lack of information, which
highlights the dependence of the SSI on information avail-
ability.
The proposed procedure is useful to minimize costs and
maximize efficiency in selecting species for forest restora-
tion so that it can be attractive to different stakeholders. It
can also be applied to the screening and selection of woody
species from a wide spectrum of other tropical and temper-
ate regions. It is useful where trees are dominant, but its
use would be limited in grasslands or other ecosystem
types where species regeneration is difficult to estimate
(Meli et al. 2013a). Further research is needed to select
appropriate species to suit the specific ecological require-
ments in other ecosystem types.
Finally, the most appropriate methodology to select tar-
get species for restoration will strongly depend on the main
objectives of any particular project. Other criteria could be
considered in the selection of target species in other case
studies, including adaptive capacity to different soils (Shar-
ma & Sunderraj 2005), other social values (cf. Moreno-
Cassasola& Pardowska 2009) or attributes such as dispersal
syndromes (Sansevero et al. 2009). Technical constraints
may be the most useful criterion in practical terms because
these can increase the costs (time, labour, materials
needed) of the restoration projects, but social criteria
should be included in all restoration efforts (Garibaldi &
Turner 2004).
Conclusions
We proposed a procedure to target species for forest resto-
ration projects that leans on five criteria related to ecologi-
cal, social and technical information. A major strength of
this procedure is that the five criteria are independent and
can be used separately in projects with different goals.
Importantly, social information based on local perception
is usually neglected in restoration projects. The high num-
ber of woody species found in the reference sites indicates
that the regional species pool for riparian restoration is
wide. To facilitate practical restoration, we identified a pre-
liminary list of tree species that are most suitable for their
reintroduction into degraded riparian zones in our study
region and similar ecological and social settings (Brudvig &
Mabry 2008).
A list of target species must be identified and used for
the initial stages of restoration of ecosystems dominated by
trees. However, the species selection criteria will depend
on the main goals of the restoration project and on infor-
mation availability. In human-dominated ecosystems or
agricultural landscapes, prioritizing social and technical cri-
teria to select species for restoration is crucial for restora-
tion sustainability. Our procedure could be adapted to
different social and ecological conditions and be enriched
as new information is generated.
Acknowledgements
We thank C. M
endez and G. Jamangap
e for field
assistance. We are very grateful to M. Gonz
alez and
Applied Vegetation Science
8Doi: 10.1111/avsc.12096 ©2014 International Association for Vegetation Science
Selecting species for forest restoration P. Meli et al.

J.A. Parrotta for comments on an early draft of this
paper. R. Chazdon and an anonymous reviewer greatly
improved the content and presentation of a previous
version of this manuscript. A Rufford Small Grant for
Nature Conservation was provided to P.M. (40.11.09).
Pemex and the WWF-FCS Alliance supported Natura.
JMRB thanks projects CGL2010-18312 (Spanish Minis-
try of Science and Education) and S2009AMB-1783
REMEDINAL-2 (Madrid Government). MMR thanks
to CONACyT, FMCN and PAPIIT-UNAM for grants
support.
References
Blakesley, D., Elliott, S., Kuarak, C., Navakitbumrung, P., Zang-
kum, C. & Anusarnsunthorn, V. 2002a. Propagating frame-
work tree species to restore seasonal dry tropical forest:
implications of seasonal seed dispersal and dormancy. Forest
Ecology and Management 164: 31–38.
Blakesley, D., Hardwick, K. & Elliott, S. 2002b. Research needs
for restoring tropical forests in Southeast Asia for wildlife
conservation: framework species selection and seed propaga-
tion. New Forests 24: 165–174.
Brudvig, L.A. & Mabry, C.M. 2008. Trait-based filtering of the
regional species pool to guide understory plant reintroduc-
tions in Midwestern Oak Savannas, U.S.A. Restoration Ecology
16: 290–304.
Chazdon, R.L. & Coe, F.G. 1999. Ethnobotany of woody spe-
cies in second-growth, old growth, and selectively logged
forests of Northeastern Costa Rica. Conservation Biology 13:
1312–1322.
Chazdon, R.L., Letcher, S.G., van Breugel, M., Mart

