Identifying areas of high invasion risk: a general model and an application to Mexico

Abstract

Maps have become a key tool to guide priorities for biodiversity conservation, the maintenance of ecosystem services, but much less so for critical action against further service loss in critical areas. Biological invasions are important disruptors of ecosystem services given that they directly or indirectly affect human well being, as they are an important cause of biodiversity loss worldwide and interfere with the provision of many ecosystem services. Here, we propose a general model to identify regions where the probability of plant invasion is higher and can cause and/or aggravate negative effects upon ecosystems. We then apply the general model to Mexico. Our model of probability of invasion considers 4 main variables: propagule availability, vegetation type, anthropic disturbance and native plant species richness. We calculated an invasion risk index combining all factors. We produced 5 maps, one for each variable and another constructed with our model of combined risk, for a grid of 0.5° × 0.5° grid across the whole country. We validated our model with State level data on exotic plants per State and obtained a significant correlation (r= 0.73, p< 0.001) between our invasion risk index derived from the model and the observed density of exotic species. Areas with greater susceptibility to invasion are closer to large human settlements and to areas of intensive agriculture. Very high risk corridors and islands were detected in our maps, as well very high risk areas in high diversity regions such as Chiapas and the Puebla-Veracruz border where we suggest attention should be focused. Our model although simple, provides a useful tool for policy design to detect areas within a specific region or country where biotic invasions are likely to have a large effect. Locating these areas is important in order to maximize return on monetary and human resources and to minimize damaging effects of plant invasions.

Full text

http://dx.doi.org/10.7550/rmb.44743
1870-3453/Derechos Reservados © 2015 Universidad Nacional Autónoma de México, Instituto de Biología. Este es un artículo de acceso abierto distribuido
bajo los términos de la Licencia Creative Commons CC BY-NC-ND 4.0.
Disponible en www.sciencedirect.com
www.ib.unam.mx/revista/
Revista Mexicana de Biodiversidad
Revista Mexicana de Biodiversidad 86 (2015) 208-216
* Corresponding author.
E-mail address: ekdelval@cieco.unam.mx (E. del-Val).
Abstract
Maps have become a key tool to guide priorities for biodiversity conservation, the maintenance of ecosystem services, but much less so for
critical action against further service loss in critical areas. Biological invasions are important disruptors of ecosystem services given that they
directly or indirectly affect human well being, as they are an important cause of biodiversity loss worldwide and interfere with the provision of
many ecosystem services. Here, we propose a general model to identify regions where the probability of plant invasion is higher and can cause and/
or aggravate negative effects upon ecosystems. We then apply the general model to Mexico. Our model of probability of invasion considers 4 main
variables: propagule availability, vegetation type, anthropic disturbance and native plant species richness. We calculated an invasion risk index
combining all factors. We produced 5 maps, one for each variable and another constructed with our model of combined risk, for a grid of 0.5º × 0.5º
grid across the whole country. We validated our model with State level data on exotic plants per State and obtained a signicant correlation (r=
0.73, p< 0.001) between our invasion risk index derived from the model and the observed density of exotic species. Areas with greater susceptibility
to invasion are closer to large human settlements and to areas of intensive agriculture. Very high risk corridors and islands were detected in our
maps, as well very high risk areas in high diversity regions such as Chiapas and the Puebla-Veracruz border where we suggest attention should be
focused. Our model although simple, provides a useful tool for policy design to detect areas within a specic region or country where biotic
invasions are likely to have a large effect. Locating these areas is important in order to maximize return on monetary and human resources and to
minimize damaging effects of plant invasions.
All Rights Reserved © 2015 Universidad Nacional Autónoma de México, Instituto de Biología. This is an open access item distributed under the
Creative Commons CC License BY-NC-ND 4.0.
Keywords: Exotic species; Invasion model; Ecosystem services; Biotic invasion
Resumen
El desarrollo de mapas se ha convertido en una herramienta clave para el mantenimiento de los servicios ecosistémicos; sin embargo, ha sido
poco utilizada para prevenir la pérdida de estos. Las invasiones bióticas son consideradas como agentes de perturbación debido a que ocasionan
importantes pérdidas en la biodiversidad e intereren con la provisión de servicios. Este trabajo propone un modelo regional para detectar áreas con
alta probabilidad de invasión por plantas. El modelo se parametriza y se valida para México considerando 4 variables: disponibilidad de propágulos,
tipo de vegetación, disturbio antrópico y riqueza de plantas nativas. Obtuvimos 5 mapas para México, uno para cada factor y otro más con el
resultado del modelo de probabilidad de invasión (cuadrícula 0.5º × 0.5º). Validamos nuestro modelo contra la densidad de exóticas por estado y
obtuvimos una correlación signicativa (r= 0.73, p< 0.001). Las regiones con mayor susceptibilidad de invasión estuvieron cercanas a grandes
ciudades y grandes extensiones agrícolas, pero también a regiones con alta biodiversidad, como Chiapas y la frontera entre Puebla y Veracruz.
Ecology
Identifying areas of high invasion risk:
a general model and an application to Mexico
Identicando áreas con riesgo elevado de invasión:
un modelo general y una aplicación para México
Ek del-Val,a,* Patricia Balvanera,a Fabiana Castellarini,a,b Francisco Javier Espinosa-García,a
Miguel Murguía,c and Carlos Pachecoa
a
Centro de Investigaciones en Ecosistemas, Universidad Nacional Autónoma de México, Antigua carretera a Pátzcuaro Núm. 8701,
Col. Ex-Hacienda de San José de La Huerta, 58190 Morelia, Michoacán, Mexico
b
Instituto Multidisciplinario de Biología Vegetal, Consejo Nacional de Investigaciones Cientícas y Técnicas,
Universidad Nacional de Córdoba, CC 495 5000 Córdoba, Argentina
c
Unidad de Biología, Tecnología y Prototipos (UBIPRO), FES Iztacala, Universidad Nacional Autónoma de México,
Av. de los Barrios 1, Los Reyes Iztacala, 54090 Tlalnepantla, Estado de México, Mexico
Received 25 February 2014; accepted 26 September 2014

E. del-Val et al / Revista Mexicana de Biodiversidad 86 (2015) 208-216 209
Barnet, & Kartesz, 2003; Villaseñor & Espinosa-García, 2004)
and the drivers underpinning invasibility have been widely
studied (Arriaga, Castellanos, Moreno, & Alarcón, 2004; Chy-
try et al., 2009; Chytry et al., 2012; Deutschewitz, Lausch,
Künh, & Klotz, 2003; Pino, Font, Carbó, Jové, & Pallares,
2005; Stohlgren et al., 2006). On the other hand, niche-based
predictions have been employed to project future distribution of
individual invasive species (Arriaga et al., 2004; Zimmerman et
al., 2011). Yet, this approach is extremely data intensive and ac-
tion cannot wait until such information is gathered for all pos-
sible invasive species in most countries.
