Yes. Under application and over application both lead to economic loss.

Fertilizer recommendation should be based on soil fertility rating and targeted yield considering constraints in achieving yield like availability of irrigation, crop variety, and climatic constraints which limit yield.

It depends upon the nature of nutrient.  For nutrients like P, K and Zn, we can use soil tests, critical limits and soil fertility ratings but for nutrients like N none of these will work. In fact, the best way to manage nutrients is to follow the principles of site-specific nutrient management.  Perhaps, we need to learn to ask the plant itself for its nutritional needs.  Soil testing for nutrients like P and K can help in taking better decisions.

Fertilizer rates needed for a crop generally depends on indigenous nutrient supply, and vary with nature of crop, its yield potential, previous crop grown and agronomic management practices.Therefore, the addition of fertilizers and manures should provide the nutrients in appropriate quantities and ratios managing the plant requirement for desired yield goals. So, what is the need of different approaches of fertilizer recommendation...

Of course, the major mineral elements N, P and K must be applied to the soil according to the plant needs and the initial richness (composition) of the soil in these elements.

Soil test -based nutrient limit frames a base for fertilizer application , depending upon the relation between amount of fertilizer applied and magnitude of crop response as a function of nature and properties of soil vis-a-via different nutrients.We have understood very little about tailoring crop nutrient requirement as per soil nutrient supply levels.This is the area we need  serious revisit. 

Soil fertility rating, critical limit based, targeted yield, nutrient index, cropping sequence

all  these important for determination of fertilizers recommendation as well as soil texture ,soil pH ,the state of many elements (if it was available or fixed on soil particles).

I agree , there has to be some rider to apply fertilizers . But , it will surely never help adding umbrella kind of  fertilizers doses year after year on the same field with the same cropping sequence. 

The conventional approaches of fertilizer recommendations based on crop responses and soil test values are often difficult to extrapolate to the areas/ regions differing in agro-climatic conditions. I think, simulation modeling can better be develop and use for the purpose.

This is going to be still tough ask , Dr Tarafdar . Just imagine the current state of art of soil testing  and its interpretation for fertilizer recommendation to different crops. You are much better placed than any one else on this issue... 

Dr. Srivastava, the advantage of these techniques are relatively cheaper, and take into account soil-plant-environment continuum. As these computer based models integrate information related to soil nutrient availability with other factors ( like management, climate etc) affecting crop growth and response to applied nutrients, even conventionally developed fertilizer recommendations can be refined using the models. Once validated, the models can be effectively used within the defined boundaries for a particular yield targets.

Dr Tarafdar , you know all these , first you develop models taking into consideration these variables and then you validate them in field , again modify , revaludate.Such exercises have no head and tail. We need a simple soil test - crop response models to be later validated in field with different crops and soils. This exercise alone needs a monumental exercise looking at the variability that we have at our hand. But none the less, those who have STCR  can afford to Dr slip such multi-factor models.. 

Excellent Discussion , I agree with the points put forward by Dr. JC Tarafdar and Dr. Srivastava. It is the need of the hour to understand the concept of fertilizer / nutrient recommendation. In one of my current project, we are in the process of developing a DSS, in which it will try to generate the recommendation, based on soil fertility rating, critical limit , STCR ( targeted yield), cropping sequence/ pattern, M&B approach, Previous crop nutrient balance. Final choice about the fertilizer recommendation will be based on the economics, yield and soil nutrient balance. Several iterations it will do and select the economically optimal dose. This has been done by validating the models (nutrient budgeting  & crop simulation models) on agro ecological zone basis. I hope with the existing techniques and concept this idea seems to be good and now we are planning to take it to the next district  

Endorsing some good viewpoints by Surendran , what is your take on  developing crop-based NUTRIENT EXPERT like what IPNI has done for crops like maize, wheat, and many more on the card including citrus..But , again without their validation under varying farming situation , whole exercise will remain futile..   and till then , soil test -based fertilizer recommendation will continue as guesstimate only ..

Thank you all for your excellent input and discussion. Definitely the present recommendations are guesstimate only as Dr. Srivastava suggested. We need further research on the aspect for efficient and cost effective recommendations. I still strongly feel that development of a good simulation model may recommended better in future.

Is it not astonishing Dr Tarafdar , that such a core issue and the very basis of fertilizer recommendation is so conspicuously missing . And , this is  where , the recommendations of soil testing laboratories are so often than not , questioned..? In one of the discussions , Dr Bijay Singh raised almost similar issue , which was so well taken..

You are absolutely right. I think, Universities should come forward to address this issue as they have large number of young budding scientists and having enormous land and farmers contact . 

The definition of maximum economic yield. needs to include environmental cost otherwise we are fooling ourselves

It differs from nutrient to another. For instance, P, K and micronutrients can depend on soil test and critical limit approaches. However, N rates can be adjusted in-season based on plant growth and gadgets like chlorophyll meters, leaf color chart and vegetation indexes. One in general should depend on site-specific nutrient management approaches whether soil-based, plant-based or mixed.

I endorse the answer given by Ali! Critical limit based fertilizer recommendations can work for P, K and micronutrients but not for N.

I personally feel that this concept is useful for categorizing the soils into responsive and non-responsive groups, but the quantification of fertilizer dose for individual situation is not possible. It is more suitable for micro-nutrient recommendations for responsive category soils.