ınez-Ramos,
M., Bongers, F. & Finegan, B. 2007. Rates of change in tree
communities of secondary Neotropical forests following
major disturbances. Philosophical Transactions of the Royal
Society B 362: 273–289.
Comita, L.S., Condit, R. & Hubbell, S.P. 2007. Developmental
changes in habitat associations of tropical trees. Journal of
Ecology 95: 482–492.
Curtis, J.T. & McIntosh, R.P. 1951. An upland forest continuum
in the prairie–forest border region of Wisconsin. Ecology 32:
476–496.
De Vos, J. 2002. Una tierra para sembrar sue~
nos. Historia reciente de
la Selva Lacandona 1950 –2000. Fondo de Cultura Econ
omica,
Mexico.
Elliott, S., Navakitbumrunga, P., Kuaraka, C., Zangkuma, S.,
Anusarnsunthorna, V. & Blakesley, D. 2003. Selecting
framework tree species for restoring seasonally dry tropical
forests in northern Thailand based on field performance. For-
est Ecology and Management 184: 177–181.
Garibaldi, A. & Turner, N. 2004. Cultural keystone species: impli-
cations for ecological conservation and restoration. Ecology
and Society 9: 1. [online] URL: http://www.ecologyandsoci-
ety.org/vol9/iss3/art1 (accessed 8 January 2013).
Gavin, M.C. 2009. Conservation implications of rainforest use
patterns: mature forests provide more resources but
secondary forests supply more medicine. Journal of Applied
Ecology 46: 1275–1282.
Holl, K.D. 2007. Old field vegetation succession in the neotrop-
ics, In: Cramer, V. & Hobbs, R. (eds.), Old fields. Dynamics and
restoration of abandoned farmland,pp.93–188. Island Press,
Washington DC, US.
Knowles, O.H. & Parrotta, J.A. 1995. Amazonian forest restora-
tion: an innovative system for native species selection based
on phenological data and field performance. Commonwealth
Forestry Review 74: 230–243.
Mart

ınez-Garza, C., Pe~
na, V., Ricker, M., Campos, A. & Howe,
H.F. 2005. Restoring tropical biodiversity: leaf traits predict
growth and survival of late-successional trees in early-suc-
cessional environments. Forest Ecology and Management 217:
365–379.
Meli, P. & Dirzo, R. 2013. Effects of grasses on sapling establish-
ment and the role of transplanted saplings on the light envi-
ronment of pastures: implications for tropical forest
restoration. Applied Vegetation Science 16: 296–304.
Meli, P., Mart

ınez-Ramos, M. & Rey-Benayas, J.M. 2013a.
Selectin g species for pas sive and active riparian restoration in
Southern Mexico. Restoration Ecology 21: 163–165.
Meli, P., Rey Benayas, J.M., Carabias, J., Ruiz, L. & Mart

ınez Ra-
mos, M., 2013b. Restauraci
on de los servicios ecosist
emicos
ribere~
nos. Meta-an
alisis global y un estudio de caso en Chiapas,
M
exico. In: Lara, A., Laterra, P., Manson, R. & Barrantes, G.
(eds.) Servicios ecosist
emicos h

