Monthly Archives: December 2017

Colony growth – putting it all together

The colony growth model is now more or less complete, I have worked up the math for how a bee population can progress thru the season, as well as the math for varroa mite populations. You can read about the details behind the math here.

So the next question, how realistic is this model, ie, does it truely represent what we see out here in the real world. The proof is in the pudding so to speak, so lets walk thru an example which is VERY typical of what we see happen with new beekeepers in our area, particularly those with an inclination toward ‘natural beekeeping’, ie, no management of varroa.

A very typical start is easy to set up. Open the colony growth model in a new tab with this link. Set the start conditions to a typical new start for our area, a 4 frame nuc started on May 15, queen starts to slow down on Sept 1 and the bees will start evicting the drones by Sept 15. The lines will show a very typical colony growth trajectory that ends with roughly 9 frames of bees for the winter cluster. Now lets assume this colony is very isolated, and catches no mites thru drift, but, it arrived with 1 mite, so select 1 mite in the starting mite population. The end result has virtually no impact on the overall colony, which will end the season with 9 frames of bees, and a 0.6% mite load with a total of 172 mites. This looks like a healthy productive colony of bees, and it’s easy now to start thinking ‘I have no mites, never saw any mites, and the bees look good without any form of treatment, I have good bees that deal with mites’. this is a strong colony with a great looking population of healthy winter bees.

So now lets follow this colony into it’s second season. Start conditions for season 2 will be 9 frames of bees in a wintered unit that starts brooding on March 1. As you can immediately see, the colony grows to the same strength is did the year prior, but gets there earlier in the season. This is an expected difference, with a larger starting population we can expect the initial brood cycles to produce more brood. But there is yet a hidden gotcha, the starting mite population is still set to one mite, and if we look at the results from the end of last year, the colony went into winter with 172 mites. Lets assume a big chunk of the mites didn’t survive the winter, so set the starting mite population to 100 mites to account for almost half of them not surviving the winter.

Now look carefully at that graph of the colony that started the year as a healthy strong colony, 9 frames of bees to begin the season, but they had 100 mites at the start too. Right around Sept 1 you see the mite population is greater than the brood population, which means virtually EVERY brood cell will be infested with a varroa mite. Those are the brood that _should_ be making up the healthy winter bee population, but, they are now emerging as sick bees infested with mite vectored virus. From the outside, this colony doesn’t look to bad in the Sept 1 timeframe, still a healthy batch of foragers coming and going, they made a decent honey crop because there was a good forager population during the honey flow, but the exploding mite population has decimated this colony. By the time they are finished brooding, there isn’t a healthy bee left in the colony, and by the mid November to early December timeframe, there are virtually no bees left.

This mirrors almost exactly what we see over and over. Somebody new to bees starts a colony and doesn’t seem to have a problem with mites in the first year, so they erroneously come to the conclusion ‘I have good bees, they deal with mites fine without a beekeeper intervention’. thru the second year they smugly pay lip service to the old timers saying ‘you need to manage varroa mites’ because, they know, those bees did fine last year with no management. And then by November we see postings on various online sites saying ‘My bees just absconded, why would they do that so late in the season?’. The answer is, they did not abscond, they died. What happened was, that varroa population was growing exponentially thru the season, and then when the time came for the bees to start raising long lived winter bees to carry the population thru the upcoming winter, they were only raising sick bees totally infested with varroa mites in the cells while developing.

So, armed with the knowledge of what happens later in the season in your colony that starts out with 100 mites, the beauty of having a model, we can experiment with options and play the ‘what if’ game. The exercise at this point then, can you find a series of treatment options that will allow your colony started with 100 mites to end the season with a healthy population of winter bees, and 200 or less mites. With the formic acid flash and OAV in your mite control toolbox, the answer is yes you can, but, it will take no less than 8 applications of OAV to keep this colony as healthy at the end of the season as it was at the start.

The real answer to mite control, start the season with fewer mites.

Dealing with varroa

Any beekeeper that has kept bees for a number of seasons will understand and know, there comes a time when your colony requires assistance in dealing with varroa, or the colony will die. This is a sad fact of modern beekeeping, but, if we are going to apply good animal husbandry concepts to our bee stock, there are times when it’s necessary to intervene and try keep our stock healthy.

To begin with, when modelling mite interventions (Treatments), I have chosen two of the common organic acid methods to introduce as tools for managing mites.