Invasion risk maps to guide priority action that can be pro-
duced with readily available information are urgent for most
countries. This is particularly true for the case of Mexico for
various reasons. First, it is a highly diverse country with little
public and governmental awareness of the threats of the bio-
logical invasions (Espinosa-García, 2009), thus, information on
areas where invasive species could have a signicant negative
effect on ecosystems and human societies are urgently needed
(IMTA, TNC, Conabio, Aridamerica, & GECI, 2008). Second,
there are well-known examples of how invasives are having a
strong effect upon biodiversity, ecosystems and human-well be-
ing (Pejchar & Mooney, 1999), e. g. the exotic water hyacinth
(Eichhornia crassipes) (Martínez-Jiménez & Gómez-Balandra,
2007; Pérez-Panduro, 1998) and the Itchgrass (Rottboellia co-
chinchinensis), considered to be one of the worst weeds in the
world (Esqueda-Esquivel, 2005; Holm, Plucknett, Pancho, &
Herberger, 1977; Medina-Pitalúa & Domínguez-Valenzuela,
2001). Third, ongoing research has already explored what are
the most important factors associated with the presence of inva-
sive species in Mexico as well as their relative importance at the
country level (Espinosa-García, Villaseñor, & Vibrans, 2004;
Villaseñor & Espinosa-García, 2004).
In this manuscript we developed a conceptual model and a
simple analytical procedure based on readily available informa-
tion for mapping invasibility. We apply the model to the case of
the whole Mexican country, and use empirical data to validate
our model. We then discuss how much was gained from this
approach and what are its limitations. We also discuss how use-
ful this map could be for other countries beyond Mexico.
Materials and methods
The conceptual model
Four main factors have been found to be among the most
important for plant invasions into a spatially explicit model of
Introduction
Mapping has become a key tool to guide priority action. Re-
cent literature shows an increasing interest in mapping ecosys-
tem services (Martínez-Harms & Balvanera, 2012). The
identication of priority areas for maintaining the provision of
ecosystem services and for exploring potential synergies or
conicts between biodiversity conservation and that of ecosys-
tem services (Martínez-Harms & Balvanera, 2012; Turner et
al., 2007) has relied on this approach. Also, recent emphasis has
been put on how much ecosystems have been impacted by hu-
man enterprise (Ellis & Ramankutty, 2008; Halpern et al.,
2008). Such maps are critical for identifying areas where resto-
ration, for instance, is most urgently needed.
Maps to guide priority action in the prevention and man-
agement of invasive species are scarce (Chytry et al., 2009;
Mgidi et al., 2007; Nel et al., 2004; Rouget et al., 2004). In-
vasive species are an increasing threat to human wellbeing
and to ecosystems in general. Mapping invasibility, dened
as the overall susceptibility of sites to invasion (Williamson,
1996), could become key tools to guide urgent preventive ac-
tions. Invasive species can cause severe shifts in ecosystems,
leading to native species extinctions, to substantial economic
loss, reductions in the ability to provide ecosystem services
and threats human health (Mack & Erneberg, 2002; Pimentel,
Zuniga, & Morrison, 2005). Today species invasions are con-
sidered as the second cause of biodiversity loss, just behind land
use change (Leung et al., 2002; Vitousek, D´Antonio, Loope,
& Westbrooks, 1996). Big shifts in native species composition
have been documented in South Africa, Australia and the USA,
where approximately 400 of the 958 species that are listed as
threatened or endangered are considered to be at risk because
of competition-with and predation by non indigenous species
(Pimentel, Lach, Zuniga, & Morrison, 2000). Species invasions
also cause substantial economic losses; Pimentel et al. (2005)
have calculated that in the US alone over $120 billion are spent
due to species invasions whereas Colautti, Bailey, van Over-
dijk, Amudsen, and MacIsaac (2006) estimated that Canada is
losing $187 million Canadian per year. Other countries such
as Mexico do not have sufcient information about the effects
of non- indigenous species on the economy, but few plant and
sh species cause severe losses (Aguirre-Muñoz et al., 2009;
Espinosa-García & Vibrans, 2009; Espinosa-García, Villase-
ñor, & Vibrans, 2009).
Invasion research is ripe for the development of invasion risk
maps to guide priority action. An increasing amount of empiri-
cal data available on invasive species, in many parts of the
world (NLWRA, 2007; Rejmánek & Randall, 2004; Stohlgren,
Nuestro modelo, a pesar de ser simple, provee una herramienta útil para diseñar políticas públicas para detectar áreas con alta probabilidad de
invasión y maximizar los recursos nancieros y humanos.
Derechos Reservados © 2015 Universidad Nacional Autónoma de México, Instituto de Biología. Este es un artículo de acceso abierto distribuido
bajo los términos de la Licencia Creative Commons CC BY-NC-ND 4.0.
Palabras clave: Especies exóticas; Modelo de invasión; Servicios ecosistémicos; Invasión biótica

210 E. del-Val et al / Revista Mexicana de Biodiversidad 86 (2015) 208-216
pirical data gathered at a larger spatial scale than the one used
for modeling. Lastly, we rened the model according to the test.
Calculating values for each variable
The territory of Mexico was divided into 861 quadrats of
0.5º × 0.5º where the values of the 4 variables were recorded.
Quadrats of regular size are widely used in biodiversity analysis
at country and world-wide geographic scales (e. g. Ellis and
Ramankutty, 2008). The calculation of the different variables
is detailed below.
1) Propagule availability index (PAI). Given that propagule
availability is related to anthropogenic activity and to roads,
we assumed that highest population densities and highest
road densities were predictors of highest anthropogenic ac-
tivity, which in turn would contribute to highest propagule
availability. We calculated a PAI proxy as the population
density (log density) per quadrant multiplied by road density
(log roads/ha), converted to positive number and then nor-
malized. Road density was obtained from Secretaría de Co-
municaciones y Transportes map (SCT, 2008) and population
densities were obtained from Mexican population census
(Inegi, 2 0 0 5b).
