The Small Indian Mongoose (Herpestes auropunctatus)
Historical spread and using correlative models to predict environmental suitability in the invasive range.
Introduction: Exotic species on a global scale
Humans have historically been the main vector for bringing species to new areas, both accidentally and intentionally. Some of these introductions have had long lasting negative impacts both for us and nature.
Number of Invasive species globally
It is well known that preventing a species from becoming established is the best and most cost effective control method (Mack et al. 2000). This usually entails bio-security and import regulations at borders, ports, or other transit hubs. Some number of species will still slip through, so early detection surveys and monitoring are also essential.
When a non-indigenous species is detected however, governments may be reluctant to provide sufficient resources for eradication without proof the species is truly a problem. Ironically, once a species is proven to be a successful invader it is often extremely expensive and difficult to manage, much less eradicate (Simberloff 1997).
Spatial Analysis to Predict Potential Invasions
To solve this issue, scientists have focused research on two main types of prediction: which species are likely to become invasive and which areas are susceptible to invasion. This information can help regulatory agencies to make informed choices when restricting import of species or allocating funds for removal of non-indigenous species (Reichard & Hamilton 1997).
Species Distribution Models (SDMs) can be useful for predicting what areas may be suitable for potential invades, providing information on not only what species to be cautious of but also which locations should be monitored. One such popular SDM is Maxent, which uses a correlative approach by matching species occurrence data with environmental layers and extrapolating based on the relationships between them.
Case Study: Herpestes auropunctatus
For my research, I will be using Maxent and evaluating if the model accurately predicts where invasive species occur outside the native range.
One critique of using these models is the underlying assumption that a species niche(s), both realized and fundamental, are conserved through space and time (Tingley et al 2014). However, growing evidence points to some invaders displaying rapid shifts in niche space that cannot be predicted by corrlative means alone. I will examine if such potential niche shifts have occurred for the small Indian mongoose, an intentionally introduced species meant to combat an unintentional introduction.
(Wikimedia Commons)
Of Rats and Mongooses: A Brief History
During the Age of Exploration (15th-17th century) as European nations expanded and formed colonies, many species introductions occurred.
Most of the new colonies were established to produce sugarcane or other cash crops. However rats (Rattus rattus), brought along on ships, caused extensive damage to these plantations.
Within the native range, mongoose were used to control agricultural pests and unwanted species such as rats, snakes, and insects. Snakes were especially reviled in this time period and mongooses were famous for being able to prey on Cobra species. Hoping to control these undesired populations, plantation owners imported this species.
Origin
1872: Jamaica
1883: Fiji
1883: Hawai'i
1902: Mauritius
1910: Croatia
1910: Ryukyu Islands
Islands Across the Globe
Origin
Kolkata (formerly Culcutta), India. Omnivorous small mammal.
1872: Jamaica
9 individuals imported to control nuisance species. Likely responsible for the extinction of the Jamaica petrel (Pterodroma caribbaea) (Collar et al 1992) . Bred and then spread throughout the Antilles archipelago and linked to declines in abundance of many native species.
1883: Fiji
Introduced to some islands of Fiji: Likely responsible for extinction of the barred-wing rail (Nesoclopeus poecilopterus). Some species of ground nesting birds now found only on mongoose free islands (Morley and Winder 2013).
1883: Hawai'i
72 individuals Introduced to the Big Island of Hawaii. Intentionally spread to Molokai, Oahu, and Maui as well. In 1952 only 30 of the endemic Nene geese, the state bird of Hawaii, were left. Though not the sole predator driving declines of this and other native ground nesting birds, mongooses contributed to declines (Hays & Conant 2007).
1902: Mauritius
Introduced to Mauritius, Mafia, and Grand Comoro. On Mauritius, mongooses threaten the endangered Pink Pigeon (Streptopelia mayeri) (Roy et al 2002).
1910: Croatia
Introduced to several small Islands in the Adriatic sea. Species such as the horned viper (Vipera ammodytes) and Balkan green lizard (Lacerta trilineata) are very rare or absent from islands with mongoose, but common on islands without mongoose (Barun et al 2010).
1910: Ryukyu Islands
4 individuals released on the southernmost point of Okinawa to help control Habu snake. They reached northernmost forests, which contain many endangered species, in 1990. 30 mongooses released on nearby Amami-Ohshima Island in 1979, where their continued expansion greatly reduced populations of the endangered Amami rabbit (Yamada and Sugimura 2004).
Islands Across the Globe
Although this list does not include every location that mongooses were introduced too, it demonstrates both how widely they were spread and how they can negatively impact native species. They have the distinction of being included on the list of the worlds 100 worst invasive species (Lowe et al. 2000). In addition to the threat they pose to wildlife, mongooses are also a human health risk as they can carry leptospirosis and rabies (Barun et al. 2011).
