Should Crocodiles Do Rows?
An important tenet in the training world is that, to avoid injury, approximately equal work should be done with "antagonistic" movements, meaning movements that move a given joint in opposing directions (often flexing and extending).

For example, bicep curls flex the elbow joint whereas triceps extensions extend it, and these two joint isolating movements are therefore "antagonistic" to each other. To qualify as "antagonistic", movements that use more than one joint (compound movements) must use the opposite functions of all the joints used. For example, pull-ups and dips use opposite functions of the elbow joint (flexion versus extension), but they both involve adduction rather than abduction at the shoulder joint (moving the arms towards the midriff rather than overhead). The overhead press (extension of the elbow and abduction/flexion of the shoulder) is "antagonistic" to the pull-up. (The press will involve greater or lesser amounts of abduction and flexion at the shoulder joint depending on the exact placement of the item being pressed, and the pull/chin-up will involve greater or lesser amounts of adduction and extension at the shoulder joint depending on the grip. At the shoulder joint, "flexion" basically means moving the arm forwards and up whereas "extension" means moving it down and to the rear.) The "antagonistic" movement to the dip would be a kind of side row that, in my experience, doesn't work very effectively.

I have a deeply embedded devotion to the notion that exercises should mimic movements that you might have to perform in "real-life" situations, however remote the chances are that your life will really throw such situations at you. For example, if you ever become stranded in the Sumatran jungle, prior practice at pull-ups and brachiation will enhance your ability to escape tigers by climbing trees and swinging from branch to branch like an orangutan.

I struggle to imagine scenarios in which a human being (or most other creatures) would need to flex the knee joint against significant resistance, as is done with leg-curl machines (or "glute-ham raises"). Our knees do flex, and wouldn't do so unless they needed to do so, but never (so far as I can see) against significant resistance. This doesn't mean that the muscles involved in knee flexion - the hamstrings and gastrocnemii (calves) - will be neglected. The former are amply recruited during movements that extend the hip joint (deadlifts and squats, for example); the latter provide stability for any activity performed on the feet. "Natural" knee-flexing movements against significant resistance require an opposable big toe, which humans have discarded. Orangutans still have them, as do other non-human great apes, but the physiology of the orangutan thigh is specially designed to cater for greater knee flexion (pdf).

And what about creatures that walk on four legs (quadrupeds), as did the sauropod dinosaurs (diplodocus, brachiosaurus etc.)? The sauropods spent the greater part of their lives extending all four of their limbs against the resistance of their colossal bulk. Did they go extinct because they failed to conceive of preacher-curl machines?

But what about crocodiles, who (along with birds) survived the extinction of their archosaurian cousins? When on land, they spend all of their time standing/walking/running in a push-up position. Surely, to avoid injury, they should spend less time lasciviously stalking prey and more time performing horizontal pull-ups or lying/prone rows or (better yet) walking tail-first while hanging upside down from monkey bars?

What fun can be had by creating senseless nonsense from strict logic!