Global environmental changes have been well documented over the years that are causing vast immeasurable ecological changes (Vitousek, 1994). One of which includes climate change, which is projected to significantly alter biodiversity, causing changes in genetic composition, species distributions and interactions, and ecosystem processes (Vitousek, 1994; Hellmann et al., 2008). Thus, researchers have centered around the impacts of climate change on species that are endemic, endangered, or economically important for the sole purpose of preservation (Hellmann et al., 2008). However, over the years it has become evident that human introductions of exotics have resulted in astonishing global changes, such as biotic homogenization, and extinction of native or endemic taxa to such an extent that it has increased concern among scientists as well as the public and governments (Pound et al., 2011). These and many more reasons have prompted researchers to consider how climate change might affect invasive species of economically important crops, species causing human disease, and other ecological and economic implications separately from native species (Hellmann et al., 2008).
Alien invasive species (AIS) are characterized as taxa that have been introduced outside their native range and exert significant negative impact on native biota, economic values, or human health (Hellmann et al., 2008). Since invasive species are usually successful and abundant, whereas many native species are vulnerable and limited, they are considered one of the fundamental drivers of loss of biodiversity and extinction at a global scale (Ficetola et al., 2007; Hellmann et al., 2008). Most studies investigating invasions intend to provide useful information to help manage invasive species and eliminate or reduce their negative impacts especially in freshwater ecosystems as they are more vulnerable to invasions because there are so many possible routes of introduction, including intended pathways like stocking, and unintentional pathways like aquarium releases (Pound et al., 2011). However, the removal of the whole invasive population can be an impossible mission, particularly for species that have established over large areas which is the reason prevention of introduction and establishment of species with a high risk of becoming invasive is crucial and considered the most cost-effective way of reducing future problems (Myers et al., 2000; Ficetola et al., 2007). Currently, billions of dollars are spent every year to try and eliminate nonnative aquatic species, but efforts to limit the introduction of invasive species or to manage established invasive populations are often delayed due to insufficient understanding or lack of projected behavior of the species (Pound et al., 2011). The invasive potential of introduced invasive species needs to be continually reassessed as global changes occur that can either hinder or help spread the species (Ficetola et al., 2007; Bradley et al., 2009). But based on several reports, climate change is not likely to hinder or diminish the impact of invasive species given that most of them tolerate a large range of environmental conditions (Hellmann et al., 2008).
Suckermouth catfishes are freshwater fish from the family Loricariidae and are naturally found in South America, Panama, and Costa Rica (Pound et al., 2010). They are ventrally flattened, their dorsal and lateral surfaces are covered with rough, hard plates, and have an inferior mouth with lips that form a sucking disc, thus its name. They are extremely popular among aquarists due to their unique appearance and tendency for cleaning algae (Norman, 1948; Pound et al., 2010). For these reasons, suckermouth catfishes have been commonly imported globally and are now established in a number of freshwater ecosystems in North America (Innes, 1948; Pound et al., 2010). Hypostomus plecostomus is the most geographically widespread as it is the most frequently imported species. These fishes create burrows (120-150 cm deep) utilized as nesting tunnels that are guarded by the males until larvae leave the burrow enabling introduced populations to become abundant in a short period of time (Burgess, 1989). The feeding and reproductive behaviors of H. plecostomus, combined with its large size and high population densities, establish substantial threats to native fish communities and to aquatic habitats. For these reasons, plans to halt their expansion and/or new introductions should be a priority.
Foreseeing and mapping the potential result of introductions based on ecological requirements of invasive species and variables that impact the probability of establishment is an important tactic to prevent invasions (Ficetola et al., 2007). This has expanded the use of species distribution models as a tool to evaluate the risk of invasion and to implement necessary proactive management strategies (Kumar & Stohlgren, 2009). These models establish relationships based on the geographical relationship between occurrences and climate conditions in the study area (Bradley et al., 2009). Even though other variables play a role such as biotic factors, like interactions with native species, soils, and land use, climate matching is, in most cases, the most important factor and is expected to lead to large-scale range shifts in species distribution (Ficetola et al., 2007; Bradley et al., 2009). Given this, data on the native range of a species can be used to model their climatic niche, and be projected at global scale to locate the areas where the likelihood of establishment will be higher (Ficetola et al., 2007). The aim of this study is to: (1) predict suitable habitat distribution for the invasive armored suckermouth catfish, Hypostomus plecostomus, using number of occurrence records, (2) build a model predicting which areas are more susceptible to a successful invasion in the United States and Central America, (3) identify the environmental factors associated with H. plecostomus’s habitat distribution; and (4) evaluate if the model correctly predicts the outcome of introductions.
