Bioversity International’s Gene Flow Risk Assessment of Genetically Engineered Crops project, funded by GTZ and realized in collaboration with CIAT and Universidad del Valle (Cali, Colombia), has got (some of) its products out. The project focused on the “likelihood of gene flow and introgression to crop wild relatives (CWR) and other domesticated species.” A book is coming, but you can see the risk maps for a number of crops online now. And there’s also a bibliography.
LATER: Jeremy points out, correctly, that “see” in the last sentence above is a bit of an overstatement. You need to do a bit more work than is perhaps implied.
I just wanted to add two things:
The first may be a bit hair-splitting, but just to point out that we try to avoid using the word “risk”. So to be correct: what we have mapped is not the “risk” but rather the probability or likelihood of gene flow and introgression. “Risk” always implies something negative – but our aim with this project was to provide scientifically-based, objective information so others can make their own judgements.
Number two: I admit that the mapping pages could be a bit more user-friendly (and we are working on it) – but apart from the gene flow maps, they are really worthwhile looking around for some additional reason: We have not only uploaded the maps here, but also the raw data that we have used for mapping the distributions of the crop wild relatives (as grid, Google Earth kml, and excel files). So before searching GBIF, SINGER and other online databases: you might save yourself some (a lot!) time if you have a look here first, because we have already done this for most of the wild relatives of 20 different crops!
Meike:
I do not understand the map for cotton: the probability of gene flow or introgression. In Brazil, what species is likely to suffer from introgression with cultivated hirsutum. The * species indicates it cannot be G. mustelinum, which is native to northeast Brazil. Why cannot someone in Columbia chart the known distribution of a species in Brazil? G. tomentosum is native to Hawaii. Does wild barbadense grow over all that red area in Brazil? Or is the red area the distribution of cultivated hirsutum and cultivated barbadense, which as we know from the literature of the last century have been introgressing for many years? Both are AADD tetraploids, and as far as we know may have been formed from the same parental species. This introgression is exploited by plant breeders.
@J. Giles Waines – Hey Giles,
Yes, the red zones in Brazil are those areas where cultivated G. hirsutum and G. barbadense – cultivated, naturalized or wild – can potentially co-occur (and as you say: in fact they often do). These are modeled distributions that are based on known germplasm records and an algorithm using 22 bio-climatic variables (rainfall, temperature, no. of consecutive dry months, etc.) at these sites. In this concrete case, the modeling is based on >300 records of G. barbadense from gene bank collections, other public online databases and herbaria. For G. mustelinum (endemic to NE Brazil) we could only find 5 reports, which is not enough for modeling its potential distribution range. If you are interested in more details on methodology, I would be glad to share a pdf file of the cotton chapter – or you wait until September and buy the book (I can get you a discount!) :)
Is maize not “sexually compatible” with Tripsacum?
Are you updating your results? if so, how often? and, is there an automatized algorithm to re-map all the stuff when more herbarium/genebank samples are added to your database? It would be interesting to see how results are improved while we get out of genebank database hell.
Moreover, did you only map wild species with more than 10 samples (or whichever data-availability threshold you chose) and left the other ones outside? Think you could use real presences of all not-modeled species (and a buffer area of, say, 100km) to detect other zones where gene flow could occur.
Finally, is gene flow more likely to occur in zones where more than 1 sexually compatible wild species is present? or is it the same likelihood if there are 100 or 1 species along with the crop in the same site? and what about of the combination different levels of “likelihood” (species from GP1, GP2) in the same site?
Hi Julian,
1) The idea is that these maps will be updated “interactively” through Google Earth: Once all kml files are online, we will invite the public to revise the modeled distributions and post comments on areas/regions where our predictions are erroneous. So in the ideal case these maps will improve over time in an interactive way, with the help of everybody interested or knowledgeable about crop production areas or distribution ranges of crop wild relatives. Yes, there is an algorithm to re-map the stuff, but it is not automated – it’s still quite a cumbersome process that requires a while to update and re-do the maps. Would be great if somebody could come up with an idea of how to automate this to make updating easier!
2) Yes, we only mapped species for which we had at least 10 reliable geographic records. The records for those species with <10 reports are available on the website, and yes you are right: they could be included relatively easily as “real presences” plus a buffer zone. Good idea, will see what we can do about it.
3) The maps are conservative. In cases where we have a combination of different levels of “likelihood†(e.g. species from GP1, GP2) in the same site, we have always used the highest category on top, i.e. if there are species present with a low and others with a high likelihood of introgression, then we only showed the highest category. The likelihood remains the same, no matter if there are 100 or 10 species of the same category co-occurring with a crop at the same site. Our categories of likelihood (low, moderate, high) are based on a set of biological and reproductive characteristics that have to coincide between the crop and a wild relative, and which are discussed in detail in the individual crop chapters in the book. The “likelihood classification” is the conclusion of this analysis.