Formic Flash:- The formic flash treatment is one commonly used in our area, it’s a single day application of formic acid to the colony. The benefit of the formic acid treatment over others, it will apparently effect mites under cappings as well as phoretic mites. One can read endlessly about efficacy, and eventually you realize, the numbers quoted in different sources are all over the map. I have chosen to model the formic flash treatment with a very high efficacy level, not because I believe it is this high, but to show how even with high efficacy on a treatment, an out of control mite population will still kill the hive even after a mite treatment is applied. For the purposes of this model, the formic flash treatment kills 90% of the mites under cappings and those in a phoretic state. This is a highly effective mite kill. But a huge caveat, reading at Randy Oliver’s site at scientific beekeeping, he tried multiple rounds of formic flash to try improve mite kill. A second round of formic a week after the first resulted in a dead colony. Reference available here.

Oxalic Acid Vapour:- The Oxalic Acid Vapour treatment is less effective than the formic treatment in that it only affects phoretic mites. Easily applied on a small scale using a wand style vaporizer that takes about 5 minutes per colony for a full treatment, or using a blower style of vaporizer that takes about a minute per hive when doing larger numbers. The OAV treatment puts oxalic acid crystals in the hive which work for about 24 hours killing phoretic mites. Literature suggests this treatment is about 95% effective, in that when properly applied it will kill 95% of the mites. As it lasts over roughly 24 hours, for the purposes of modelling, we kill off 95% of phoretic mites on the day it is applied, and also get 95% of the mites emerging over the next 24 hours.

The OAV treatment only gets phoretic mites, and one strategy to help with this problem that I’ve read about online a number of times is folks are trying 3 treatments a week apart with the expectation this will get all of the mites thru a brood cycle. That would be true if the mite brood cycle turns over on a weekly basis, but, it doesn’t, it turns over on a 5 day period, the amount of time a mite remains phoretic. I added two more OAV options to the model, one of them applies 3 treatments at a 1 week interval, and another that applies 4 treatments at 5 day intervals. for those with a small number of hives, these may be viable options for mite control. If your colony count is such that you cant get thru them all in 5 days of work, then these become less viable options.

Modelling Varroa

So the bee population model is more or less complete, but it still is not representative of the real world. Out here in the real world, we have the varroa mite. To properly model the growth of varroa within a colony, we first need to understand the life cycle of this critter.

After reading endlessly in various literature on the subject, my conclusion is, the life cycle of the mite is fairly well understood, and is dramatically effected by what type of cell a given foundress mite enters. We need some numbers to realistically place timeframes on varroa development, a good reference is found here.

The phoretic period may last 4.5 to 11 days when brood is present in the hive or as long as five to six months during the winter when no brood is present in the hive. Consequently, female mites living when brood is present in the colony have an average life expectancy of 27 days, yet in the absence of brood, they may live for many months.

To get a handle on the reproductive success of those mites, another quote from the same article

Considering mortality in brood cells and improper mating, the average foundress mite produces about one offspring per worker cell she enters, and about two offspring per drone cell. Drones take longer to develop so more mites are produced in drone cells.

To try model these numbers is fairly strait forward. When a mite emerges we keep them in a phoretic state for 4 days. Starting on day 5, we assume that half of the mites available for going into cells will successfully find their way into a cell to try and reproduce. A fertile varroa mite going into a worker cell will produce one offspring, so two mites will emerge. That original foundress has been under the cap for 10 days and spent 5 days phoretic, so is now 15 days old, and after another 5 days of phoretic behaviour will enter another cell, so we have two fertile mites entering worker cells at this time. 10 days later we will have 4 mites emerge, but the original foundress is now fully aged and dies of age. The net result is, after two varroa brood cycles in worker brood, a single foundress mite has resulted in 3 mites in the colony.

Things change when there is drone brood available, literature suggests the varroa much prefer to hop into a drone cell over a worker cell. When a varroa mite enters the drone cell, she remains under the cap for 14 days, and will produce two viable daughters. After a phoretic period, all 3 of these mites will enter drone cells, and all 3 produce 2 more viable daughters, for a total of 9 mites emerging on the second round. At this time the original foundress mite dies of age, but leaves behind 8 viable mites as daughters and grand daughters.

As this math shows, there are two very distinctly different details when comparing the varroa life cycle to the honeybee life cycle. The bee population is based on a single queen laying eggs, and will grow in a linear fashion limited by the rate of egg laying of the queen when not limited by temperatures for brood incubation. The varro life cycle is shorter, and not limited by a single queen laying eggs, it grows exponentially rather than linearly because all fertile varroa mites are producing offspring.