2) Biophysical condition index (BCI). The conditions that can
potentially contribute to establishment and performance of
exotic species were assessed using a potential vegetation
map of Mexico proposed by Rzedowski (1978) that employs
9 vegetation categories. Such potential vegetation categories
were ranked from 1 (less invasible) to 9 (highly invasible)
based on general conclusions from previous investigations
(Holdgate, 1986; Lonsdale, 1999) and from an assessment
coordinated by F. J. Espinosa-García. We used the following
categories and ranking (in parenthesis the relative coefcient
of invasibility assigned to each vegetation type): wet tropical
forest (1), subtropical wet forest (2), cloud forest (3), decidu-
ous tropical forest (4), temperate forest (5), thorn forest (6),
aquatic vegetation (7), scrubland (8) and pasture (9). We cal-
culated the area covered by each vegetation type per quad-
rant, multiplied by the corresponding vegetation ranking and
added these up to have one number per cell. Since we wanted
to keep our model as simple as possible, and the ranking
among land cover classes with an interval scale is practically
impossible for the entire country of Mexico, we decided to
use our ordinal scale as a proxy of an interval scale. We did
not nd a more parsimonious option to do the ranking.
3) Disturbance index (DI). We assessed habitat disturbance
through intensity of land transformation. Land use and land
cover information was obtained from the most recent and
most detailed vegetation and land use map from Mexico
(Inegi, 2005a). Land use and land cover categories were
clumped into 8 groups, and each group was assigned a coef-
cient of disturbance (between 1 and 7); we assumed that
the more heavily transformed the more severe the change
in the disturbance regime relative to conserved conditions,
and thus the highest the probability of suffering invasions
from exotic species. The coefcients were chosen from an
invasibility (Chytry et al., 2008; Eschruth & Battles, 2009;
Lonsdale, 1999).
Propagule availability regulates the frequency of arrival
events and the amount of seeds or individuals of a given exotic
species arriving to a particular place. Propagule availability has
been linked with local roads and highways, and invasion by ex-
otic plants has been shown to be facilitated by the proximity to
roads in wetlands (Choi & Bury, 2003) and semiarid grasslands
(Gelbard & Belnap, 2003); the importance of such roads in the
maintenance of invasive populations and as a conduit for their
dissemination is widely accepted (Christen & Matlack, 2009;
Forman, 2000; Gelbard & Belnap, 2003). Furthermore, roads are
associated with habitat destruction, which paves the way for inva-
sions (Forman & Alexander, 1998). Also, human activity favors
accidental introductions and deliberate plantings of ornamental
or domesticated plant species that may become feral (Mack &
Erneberg, 2002). Thus towns and cities and road edges become
repositories of non-native species and sources of propagules that
are dispersed by vehicle adhesion at short or long distance (von
der Lippe & Kowarik, 2007; Wichman et al., 2009).
Invasibility depends on habitat characteristics or biophysical
conditions. Some systems have been suggested to be more
prone to species invasions than others, yet the reasons behind
these trends are still not well understood; also, it is known that
habitat type interacts with other invasion drivers (Vila, Pino, &
Font, 2007; Vila et al., 2008). It is very difcult to infer pattern
from process, and invasibility as a habitat property is confound-
ed with propagule pressure and the attributes of the invading
species themselves (Lonsdale, 1999). Nevertheless habitat is a
better correlate of the level of plant invasion than isolated envi-
ronmental variables (Chytry et al., 2008; Lonsdale, 1999).
The disturbance regime has also been recognized to be one
of the main factors promoting plant species invasions (Daehler,
2003; Espinosa-García et al., 2004; Vila et al., 2007). The larg-
er the departure from the natural disturbance regime (i.e. habi-
tat transformation), the larger the non-native species richness
(Chytry et al., 2008; Daehler, 2003; Espinosa-García et al.,
2004).
Invasibility is also related to native plant richness (Chytry et
al., 2008). Very consistent correlations across the world have
shown a very robust positive correlation between native species
richness and non-native species richness, particularly at large
spatial scales (e. g., Espinosa-García et al., 2004; Lonsdale,
1999; Stohlgren et al., 2003).
Applying the conceptual model to mapping invasion risk
inMexico
Using the arguments presented above we developed an inva-
sion risk model, parametrized it for Mexico and validated the
model. Parameterization involved: i) calculating values for
propagule availability (road and population density), biophysi-
cal conditions (habitat type), disturbance regime (habitat trans-
formation), and native species richness for spatial units in allthe
country; ii) calculating an invasion risk index adding up all
factors, and iii) mapping the resulting predicted values. Valida-
tion then involved testing the predictions with independent em-

E. del-Val et al / Revista Mexicana de Biodiversidad 86 (2015) 208-216 211
components of the IRI to the nal IRI for each grid cell to assess
for potential overrepresentation of any of these components.
Results
Emerging patterns
The map produced to assess propagule availability of inva-
sive species (PAI) showed the highest scores close to large ur-
ban centers like México City, Monterrey, Guadalajara and
Tijuana (Fig.1A).
The map of the biophysical conditions that promote invasi-
bility (BCI; vegetation types) showed a different pattern. North-
ern Mexico appears more prone to invasion followed by central
Mexico and the Pacic Coast following this index (Fig.1B).
The disturbance index (DI) map also showed the highest
score in cells situated in the proximity of cities but regions with
technied agriculture like the Veracruz plateau and the Sinaloa
elds also scored high. Places with the lowest scores were Baja
California Sur and the Chihuahuan desert that are dominated
by primary vegetation (Fig.1C).
The species richness (SRI) map showed a concentration of
high species richness in the southern portion of the country with
2 additional hotspots in the Nuevo León-Tamaulipas southern
border and in the Sonora-Chihuahua-Sinaloa border (Fig.1D).
The nal invasion risk (IRI) map (Fig.2) showed that central
Mexico appears to be the area with highest risk. Invasion prob-
ability is also concentrated close to large urban centers such as
Tijuana, Monterrey and Guadalajara. Surprisingly, Southern
Chiapas, northern Oaxaca and central Veracruz scored high
due to the role played by species richness.
