As the Ljosalfar cannot use the World Spell March of the Trees until they have spent nearly 2000 beakers in research [ Ancient Chants (210 beakers), Mysticism (336), Hunting (604), Ways of the Forest (840) ] I would say the Elves' World Spell better be powerful or why bother?
Then, once you get your World Spell capability, you need Philosophy (672) and Priesthood (1344), another 2000 beakers total, to be able to recruit Priest of Leaves (120 Hammers) which are built at a Temple of Leaves (120 Hammers) in order to get your blooming forests going so as to be sure to have enough Treebeards worth summoning.
Again, the World Spell better be powerful, I only wish it were true in reality as some might think it is in theory.
The theory might suggest that the March of the Trees is so powerful that the Elves should have been limited to only a percentage of their forests spawning Treants. Perhaps if an Elven nation had a large empire completely or even mostly forested, it might have (for five turns) a strong, slow-moving army that could stop many invaders. (If the Elven nation were so large and so forested, it'd also be much less likely to need this army anyway but... you know... they might want to have some fun.)
However, the reality is you rarely have most (or even much) of your lands forested when you need the spell, the Treants don't live long, they're often too far away from the area needed (arriving just as their short life-span dissipates), and you lose all your ancient forests' (they become new forests) economic benefits in casting the spell.
It's a spell best used in desperation by an empire falling to a super invader or by an empire positioned by size or geography to win by knockout without regard to the economic effects.
I don't understand why someone would play the Ljosalfar and then just basically "get rid" of the World Spell by using its potentially powerful and civ-saving effects to clear out lizardmen and barbarians? Even if one did have access to it with an Advanced Start, it'd be like taking a shotgun to get rid of moths or mosquitoes.
If your civilization is so small that lizardmen and barbarians are such a threat that they're possible game enders (an idea I don't buy 'cause even at the Immortal/Deity level I play on, raging barbarians are an annoyance but unlikely "game enders" to a player with reasonable caution), then your access to Treants from the World Spell would be small also since you'd be unlikely to have much land, and even less forests. Sure you'd kill any barbarian that you could catch for five turns but then you'd have used up your "ace in the hole."
The idea that the temporary Treant army of five turns would be an unacceptable maintenance problem slowing research or preventing regular units from earning combat promotions (because the Treants were making the kills) is problematic: a desperate civ wouldn't care about the cost, and a powerful one could afford a five turn Treant mob. And opportunities for combat promotions? They aren't hard to find.)
My practical experience with the March of the Trees? On the Immortal/Deity levels, I've used it mainly for desperate attempts to stop powerful invaders or in the infrequent case when I've had several rival cities very close to my civilization's forested borders (and vulnerable to slow-moving Treebeards.) Often I don't need the spell at all, and use it only because the win's inevitable anyway, so it'd be unlikely I needed the insurance.
The March of the Treants is an impressive but temporary spell, usually costly to acquire and very dependent on geography and forested areas. Casting the spell makes your valuable ancient forests into new bloom forests and your Treant recruits are often not where you need them. And when you're faced with AI civilizations that can field huge stacks of champions or beastmasters or pyre zombies, etc. while your Treants are often scattered far and wide, it's not an automatic win or guaranteed salvation to cast "March of the Trees".
I like having it as another tool in the arsenal but it's not usually the knockout punch we'd like it to be.