That kind of interplant communication is subject to a prisoners'-dilemma-like payoff matrix though. Yes, a plant species evolving a co-operative strategy like that will do well, in aggregate, but individual plants which 'defect' and advertise to other plants that the nutritious soil they have found is poisonous and that other plants should keep away will outcompete its peers - so defector genes will prosper in the population - except in the face of plants which ignore their peers' chemical signals.
Overall, seems likely to be selected against as a strategy for individual organisms.
I think that the effectiveness of the strategy will depend on the specific plant's pollination/seed dispersal mechanism. And it gets really weird when you consider that you can have plants that over multiple seedling generations (years really) that have a mix of clonal colonies and sexually reproduced offspring dispersed around them.
To the original question, one obvious difficulty is considering how a mechanism encoding directionality in a chemical signal can evolve. It's certainly possible that these fungal networks are weakly directional, and it's also possible for a plant to evolve some sort of gradient sensing mechanism if its root network is sufficiently wide, but it -seems- unlikely. I would totally love to see if we can find an example of it though - it would be sooo cool.
What about a population of 100 such plants where the defector gene is very unlikely to surface for whatever reason, and the plants are able to cooperate and that population becomes stronger for it, versus a population of 100 similar plants except the defector gene might be rather likely to show up, making the population weaker.
Couldn't the population without the defector gene, eventually outcompete and extinguish the population where it can appear?
Overall, seems likely to be selected against as a strategy for individual organisms.