Many very high invasion risk (VHIR) areas (II= 0.62-1) ap-
peared isolated, while VHI corridors are evident along the Neo-
volcanic belt (Estado de México, Distrito Federal, Hidalgo,
Puebla, Tlaxcala and Veracruz) for temperate areas. The Neo-
volcanic belt has other VHIR areas in Michoacán, Jalisco and
Colima, neighboring with high invasion risk (HIR) (II= 0.40-
0.61) areas. If these HIR cells were to change their status to
VHIR, then the whole Neo-Volcanic belt would have 2 of the
most important commercial seaports at every end: Manzanillo
and Veracruz. Seaports, airports and border-crossing terrestrial
ports function as gateways for invasive species. Once estab-
lished, these species disperse easily along corridors such as
those formed by VHIR areas.
There is another VHIR corridor for wet tropical lowlands of
southern Veracruz, Chiapas and Tabasco, with the Veracruz
seaport at the north extreme and several border terrestrial ports
at the southeastern extreme. The temperate Chiapas highlands
appear very highly invasible, but they are not connected with
other temperate VHIR areas.
Testing the predictions with empirical data and refining the
model
In our predicted index values at the State level (Table 1), we
found that Distrito Federal and Tlaxcala had the highest indices
expert assessment. From low to high intensity, categories
(in parenthesis the coefcients) were: primary vegetation
(1), secondary vegetation (2), forests plantations (3), induced
pastures (4), rain-fed agriculture (5), intensive irrigated ag-
riculture (6) and human settlements (7). For each quadrant
we calculated the surface covered by of each land use type,
multiplied it by the coefcient assigned and added them up
for each quadrant considering all land use types within it.
4) Species richness index (SRI). Species richness of ower-
ing plants was estimated using the best available oristic
database for Mexico that contains herbarium records, the
World Network for Biodiversity Information (Red Mundial
de Información sobre Biodiversidad, REMIB; http://www.
conabio.gob.mx/remib). To rene the obtained richness
values, the database for which 1º × 1º grid (as reported in
Villaseñor, Maeda, Rosell, & Ortiz, 2007) was modied
using Asteraceae and Fabaceae richness values that have
been calculated and validated (Villaseñor et al., 2007). The
total species number included in our model is 2,848 Astera-
ceae and 1,543 Fabaceae. While the Villaseñor database
is available for a 0.5º × 0.5º grid, it has been most widely
used for the 1º × 1º grid to avoid noise from unsampled or
incompletely sampled cells. Thus, to maintain data accuracy
we used this layer of information based on a 1º × 1º grid and
not for a 0.5º × 0.5º grid as we did with the other 3 layers.
5) Invasion risk index (IRI). Our nal index was a combina-
tion of the 4 variables. Although we know these may not be
equivalent, we decided not to assess a differential contribution
to each one since there is no agreement upon which is more
determinant of plant invasion (Chytry et al., 2008; Eschruth
& Battles, 2009; Lonsdale, 1999). Values obtained for each of
the 4 variables were normalized assigning the highest score
to 1 and the lowest score to 0 to generate an equivalent scale
among variables. We ended up with the following index:
IRI= PAI+BCI+DI+SRI
We calculated an index per each cell of the grid and pro-
duced a nal invasibility map of 0.5º × 0.5º for Mexico. Differ-
ent layers were incorporated using Arcgis 9.3 (ESRI). Because
all the 4 variables are normalized to the unit, and the sum of the
indices magnitude was also normalized, the theoretical values
of IRI runs from 0 through 1, a higher value of the IRI indicates
a site with a higher degree of invasibility.
Model validation and refining
To validate our model we used a readily available database on
the density of recorded plant exotics per state from Villaseñor
and Espinosa-García (2004). We calculated the mean invasion
risk index per Mexican State, based on the data of all grid cells
found within such state. We then correlated our predicted values
with actually observed ones. Based on these values, we decided
to optimize our model by removing the BCI component from the
IRI. We recalculated the new average IRI per state without the
BCI and correlated it again with the observed density of non-na-
tive plants per state. We explored the contributions of each of the

212 E. del-Val et al / Revista Mexicana de Biodiversidad 86 (2015) 208-216
(0.77 and 0.53, respectively) while Quintana-Roo and Campeche
had the lowest (0.32 and 0.34 respectively).
The synthetic invasion risk index we developed appears to be
a good predictor of how many introduced plant species have
been found in a State. We found a good correlation between
recorded introduced plant species by State and the predicted
invasion risk index we propose (Fig.3A; r= 0.73, t(1,30) = 5.88,
p<0.001). The correlation was clearly improved by using the
new Invasion Risk Index that does not include the BCI compo-
nent (Fig.3B, r= 0.82, t(1,30 )= 8.11, p< 0.001).
In order to check if the specic contribution of any compo-
nent of the IRI to the nal IRI was not biased towards one of
them, we identied those cells where any component contrib-
uted in more than 50%, more than 75% and more than 90% to
any cell (supplementary material). We found that propagule
availability index contributes more than 50% and less than 75%
of our nal index in 33.4% of cells, the disturbance index con-
Figure 2. Invasion risk map of Mexico. Invasion risk index map for Mexico
showing the final scoring for each 0.5º × 0.5º plot. The index ranges from 0
(low invasibility) to 1 (high invasibility) and thus areas in red a re potentially
more invaded than areas in yellow.
Figure 1. Variables for the invasion risk model. Maps showing the 4 variables used to construct the invasion risk model for a 0.5º × 0.5º grid in Mexico. A),Usa-
ge index map based on human population density and road densities; B), biophysical conditions index map based on the Rzedowski´s potential vegetation cate-
gories (1978); C), disturbance index map based on land use types, and D), species richness index map for Mexico (SI) based on a 1º × 1º grid. The square on the
left-down corner of each map is a zoom of the Mexico City and adjacent areas.