To model varroa growth, we know the mites prefer a drone cell so it would be easy to just place all of the fertile varroa into drone cells when they are available, but, this is not realistic. There are 10 worker cells open on capping day for every drone cell that is open, not all of the varroa will find a drone cell. To account for this, we have prefererred drones when it’s time for varroa to enter cells, but assume only half of them find a drone cell, the other half entering cells will end up in a worker cell. This is the basis on which the varroa growth has been incorporated into the colony growth model.

Introducing varroa into the colony growth model does introduce another new concept, that of the ‘sick bee’. We know that bee virus are vectored by the varroa mite, and a colony with extremely high varroa levels will show lots of sick bees in the form of deformed wing virus and other inflictions. To account for this, a new type of bee has been incorporated into the population model, the ‘sick bee’. Any bee emerging from a cell that was populated with a varroa mite during the brood development is not placed in the normal bee population to graduate from nurse to wax maker and on up to forager. Instead, they are placed into the ‘sick bee’ population, ie, deformed wing etc. I cant find any suitable references in the literature to suggest how long a sick bee will live, but, we chose a rather arbitrary ‘it make sense to me’ way of handling the sick bees. If the bee is inflicted with deformed wing, it can still manage to crawl around on the frames, clean cells, etc. Where trouble begins for that bee is when the time comes to orient and graduate to foraging, instead of flying out of the hive, it ends up crawling out because it cant fly. Once a sick bee crawls out of the hive, it’s gone, so we have arbitrarily set the numbers so that sick bees crawl out and die at an age of 25 days, shortly after they should have graduated from being a house bee to a foraging bee.

The final tweak to the handling of varroa and sick bees came from a comment I saw in an online video by Jamie Ellis from the University of Florida. His comment was, when varroa population gets large, some cells will end up with two or more varroa feeding on the pupae in that cells. With two or more varroa feeding on a pupae, that bee will be dead or very close to it when it emerges. This final detail explains one symptom we often see when a hive crashes due to varroa load, we see numerous bees that were in the process of emerging and never completely got out of the cell when doing the post mortem. After adding this little bit into the math for handling varroa populations as they explode, the graphs mimic almost perfectly what we see from hives crashing due to varroa loads.

Expanding on the colony growth model

The first run at modeling bee colony population growth through a season was meant to validate some of the math and get a rough idea of how it would all work. Once the framework was in place the job becomes one of accounting for more details. The single biggest detail missing from the original math set was incorporating drones into the colony growth.

Much of the reading I’ve done both online and in books, many folks tend to view drones in a colony as a waste of resources, they produce no honey and do no work in the colony. All they do is eat, and fly out. This view may be correct for folks that have an outlook of ‘honey produced this season’ and they buy in all the queens they need over time. But if we raise our own stock and have an outlook that looks beyond the results of this year, the drone population we raise in this season is a very important component of our results the following season. Those drones will mate with the queens we raise this season, so they are providing half of the genetic input to our bee population next year. So while some folks view drones as a drag on the colony, my own person view is, our drone crop this year is responsible for a good honey crop next year. Another place where drones actually help the colony is during the spring buildup. While the drones are out flying during the day, overnight they are in the cluster, and that cluster is incubating the early brood rounds when nights are cold. The drone population can and does help incubating brood overnight.

The drones live on a different life cycle as compared to the worker bees, and it is very important to model this different cycle correctly. A drone egg is laid, then emerges as a larvae 3 and a half days later, just like the worker bee. The drone cell is capped on day 10, so it sits open for a day longer than a worker, then emerges on day 24. This is a critical difference as the drone cell is capped for 14 days vs the 11 days for which a worker is capped. After the drones emerge, they spend a week or two hardening and maturing in the colony before they start making regular afternoon flights to the drone congregation area. This is another important detail, because it tells us about our ability to successfully mate a new queen. You cannot successfully mate a new queen till a couple weeks after you see the first drones on frames in the colony.

So, when do the bees start raising drones ? Just about everything I’ve read on the subject suggests that the bees will start raising drones later than when they start raising workers during the early spring buildup. But this is not what we see in our hives here in Campbell River. Our bees typically start the first round of brood in the mid February timeframe, and that’s about the time we will start to consider spring feed in the form of patties. We dont normally go deep into the hives lifting frames to inspect until mid to late March. On the late March first inspection, we often see some drones walking on the frames, not a lot, but there are some. If we do the math on drone development time, seeing drones on the frames in mid March suggests they were laid as eggs in mid to late February. I am convinced the first drone eggs are indeed placed in cells as soon as the first round of replacement bees is started.