Methods: Data and Analysis of Mongoose Distribution
All occurrence data were sourced from the Global Biodiversity Information Facility (GBIF), and consist of museum records, I-Naturalist observations, and various publicly shared agency or research data (GBIF.org). Climate layers are from the WorldClim data set, and consisted of 7 variables: annual mean temperature, maximum temperature, minimum temperature, annual temperature range, annual precipitation, precipitation of wettest month, and precipitation of driest month (Hijmans et al. 2005, Fick and Hijmans 2017). Hawaii land cover and road maps are primarily from the Hawaii Statewide GIS Program from the Office of Planning, with some base maps sourced directly from ESRI/ArcGIS Online.
The native distribution extends from Iraq to Myanmar, covering Iran, Afghanistan, Pakistan, Northern India, Nepal and Bangladesh (Gilchrist et al. 2009). Some sources also include Thailand, Malaysia, Indonesia, Laos, Cambodia, Vietnam and parts of China.
My focal invasion is the Hawaiian Islands. Mongooses were introduced to and became established on Oahu, Molokai, Maui and Hawai'i. There have been some sightings and two mongooses trapped on Kauai, but they are not believed to be established
Example of one of the seven layers used for Maxent, shown here at global scale. Polygon vector layers were used to divide environmental rasters into native and invasive ranges. This layer is annual precipitation (10m scale ~340 km^2).
Maxent results showing suitability of native range and then predicting for the Hawaiian islands. This analysis was performed using WorldClim 30s layers (~1km^2). AUC, the probability that a randomly chosen occurrence point is ranked higher than a randomly chosen background point, was 0.738, and the maximum test sensitivity plus specificity threshold was 0.37.
With the suitability map from the Maxent prediction, I used GBIF occurrence data from Hawaii (n=463) and extracted suitability. Mean suitability of mongoose sighting points was 0.50, which is greater than the threshold of 0.37.
To test the Maxent prediction on a global scale I used the 30s (~1km^2) environmental layers and projected it onto WorldClim 10m (~340 km^2) layers. Precipitation during the wettest month was the most important variable contributing to the Maxent model.
Using the same process as for Hawaii, I looked at predicted suitability at mongoose sighting locations across the world. The data publicly available were concentrated in island areas, particularly the Caribbean and the previously mentioned mongoose introduction sites. The mean suitability was 0.46. The AUC for this run was 0.777, and the maximum test sensitivity plus specificity threshold was 0.44.
To get another viewpoint on Maxent output, I ran the model with the 30s WorldClim layers, but this time using the occurrence data from Hawaii. This output predicted much lower suitability for interior areas of the islands but identified very similar areas as highly suitable. The model also differed in that it identified minimum temperature as the most important variable. AUC was 0.876 and the maximum test sensitivity plus specificity threshold was 0.283.
Interpretation and Discussion
In general my results suggest that Maxent is an ok estimator of suitable habitat for mongooses. My AUC values were not ideal, but areas identified as favorable for mongooses did correspond with mongoose sighting data.
Below is a map from Louppe et al. (2020) that used nine different SDMs, including Maxent, to analyze mongoose favorability both now and in 2050 using the WorldClim future climate predictions. While this study accounted for climate change and was much more sophisticated, my limited analysis is largely in alignment with their predictions.
Louppe et al. 2020
A discrepancy that needs to be addressed is the Maxent output for Hawaii when using the native range occurrences vs occurrences in Hawaii itself. If taken at face value, this would suggest mongooses have a smaller environmental niche in the invaded range. Looking at the sources of each observation, however, indicates a bias in the sample.
A large number of occurrences for Hawaii come from I-naturalist observations, while the native range data sources are more varied and often from research projects. The map below shows locations of mongoose sightings relative to roads and general vegetation type. A histogram of the distance from each sighting to a road shows the relationship clearly.
Bias of Mongoose observations in Hawaii to human population areas
Conclusion
While Maxent is a powerful tool, using it correctly is critical to producing meaningful results. Bias in occurrence data is extremely impactful to correlative models as it is the only input related to the species.
Garbage in -> Garbage out
Most modern studies make use of several different SDMs as each functions a little differently. It is also becoming more common to include mechanistic SDMs which are generated based on physiological and trait data of a species. These mechanistic models may allow for the best predictions of the potential distribution of invasive species as they are based on the fundamental niche instead of a species realized niche (Tingley et al. 2014).
While limited in scope this analysis suggests that mongooses have the potential to successfully invade many more locations around the world. In areas where the species has managed to reach continents from islands it was introduced to, such as in the Adriatic, it has established dense population centers and expanded its range rapidly (Cirovic et al. 2011). This highly adaptable species has certainly earned its place as one of the world's worst invasives, but is a valuable case study for researchers trying to understand and prevent future invasions.