Still it is important to keep in mind that the maps show a conservative “worst-case” scenario. Because – in addition to the mainly sexual/reproductive characteristics of the plant species involved – we assume that a number of other conditions are in place that favour gene flow and introgression:
i) The distance between crop and wild relatives has to be such that pollen exchange via wind or insects is a real possibility,
ii) Flowering times of the species in question have to co-incide, and these vary regionally as well as locally and are influenced not only by genotype but also by environmental conditions,
iii) Other biotic and abiotic conditions need to be in place to allow for pollen exchange (e.g. presence and abundance of pollinator species, wind conditions (speed, direction), air humidity, pollen viability, etc.),
iv) The “nature†of the GM trait needs to favour introgression in the long term, i.e. the trait has to be neutral (at minimum) or imply a selective advantage to the recipient plant (e.g. improved fitness), otherwise natural selection would work against permanent introgression over generations,
v) Local management practices have to favor survival of the resulting hybrids (e.g., if hybrids grow up in a crop field, they may often be harvested with the crop and thus not have a chance to reproduce and establish themselves in the wild).
So to answer the question: “Does gene flow or introgression in reality occur at a specific place within the areas highlighted in the maps?†all of these above listed conditions (and probably others which I may just have forgotten here) need to be in place. And that cannot be assessed with a global mapping exercise – but has to be evaluated locally, on a case-by-case basis.
Hi Jacob,
Hybridization of maize with Tripsacum species is not known to occur in the wild, except for one case: T. andersonii arose from a unique spontaneous hybridization event between two Tripsacum species and Zea luxurians, but this hybrid is sterile. Crosses between maize and some other Tripsacum species have been achieved in the laboratory, but they require artificial techniques and none of them would occur spantenously in the wild.
@Meike – T. dactyloides and Z. diploperennis can be crossed without special techniques. Z. diploperennis is compatible with maize. I don’t know if the hybrids can be backcrossed with their progenitors.
But I guess you are right. Introgression is a chance event, however, so in the future you may be adding more species to the maps.
Hi Jacob,
In our study, we focussed on hybridization events that involve the crop (i.e. domesticated Z. mays) and wild relatives, since this is where genetic transformations will have implications for (agro-)biodiversity. Hybridizations between two crop wild relatives (e.g., Z. diploperennis and a Tripsacum species) may be relevant if one of the two species could act as a “bridge-species” to the crop. This might in fact apply to Z. diploperennis, and there have been claims that crosses between T. dactyloides and Z. diploperennis (so-called “tripsacorn”) have been used to transfer genes from Tripsacum into domesticated maize. However, this study has been challenged due to the lack of molecular evidence, and to my knowledge so far other scientists have not been able to replicate these results. Of course this does not exclude the possibility of such an event happening naturally in the field, but the likelihood appears to be extremely low.
Maybe I will try to replicate the experiment.
@Meike – I googled and read a bit more.
Randolph found that hybridization under natural conditions is rare and produces infertile plants (but vegetative reproduction could play a role). Experimental studies, however, have produced more hybrid plants and some were (partially) fertile. It depends on the chromosome combination of the resulting hybrid.
Besides this, I found a reference to a wild Zea mays x Tripsacum dactyloides hybrid found in Venezuela (De Wet et al. 1981).
Even though the likelihood of this is clearly less than for Zea species, it seems that it cannot be totally ruled out.
Randolph, LF. 1955. “Cytogenetic aspects of the origin and evolutionary history of corn.” Sprague (ed.) Corn and Corn Improvement, 16-61.
De Wet, JMJ, DH Timothy, KW Hilu, GB Fletscher. 1981. Systematics of South American Tripsacum (Gramineae). American Journal of Botany 68(2), 262-276.
@Meike – Yes, a manual mapping process would be quite long, but some scripting could make your life a lot easier. For info, we have here at CIAT two Java tools to create KMLs from ASCII grids and shapefiles, and they can be used in command line mode.
Furthermore, how are these results being changed by, for example, land use and climate change? crop wild relative niches are likely to be impacted by both of those two processes, and in some cases, highly endangered, so, is geneflow being favored or reduced? where? and how much?
@Julian – Hey Julian, definitely both climate change and land use changes will affect crop production areas and the distribution ranges of crop wild relatives. Most likely they will also influence gene flow and introgression rates, e.g. by affecting the niches of pollinators, flowering times etc. The maps we produced are “snapshots” of the current situation, but different climate change scenarios could of course be modeled – If I’m not mistaken, CIAT is doing that already for several crop wild relatives.
@Meike
Hi Meike,
It is nice to see the maps from your project up on the web. It wasn’t too hard to get to the information.
With a background in documenting barley genetic resources, it was also pleasing to be able to download a spread sheet with the passport data that you used for Hordeum vulgare ssp. spontaneum [wild barley].
There are more data points available than can be provided by ICARDA, since their database excludes certain countries for political reasons.
The European barley database has attempted to combine the available geographic data sets for barley germplasm, including the wild relatives, so the site would be well worth a visit to fill in some of the blanks on your present map. [use the link to the old version http://barley.ipk-gatersleben.de/ebdb.php3%5D.
A unlimited search for collector Dinoor provides to the patient inquirer ;) the John Innes collection of H. vulgare ssp. spontaneum that are missing on your map [ca 200 collecting sites].
Kind wishes
Dirk
@Dirk Enneking – Thanks Dirk, will take this into account! Best, Meike