A B
C D
Usage Index (UI)
0.00 - 0.03
0.04 - 0.10
0.11 - 0.12
0.13 - 0.14
0.15 - 0.30
Biophysical Conditions Index (BCI)
0.00 - 0.03
0.04 - 0.06
0.07 - 0.08
0.09 - 0.10
Disturbance Index (DI)
0.01 - 0.06
0.07 - 0.11
0.12 - 0.17
0.18 - 0.23
0.24 - 0.40
Species rinches Index (SI)
0.00 - 0.02
0.03 - 0.05
0.06 - 0.09
0.10 - 0.20
Invasion risk
0 - 0.14
0.15 - 0.31
0.32 - 0.45
0.46 - 0.61
0.62 - 1
States

E. del-Val et al / Revista Mexicana de Biodiversidad 86 (2015) 208-216 213
ity areas close to urban centers and predicted high invasibility
in lowland areas contrasting with low invasibility in the boreal
and mountain regions across the continent. Climate, habitat and
landscape diversity, and man-induced disturbance are the most
important factors explaining alien diversity in Spain (Pino et
al., 2005), Great Britain (de Albuquerque, Castro-Díez, Rodrí-
guez, & Cayuela, 2011), and USA (Guo, Rejmánek, & Wen,
2012). However a recent study on future plant invasion patterns
in Europe (Chytry et al., 2012) found that levels of invasion will
likely increase in northern Europe. In Mexico there is only
1study concentrating on the probability of invasion by Buffel
grass, showing that the probability of invasion is concentrated
in northern Mexico (Arriaga et al., 2004).
Real plant invasion threat may in fact be larger than that
shown in our maps. Given that plant species richness is a pre-
dictor of invasibility, and that large areas in Mexico are not well
explored botanically, we could be underestimating the threat,
turning prevention action in key areas as important to attack
potential areas where invasion could be larger than shown in
our model.
Our model showed very high correlations between our pre-
dicted values and the observed ones, especially when the infor-
mation about biotic conditions (vegetation type) was removed.
Correlations found in other studies for validating predictive
ecosystem services maps with readily available data can be as
low as 0.13 for poor ts (Bowker, Miller, Belnap, Sisk, & John-
son, 2008) and starting at 0.3 for barely adequate t, to 0.5 to
tributes more than 50% and less than 75% of our nal index in
9.9% of cells and S index contributes more than 50% and less
than 75% of our nal index only in 1.85% of cells. Also the PA
index only contributes more than 75% in 1% of cells, D index
only contributes more than 75% in 0.8% of cells and S index do
not contribute more than 75% in any cell. Only the disturbance
index contributed in 90% in 0.6% of cells (5 cells).
Discussion
Maps for invasion risk at a country level are rare and in this
work we present an approach to build them. Our model pro-
vides a tool to identify sites with a high potential abundance of
exotic plant species, and where the problem of invasive species
could be causing great harm to ecosystems, as well as econom-
ic losses and threats to human health.
We detected a pattern of very high invasion risk corridors
that are potentially useful to set monitoring and control priori-
ties for Mexico. Also Mexican urban areas, among other high-
lighted areas in the southeast, appear as hotspots for invasion.
This information is useful to report to policy makers (both local
and national) in order to concentrate efforts and economic re-
sources in high scored areas to monitor and eradicate danger-
ous invasive species.
Previous efforts of mapping invasibility in Europe (Chytry et
al., 2009; Deutschewitz et al., 2003) also showed high invasibil-
Table 1
Mean values of the invasion risk index and its 4 components calculated for all Mexican States, also showing land surface (km2) for reference. Values go from 0 to 1
State Surface (km2)Usage index Biophysical
conditions index
Disturbance index Species richness
index
Invasion risk index
Quintana Roo 39,147 0.118 0.019 0.10 6 0.028 0.320
Campeche 51,139 0 .12 0.023 0.125 0.025 0.342
Yucatán 37,4 25 0.123 0.031 0.156 0.025 0.4 01
Baja Califor nia Sur 73,964 0.108 0.083 0.055 0.019 0.470
Sonora 180,937 0.102 0.079 0.084 0.028 0.488
Durango 122, 031 0.112 0.0 75 0.095 0.024 0.490
Tabasco 24,70 2 0.129 0.036 0.2 0 .0 41 0.493
Chihuahua 24 6,991 0 .104 0.0 81 0.084 0.025 0.494
Coahuila 150,670 0.104 0.088 0.073 0.025 0.508
Baja Californ ia 73,566 0.102 0.086 0.098 0.019 0.509
Nayarit 27,7 71 0.124 0.049 0.149 0.054 0.509
Sinaloa 56,801 0 .123 0.053 0 .16 0.0 47 0.522
Guerrero 63,609 0.12 8 0.046 0.156 0.069 0.538
Michoacán 58,30 0 0.12 6 0.05 0.176 0.054 0.539
Zacatecas 74, 502 0.11 0.08 0.14 0.023 0.540
Nuevo León 63,615 0 .108 0.085 0.127 0.024 0.545
Chiapas 73,46 4 0.131 0.033 0 .175 0.091 0.548
Veracruz 71,470 0.133 0.023 0.236 0.079 0.550
Oaxaca 93,707 0.12 6 0.042 0.157 0.086 0.552
San Luis Potosí 60,463 0.114 0.08 0.132 0.036 0.564
Jalisco 77,9 53 0.127 0.054 0 .168 0.064 0.567
Tamaulipas 79,404 0 .114 0.073 0.173 0.039 0.578
Colima 5 ,752 0.132 0.044 0 .192 0.089 0.601
Puebla 34 ,119 0 .14 0.053 0.22 0.0 78 0.646
Guanajuato 30,336 0.138 0.072 0.222 0. 0 41 0.647
Aguascalientes 5,560 0.136 0.083 0. 213 0.034 0.664
Morelos 4,862 0.158 0.048 0. 251 0.0 74 0.667
Querétaro 11, 604 0 .135 0.078 0.179 0.074 0.683
Hidalgo 20,653 0.139 0.062 0.215 0.085 0.684
México 22,227 0.158 0.06 0.234 0.069 0.688
Tlaxcala 3,982 0 .166 0 .061 0.276 0.068 0.733
Distrito Federal 1,487 0.21 0.065 0.242 0.072 0.777

214 E. del-Val et al / Revista Mexicana de Biodiversidad 86 (2015) 208-216
well established that some species act as ecosystem engineers
(sensu Jones, Lawton, & Shackak, 1994) therefore altering eco-
system dynamics in essential features while other do not have
great impact on ecosystems. For example invasion by saltcedar
(Tamarix ramosissima) in northern Mexico may have exacer-
bated outcomes because it can alter water ow of large areas
along the riverbanks (Zavaleta, 2000). On the other extreme,
weeds that live in small populations could add biodiversity
without altering ecosystem dynamics (Espinosa-García et al.,
2004). Yet, considering the whole load of exotic plants is sup-
ported by the fact that there is a good correlation between the
number of exotic plant species present in a particular site and
the number of noxious exotic plant species (Rejmánek & Ran-
dall, 2004).