MATERIALS AND METHODS
Occurrence data
Global occurrence data (latitudes and longitudes of point occurrence records) of H. plecostomus were obtained from the Global Biodiversity Information Facility (GBIF, www.gbif.org) (accessed February 2018) (Figure 1). The occurrence data were imported into Microsoft Excel 2001® and distribution points were checked for duplicates and obvious errors resulting from inaccurate georeferencing.
Environmental variables
Nineteen bioclimatic variables were utilized as potential predictors of the H. plecostomus habitat distribution from the WorldClim database (Hijmans et al., 2005; http://www.worldclim.org/bioclim.htm) (Table 1). These climatic variables represent annual trends (e.g., mean annual temperature, annual precipitation), seasonality (e.g., annual range in temperature and precipitation) and extreme or limiting environmental factors (e.g., temperature of the coldest and warmest month, and precipitation of the wet and dry quarters) that have a resolution of approximately 1×1 km2.
Modeling procedure
Present and projected future potential distributions for H. plecostomus were modeled using maximum entropy modeling (Maxent) (Phillips et al., 2006). Maxent is a machine learning method that estimates the distribution of a species by finding the probability distribution of maximum entropy subject to environmental constraints. Maxent has been found to perform best among many different modeling methods since it requires only species presence data (not absence) and can incorporate interactions between different variables such as the environmental variables for the study area (Elith et al., 2006). The free available Maxent software, version 1.4, which generates an estimate of probability of presence of the species that ranges from 0 to 1, where 0 being the lowest unsuitable habitat and 1 the highest optimal habitat was used. The occurrence records and the 19 bioclimatic variables were used in Maxent to model potential habitat distribution for H. plecostomus (Figure 2). The model was developed on the basis of distribution records within the native range, and run over the study area. Then, the model was projected to evaluate the environmental suitability of each grid cell. For the future climate projections (2050) two General Circulation Models (GCMs) were used: the Centre National de Recherches Météorologiques (CNRM-CM5) and the Hadley Centre Global Environment Model (HadGEM2-ES). To assess the predictive power of the models, the area under the curve (AUC) of the receiver operator characteristic was used. AUC plots in Maxent are created by graphing sensitivity (the number of false positives) against specificity (the number of false negatives). Values typically range from 0.5 (indicating random discrimination) to 1 (the theoretically perfect result).
RESULTS
Maxent observed a positive relationship between probability of occurrence of H. plecostomus and three variables: BIO1, BIO6, and BIO11. The jack-knife procedure suggested that annual mean temperature was the variable having most predictive power, while annual precipitation and precipitation of wettest quarter were the least important (Figure 3). The AUC of model for the calibration area was 0.981. Models of future distribution for the suitable habitat of H. plecostomus show a shift to northern areas in USA and towards the west in Mexico, with projected suitable habitat in Oklahoma, Kansas, Louisiana, Georgia, Yucatan, and Campeche to name a few (Figure 4, 5). However, distributional shifts northward showed marked differences in habitat suitability between the different climate change models and scenarios. For example, CNRM-CM5 and HadGEM2-ES models showed differences in regions of suitable habitat in the northeastern part of Texas. The HadGEM2-ES model shows a much more wider spread of suitable habitat than the CNRM-CM5 model. Overall, both thresholds predicted an expansion of the suitable habitat by 2050.
DISCUSSION
This study showed that the habitat distribution patterns for invasive armored suckermouth catfish, H. plecostomus, can be modeled using a small number of occurrence records and environmental variables using Maxent. This study provides the first predicted potential habitat distribution map for an invasive freshwater species in the United States and Central America. Since Maxent is mapping the fundamental niche (different from occupied niche) of the species using bioclimatic variables, the suitable habitat for H. plecostomus may be over predicted in some areas (Murienne et al., 2009). However, the information produced during this study is timely and highly relevant given the potential threats H. plecostomus’s may have on potential habitats and to the overall biodiversity in these freshwater ecosystems. Potential and documented impacts of H. plecostomus include reduction of food and physical cover available for the aquatic insects and other benthic fishes, decrease native populations by incidentally ingesting their eggs, displace smaller or less aggressive benthic fishes that could lead to a collapse in freshwater fisheries and a decrease in biodiversity (Hubbs et al., 1978; Pound et al., 2010). They may also pose mortal danger to birds attempting to feed on them due to their defensive erection of dorsal and pectoral spines (Bunkley-Williams et al., 1994). H. plecostomus occasionally bury their heads in the substrate and lash their tails that can uproot or shear aquatic plants (Walker, 1968). This would affect local plant species by diminishing their abundance and benefit non-native plant species. The burrows of H. plecostomus can potentially compromise shoreline stability, increase erosion and suspended sediment loads (Nikolsky, 1963). The potential habitat distribution map for H. plecostomus can help in planning land use management in order to implement proactive management strategies necessary to sustain our freshwater fisheries.