ıdricos en Am
erica Latina y el Car-
ibe,pp.39–58. Red ProAgua –CYTED, Valdivia, CL.
Mendoza, E. & Dirzo, R. 1999. Deforestation in Lacandonia
(southeast Mexico): evidence for the declaration of the
northernmost tropical hot-spot. Biodiversity and Conservation
8: 1621–1641.
Millennium Ecosystem Assessment (MEA) 2005. Ecosystems and
human well-being: a framework for assessment. Wetlands and
water. World Resources Institute, Washington, DC, US.
Moreno-Cassasola, P. & Pardowska, K. 2009. Useful plants of
tropical dry forest on the coastal dunes of the center of Vera-
cruz State. Madera y Bosques 15: 21–44.
Parrotta, J.A. & Knowles, O.H. 2001. Restoring tropical forests
on bauxite mined lands: lessons from the Brazilian Amazon.
Ecological Engineering 17: 219–239.
Parrotta, J.A., Turnbull, J.W. & Jones, N. 1997. Catalyzing native
forest regeneration on degraded tropical lands. Forest Ecology
and Management 99: 1–17.
Pausas, J.G. & Lavorel, S. 2003. A hierarchical deductive
approach for functional types in disturbed ecosystems. Jour-
nal of Vegetation Science 14: 409–416.
Sansevero, J.B.B., Prieto, P.V., Duarte de Moraes, L.F. & Pena-
Rodrigues, P.J.F. 2009. Natural regeneration in plantations
of native trees in lowland Brazilian Atlantic forest: commu-
nity structure, diversity, and dispersal syndromes. Restoration
Ecology 19: 379–389.
9
Applied Vegetation Science
Doi: 10.1111/avsc.12096©2014 International Association for Vegetation Science
P. Meli et al. Selecting species for forest restoration

dos Santos, R., Citadini-Zanette, V., Leal-Filho, S. & Hennies,
W.T. 2008. Spontaneous vegetation on overburden piles in
the coal basin of Santa Catarina, Brazil. Restoration Ecology 16:
444–452.
Sharma, D. & Sunderraj, S.F.W. 2005. Species selection for
improving disturbed habitats in Western India. Current Science
88: 462–467.
Siebe, C., Mart

ınez-Ramos, M., Segura-Warnholtz, G.,
Rodr
ıguez- Vel
azquez, J. & S
anchez-Beltr
an, S. 1995. Soil
and vegetation patterns in the tropical rainforest at Chajul,
Chiapas, Southeast Mexico. In: Simmorangkir, D. (ed.), Pro-
ceedings of the International Congress on Soils of Tropical Forest
Ecosystems, 3rd Conference on Forest Soils, pp. 40–58. Mulawar-
man University Press, Samarinda, ID.
Slocum, M. 2001. How tree species differ as recruitment foci in a
tropical pasture? Ecology 82: 2547–2559.
Society for Ecological Restoration International Science & Policy
Working Group (SER) 2004. The SER International Primer
on Ecological Restoration. www.ser.org & Tucson: Society
for Ecological Restoration International.
Turnbull,L.A., Crawley, M.J. & Rees, M. 2000. Are plant popula-
tions seed-limited? A review of seed sowing experiments. Oi-
kos 88: 225–238.
Vœk, R.A. 2004. Disturbance pharmacopoeias: medicine and
myth from the humid tropics. Annals of the Association of
American Geographers 94: 868–888.
Young, T.P., Petersen, D.A. & Clary, J.J. 2005. The ecology of res-
toration: historical links, emerging issues and unexplored
realms. Ecology Letters 8: 662–673.
Zimmerman, J.K., Pascarella, J.B. & Aide, M.A. 2000. Barriers to
forest regeneration in an abandoned pasture in Puerto Rico.
Restoration Ecology 8: 350–360.
supporting information
Additional supporting information may be found in the
online version of this article:
Appendix S1. Participatory interviews with local com-
munities and social value data.
Appendix S2. Technical constraints, methods and data.
Appendix S3. List of abbreviations.
Table S1. Species list in the reference sites.
Table S2. Species list in the disturbed sites.
Table S3. Data on importance index, natural regeneration
potential and habitat breadth.
Applied Vegetation Science
10 Doi: 10.1111/avsc.12096 ©2014 International Association for Vegetation Science
Selecting species for forest restoration P. Meli et al.