The information provided by this map can guide action in a
country with incipient information about invasive plant species.
Raising awareness on government and society of key highlight-
ed areas is much needed. The mapping initiative presented in
this paper provides a framework to evaluate invasion risk at
regional scales. Our invasion risk model simplied to 3vari-
ables that are easily obtained, give good estimations of exotic
plant species densities at the state level in Mexico. Since species
invasions are believed to be the second cause of biodiversity
loss globally, the cause of many economic losses and impacts
on the human welfare, our model could be used as a tool to
prioritize resources where invasion risk is high and material
resources are limited.
Acknowledgements
We would like to express our gratitude to Urani Carillo for
help with data processing; Heberto Ferreira, Alberto Valencia
and Atzimba López provided logistical support to enable com-
munication and data sharing among the co-authors. Conabio
funded this research through project FQ003 to P. Balvanera.
P.Balvanera ackowledges PASPA-UNAM for sabbatical fel-
lowship.
0.7 for excellent ts (Bowker et al., 2008; Eigenbrod et al.,
2010). Much higher correlations could not be expected given
that not all States have been equally sampled. States with high-
er invasion risk show the greatest density of reported exotic
plants (i. e. Distrito Federal, Tlaxcala, Hidalgo, Mexico). There-
fore our model appears to be a good predictor for number of
invasive plants at the state level in Mexico.
The model validation performed may be limited by the reso-
lution used. While validation at the 0.5º × 0.5º grid level would
be needed, no data was available. On the other hand, it is well
known that resolution can change results from models based on
proxys, as has been observed for the case of ecosystem services
(Eigenbrod et al., 2010). Nevertheless, in this case our model
had the highest resolution, and was then averaged for each state
for the validation. It has been shown that predictive maps for
ecosystem services have low error at broad resolutions (e. g.
grids cells that are 20 km wide), such as the ones used here
(ours are 25 Km wide), but not so much at ner scales (e. g.
2km wide).
The model developed here may be useful for other countries
or regions, but we strongly suggest that sensitivity analyses for
each particular site and supporting empirical data will be im-
portant for decision making along with mapping services for
conservation.
The invasion risk map presented here has limitations, as
most models based on proxies can have, particularly since some
exotic species introduced on purpose are not always regarded as
harmful and different stakeholders may have different appre-
ciations of the same situation (Talliset al., 2012). For example in
northern Mexico there is a conict of interests with the exotic
Cenchrus ciliaris (Buffel grass) since it is highly appreciated by
ranchers because it provides large quantities of fodder in very
dry areas. Yet, this species promotes re, causing native species
displacement or even extinction that concern conservationists
(Arriaga et al., 2004; Búrquez, Millar, & Martínez-Irízar, 2002;
Franklin et al., 2006). Such conicts cannot be predicted here.
This model, considers all species to be equivalent on their
impacts on ecosystems, a highly unlikely scenario. It has been
Figure 3. Invasion risk index vs. Introduced species density. Correlation between mean invasibility per state and actual density of introduced species per state
from Espinosa-García et al., 2004. A), Invasion risk Index using all the variables; B), index without the index for biophysical conditions (BCI) based on vege-
tation type.
Log10 Introduced species density
0 0.1 0.2 0.3
y = 3.6859x – 10.461
R2 = 0.53571
0.4 0.5 0.6 0.7 0.8 0.9
0
–2
–4
–6
–8
–10
Invasibility index
A
Log10 Introduced species density
0 0.1 0.2 0.3
y = 2.8694x – 9.8811
R2 = 0.68667
0.4 0.5 0.6 0.7 0.8 0.9
0
–2
–4
–6
–8
–10
Invasibility index
B

E. del-Val et al / Revista Mexicana de Biodiversidad 86 (2015) 208-216 215
Controlling them in North America (pp. 23–32). Tucson: Arizona-Sonora
Desert Museum Press.
Espinosa-García, F. J., Villaseñor, F. J., & Vibrans, H. (2004). The rich
generally get richer, but there are exceptions: Correlations between
species richness of native plant species and alien weeds in Mexico.
Diversity a nd Distributions, 10, 39 9 –4 07.
Espinosa-García, F. J., Villaseñor, F. J., & Vibrans, H. (2009). Biodiversity,
distribution, and possible impacts of exotic weeds in Mexico. In T. van
Devender, F. J. Espinosa- García, B. L. Harper-Lore, & T. Hubbard
(Eds.), Invasive plants on the move. Controlling them in North America
(pp. 43–52). Tucson: Arizona-Sonora Desert Museum Press.
Esqueda-Esquivel, V. A. (2005). Efecto de herbicidas sobre plantas y semillas
de Rottboellia cochinchinensis (Lour.) W. Clayton, en caña de azúcar.
Agronomía Mesoamericana, 161, 45–50.
Forman, R. T. T. (2000). Estimate of the area affected ecologically by the road
system in the United States. Conservation Biology, 14, 31–35.
Forman, R. T. T., & Alexander, L. E. (1998). Roads and their major ecologica l
effects. Annual Review of Ecology and Systematics, 29, 20 7–231.
Franklin, K. A., Lyons, K., Nagler, P. L., Lampkin, D., Glenn, E. P., Molina-
Fraener, F., et al. (2006). Buffelgrass (Pennisetum ciliare) la nd
conversion and productivity in the plains of Sonora, Mexico. Biological
Conservation, 127, 62 –71.
Gelbard, J. L., & Belnap, J. (2003). Roads as Conduits for Exotic Plant
Invasions in a Semiarid Landscape. Conservation Biology, 17, 420– 432.
Guo, Q., Rejmánek, M., & Wen, J. (2012). Geographical, socioeconomic,
and ecological determinants of exotic plant naturalization in the United
States: insights and updates from improved data. Neobiota , 12, 41– 55.
Halpern, B.S., Walbridge, S., Selkoe, K.A., Kappel, C.V., Micheli, F., et al.
(2008) A global map of human impact on marine ecosystems. Science,
319, 948–952 .
