I've heard that in harsh conditions more human baby boys are born than girls because a bigger ratio of them will die before reproductive age. But in which aspects of the random gene combination while conception can harsh conditions (and which of their characteristics) actually can make a difference?
Issues with the post
Sexual selection is a mode of natural selection where members of one biological sex choose mates of the other sex to mate with (intersexual selection), and compete with members of the same sex for access to members of the opposite sex (intrasexual selection). This question has nothing to do with sexual selection.
What do you mean by
random gene combination? More globally I don't understand the last sentence.
As a consequence, I don't understand the question. IMO, the post should be closed as unclear. However, I can imagine a question that I would consider interesting. I am answering to the question
Is it possible that selection yield to a sex-ratio that deviates from 1:1 at equilibrium? How does it relate to environmental conditions?.
Evolution of sex-ratio under a standard Fisherian scneario is often refered to as Fisher's Principle.
In the vast majority of strictly sexual species, every individual has exactly one mother and one father. It means that at equal sex-ratio the average relatedness of all offspring and all males in the previous generation is equal to the average relatedness of all offsprings and all females int he previous generation. When the sex-ratio is biased toward more females (for example), males have on average a higher reproductive success than females. As a consequence individuals that produce more males will increase their fitness and will increase in fecundity. As a consequence, the adult sex-ratio will always go back to 1:1 (fifty-fifty).
Adult vs baby sex-ratio
Now, it is important to understand that the adult sex-ratio is not necessarily the same as the babies sex-ratio. If typically, there is higher mortality in males than in females (could happen depending on the species and the environment), then selection would favour a bias sex-ratio toward males in babies. The adult sex-ratio will in any case be 1:1
Kin/group selection (non-fisherian scenario)
See for example the case of social spider Aviles (1993).
What is good for the population?
Thinking about the same problem but thinking about what is good for the population (rather than for the individual). For the population, the limiting factor in the number of babies produced are often the females. As a consequence, it would be good for the population if the sex-ratio were biased toward females.
When can alleles that are beneficial for the population overweight what is beneficial for the individuals
Under a standard fisherian scenario this cannot happen though selection will increase the frequency of alleles that spread the best and not the ones that are good for the population. However, typically when the relatedness within the population is high enough (you should read about group selection and kin selection) such trait that are good for the population may evolve.
I recommend that you take some time an introductory class to evolutionary biology such as Understanding Evolution for example. It is fast and you will learn more and faster than what you can learn here with your question. I think I already gave you this suggestion a number of times.