How many drones do the bees raise ? Again, reading literature provides a wide range of numbers, really depends on ‘which book did you read’ to get a handle on that number. I’ve seen numbers as low as 5 percent, and as high as 20 percent. Our own experience in looking at colonies where we place a drone frame, bees tend to fill one side completely, and the second side partially, which works out to approximately 10 percent of the brood is drone cells.

Another detail that we need to account for is drone eviction. It’s well known, as we get into the later part of the season, worker bees evict the drones from the colony. With no basis other than ‘it makes sense’, we need to consider another important date then when modelling hive populations. If the bees are evicting the living drones on a given date, when did the queen stop laying eggs in drone cells ? It takes 24 days for a drone to develop from egg to emerging bee, so it does make sense to assume that no more eggs are placed in drone cells when we are 24 days before the date at which the bees will evict drones.

After going over all of the numbers I’ve seen over time, and trying to make a realistic mathematical model for colony development, I chose to model drones by having the queen place 10% of the eggs into drone cells while she is laying eggs, and stops laying in drone cells 24 days prior to the date when the bees evict the drones. The way eviction is modelled, on days after the start of the eviction process, half of the remaining drone population gets evicted from the colony, resulting in a drone population that declines rapidly once eviction starts.

Hive models

So Ian is showing the work in progress hive model on Youtube now.

Find it here:- Hive model

We have seen in numerous presentations the population dynamics chart originally produced by Randy Oliver with respect to bee population growth and dynamics in a healthy bee colony. It was a bit inspiring, but, we wanted something that would allow us to modify the start conditions and see how different start conditions would change the dynamics.

Honeybee duties are based on the age of the bee, there is a pretty good description found on Wikipedia.

The way the calculations work is strait forward. The starting population is distributed by age based on the start condition. A package is an even spread of bees of all ages. A wintered unit starts out with all winter bees, and a nuc starts with a population of all house bees, along with the number of brood frames selected, with the brood spread evenly over all ages.

When the simulation runs, the date is advanced one day at a time, and all the bees / brood are aged by a day, then tally up how many in each age group to plot population of that group. During the process realistic restrictions are incorporated, ie the queen doesn’t lay more eggs than the current population can support for feeding and incubating brood.

Some assumptions are made during the simulation. It is assumed the bees will have all the necessary protein and carbohydrates available for feeding the brood, if not available naturally then they should be beekeeper provided. It is also assumed there is always enough comb available for eggs to be laid at the best rate the queen is capable of. During the buildup, it’s also assumed that a queen doesn’t go from 0 to thousands of eggs overnight, it takes time for the rate of eggs laid to ramp up. An arbitrary number was chosen, on any given day queen will be capable of at least a couple hundred eggs, and the rate of eggs being laid can increase by 25% day over day, so when the simulation first starts, there is a ramp on the rate of eggs going into cells.

Winter bees are a special case, and there is really no good numbers available in literature for modelling the winter honeybees. But we can make some intelligent guesses. Looking at the division of labour by age, bees are nursing brood from ages 3 thru 11 days for a total of 8 days. Larvae is open from day 3 thru 8 for a total of 6 days, so the ratio of nurse bees to open larvae is 4/3 in a hive with a steady state population at maximum potential. During the fall slowdown there is a period where the ratio of nurses to open larvae gets much larger, so we have a surplus of nurse bees that have the body fats of the freshly emerged bees, but are not expending them nursing new larvae. These are the bees we allocate to the ‘winter bees’ population. Conversly in the spring, when winter bees are pressed into nursing duty, that starts the aging clock for them to age out on the normal bee age cycle. During spring buildup, the way this is modelled, when there is a shortage of nurse bees for the open larvae, bees are taken from the winter bees category and placed into the nurse bee role at age of 4 day, and then allowed to age out in the normal progression.

Ofc, we all know, there is bee die-off thru the winter, not all of the winter bees survive thru till the spring. Again, to simulate this I have found no strong references in the literature for death rates, so we go back to our own experience and look at what we’ve seen in colonies in our back lot. My best guesstimate for that is, winter bees die off at approximately 10% per month.

On the ‘To-Do’ list. The next addition will be modelling growth of the drone population along with the worker bees. When that’s done, plan is to home in on the biology of varroa mites, and introduce a varroa mite population that runs in conjunction with the bee population model.