Holdgate, M. W. (1986). Summary and conclusions: characteristics and
consequences of biological invasions. Philosophical Transactions of the
Royal Society of London Series B-Biological Sciences, 15, 733 –742.
Holm, L., Plucknett, D. L., Pancho, V., & Herberger, P. (1977). The world’s
worst weeds: distribution and biology. Honolulu: University Press of
Hawaii.
IMTA (Instituto Mexicano de Tecnología del Agua), TNC, Conabio,
Aridamerica, & GECI (2008). Especies invasoras de alto impacto a
la biodiversidad: prioridades en México. Jiutepec, Morelos: IMTA
(Instituto Mexicano de Tecnología del Agua).
Inegi (Instituto Nacional de Estadística y Geografía) (2005a). Carta de uso
de suelo y vegetación. Serie III, escala 1:250,000. Conjunto de Datos
Vectoriales. Aguascalientes: Inegi.
Inegi (Instituto Nacional de Estadística y Geografía) (2005b). Instituto
de Integración territorial del conteo de población y viviend a Inegi.
Aguascalientes: Inegi.
Jones, C. G., Lawton, J. H., & Shackak, M. (1994). Organisms as ecosystem
engineers. Oikos, 69, 373–386.
Leung, B., Lodge, D. M., Finnoff, D., Shogren, J. F., Lewis, M. A., & Lamber ti,
G. A. (2002). An ounce of prevention or a pound of cure: bioeconomic
risk analysis of invasive species. Proceedings of the National Academy
of Science, USA, 269, 2407–2413.
Lonsdale, W. M. (1999). Global patterns of plant invasions and the concept of
invasibility. Ecolog y, 80, 1522–1536.
Mack, R. N., & Erneberg, M. (2002). The United States naturalized flora:
Largely the product of deliberate introductions. Annals of the Missouri
Botanical Garden, 89, 176 –189.
Martínez-Harms, M. J., & Balvanera, P. (2012). Methods for mapping
ecosystem service supply: a review. International Journal of Biodiversity
Science, Ecosystem Services and Management, 8, 17–25.
Martínez-Jiménez, M., & Gómez-Balandra, M. A. (2007). Integrated control
of waterhyacinth in Mexico by using insects a nd plant pathogens. Crop
Protection, 26, 1234 –1238.
Medina-Pitalúa, J. L., & Domínguez-Valenzuela, J. A. (2001). Rottboellia
cochinchinensis, un pasto fuera de la ley. Revista de la Asociación
Mexicana de la Ciencia de la Maleza, 1, 15–18.
Mgidi, T. N., Le Maitre, D. C., Schonegevel, L., Nel, J. L., Rouget, M.,
& Richardson, D. M. (2007). Alien plant invasions – incorporating
References
Aguirre-Muñoz, A., Mendoza, R. A., Arredondo, H., Arr iaga, L., Campos,
E., Contreras-Balderas, S., et al. (2009). Especies exóticas invasoras:
impactos sobre las poblaciones de f lora y fauna, los procesos ecológicos
y la economía. In J. Saruk han (Ed.), Capital Natural de México. Estado
de conser vación y ten dencias de cambio (Vol. II) (pp. 277–318). Mexico
D. F.: Conabio.
Arriaga, L., Castellanos, A. E., Moreno, E., & Alarcón, J. (2004). Potential
ecological distribution of alien invasive species and risk assessment:
A case study of buffel grass in arid regions of Mexico. Conservation
Biology, 18, 1504 –1514.
Bowker, M. A., Miller, M. E., Belnap, J., Sisk, T., & Johnson, N. (2008).
Prioritizing conservation effort through the use of biological soil crusts
as ecosystem fu nction indicators in an arid region. Conservation Biology,
22, 1533–1543.
Búrquez, A., Millar, M., & Martínez-Irízar, A. (2002). Mexican grasslands,
thornscrub, and the transformation of the Sonoran desert by invasive
exotic buffelgrass (Pennisetum ciliare). In B. Tellman (Ed.), Invasive
exotic species in the Sonoran region (pp. 147–156). Tucson: University
of Arizona Press.
Choi, Y. D., & Bury, C. (2003). Process of f loristic degradation in urban
and suburban wetlands in northwestern Indiana, USA. Natural Areas
Journal, 23, 320–331.
Christen, D. C., & Matlack, G. R. (2009). The habitat and conduit functions of
roads in the spread of three invasive plant species. Biological Invasions,
11, 453–465.
Chytry, M., Jarosik, V., Pysek, P., Hajek, O. J., Knollová, I., Tichy, L., et al.
(2008). Separating habitat invasibility by alien plants from the actual
level of invasion. Ecology, 89, 1541–1553.
Chytry, M., Pysek, P., Wild, J., Pino, J., Maskel, L. C., & Vila, M. (2009).
European map of alien plant invasions based on the quantitative
assessment across habitats. Diversity and Distributions, 15, 9 8–107.
Chytry, M., Wild, J., Pysek, P., Jarosik, V., Dendoncker, N., Reginster, I., et
al. (2012). Projecting trends in plant invasions in Europe under different
scenarios of future land-use change. Global Ecology and Biogeography,
21, 75 –87.
Colautti, R. I., Bailey, S. A., van Overdijk, C. D. A., Amudsen, K., & MacIsaac,
H. J. (2006). Characterised and projected costs of nonindigenous species
in Canada. Biological Invasions, 8, 45 – 49.
Daehler, C. C. (2003). Performance comparisons of co-occur ring native and
alien invasive plants: Implications for conservation and restoration.
Annual Review of Ecology, Evolution and Systematics , 34, 183–211.
de Albuquerque, F. S., Castro-Díez, P., Rodríguez, M. A., & Cayuela, L.
(2011). Assessing the influence of environmental and human factors on
native and exotic species richness. Acta Oecologica, 37, 51 –57.
Deutschewitz, K., Lausch, A., Künh, I., & Klotz, S. (2003). Native and alien
plant species richness in relation to spatial heterogeneity on a regional
scale in Ger many. Global Ecology and Biogeography, 12, 29 9 –311.
Eigenbrod, F., Armsworth, P. R., Anderson, B. J., Heinemeyer, A., Gillings, S.,
Roy, D. B., et al. (2010). The impact of proxy-based methods on mapping
the distr ibution of ecosystem services. Jour nal of Applied Ecology, 47,
377–385.
