About the first question, the most important goal in Marathon is to reduce the difference between the 2 thresholds (AT of about 2 mmol, and AnT of about 4 mmol), that, when you are really prepared in your SPECIFIC POWER MARATHON ENDURANCE are very close (under 5% in the best runners). What does it mean 5% ?
You have a marathon runner, 4 months before his race. You test your runner (that has a PB of 63:20 in HM = 3:00 / km)
and find that, at 3:00 pace (after 20/25 min), he has a level of lactate of about 4 mmol. You can really suppose that this athlete, well prepared, can run a full marathon in 2:13 (that is 3:09 pace). If you think that the relation between 63:20 and 2:13:00 is not correct, is only because you type of training marathon is not correct. Of sure, every athlete talented for long run, able running a HM at 3:00 / km pace, must be able running in 2:13:00 with a correct training (sometime, TRAINING HARD IS NOT TRAINING WELL...).
Anyway, if 3:00 pace is 100% (18.0 every 100m), 3:09 is 95% of the same speed.
But, if you test your runner 4 months before the race, you find that running at 3:09 can provoke a lactate level of about 3 / 3.3, so higher than what is possible to accept for running full marathon. You must remember that, when you produce more lactate, is because your consumption of glycogen is growing, and this fact means that your tank of glycogen cannot last more than 1:30:00.
If you test the level of 2 mmol (for example, using the FARAGGIANA - GIGLIOTTI test, consisting in 6 x 2000m even pace, with very short recovery (may be 40 sec, only the time for taking blood from the ear), increasing speed every time (for example, in the above case : 6:50 / 6:40 / 6:30 / 6:20 / 6:10 / 6:00, more a last 1200m free faster possible), you can find (it's an example) values of 1.8 mmol - 2.0 - 2.4 - 2.8 - 3.4 - 4.0. So, you can suppose that, at that time, the speed for producing 2 mmol (good for running full marathon) can be about 3:20 (a difference of 11% from HM speed), and this athlete, also with long run, cannot run a marathon under 2:20.
One of the systems for reducing the difference of speed from 2 to 4 mmol (conventional values) is to work very much using the speed of recovery, of course running fast.
What example we can do for building a correct consumption of fuel ? For instance, using 1000m at alternated speed :
4 MONTHS BEFORE MARATHON :
You start alternating a fast km (at the speed of 4 mmol, may be 3:05 at that moment) to a slower km of recovery (at a speed under 2 mmol, may be 3:40)
During the first period, if you use this type of training 3 times, every time you KEEP CONSTANT RECOVERY SPEED, and try to improve TIME OF TEST, for improving your ANAEROBIC THRESHOLD (Resistance is the BASE for increasing POWER).
At the end of the period, you are able running in 3:00/3:40 (average 3:20)
3 MONTHS BEFORE MARATHON :
You KEEP CONSTANT THE SPEED OF TESTS, and try to improve SPEED OF RECOVERY. You are able to improve your AEROBIC THRESHOLD, keeping at the same level the AnT (Power is the BASE for increasing SPECIFIC ENDURANCE).
At the end of the period, you are able running in 3:00/3:30
(average 3:15)
2 MONTHS BEFORE MARATHON :
You work on speed of tests (till 2:58) during the first 2 sessions, and on speed of recovery (3:26) during the third session. So, at the end, you run in 2:58/3:26 (av. 3:12)
LAST MONTH :
You can lose little power, as when you are really ready for full Marathon, you cannot run your PB in Half. If this happens, there's some mistake in your training. Last period has the focus of INCREASING SPECIFIC MARATHON ENDURANCE, bringing the speed of 2 mmol (marathon) very close to the speed of HM. So, you can run a little slower your tests (3:00), working for improving speed of recovery (at the end, 3:18). In this case, your average is 3:09, and you can identify it in your Marathon Pace, using of course even pace.
Of course, we don't use only this work, but use also more long run, specific marathon endurance workouts (for ex., 4 x 5000m at 102% of your MP, recovering 1 km in 3:30), always having inside a SPECIFIC ELEMENT : Marathon Speed.
The most important thing, in this project, is to control a correct recovery. So, a big modulation becomes the most important way of training. When you go to specificity, YOU MUST BE FRESH IN MUSCLES AND IN BRAIN FOR RUNNING AT Mar. Pace. So, recovery is very important. You not only can, but MUST recover running big volume at low intensity. MORE YOU RUN SLOWLY IN YOUR RECOVERIES DURING SPECIFIC PERIOD, MORE YOU CAN RUN FAST FOR LONG TIME DURING YOUR SPECIFIC WORKOUTS. Don't forget that the Specificity, in Mararathon, is a Specificity of Extension.
Regarding the other question, with specific training for 5000/10000m (and also steeple), athletes can increase their capacity of running in Steady State with a very high level of lactate. For example, in a 10000m runner (in 27:30), if we can share the race (supposing that is at even pace) in parts of 2000m, we could find (after every 2000) lactate of 6 mmol (5:30) - 8 mmol (11:00) - 8.5 mmol (16:30) - 9 mmol (22:00) and a final of 12/14 mmol, due not only to a final increase of speed, but also (supposing yet even pace) to the difficulty in keeping a good Steady State longer than 15/17 min. This type of SS is called MAX LASS (Maximum Lactate Steady State), and for top runners can be about 12/13 mmol lasting 6/7 min at the speed of 5000m (2:30/2:35), and about 8/9 mmol lasting 15/17 min at the speed of 10000 (2:38/2:42).
One of the reason of variations of speed in long tests on track is to improve this attitude.