Images that didn't make the cut but were produced for this project:
(MANgeese credit RS)
Sources:
GIS Layers: https://planning.hawaii.gov/gis/download-gis-data/, https://www.worldclim.org/data (paper citation below)
Occurrence data: GBIF.org (02 May 2021) GBIF Occurrence Download https://doi.org/10.15468/dl.nazppb
Images: Wikimedia Commons: Chung Bill Bill, General Douglas Hamilton, and 環境省.
References:
Barun, A., Simberloff, D. & Budinski, I. (2010). Impact of the small Indian mongoose on native amphibians and reptiles of the Adriatic islands, Croatia. Animal Conservation, 13: 549–555.
Barun, A., Hanson, C.C., Campbell, K.J., and Simberloff, D. (2011). A review of small Indian mongoose management and eradications on islands. Island invasives: eradication and management. IUCN, Gland, Switzerland, 17–25.
Collar, N.J., Gonzaga, L.P., Krabbe, N., Madrono Nieto, A., Naranjo, L.G., Parker III, T. A. and Wege, D.C. (1992). Threat- ened birds of the Americas. Smithsonian Institution Press, Washington, D.C.
Gilchrist, J.S., Jennings, A.P., Veron, G., Cavallini, P. (2009) The Herpestidae. In: Handbook of mammals of the world. Volume 1: Carnivora. Wilson, D.E., and Mittermeier, R.A. (eds). Lynx Edicions, Barcelona, pp 262–328.
Hays, W.S.T., and Conant, S. (2007). Biology and Impacts of Pacific Island Invasive Species . 1 . A Worldwide Review of Effects of the Small Indian Mongoose , Herpestes javanicus ( Carnivora : Herpestidae ). Pacific Science, 61(1): 3-16.
Fick, S.E., and Hijmans, R.J., (2017). WorldClim 2: new 1km spatial resolution climate surfaces for global land areas. International Journal of Climatology 37 (12): 4302-4315.
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A. (2005). Very high resolution interpo- lated climate surfaces for global land areas. International Journal of Climatology, 25:1965– 1978.
Louppe, V., Leroy, B., Herrel, A. & Veron, G. (2020). The globally invasive small Indian mongoose Urva auropunctata is likely to spread with climate change. Sci. Rep. 10, 1–11.
Lowe, S., Browne, M., Boudjelas, S., De Poorter, M. (2000). 100 of the World’s worst invasive alien species: A selection from the global invasive species database. Published by The Invasive Species Specijalist Group (ISSG) a specialist group of the Species Survival Commision (SSC) of the World Conservation Union (IUCN).
Mack, R.N., Simberloff, D., Lonsdale, W.M., Evans, H., Clout, M., and Bazzaz, F.A. (2000). Biotic Invasions: Causes, Epidemiology, Global Consequences, and Control. Ecological Applications, 10(3): 689–710.
Morley, C.G. & Winder, L. (2013). The Effect of the Small Indian Mongoose ( Urva auropunctatus ), Island Quality and Habitat on the Distribution of Native and Endemic Birds on Small Islands within Fiji. PLoS ONE 8: 1–11.
Cirovic, D., Rakovic, M., Paunovic, M. & Milenkovic, M. (2011). Small Indian Mongoose Herpestes auropunctatus ( Herpestidae , Carnivora ): an invasive species in Montenegro. Biol Invasions 13: 393–399. doi:10.1007/s10530-010-9831-7.
Reichard, S., and W. Hamilton. (1997). Predicting invasions of woody plants introduced into North America. Conservation Biology 11:193–203.Reichard, S., and W. Hamilton.
Roy, S.S., Jones, C.G., and Harris, S. (2002). An ecological basis for control of the mongoose Herpestesjavanicus in Mauritius: is eradication possible? Article in: Turning the tide: the eradication of invasive species. IUCN SSC Invasive Species Specialist Group. IUCN, Gland, Switzerland and Cambridge, UK. Veitch, C.R., and Clout, M.N. (eds.) Pages 266-273.
Simberloff, D. (1997). Eradication. Article in: Strangers in paradise. Simberloff, D., Schmitz, D.C., and Brown, T.C. (eds.) Island, Washington, D.C., USA. Pages 221–228.
Tingley, R., Vallinoto, M., Sequeira, F. & Kearney, M. R. (2014). Realized niche shift during a global biological invasion. Proc. Natl. Acad. Sci. U.S.A. 111, 10233–10238 .
Yamada, F. & Sugimura, K. (2004). Negative Impact of an Invasive Small Indian Mongoose Herpestes javanicus on Native Wildlife Species and Evaluation of a Control Project in Amami-Ohshima and Okinawa Islands, Japan. Global Environmental Research, 8: 117–124.