Ellis, E.C., & Ra mankutty, N. (2008). Putti ng people in the map: anth ropogenic
biomes of the world. Frontiers in Ecology and the Environment , 6, 439-
447.
Eschruth, A. K., & Battles, J. J. (2009). Assessing the relative importance of
disturbance, herbivory, diversity, and propagule pressure in exotic plant
invasion. Ecological Mo nographs, 79, 265–280.
Espinosa-García, F. J. (2009). Invasive weeds in Mexico: overview of
awareness, m anagement and legal aspects. In Proceedings of the Weeds
Across Borders 20 08 Conference. Banff, Canada: Alberta Invasive
Plants Council.
Espinosa-García, F. J., & Vibrans, H. (2009). The need of a national weed
management strategy in Mexico. Invasive plants on t he move. Controlling
them in North America. In T. van Devender, F. J. Espinosa- García,
B. L. Harper-Lore, & T. Hubbard (Eds.), Invasive plants on the move.

216 E. del-Val et al / Revista Mexicana de Biodiversidad 86 (2015) 208-216
Stohlgren, T. J., Barnet, D. T., & Kartesz, J. (2003). The rich get richer:
patterns of plant invasions in the United States. Frontiers in Ecology and
the Environment, 11, 11–14.
Tallis, H., Ruckelshaus, M., Plummer, M., McLeod, K., Guerry, A., Andelman,
S., et al. (2012). New metrics for managing and sustaining the ocean's
bo u nt y. Marine Policy, 36, 303–306.
Turner, W. R., Brandon, K., Brooks, T. M., Costanza, R., da Fonseca, G. A. B.,
& Portela, R. (2007). Global conservation of biodiversity and ecosystem
services. Bio-Science, 57, 868–873.
Vila, M., Pino, J., & Font, X. (2007). Regional assessment of plant invasions
across different habitat types. Journal of Vegetation Science, 18, 35–42.
Vila, M., Siamantziouras, A. S. D., Brundu, G., Camarda, I., Lambdon, P.,
Médali, F., et al. (2008). Widespread resistance of Mediter ranea n island
ecosystems to the establishment of three alien species. Diversit y and
Distributions, 10, 113 –123.
Villaseñor, J. L., & Espinosa-García, F. J. (2004). The alien flowering plants of
Mexico. Diversity and Distr ibutions, 10, 113–12 3.
Villaseñor, J. L., Maeda, P., Rosell, J. A., & Ortiz, E. (2007). Plant families as
predictors of plant biodiversity in Mexico. Diversity a nd Distributions,
13, 871–876.
Vitousek, P., D´Antonio, C. M., Loope, L. L., & Westbrooks, R. (1996).
Biological invasions as global environmental change. American Scientist,
84, 468–478.
von der Lippe, M., & Kowarik, I. (2007). Long-distance dispersal of plants
by vehicles as a driver of plant invasions. Conservation Biology, 21,
986–996.
Wichman, M. C., Alexander, M. J., Soons, M. B., Galsworthy, S., Niggemann,
M., Hails, R. S., et al. (2009). Human-mediated dispersal of seeds over
long distances. Proceedings of the Royal Society of London Series
B-Biological Sciences, 276, 523 –532.
Williamson, M. (1996). Biological invasions. London: Chapman and Hall.
Zavaleta, E. S. (2000). The Economic value of controlling an invasive shrub.
Ambio, 29, 462–467.
Zimmerman, H., Von Wehrden, H., Damascos, M. A., Bran, D., Welk, E.,
Renison, D., et al. (2011). Habitat invasion risk assessment based on
Landsat 5 data, exemplified by the shr ub Rosa rubiginosa in souther n
Argentina. Austral Ecology, 36, 870–880.
emerging invaders in regional prioritization: a pragmatic approach for
Southern Africa. Journal of Environmental Management, 84, 17 3 –187.
Nel, J. L., Richardson, D. M., Rouegt, M., Mgidi, T. N., Mdzeke, N., Le
Maitre, D. C., et al. (2004). A proposed classification of invasive alien
plant species in South Afr ica: towards prioritising species and areas for
management action. South African Jour nal of Science, 100, 53– 64.
NLWRA (National Land and Water Resources Audit) (2007). Assessing
invasive plants in Australia. Canberra: National Land and Water
Resources Audit.
Pejchar, L., & Mooney, H. A. (1999). Invasive species, ecosystem services and
human well-being. Trends in Ecology and Evolution, 24, 497–50 4.
Pérez-Panduro, A. (1998). Primera experiencia exitosa de control biológico de
lirio acuático en México. El Entomófago, 8, 3 –4.
Pimentel, D., Lach, L., Zuniga, R., & Morrison, D. (2000). Environmental and
Economic Costs of Nonindigenous Species in the United States. Bio-
Science, 50, 53 – 65.
Pimentel, D., Zuniga, R., & Morrison, D. (2005). Update on the environmental
and economic costs associated with alien-invasive species in the United
States. Ecological Economics, 52, 273–288.
Pino, J., Font, X., Carbó, J., Jové, M., & Pallares, L. (2005). Large-scale
correlates of alien plant invasion in Catalonia (NE of Spain). Biological
Conservation, 122, 339–350.
Rejmánek, M., & Randall, J. M. (2004). The total number of natura lized
species can be a reliable predictor of the number of alien pest species.
Diversity a nd Distributions, 10, 367–369.
Rouget, M., Richardson, D. M., Nel, J. L., Le Maitre, D. C., Egoh, B., & Mgidi,
T. (2004). Mapping the potential spread of major plant invaders in South
Africa, Lesotho and Swaziland using climatic suitability. Diversity and
Distributions, 10, 475–484.
Rzedowski, J. (1978). Vegetación de México. México D. F.: Limusa.
SCT (Secretaría de Comunicaciones y Transportes) (2008). Mapa nacional de
comunicaciones y transportes. Coordinación General de Planeación y
Centros SCT. Mexico D. F.: Dirección General de Planeación, Dirección
de Estadística y Cartografía, Subdirección de Cartografía (SCT).
Stohlgren, T. J., Barnet, D. T., Flather, C., Fuller, P., Peterjohn, B., Kartesz, J.,
et al. (2006). Species richness and patter ns of invasion in plants, birds,
and fishes in the United States. Biological Invasions, 8, 427 –4 47.